PART THREE
ANALYSIS

Seven
Public and Private Conceptions of Safety

In the abstract, public and private standards-setting are easily equated. Both ostensibly involve similar objectives and entail similar tasks, although differences in political dynamics obviously slant outcomes accordingly. In this respect, standards-setting boils down to a question of strictness or, in more economic terms, what balance to strike between probable costs and benefits. The case studies suggest, however, that regulatory philosophy circumscribes the range of possible outcomes in both sectors and that different conceptions of safety regulation prevail in the public and private sectors. Estimating the costs and benefits of possible action is practically the last step in setting safety standards. Several important tasks precede the consideration of any specific proposals. First, the organization must define "the problem" or problems to be addressed. Then it must conceptualize a range of possible solutions.

The case studies indicate that decisionmakers in the public and private sectors view the same "problems" quite differently. The difference is not just in what should be done about a problem, but whether anything should be done at all. Taking a paternalistic view of safety regulation, the public sector is much more likely to favor standards to protect people against their own folly or mistakes. Even when public and private organizations address the same or similar "problems," the range of acceptable solutions can vary considerably. Some solutions, never seriously considered by one sector, are routinely favored by the other. These conceptual preferences are not arrived at in a deliberative

fashion. In fact, they are almost never openly discussed. Rather, these working assumptions are reflections of a shared regulatory philosophy . This chapter describes several manifestations of regulatory philosophy and analyzes how the professional ethics of engineers and lawyers influence the framework for setting standards in the two sectors.

Paternalism and Problem Definition

The first step in writing a safety standard is determining the scope of "the problem" to be addressed. As Charles Lindblom and David Cohen point out, "we do not discover a problem 'out there'; we make a choice about how we want to formulate a problem."^[1] That choice reflects certain values and in turn constrains the realm of possible solutions. Ideally, those defining public "problems" base their analysis on "feelings of distress, or discontent, or annoyance, or unhappiness of some subset of the citizenry."^[2] But the more technical or obscure the issue, the weaker the links to democratic problem definition. Professionalism appears to fill the gap.

Defining "the problem" is a more complicated matter than is often imagined. Deciding, for example, to address safety concerns in grain elevators is only the starting point in defining "the problem." Usually there are many hazards related to a single process or product, only some of which are considered appropriate topics for regulation. In the grain elevator case, OSHA considered housekeeping a much more significant problem than NFPA. With aviation safety, both sectors had similar concerns about fire extinguishers, but only the FAA worried about smoke detectors. Of course, there is a danger that any observations from the case studies are colored by the interaction unique to "paired" cases. In other words, OSHA's definition of the problem might be contingent on NFPA's. As it turns out, OSHA and the FAA acted almost independently of NFPA; but the CPSC acted in response to UL and AGA Labs. (The special issues attendant to standards aimed at complementing or supplementing standards in the other sector are taken up in the next chapter. As explained below, however, these interactive cases seem to accent, rather than distort, the differences in public and private regulatory philosophy.)

Paternalism is one of the distinguishing tenets of the CPSC's regulatory philosophy. Most injuries related to the woodstove and gas space heater should be blamed on the user, not the product. Properly installed woodstoves pose very little risk; so do space heaters that are operated

properly. But protecting people against their own mistakes is part of the mission of the CPSC staff. Private standards-setters, by contrast, are loath to recognize such safety "problems." They are also loath to discuss the topic of which hazards are appropriate for regulation. It would be inopportune for UL to state baldly that a significant hazard scenario is "the consumer's fault." As a result, some of the arguments about paternalism are cast in technical terms. GAMA's stated opposition to the CPSC's gas space heater standard concentrated on alleged technical problems with the oxygen detection sensor. None of these problems materialized in the four years after the rule was adopted, suggesting that nontechnical concerns were probably the underlying motivation.

Although a UL engineer allows that "there are occasional arguments within the organization about how forgiving products should be," the answer is almost always less forgiving than what the CPSC would require. Why should a metal chimney have to tolerate a creosote fire, UL wonders, when the consumer can prevent such an occurrence by cleaning the chimney at proper intervals? Similarly, why should kerosene heaters be tested for "flare up," when the problem stems from the use of improper fuel mixtures?^[3] Surface temperatures are another case in point. The CPSC has long worried about the surface temperatures of products such as gas space heaters, kerosene heaters, furnaces, stoves, and small kitchen appliances (such as toasters). Reluctantly, UL has toughened some of its surface temperature requirements in response to CPSC pressure. In the case of kerosene heaters, however, where the effort would require substantial reengineering, a UL representative candidly testified that "it's going to take a kind of change in our philosophy of approach" to address surface temperatures.^[4]

Private standards-setters are most likely to protect against misuse when it involves children or the elderly. Private testing labs subject many consumer products to an articulate probe test intended to simulate an overly curious finger probing inside protective grates or openings in the appliance casing. The AGA/ANSI standard for space heaters simulates this hazard and the possibility of clothing ignition caused by contact with the heater. AGA Labs disagrees with Consumers Union about test methods—the proper fabric to use when testing flammability and the correct shape of the articulate probe—but both agree that these mishaps merit some protection.

Nevertheless, significant differences remain in how the two sectors view the responsibility of consumers. UL and AGA usually certify safety under conditions that assume full compliance with both the manufac-

turer's instructions and any applicable installation codes. AGA assumes, for example, that gas space heaters (1) will not be operated in bedrooms (because that is prohibited by the National Fuel Gas Code) and (2) will always be operated with proper ventilation (in accordance with the manufacturer's instructions). Injuries that might be caused by the failure of an adult to read the instructions or install an appliance with professional guidance are of little concern to AGA. Its regulatory philosophy does not encompass such paternalism. The CPSC, in contrast, takes the responsibility of manufacturers much more seriously than it takes the responsibility of consumers. When the CPSC staff drafted a proposal for the criteria to use "in considering requests for endorsement or recognition" of private standards, one of the only substantive conditions was that "information in the [standard] does not place the burden of safe use of the product onto the consumer in such a manner as to exculpate manufacturers or distributors from any liability associated with their product's use by the consumer."^[5]

These different worldviews were highlighted clearly in the case of gas space heaters. According to a member of the Z21.11.2 committee, carbon monoxide poisoning occurs because "people do not use and maintain the product correctly." The CPSC had a different view. "Consumer misbehavior was not a problem with the unvented heater," a former CPSC commissioner asserts; "it was a dangerous product." The private sector's position is factually correct. No one has ever been poisoned by carbon monoxide while using a space heater correctly. But whether it is desirable to require the ODS depends on your philosophical view of government paternalism and individual responsibility. Thus, in the case of gas space heaters, woodstoves, and many other products, the CPSC sees problems where UL does not.

Defining the Range of Acceptable Solutions

The case studies suggest that even when public and private organizations address the same or similar "problems," the range of acceptable solutions varies considerably. These differences involve what might be called solution definition, a process analogous to problem definition. Defining the range of acceptable solutions is also a matter of philosophy, beyond the scope of the deliberative process. NFPA does not spurn housekeeping rules for economic or functional reasons. In fact, several firms require in-house what they so strongly oppose in external standards imposed by OSHA or NFPA. The objection is to any industry-

wide standard infringing on what is perceived as management prerogative. The same phenomenon exists with matters of less consequence than housekeeping. UL avoids certain types of warning labels, not because they would be too costly, but because they conflict with its philosophy of regulation. Similarly, most private standards-setters refuse to apply standards retroactively. The logic is predominantly economic. But NFPA's position, stated in the grain elevator proceedings, against making retroactive any provision requiring "even a dime of investment" suggests that more than economics is at issue.

Perhaps the most familiar example of "defining the solution" to a safety problem is deciding whether standards should specify design requirements (such as metal thickness) or performance levels (such as an "action level" for housekeeping).^[6] Another design choice concerns the extent to which standards should (a) modify products or processes or (b) communicate information about risk to the consumer or user. These alternatives are not mutually exclusive, though they are often treated that way. The difference in how public and private standards-setters tend to view these questions is not well understood. Government is often accused of favoring design requirements over performance standards, adopting rigid rules that stifle technological innovation. Business interests, on the other hand, are presumed to favor performance standards. The case studies belie these characterizations. In none of the four cases was the government standard as detailed or lengthy as the private standard. UL and AGA standards are filled with detailed design requirements, and these requirements meet with the general approval of industry. In contrast, OSHA's "action level" for grain elevator housekeeping—a pure performance standard—was strongly opposed by industry.

The case studies suggest many more subtle differences in public and private conceptions of safety regulation. Five examples are detailed below. Some apply exclusively to consumer products, others to industrial processes and practices; a few apply to both. The discussion that follows this description unifies these seemingly disparate examples by considering the common links to professional ethics.

Banning a Product or Process

Perhaps the most fundamental difference in public and private conceptions of safety regulation is that only the government gives serious consideration to prohibition as a regulatory strategy. The private sector

tends to assume that the object of regulation is socially desirable and is (or can become) sufficiently safe to be "acceptable." UL begins with the assumption that a product is not inherently or unreasonably safe. If it is too unsafe, UL will not list it. But the issue rarely comes up. The electronic bucking bronco is the only example that several UL officials could name of a product that UL would not accept for testing. In contrast, public agencies generally assume practically the opposite: that the object of regulation is suspect and may not be sufficiently safe to be socially acceptable. The CPSC space heater rule began as a proposed ban on all space heaters. More recently, the agency proposed a ban on certain all-terrain vehicles.

Participants in private standards-setting, cognizant of the antitrust law, argue that private standards are never intentionally prohibitory. They rarely are—at least explicitly. But all standards of any substance prohibit something. This clouds the distinction between public and private conceptions of safety regulation. Both sectors utilize prohibition in some form. The difference in perspectives is more a matter of degree than a difference in kind. Nevertheless, the distinction is important because the public sector is more likely to favor this strategy.

Private standards-setters prohibit all sorts of specifics through requirements that alter products or processes incrementally. They often prohibit specific product features such as power lawn mowers without an automatic shutoff and foot guard. But they are reluctant to prohibit at a more general level. On rare occasions, however, they prohibit an entire line or type of product. The ANSI/AGA standard for gas space heaters falls somewhere in between: regulating enough specific product features to prohibit a whole type of heater, the small porcelain bathroom model. The prohibition is not direct, however; the porcelain model simply could not meet all of the provisions added to the standard. Although that model has fallen by the wayside, other types of space heaters have taken its place. A representative of the gas appliance manufacturers describes it as a "drop-out ban." Prohibition at a more general level is out of the question.

Forcing Technology

Public and private standards-setters also have fundamentally different conceptions of technology. They disagree about how the state of technology limits the range of possible provisions in safety standards. To the private sector, technology is usually seen as a constraint. It is a given.

Government agencies are far more willing to "force" technology through requirements that are unattainable with technology currently in use.^[7] The private sector, by contrast, seems unwilling to give technology even a gentle nudge. To private standards-setters, the current state of technology refers to what is generally in use, not to what is close at hand or on the so-called cutting edge. Hence, the ANSI/AGA committee was unwilling to require the oxygen depletion sensor on space heaters even after it was marketed by one company. The technology was still considered "speculative" and "unproven," so the committee deemed it an optional portion of the standard. In both of those cases, the government illustrated a willingness to do what the private sector would not. The FAA and the CPSC considered, and later required, solutions that moved beyond the technologies that were widely available. There was a feeling at the FAA, according to one staff member, that "the market would respond to the need for smoke alarms designed especially for aircraft." Similarly, when questions were raised about the feasibility of the largely unproven oxygen depletion sensor, CPSC staffers were equally convinced that the market would adapt.

Work Rules and Other Operational Controls

Safety standards are often thought of as technical in nature, taking the form of either performance or design requirements. But technical provisions are only part of safety regulation. Work rules, maintenance schedules, and other operational controls play an important role, especially in the kind of standards written by NFPA. To be effective, technical standards must work in tandem with operational controls. Sometimes, operational controls can even substitute for engineering solutions.

To more fully understand the distinction between engineering standards and operational controls, consider the example of safety standards for a hydraulic system.^[8] Safety standards could incorporate any of the following four types of controls. First, a technical performance standard might specify that the bursting strength of hydraulic lines be equal to or greater than four times the working pressure. Second, a technical design standard might require that hydraulic lines be equipped with safety valves to prevent loss of system pressure in the event of a ruptured line. Third, behavioral rules could be adopted instead of these technical solutions. For example, the safety standard might include the following operational control: the hydraulic system must be inspected

for leaks after each one hundred hours of operation. Finally, a work rule could be adopted requiring that hydraulic controls be checked every time the machine is started. These technical and behavioral approaches all provide possible solutions to identified hazards. The prevailing conception of safety regulation in the private sector generally excludes the latter two approaches. In response to pressure from business, ANSI formed a Safe Work Practices Task Group aimed at ensuring that "safe work practices are not included in the body of standards."^[9] This philosophy accounts for why the appendix contains important provisions in both NFPA standards studied. It also helps account for the storm of protest over housekeeping. Maintenance procedures (particularly housekeeping practices) clearly have a significant effect on safety, but the prevailing view at NFPA is that housekeeping should be left entirely to managerial discretion. Standards for almost any operational controls, even those intended to complement specific technical requirements, make private standards-setters uncomfortable. We try to avoid "'how to' standards," notes a committee member. The provisions of NFPA 61B that address operating procedures are vague and relegated to the appendix, where all provisions are couched in the disclaimer that "this Appendix is not part of the requirements of this NFPA document but is included for information purposes only." The same is true of NFPA 408 for aviation fire safety. Several committee members agree that training requirements are vitally important to the successful use of Halon extinguishers. The standard even includes a (vague) chapter on flight crew training, something quite unusual for the NFPA. More typical is what was relegated to the appendix: "Although not required by this standard, it is highly recommended that live fire training on representative aircraft fires be conducted for all flight crew members during both initial and recurrent training sessions."^[10] In contrast, public standards-setters embrace operational controls. The FAA tells commercial airlines "how to" do all sorts of operations and routine maintenance. The essence of the OSHA standard for grain elevators, the "action level" for housekeeping, is an operational control. So, too, are provisions on safety training, evacuation drills, and hot work permits.

Information Disclosure and Labels

Public and private standards-setters also appear to have different positions concerning information disclosure and warning labels. Government sometimes relies on warning labels as its major regulatory strat-

egy. The woodstove labeling rule relies exclusively on information disclosure as a regulatory strategy. The agency defended this standard over the private alternatives on the grounds that its provisions had stronger language and were more likely to be effective. Labeling provisions were also a significant part of the gas space heater rule. In both cases the label was intended to change people's behavior. The private sector is less enamored of information disclosure and does not share in the public sector's penchant for trying to change people's behavior. There is a sense among many private standards-writers that warning labels should not even be included in standards. Historically, many of these organizations shied away from warning labels entirely. The labeling and marking requirements for the gas space heater filled less than a page in the 1963 version of Z21.11.2. There are almost five pages in the 1983 version. Warning labels and other instructions have become more prominent in private standards, but they tend to be written by lawyers who do not otherwise participate in the standards-writing process. A representative on AGA's unvented space heater committee takes labeling questions to in-house counsel at his firm before attending meetings. The Standards Department at UL oversees the warning and labeling provisions in all UL standards, but the engineers do everything else. These provisions are added more to provide a defense against lawsuits than to prevent injuries. It is inconceivable that UL would respond to a problem it considered serious by changing the labeling requirements. Only the government seems intent on what a former CPSC commissioner refers to as a fool's errand: trying to change the behavior of millions of consumers.

When Regulations Take Effect

Finally, public and private standards-setters have disparate views about when standards should take effect. This issue was controversial in all four cases. Government tends to favor the earliest possible effective date, being intolerant of industry claims concerning the feasibility of compliance. Like the stereotypical boss, they want it done yesterday. In contrast, it is outside the realm of possibility for private standards-writers to make standards effective immediately. They tend to allow much more lead time than government. "As long as one company has a device," complains a UL official, "the CPSC is inclined to make it an immediate requirement." The problem, in his view, is twofold. First, the device has not always been evaluated by an independent organization,

so its usefulness is in doubt. Second, "other companies may not even have a prototype," meaning that an immediate requirement would grant a temporary monopoly to one firm. UL takes its cues from the market. Until something is widely available, particularly from more than one supplier, UL is reluctant to require it.

Government sometimes even favors the past over the present—making standards retroactively effective. Most private groups have an explicit policy against adopting retroactive standards. This was the most critical difference between the public and private standards for grain elevators: NFPA 61B applies only to facilities built after the standard was adopted, while the OSHA rule applies retroactively. This means that 61B affects, at most, only a few percent of all grain elevators. Changing that provision would probably have greater safety implications than any other single change in the NFPA standard. But the idea of such a change is outside the realm of conceivable solutions and has never been seriously discussed. Even when NFPA made a rare exception, applying its vague housekeeping requirements retroactively, there was opposition on principle. One committee member recalls that he "didn't want to be on record as requiring this even though [he] agreed with it." His rationale: "Let OSHA do it."

Regulatory Philosophy and Professional Ethics

These observed differences in regulatory philosophy appear to be more institutional than transient. That is, they were not produced by situational bargaining or compromise. Rather, they are working assumptions that are rarely the topic of discussion. The case studies suggest that two institutional features help shape the prevailing regulatory philosophy: professional ethics and organizational self-interest. (Organizational self-interest varies by regulatory environment—the topic of chapter 9.) The remainder of this chapter focuses on the important role of professional ethics in distinguishing public and private conceptions of safety.

In his collection of studies on public regulation, James Q. Wilson maintains that specific types of employees are important in shaping regulatory behavior. One category discussed by Wilson is professionals. Professionals "have distinctive ways of thinking about problems," and they care deeply about "the maintenance of professional esteem."^[11] Of course, no single profession dominates government. Rather, as Frederick Mosher contends, a "very wide variety of professions and profes-

sionals in diverse fields" influence the leadership of government. However, in well-established agencies, Mosher continues, there is often "a single occupational group whose knowledge, skills, and orientation are closely identified with the missions and activities of the agency."^[12]

A recent survey of "risk professionals" involved in environmental issues suggests that diversity reigns in this field, which accounts for a wide array of professionals from lawyers and economists to engineers and toxicologists.^[13] The trends observed by Thomas Dietz and Robert Rycroft provide a solid basis for describing the prevailing ideology in the government agencies examined in the case studies. These risk professionals share an ideology described below as the "enforcement ethic." This ideology emphasizes formal legal rules. It is often associated with lawyers, although many nonlawyers share its premises. But this study did not consider private standards-setting. Rather, it examined professionals housed largely inside the Washington Beltway—in government agencies, law firms, consulting firms, environmental organizations, and trade associations

Beyond the Beltway, where most private standards-setting actually occurs, the patterns of professionalism are distinctly different. Engineers are much more dominant. UL has several hundred engineers and only a handful of lawyers (or nonlawyers fitting the profile of "risk professionals"). Engineers outnumbered all other professions on the technical committees for NFPA 408, 61B, and ANSI Z21.11.2. None of these committees includes a lawyer or an economist. A study for the Administrative Conference makes special note of the "surprising lack of lawyers" in many private standards-setting committees.^[14]

The Engineering Ethic

Engineers have a distinct ethic, and it appears to explain several aspects of the regulatory philosophy observed in the private sector. The contours of the engineering ethic are sketched below, following two caveats about this analysis. First, the forces of professionalism vie with other political and economic influences operating on private standards-setters. Many participants simply register "directed votes" at committee meetings—votes that reflect the economic interest of their employers. Others operate with more independence, however, acting without specific instructions or in arenas where they have no direct stake. The fire protection engineer for a major airline participates not only in NFPA committees for aviation safety but also in those for railroad fire safety.

Obviously, proprietary considerations play no role in the latter. Similarly, there is no commercial link between Factory Mutual and the subject matter of NFPA 408, even though a Factory Mutual employee played a pivotal role in developing the standard. The FAA employee on the committee, although listed with his government affiliation, is actually an individual member who "joined for [his] professional development." A pilot for another major airline remained active on several NFPA committees long after his retirement. Although it is rare that engineers operate with such free rein, the professional influences apparent in such situations probably affect other standards as well.

Second, generalizations about groups as large as the engineering profession are necessarily stereotypes—they are suggestive of general themes but should not be taken too literally. Obviously, engineers do not all share the values discussed below. There are competing ethics in the private sector; but it would take dramatic changes in the structure of standards-setting or the profession of engineering for them to dominate. One competing ethic comes from science. Though the distinction between science and engineering has lessened as engineering schools have become more "scientific," there remains a distinct difference in how engineers and scientists approach standards-setting.^[15] Researchers at the National Bureau of Standards, for example, sometimes take "scientific" positions that the engineers at AGA or UL consider unrealistic. This kind of conflict, which often pits principle against practicality, has been documented in other policy arenas where scientists and engineers interact.^[16]

There are also factional divisions within the engineering profession—a profession more fragmented than is often realized.^[17] The values described in this chapter characterized most of the engineers involved in the four case studies, but a small fraction of the profession espouse a different value system. The prevailing ethic would certainly change if these "human factors engineers" played a more prominent role in the process.^[18] This seems unlikely, however, as the field is looked upon with suspicion by many engineers and rarely works its way into engineering education.

Instead, the prevailing view, seldom discussed but consistently applied, is that safety standards should not be paternalistic, "engineering for" things such as consumer misuse and poor maintenance. There are several reasons why engineers as professionals might take this view. In some cases, they may have a trained incapacity to recognize the problem. It was not until the American Gas Association Labs conducted

some tests with consumers that its engineers realized that the lighting instructions for gas appliances were unclear to almost everyone but the gas appliance engineers who wrote them. The rigors of product testing can cause a similar failure of imagination. UL tests woodstoves under such carefully controlled conditions that its engineers have been unable to create creosote. It is hard for UL's engineers to conceive of a problem they have not been able to recreate. It follows that UL is less concerned about metal chimney tolerances than the National Bureau of Standards, which has not only created creosote but has measured the intensity of subsequent chimney fires.

Although this may explain differences in recognizing problems, it says little about those situations in which a problem has been recognized (or placed on the public agenda) but remains unaddressed by the private sector. UL may not appreciate why people have chimney fires, but it certainly knows that it happens frequently. Here, too, professional norms may explain why some of these issues go unaddressed. UL's engineers have a very different sense of professional duty and responsibility than the lawyers and other standards-writers in government agencies. Engineering to prevent misuse, at least at some point, is abhorrent to the professional engineer. Samuel Florman, an engineer and author, deplores those "who seek salvation for society through the moral conversion of the engineer."^[19] "Why design a toaster that cannot burn you," asks a UL engineer, "when you can achieve the same result by not touching it when it is hot?" Expecting a "certain amount of prudence" by people, as another UL engineer puts it, is part of the engineering ethic.

Engineers generally consider issues such as warning labels or standards to battle product misuse as "political." As such, they fall outside the "technical" arena in which engineers operate most comfortably.^[20] Engineers involved with grain elevator safety have a similar view: regulating safety devices is their job, but housekeeping is a "management" issue. The prevailing professional norm is to leave politics to the politicians (who are, quite frequently, lawyers) and management to the managers. Lawyers may mandate housekeeping or add warnings to products, but engineers are not in the business of trying to change people's behavior.

The engineering ethic also affects how private standards-setters define the range of acceptable solutions. Engineers appreciate the benefits of technology but, contrary to popular opinion, are often humble about its prospects. New technologies must be proven to be accepted by en-

gineers. Smoke detectors for airplanes might, be a good idea, but engineers want proof. They are all too aware of the factors that could cause malfunctions in the aviation environment. Accordingly, private standards-setters are slower to adopt certain safety measures. It takes more than a prototype or a good analogy to change a private standard. Caution does not necessarily mean restraint. Although engineers may be slower than others to endorse new technology, they are often less restrained when safety issues are cast in economic terms. To many engineers, "feasible" has only a technical meaning, not an economic one.^[21] "There are those of us who go overboard," admits a professional safety engineer who sits on several NFPA committees. "It is easy to do in the name of safety." This is apparently why NFPA 408 is more demanding than the FAA standard for hand-held fire extinguishers. When deciding the appropriate number and type of extinguishers, the engineers who drafted NFPA 408 turned only to professional norms. The additional cost of Halon extinguishers was not an issue; neither was the marginal benefit of adding more extinguishers, Halon or otherwise.

The Enforcement Ethic

The engineering ethic does not pervade the public sector. With the exception of OSHA, where a fire protection engineer was in charge of the grain elevator rule, none of the key rulemaking personnel or official decisionmakers were engineers. Lawyers played an important role at the CPSC and OSHA. Their influence was less great at the FAA, where the Office of General Counsel reviewed the rule but did not influence the content. The prevalence of lawyers does not of itself denote a commonality of values like that which prevails in the engineering profession. Encompassing both the plaintiffs' and the defendants' bar—ardent adversaries in litigation concerning public safety—the profession is too fragmented to have common values on questions such as how to divide responsibility between consumers and manufacturers. There appear to be common values among most public standards-setters, however, and that regulatory ethos is described below as the "enforcement ethic." This ethic appears to infuse the hearts and minds of many government lawyers and other rulemaking personnel in the public sector. The enforcement ethic seeks to eliminate harm wherever possible; it tolerates few excuses, and demands near-total compliance.

The enforcement ethic is legalistic. It favors rules as a response to problems. Just as personal injury lawyers seek a remedy for every in-

jury, some agencies propose a standard for every identified hazard. The process is routinized at the FAA, where accidents beget regulations. National injury estimates drive the CPSC. The magnitude of the hazard matters more than the cause. Fault and responsibility on the part of the consumer do not play an important role in this value system. Public agencies seek to prevent injuries, whether caused by misuse, poor management, or product design.

The enforcement ethic also tolerates few excuses, and business concerns are generally not among them. Some CPSC commissioners seek to determine whether issues are "technical" or "economic." Technical arguments can forestall action; economic arguments rarely do. The CPSC staff had little patience for industry arguments concerning the effective date of the woodstove and space heater rules. In both cases, industry was concerned about disposing of existing inventories before the rule took effect. Privately, CPSC staff members admit that they did not consider the concern a valid one. The FAA summarily dismissed the argument that false alarms from smoke detectors might panic passengers. "That is not a safety issue," according to an FAA staff member.

Finally, the enforcement ethic stresses compliance at all costs. Total compliance is often an end rather than a means. The CPSC staff was inclined to go ahead with a woodstove labeling rule so long as there was anything short of total compliance on the private side. "I don't buy the argument about 85 percent compliance [with the UL standard]," commented one staff member, "because it says nothing about the other 15 percent." Undue emphasis on total compliance led the CPSC to adopt a woodstove labeling rule that had doubtful marginal benefits and a minuscule chance of actually improving compliance. Similarly, for the rulemaking staff at OSHA it mattered not that the cost of proposed safety measures would be highest at those facilities where the benefits would be lowest—country elevators. To exempt smaller facilities from housekeeping requirements would have gone against the enforcement ethic, and the idea of doing so was never seriously entertained by the OSHA staff.

Eight
Regulatory Decisionmaking:
Public and Private Standards in Action

Given these differences in regulatory philosophy, public and private standards-setting systems, even when they address the same general subject, seem somewhat like regulatory apples and oranges. Though similar in form and purpose, they are difficult to compare in substance. Economists and policy analysts, unfazed by this predicament, look to the bottom line. They compare the seemingly incomparable by thinking in terms of aggregate costs and benefits. This approach facilitates the normative comparison of outcomes, an important element in evaluating public and private standards, but it obscures the nature and workings of the decisionmaking process. The private sector almost never engages in cost-benefit analysis. In contrast, the public sector produces plenty of cost-benefit analysis—it is required to by law—but mandated analysis tends to provide rationalizations rather than bases for decisions. The task is further complicated, of course, by the uncertainties attendant to complex regulatory decisions.

This chapter examines the outcomes of public and private standards-setting and the nature of the underlying decisionmaking process, with the goal of evaluating the relative performance of the two systems. The cases partially confirm the conventional view that public standards are stricter than private ones and are sometimes prone to overzealousness. But there are contrary indications as well, suggesting that the private sector is more diverse than is commonly presumed. The private sector even exhibits surprising tendencies toward strictness on occasion. Un-

derstanding these patterns of behavior is critical to predicting how the private sector might act in the future. This chapter analyzes the cases in two ways: first, in terms of the tendencies of each sector in handling the uncertainties about regulatory costs and benefits; second, on the basis of overall reasonableness, making some important assumptions explained below. Both approaches suggest that private standards-setting is more diverse than is often imagined. Various explanations for this regulatory behavior are ventured in chapter 9.

Decision Rules and Routines

Within the constraint of regulatory philosophy there is considerable room for discretionary judgment. Given the inherent uncertainties in setting safety standards, however, it is unclear how standards-setters structure and exercise this discretion. In other words, what are the rules of evidence, formal and informal? Some version of cost-benefit analysis is the formal answer in the public sector, although agencies are often accused of circumventing the requirement. The mystery widens in the private sector, where cost-benefit analysis is almost never done. Private safety standards often precede any information on accidents or injuries. UL listed woodstoves for decades without the benefit of any injury estimates. So what does the private sector do instead? How does it actually make decisions? The cases suggest that heuristics and other standard operating procedures have evolved to simplify the resolution of these complex safety questions in the private sector.

One technique is to defer to "professional judgment" within certain constraints. In other words, so long as particular provisions do not cost too much, engineers are given free rein in making the decisions. This strategy requires only rough estimates of costs and not necessarily any formal assessment of benefits. It also results in a substantial amount of guesswork. UL prefers to call this "engineering judgment." Others use the phrase "educated guess," with the stress on educated . In either case, the implication is that guesses are guided by a combination of education, experience, and values. This explains both the importance of engineering ethics and the potential for unreasonably strict (but economically inexpensive) provisions in private standards. More costly and salient issues are not resolved so informally; they are subject to the political process described in the next chapter.

A related approach that also does not require extensive information about costs and benefits is to concentrate on ominous hazards such as

electrocution and amputation. As a staff engineer at NFPA put it: "I can tell you how to protect against a hazard. I can't tell you how likely it is to happen." With ominous hazards, the probabilities are unimportant. Participants in both public and private standards-setting express the sentiment that certain hazards are obviously worth regulating. "One case can give you the answer," explains a UL employee. "Somebody loses a finger, let's fix it. I don't need information on a thousand cases." A former CPSC employee agrees: "If the problem is really significant, like the amputation of fingers or hands in the snowblower, you don't have to do a cost-benefit analysis to say that a twenty dollar control is going to pay for itself."

"Through almost intuition you can come up with cost-benefit analysis," explains an NFPA employee, who boasts that this intuition resulted in the requirement for ground-fault circuit interrupters. These devices, added to the National Electric Code in 1975, automatically trip the circuitbreaker when there might otherwise be serious or fatal shock. The irony is that a subsequent cost-benefit analysis conducted by the National Bureau of Standards casts doubt on the wisdom of the standard, concluding that the cost per life saved could exceed $7 million.^[1]

Considering these tendencies in two dimensions—that is, separating estimates of cost from evaluations of benefits—suggests that there are different patterns of preference in the public and private sectors concerning whether and when to err on the side of safety. Four combinations of estimation errors are possible. Two of these patterns are suggestive of regulatory outcomes; the other two are ambiguous. A standards-setting system might tend to under estimate costs and over estimate benefits. Because both estimation errors favor regulation, the resulting standards, relative to other combinations of estimation errors, would tend to be the strictest, bordering on overly protective (see table 6). Public standards-writing is generally thought to have these characteristics. Conversely, a standards-setting system in which costs are over estimated and benefits under estimated would tend to be most lenient, with a danger of being too lax. That is the conventional wisdom about private standards-setting. The other combinations of estimation errors are ambiguous because the errors are in different directions.

Two caveats are necessary before carrying this analysis further. First, these descriptions are not meant to be pejorative. Overestimation, in this context, is not deliberate puffery, nor is underestimation intentionally deceptive. Rather, these terms describe tendencies that are akin to

legal rules of evidence—they encompass various presumptions and informal rules concerning the burden of proof and the resolution of scientific disputes under uncertainty. Second, the typology illuminates differences in the direction, not the magnitude, of these tendencies. Standards classified in any of the four quadrants might be socially desirable, depending on the magnitude of the errors. Even with this ambiguity, the observed tendencies in the two sectors are quite revealing.

Moving first to how costs are estimated in both sectors, it is widely assumed that private standards-setters err on the high side, while government generally errs on the low. Private organizations have first-hand knowledge of costs and are likely to be sensitive to them, erring in the direction of the industry's concerns about profitability. In contrast, government lacks this first-hand knowledge and is likely to discount industry estimates of costs for fear they have been exaggerated for strategic purposes. Government also has no direct interest in, and often little actual concern about, how regulations affect profitability.

This much of the conventional wisdom is largely supported by the case studies. Government probably underestimated costs in all four cases. OSHA's estimates of the costs of the grain elevator rule were widely criticized as being too low. The cost estimates in the other cases, although much closer to the industry consensus, involved various minor errors in the "low" direction. The CPSC used a low figure for testing costs in the woodstove proceeding. The FAA assumed that airlines would buy the most inexpensive smoke detectors and, more significant, did not even calculate the additional training costs of Halon.

As expected, a tendency to underestimate costs does not characterize the four private cases. What occurred instead, however, is not readily apparent from the record. Documentation is scant, and many participants obscure the issues because arguments "against safety" are considered either unsavory or impolitic.^[2] The one apparent exception is NFPA 408, where costs were never explicitly discussed. The standard was set practically without regard to cost. Nevertheless, it appears that the NFPA Agricultural Dusts committee tended to overestimate, not underestimate, the cost of safety measures. Members of the National Grain and Food Association surely did so in lobbying OSHA.^[3] Manufacturers generally agree that UL and AGA "understand" the cost of various proposals and do not tend to underestimate them.

The surprise is on the benefit side, where it is generally assumed that private standards-setters underestimate benefits while government overestimates them. This characterization appears to fit government. Estimates of benefits in the woodstove labeling case, for example, were wildly exaggerated. W. Kip Viscusi argues that the estimated benefits of the gas space heater standard were also exaggerated, although the mistake was probably not very large.^[4] There is good reason to believe that the FAA and OSHA overestimated benefits as well. The private sector, by contrast, did not behave as expected. In only one of the four cases did the private sector play down the benefits of taking additional safety precautions. Members of the NFPA committee for grain elevator safety seem resigned to the notion that few, if any, safety measures can affect the number of explosions. But in the three remaining cases, there was a marked tendency for the private sector to err on the "safe" side. This tendency varied by hazard but was unmistakable in overall terms. There is no substantive basis for believing that most aspects of UL 1482 have any real effect on woodstove safety. Similarly, the AGA/ANSI standard for gas space heaters is filled with requirements that have no obvious effect on safety. The same is true of the NFPA standard for aviation fire extinguishers. All available evidence suggests that the benefits of adding more Halon extinguishers and increasing the training procedures for personnel are likely to be very small. That was not the view of most NFPA committee members, however, who tacitly assumed that the benefits were worth the cost.

In short, government standards-setting was true to form, tending in all cases to the combination of errors that promotes overprotective regulation. By contrast, the private standards-setters were all over the map (see table 7). Only one of the four cases, grain elevators, falls into

the category where underprotective regulation is most likely. In two cases, the results were ambiguous. Although it appears that UL and AGA tend to err in the favor of industry when estimating the cost of safety measures, they also erred on the side of safety in estimating the benefits of various provisions for woodstoves and gas space heaters. Finally, one of the private standards, NFPA 408, falls into the quadrant dominated by government. Based on the available evidence, this standard probably errs farthest in the direction of overprotectiveness, mandating even more than the FAA in an area where the likely marginal benefits are, by any reasonable measure, minuscule. The unexpected diversity of private standards carries over to the normative evaluation of outcomes.

Measures of Overall Performance

Distinct regulatory philosophies confine and direct the task of setting standards. Overlapping but different decisionmaking rules characterize the private and public sectors. But outcomes are not as easily differentiated as the philosophies and operating procedures that produce them. There are four reasons why it is difficult to draw policy conclusions from regulatory philosophy alone. First, regulatory philosophy con-

strains but does not determine outcomes. It leaves plenty of room for discretionary decisions. While regulatory philosophy keeps UL from addressing the creosote problem, it leaves considerable latitude in setting other requirements in the standard. These discretionary decisions result in a variety of outcomes. UL 1482 is probably too strict on glass doors, but too lax on metal grates. Similarly, the stability requirements in ANSI Z21.11.2 are undoubtedly stringent, while the clothing ignition test is not.

Second, however distinct these philosophies, the universe of potential outcomes overlap. The public and private sectors are capable of producing very similar outcomes. The FAA and NFPA, acting practically in isolation of each other, developed several similar provisions for aviation fire safety. There were pockets of agreement in the other cases as well—areas where both sectors agreed on the content of particular provisions. The CPSC approved the basic technical and performance provisions in UL 1482 and most of those in Z21.11.2. OSHA did not quarrel with the NFPA provisions for fire safety in grain elevators.

Third, even when regulatory philosophies dictate distinct regulatory outcomes, the normative implications are unclear. The same regulatory philosophy can perform well in one circumstance and poorly in another. Paternalism led the CPSC to adopt a seemingly desirable standard for gas space heaters, but an ill-advised one for woodstoves. Similarly, refusing to make standards retroactive practically gutted NFPA 61B, but had little adverse effect on NFPA 408 or UL 1482. The only way to evaluate regulatory philosophies, then, is by examining outcomes and seeking to understand the conditions under which they perform best.

Finally, it is difficult to draw substantive conclusions about regulatory philosophy because it is multifaceted and signals sometimes conflict. Safety standards are usually polycentric in nature, raising an assortment of complex issues. Under these circumstances, the manifestations of regulatory philosophy are varied and sometimes contradictory. The engineering ethic, for example, helps explain UL's reluctance to address the creosote problem in woodstoves, but it also explains its more stringent structural requirements. In sum, regulatory philosophy shapes the standards-setting process in several important ways, but the normative implications of these differences cannot be stated in simple terms that clearly favor one form over the other.

What is needed, then, is a measure of overall performance to facilitate the comparison of public and private outcomes. Whether that is possible strikes at the heart of a long-standing debate about process

versus substance. Dispensing with pesky substantive issues, those disposed to the legal perspective turn exclusively to process (that perspective is examined in chapter 11). Those venturing substantive conclusions usually couch the analysis in terms of either strictness or reasonableness. Both concepts encompass important social considerations. Strictness generally refers to absolute benefits; reasonableness, to the relationship between benefits and costs. These frames of reference differ significantly, creating conflicting impressions of many standards. Public standards are usually considered stricter, but less reasonable, than private ones. That is, they probably generate more absolute benefits than private standards, but at a cost higher than many considered acceptable. By contrast, almost no one argues that private standards impose unreasonable costs. But reasonableness is thought to come at the expense of strictness. Obviously, these substantive measures should be merged in some manner. Standards should be compared through an aggregation of strictness and reasonableness.

In theory, cost-benefit analysis facilitates the task. But there are practical and philosophical objections to even the roughest forms of cost-benefit analysis. The dearth of reliable data, particularly on benefits, dims the potential for such analysis. Part of the problem is forecasting. Much depends on implementation. For example, OSHA's "action level" for grain elevator housekeeping will produce benefits if it is implemented through a reasonable inspection scheme. It will foster unreasonableness if implemented poorly. On a more philosophical level, cost-benefit analysis requires that monetary values be placed on life and limb, something both methodologically difficult and politically explosive. Cost-benefit analysis is out of the question for current evaluative purposes. The data necessary to support it are not available. Nothing remotely resembling cost-benefit analysis was undertaken by any of the private standards-setters. And the economic analyses conducted by OSHA, the CPSC, and the FAA are easily faulted for reasons elaborated in the case studies.

One way to evaluate outcomes directly despite the significant uncertainties about costs and benefits is to indulge in liberal evidentiary presumptions and seek only to separate the obviously bad outcomes from the possibly good ones. In those terms, a standard is within the "zone of reasonableness" if there is credible evidence that benefits are (1) nontrivial and (2) not significantly in excess of costs. This overly inclusive notion of "reasonableness" avoids the almost intractable disputes about the precise magnitude of uncertain benefits and costs. This approach also permits a surprising number of normative conclusions

about the cases. The results, based on the summary evaluation in the case studies, are summarized and explained below:

On the public side, two of the standards were clearly too strict. There was no evidence that the CPSC's woodstove labeling rule would result in any measurable benefits; and fire protection engineers agree that the FAA's standard for aviation fire safety is very unlikely to generate benefits in excess of cost. The other two public standards were well within the zone of reasonableness. The need for grain elevator regulation is supported by the evidence, and the OSHA standard, although it could be improved, is reasonable in several respects. The CPSC's standard for gas space heaters is probably the best of the bunch. It seems to be responsible for the widespread use of ODS technology, an inexpensive and effective method for dealing with a problem of uncertain dimensions. The private sector was also divided. Two of its standards were clearly undesirable, although one was too strict and the other too lax. The NFPA went even further overboard than the FAA on aviation safety. NFPA officials privately admit that these standards are not supportable in economic terms. The NFPA also missed the mark on grain elevators, adopting a standard that is so weak that it barely addresses the most serious problem: grain dust.

Two tentative conclusions can be drawn from this normative evaluation. First, neither sector is clearly better than the other. Both sectors appear as capable of failing as they are of succeeding, although the public sector seems more likely to be overly strict, while the private sector is more likely to be too lax. Nevertheless, private standards should not be rejected solely on the theory that they tend to be underprotective . Sometimes they are not. This analysis also suggests that there is no simple answer to the question whether the public or private approach is generally better. Private sector behavior is too varied. NFPA 408 and 61B are cases in point. Second, the cases suggest that particular issues can foster distinct patterns of regulatory behavior. It is

no coincidence that both the public and private standards for aviation fire safety seems "overly strict."

Both of these observations underscore the importance of understanding the reasons behind the observed results—the subject of the next chapter. Given the unusual nature of the "paired" case studies, however, it is appropriate to consider first whether these observations are intrinsic to overlapping standards. In other words, to what extent is the behavior of either sector affected by the activities of the other?

Accounting for Regulatory Overlap

As with most instances of overlapping standards, the private standard came first in all four cases. The most pertinent question, then, is whether the public sector tailored its behavior to the private standard. How the private sector responded to government intervention is also important, given widespread concern that public standards will "drive out" private ones.

In two of the cases, grain elevators and aviation fire safety, the public sector acted almost independently of the private sector. In neither instance was the public sector trying to influence the private. OSHA did not seriously pursue a strategy of trying to improve NFPA 61B, and the FAA did not even know about NFPA 408 until well into its own rulemaking proceedings. Therefore, the observations from those cases can safely be said to represent inherent traits, not ones unique to overlapping standards.

The most curious aspect of these cases concerns the private sector, which apparently did not alter its behavior in response to the threat of public intervention. The NFPA did not respond to the FAA, other than to send a brief letter to the public docket. Several committee members interviewed while the FAA regulation was in the proposal stage knew almost nothing about the regulation. In contrast, committee members considering revisions to NFPA 61B in July 1985 were acutely aware of the proposed OSHA rule. But they, too, took little action aimed at forestalling OSHA regulation. The committee seemed resigned to government regulation and was not willing to enact serious grain-dust controls in order to prevent it. This relatively independent behavior bolsters the conclusions already set forth about public and private standards-setting, but it does not necessarily bode well for public policy. The government's failure to exploit existing standards or to try to in-

fluence them seems wasteful. The FAA's actions certainly violated the OMB directive to consider existing private standards before adopting government ones.

Both cases involving the CPSC were interactive from the start. The agency acted with full knowledge of the relevant private standards, attempting in both cases to supplement them with beneficial government regulation. The private sector responded in both cases as well. To what extent does this interaction alter the conclusions about public and private behavior? It is impossible to know without somehow comparing this "interactive" behavior to cases in which there is no private standard. Some effects are obvious, however. The CPSC tailored its regulations to supplement private standards. Whether it would have proposed more comprehensive standards in the absence of private ones is uncertain, but the agency would have been hard pressed to justify the technical judgments contained in, say, UL 1482. A close examination of these cases suggests that the most significant effects of this interactive behavior went largely unnoticed. In the case of woodstoves, not only did UL alter its warning labels to placate the CPSC, but the percentage of certified woodstoves rose dramatically after the CPSC proposed its regulation. With gas space heaters, significant changes were made in ANSI Z21.11.2. The small porcelain bathroom heater was eliminated as a result of numerous improvements in structural requirements and performance standards. These changes in private standards should rightly be credited to government action, and conclusions about the private sector must be adjusted accordingly.

In sum, the CPSC cases suggest that the power of government to influence private standards is significant. Unfortunately, government's capacity for recognizing these improvements is apparently not as large. In neither case did the government clearly recognize the improvements in the private standards. Instead, driven by an enforcement ethic that was apparently unaffected by the private standards, government went ahead with regulations that produced minimal marginal benefits. The next chapter looks beyond these interactive effects to broader explanations of the differences in decisionmaking and performance.

Nine
Explaining Regulatory Behavior

Whether public or private, safety standards generally impose concentrated costs and confer dispersed benefits. "It may be astonishing that [regulation] of this sort is ever passed," observes James Q. Wilson, who advances a theory of entrepreneurial politics to explain its emergence in the public sector.^[1] The theory, which relies on latent public opinion and anti-business sentiment, helps explain the government's behavior in at least three of the four case studies. Aviation safety seems to be a case unto itself. Wilson's theory does not explain private regulation, however. Private standards are not foisted on the private sector the way public standards are. Instead, private regulation is usually explained by some form of "capture" theory. The notion of capture is almost redundant, since private standards-setting is a form of self-regulation.^[2] Private standards are often thought to reflect "the lowest common denominator" or "the path of least resistance." But the varied and sometimes stringent behavior observed in the private sector challenges these notions. The expected politics of private regulation prevailed in only one case, resulting in the lax standard for grain elevators. Moreover, there are puzzles about the behavior of the private sector in all four cases. Why was the Z21.11.2 committee willing to upgrade its requirements, substantially adding to retail costs, but unwilling to add the inexpensive oxygen depletion sensor? Why are certain provisions in UL 1482 and NFPA 408 overly stringent in the view of independent experts with no

commercial axe to grind? Moreover, why weren't economic concerns ever voiced when NFPA 408 was upgraded?

This chapter explores two traditional explanations for regulatory behavior, along with three more powerful ones suggested by the case studies. The most conventional explanation of private regulation builds on the metaphor of the path of least resistance. Forestalling government regulation is the other motive most commonly attributed to the private sector. Other explanations suggested by the cases include: the political economy of product testing, the peculiar politics of aviation safety, and the legal constraints of the regulatory environment.

The Political Economy of Safety Regulation

The Path of Least Resistance

Private standards-setting is widely thought to be controlled by those who want the least done. Most private standards are "consensus" standards, a process noted for the rule that there be no "unresolved negatives" among those participating. This process seems practically designed to ensure that standards-setters follow the path of least resistance. The need for consensus, Eads and Reuter argue, leads to a "watering down" of many standards.^[3] The process is abetted, others add, by the lack of consumer representatives agitating for increased safety.

The case studies provide selective support for the conventional wisdom. Business interests often express concerns about the cost of upgrading standards. Manufacturers of woodstoves and gas space heaters complained that the costs of certain UL and AGA requirements were unjustifiably high. Grain elevator operators sing the same refrain. There may be a realm in which engineers reign, but it is most likely to encompass provisions that would not add costs perceived to be significant by producers.^[4] "People who don't have instructed votes still know who they work for," observes an NBS official familiar with many private standards-setters.

Concerns about "imbalance" among participating interests also find support in the cases. Organized consumer groups had a meek voice compared to manufacturers in all of the private cases. "Numbers count," according to an engineer at Consumers Union, "and I have been at a lot of meetings where I am the only negative vote—and that is the end of it." UL and some ANSI-sponsored committees pay lip service to "consumer participation," but consumer groups actually play only a

minimal role in the process.^[5] It is not that private standards-setters specifically exclude consumers—although attorneys have occasionally been excluded—so much as that consumers are unorganized and rarely have the resources to participate.^[6]

But the "path of least resistance" is not a predictive explanation, since it says nothing about when resistance will or will not occur. Moreover, it leaves unanswered the question why pro-safety results sometimes obtain over the objections of some business interests. The cases suggest two reasons why the lowest common denominator might not prevail: (1) there are unlikely pro-safety interests in the private sector, and (2) there are institutional impediments to the "watering down" process hypothesized by Eads and Reuter.

Unlikely Safety Interests . Although the range of interests on most committees is limited by the paucity of consumer participants, the cases highlight several largely unrecognized advocates of safety in the private sector. For example, representatives of the insurance industry have an interest in controlling losses. Insurance representatives pushed UL to upgrade its requirements for metal chimneys.^[7] Vendors of safety equipment also carry the pro-safety flag. Manufacturers of dust-control equipment speak out for improved housekeeping measures on the NFPA Agricultural Dusts committee. Labor representatives on some committees also favor various safety measures. The Air Line Pilots Association is active on NFPA's aviation committees; a representative from a Grain Workers Local sits on its Agricultural Dust Committee.

Installers and servicers also care about safety, often more than they care about cost or selection. An outspoken member of the National Chimney Sweep Guild was almost successful in getting NFPA to ban large stoves that require clearances of more than the traditional thirty-six inches.^[8] Representatives of gas utilities, whose employees must confront the victims of product injuries, frequently argue in favor of additional safety features for gas appliances. The J. C. Penney testing lab, designed to incorporate considerations about the consumer into purchasing decisions, also suggests occasional improvements in UL standards. Finally, "the fire service" itself, as it is known around NFPA, is well represented on NFPA committees. Firefighters spread what the NFPA president recently called "the gospel of fire protection."^[9] A Washington lawyer familiar with private safety standards agrees that "most NFPA committees have a bias toward the fire services." Needless to say, cost-benefit analysis is not one of the commandments.

The most intriguing examples of unlikely safety interests are committee members who appear to act against the interest they represent. Staff members at AGA Labs, for example, favored a ban on space heaters, while the AGA at large—overseer of the labs—opposed it.^[10] The representative of a major airline takes positions on NFPA aviation committees that are sometimes opposed by his superiors. The simple explanation is that an "interest" is rarely homogeneous. Just as consumers are not interested only in safety, manufacturers are not interested only in costs. Take a commercial airline, for example. Part of the organization is interested in keeping operating costs down, while other parts are interested in reducing insurance and liability costs. The latter are likely to favor increased expenditures for safety. The Maintenance Department probably wants a larger budget, and the Marketing Department might be interested in adding those safety items that matter most to passengers (for example, smoke detectors but not fire-blocking seat covers). In the case of NFPA 408, the airline representative cited above works in the Engineering and Safety Department. His professional interest is in safety, with little regard for cost. And since it is more likely that a safety engineer than an account executive will sit on the relevant standards-setting committee, organizational interests are likely to be subtly slanted in favor of safety in many cases.

Institutional Considerations . The second problem with "capture" theory is more fundamental: there are institutional reasons why standards-setting does not always reflect the conventional wisdom. Primarily, standards used by product testing labs are not written by "consensus." They are written by in-house engineers and only later sent through a "canvass" process leading to ANSI approval and the designation "national consensus standard." Fatal to the "watering down" theory, delays in the ANSI canvass process have no effect on the testing labs. UL uses (and revises) its standards long before they are distributed by ANSI for comment and review. UL officials privately confirm that failure to achieve ANSI approval would not affect this situation. In other words, UL is happy to go ahead without ANSI's imprimatur. Accordingly, no participants have an influence akin to veto power. While UL seeks comments from industry, it also adopts standards that do not enjoy the "consensus" of all manufacturers. The "least common denominator" argument also lacks an explanation for the role of the staff. A study of the Association of Home Appliance Manufacturers concluded that the staff has its own sense of mission and its own in-

terests, pursuing some goals that are not necessarily in the interest of the general membership.^[11]

Forestalling Government Regulation

Private regulatory behavior that does not follow the path of least resistance is often explained as an effort to forestall government regulation. This theory allows for stricter results without attributing altruistic motives to the private sector. The theory both overexplains and underexplains private regulatory behavior. It overexplains to the extent that virtually any outcome, lenient or strict, can be attributed to the threat of government regulation. As David Garvin states:

Only if firms were able to identify both the standards that would be established under government regulation and the precise amount that voluntary standards could fall below this level without inducing government action would [the lowest possible level] automatically follow. This information is seldom available. As a result, risk-averse firms, anxious to deter government regulation, might unknowingly set standards as rigorous—or, conceivably, even more stringent than—those that the government would demand.^[12]

Accordingly, the theory is not predictive either. Almost any outcome might result from the desire to forestall government regulation. The "forestalling theory" also underexplains private regulatory behavior, since private standards generally predate and significantly outnumber government regulations. The threat of CPSC regulation, for example, cannot possibly be relevant to more than a small number of UL standards.

Even if the forestalling theory does not have wide application, it is most likely to apply to paired cases. Although private standards usually predate public ones—as they did in all four of the case studies—it seems likely that the private sector might upgrade its standards in response to the threat of government regulation. At first glance, the private participants seemed to be motivated by the possibility of government influence. UL changed its labeling requirements in response to the CPSC. The same motive helped move AGA Labs to upgrade Z21.11.2. Even the NFPA Agricultural Dusts committee upgraded 61B in a few respects with the hope of forestalling OSHA. But these motivations were not particularly strong. Although UL changed its labeling provisions, the bulk of the standard has nothing to do with labeling or the threat of CPSC regulation. Committee members considering revisions to NFPA

61B seemed resigned to government regulation and certainly were not willing to enact serious grain-dust controls in order to prevent it.

Forestalling government regulation was a dominant influence in only one case: gas space heaters. The AGA/ANSI standard was upgraded in several significant ways while government contemplated regulatory action. But, as with NFPA and grain elevators, the private sector was unwilling to mandate what the government obviously wanted: an ODS requirement. It added that only after the CPSC did. This is not to say that forestalling government regulation is always a weak motive. The case studies suggest that this motive is largely subordinate to other factors, such as the demand for private standards. That demand and the broader political economy of private regulation suggest that product testing labs have different motivations than other private standards-setters. Product testing labs have a financial interest in eliminating competing government regulation, before or after the fact. That is why AGA/ANSI adopted the CPSC's ODS requirement and why UL was willing to alter its labeling requirements. Other standards-setters seldom have this strong an interest. The political economy of product testing is discussed below, followed by consideration of the peculiar politics of NFPA and the special problem of aviation safety.

The Business of Product Testing

Hypotheses based on the assumption that private standards-setting involves some form of collective deliberation and democratic decision-making are inappropriate for the hundreds of standards developed by testing labs. Unlike government agencies, these organizations do not operate on an issue-by-issue basis. Standards are an ongoing business. And being in the business of certification affects the approach one takes in promoting the goal of product safety. Motivated by the financial interest in certification and bolstered by the market power attendant to such a position, testing labs have different regulatory incentives than are often attributed to the private sector. Testing labs have unusual incentives for making standards demanding, although they experience countervailing forces as well. The tendency toward stringency is particularly strong when it comes to compliance.

Testing labs generally do not develop standards unless there is a proven and profitable market.^[13] These labs are in a responsive position, responding to a demand for a given standard. UL does not seek out

clients, clients seek out UL. They come for various reasons. Sometimes product certification is required by law (as in jurisdictions mandating listed woodstoves), sometimes the demand originates with product liability insurers or retail stores, and on occasion an industry voluntarily seeks certification to bolster consumer confidence.^[14] These forces combined to compel compliance with the Z21 standard. They were weaker in the case of woodstoves, probably owing to poor building code enforcement, although the pressures for compliance increased over time.

The economic realities of product testing help explain one of the major differences in public and private conceptions of safety regulation. To private testing organizations, it is almost always a "given" that the product will be marketed. UL and AGA Labs never seriously consider whether a product is best banned altogether (or "unlisted," in lab parlance). Even though some staff members at AGA Labs felt that unvented gas space heaters should be banned, it was a foregone conclusion that AGA Labs would continue to list them. The certification business can be very profitable. It is not in the interests of AGA Labs or UL to eliminate a testing market.

Testing labs are also particularly sensitive to the cost of developing and implementing standards for certification. This business equivalent of cost-benefit analysis further limits the kinds of standards that get written. The process is akin to the problem of "orphan drugs."^[15] Insufficient demand for certification causes testing labs to overlook certain problems. Aviation safety is a case in point. There are sound scientific arguments for testing aviation fire extinguishers to special standards. UL advanced these arguments when opposing the FAA rule. UL has even received requests to develop such standards. But from UL's point of view, the limited market for aviation fire extinguishers does not justify the cost of developing special standards and test procedures.

Considerations attendant to the business of product testing help explain other aspects of UL's problem. In order to maintain support and credibility for their business, testing labs must take care that their test methods are defensible, reproducible, and relatively inexpensive. Developing test methods is inherently difficult, since the goal is a simple procedure that provides information useful in a complex world where products are used (and misused) in countless ways. UL and other testing labs generally handle the problem by deferring to accepted definitions of proper use. This means following the applicable use and installation codes as well as the manufacturers' instructions. This helps explain why UL is so reluctant to factor consumer misuse into its standards. There

is often no defensible way to do it, especially when the misuse concerns the most basic conditions of installation and use. UL tests kerosene heaters, for example, using certified K-1 kerosene fuel as specified in the manufacturer's instructions. Testing for the "flare up" problem would require the addition of some amount of gasoline to the fuel. Should this be trace amounts, as would occur if a consumer put kerosene into an old gas can, or some larger amount that would simulate even grosser misuse? There is neither an obvious nor an easily defensible answer. Similar problems are posed by the fact that misuse can occur in a variety of different ways. Should UL also add tests for other improper fuel mixtures? Including "representative" examples of misuse could expand the cost of testing significantly, raising objections from manufacturers. It could also force some products from the market entirely, eliminating the market for certification. If enough gasoline is added to the fuel, all kerosene heaters will fail the test.

Simulating misuse can also be technically difficult, particularly if the hazard is related to poor maintenance. For example, it would be difficult to design practical and reproducible tests to simulate the kind of lint buildup that can, after six to eight months, impede the performance of a gas space heater. Creosote buildup poses a similar problem. It would be extremely time-consuming to build up actual creosote for testing. Yet without such an approach, there would surely be criticism of the test method intended to simulate the problem.^[16] In sum, organizations such as UL sometimes resist recognizing misuse problems because of the implementation problems that would occur in testing. Product certifiers must think ahead to the implementation problems involved in developing test methods. Accordingly, they are sensitive to problems that the CPSC often overlooks in its desire to "do something."

Beginning with a demand for their service also endows the testing labs with certain regulatory power. It may be certain that the product will be listed, but the listing agency has considerable discretion in determining the requirements. There is almost never competition between labs using different testing requirements. Hence, a manufacturer who disagrees with, say, the UL standard for woodstoves has little choice but to comply. A lawyer who criticizes many private standards for being too lenient agrees that "the UL staff, from a professional point of view, can often set a level just because they think it is right." Market power does not create unbounded discretion, however. Opinions vary about the power of the staff at organizations such as UL. Most observers agree that proposed standards must enjoy the support of a critical mass of

manufacturers. Even though UL has market power, it cannot afford to lose industry support. The credibility of the UL label would be called into question if firms were moved to challenge its standards openly. Accordingly, UL engineers speak of "what they can get away with" in their safety standards.

Several additional factors constrain the commercial testing labs. Product certification also involves a considerable degree of deference to manufacturers. The only way UL can provide certification quickly, particularly for new products, is to incorporate many prevailing business practices into its standards. The impression of most observers is that AGA and UL can "push" a recalcitrant firm marketing a product below widely accepted levels of safety, but must "follow" the determination of what is generally acceptable. "We are always following the crowd," comments an engineer at AGA Labs. The ability of testing labs to "push" is also affected by the availability of information. UL and AGA do not write product standards so much as review existing products. The certification business is prospective; products are often certified before they are marketed. Armed with specific information about product problems, UL is most effective in gaining overall improvements in standards. Absent such information, however, anyone urging a change in UL standards is likely to be rebuffed for failing to present evidence "from the field."^[17]

Product certifiers take a similar view of research: it is expensive and will not be undertaken unless the payoff is clear. Research projects are occasionally undertaken that might improve numerous standards at once, such as research on shock hazards and sharp surfaces. But research to improve some aspect of a single standard is much less likely. UL is not inclined to, investigate the qualities of catalytic combustors for woodstoves unless industry sponsors the work. UL 1482 will continue to generate revenue without this research.

A final aspect of product certification that distinguishes it from other private standards is compliance. Compliance is the business of product testing. The income received from these "follow-up" services, combined with the threat of liability if products do not actually meet the stated requirements, prompts the testing labs to implement comprehensive inspection schemes. Testing labs go to extraordinary lengths to ensure that products bearing their label comply with their standards. UL inspects producers at least four times a year, regardless of their track record—an approach that far exceeds any public enforcement scheme. But product testing labs are also more willing to phase in requirements

over time. Since they demand total compliance from their clients and charge dearly for certification, the best way for testing labs to placate angry customers is to allow ample time for compliance. Unlike the CPSC, UL is usually receptive to adjusting the effective date to production cycles, assuring minimal problems with retrofitting or disposing of existing inventories.

NFPA Standards: Economics, Politics, and Process

While independent labs are both bolstered and constrained by the market for certification, the forces affecting membership organizations are clearly different. But the underlying reality, as observed by Anthony Downs, is essentially the same: "No bureau can survive unless it is continually able to demonstrate that its services are worthwhile to some group with influence over sufficient resources to keep it alive."^[18] This support need not come directly from the most obvious source, the membership. Membership dues account for 41 percent of ANSI's revenues; the sale of publications accounts for 54 percent.^[19] Other membership organizations are even more dependent on those purchasing their standards than on those writing them. Membership dues account for less than 8 percent of NFPA's revenues; publication sales account for almost 60 percent.^[20] Sufficient support from those who rely on standards can empower private standards-setters vis-à-vis their membership.

Beyond these market-based forces, other sources of support "such as the state subsidizing the organization or legally enforcing its policies" can supplement and transform constraints imposed by the membership.^[21] As a result, the demand for NFPA standards, and hence the nature of NFPA politics, is extremely variable. The National Electric Code, given the force of law in virtually every state in the country, generates intense interest from a wide range of groups. So does the Life Safety Code. In those instances, NFPA standards-setting is apparently characterized by lively interest-group politics. The health care industry battles with the building trades; the makers of plastic pipe oppose those who sell metal. Both standards were the source of scandals earlier this decade when specific commercial interests attempted to "pack" the annual convention to change the standard through an unusual floor vote.^[22] NFPA has since vested greater authority in its Standards Council, making it impossible to write code text on the convention floor.

This is not to say that NFPA standards are merely a reflection of the

most powerful interests on the relevant technical committees. First, these committees have a "balanced" membership designed to prevent any single group from dominating. Some of these interests, as discussed earlier in this chapter, are advocates of safety regulation. Second, installation and use standards permit a modicum of leeway unavailable to testing labs. Use standards are broader in scope than product standards, so that a demand for the standard will still exist even if it prohibits certain products or practices. Although NFPA desires to sell as many standards as possible, and this certainly affects which standards get written and how often they are updated, it is not beholden to particular interests in the same way as product certifiers. Accordingly, the National Fuel Gas Code prohibits propane cabinet heaters (and could do likewise with gas space heaters) without affecting the demand for the standard.

Unfortunately, the demand for other NFPA standards, including 61B and 408, is considerably weaker than that for the National Electric Code. The NFPA liaison to the grain elevator committee could not name a single organization or firm that actually uses NFPA 61B. The only apparent interest in NFPA 408 is from foreign countries. The stakes are much lower for standards with such little demand. These standards are more likely to fall behind the times as weak demand is reflected among the committee members. NFPA is also unlikely to pour organizational resources into standards that generate little income. The resulting politics are quite varied. But without a strong demand for the standard—created through, say, use by some government agency—NFPA committees lose an important source of leverage in determining how strict to make standards.

The politics of NFPA regulation also differ from what Wilson has described on the public side in one glaring respect: antibusiness sentiment does not drive private regulation. As a consequence, private regulation is largely unaffected by the injury statistics and major catastrophes that entrepreneurial politicians use to mobilize public support for regulation.^[23] NFPA investigated the Air Canada incident, but there were no apparent repercussions for any of the aviation safety committees. The series of grain elevator explosions in 1977 probably prompted the NFPA Agricultural Dusts committee to meet—revision of the standard was already overdue—but the explosions did not cause significant changes in the standard. The limited role of public sentiment in private regulation seems to account for the less hurried and often more thoughtful approach to standards-writing in the private sector. The

kind of pressure that motivated the FAA staff to want, above all else, to complete the rulemaking proceeding for aviation fire safety does not exist on the private side. The reduction in hastiness might come at the expense of timeliness, however. Lacking external pressure to "do something," the private sector sometimes does nothing at all. NFPA 408 was so neglected that by the late 1970s it was considered practically useless.

Lacking what Wilson calls the most potent method for overcoming opposition to regulation—the public sentiment that follows any catastrophe—the NFPA committees apparently do whatever can be accomplished through political compromise. This suggests a grim reality, with standards taking whatever shape business interests desire. Hence the conventional wisdom that private standards-setting organizations are "captured." The grain elevator case supports this hypothesis. The National Grain and Feed Association opposes any regulation of grain elevator housekeeping, and NFPA 61B is obliging. But this hypothesis has limited explanatory power. The political "compromise" reached in aviation safety, and possibly in many other NFPA standards, is surprisingly strict. More NFPA standards should be examined before general conclusions are drawn about the politics of NFPA standards-setting in specific, and of membership organizations in general.

Risk and Culture: The Peculiar Fear of Flying

The politics of aviation safety are unusual, possibly unique. The political forces favoring increased public regulation are formidable, particularly during a proclaimed era of deregulation. Many congressmen favor increased safety regulation, practically without regard to cost. (They travel frequently by air, and so do their most influential constituents.) The NTSB, a zealous advocate of increased regulation, provides Congress with recommendations that carry the weight of "expert advice." The powerful Air Line Pilots Association also joins several unions representing flight attendants in urging Congress to adopt regulations proposed by the NTSB. Virtually nobody but the Airline Transportation Association argues against increased expenditures for aviation safety, and its voice is muted by the fact that many airlines generally favor government regulation. That is, they often prefer to have the government specify necessary safety precautions than to decide themselves. There is also widespread popular support for airline safety. It is one of those rare issues that motivates hundreds of citizens to write cards and letters directly to an administrative agency in support of new

regulations. More common are the CPSC proceedings, which attract little public interest.

Aviation safety is, quite simply, a political sacred cow. This explains much of what happened with the respective NFPA and FAA standards. The unusual alliance of political interests favoring aviation safety regulation accounts for the FAA proposal and its quick adoption. Mobilized by the Air Canada fire, these forces motivated Congress to establish strict oversight procedures in 1984 to ensure the quick enactment of various regulations, including the fire extinguisher and smoke detector rule. The special status of aviation safety apparently dulled the OMB's normally critical senses about new regulations. There were no challenges to the cost-benefit analysis, no delays to seek additional justifications or modifications from the agency. Similarly, without serious objection, NFPA enacted regulatory provisions that committee members agree will cost more than what can be justified rationally. But apparently the fear of flying is pervasive. The politics of aviation safety produces regulatory outcomes unlike those commonly associated with social regulation.

Legal Constraints on Standards-Setting

Political and economic factors only partly explain regulatory behavior. The law is a powerful external influence, placing very different constraints on the two sectors. Only private standards-setters are subject to the antitrust laws—unless they can cloak their activity under the guise of a "public purpose." And lacking the sovereign immunity defense, private organizations are far more vulnerable to liability suits than their public counterparts. But public agencies are subject to more frequent and intrusive judicial review of their regulatory decisions. These legal influences, explored below, help explain the observed differences in regulatory philosophy.

Antitrust Law

Critics of private standards charge that the activity is prone to anticompetitiveness and that, owing to lax antitrust enforcement, the law does not sufficiently discourage such activity.^[24] Proponents counter that the combination of liability law and antitrust law keeps private standards within an acceptable range of results. The argument goes as follows: private standards must be reasonable because liability attaches if a

product is not reasonably safe, while the antitrust law assures that standards will not be arbitrary or anticompetitive. The case studies suggest that antitrust concerns are voiced by committee members more openly than those about liability, but their perceptions of antitrust law are often uninformed.

The accusation that anticompetitive purposes lurk behind private standards is understandable given the unsavory history of product standards. Standardization schemes involving everything from ice cream to carpets were struck down by the courts in the 1940s as thinly veiled efforts at price fixing.^[25] AGA Labs was implicated in one of the most famous legal decisions in the field, Radiant Burners v. Peoples Gas, for allegedly withholding product certification for anticompetitive reasons.^[26] Economic considerations certainly influenced the AGA/ANSI committee's decision about requiring the ODS device when there was only one supplier. Ironically, while pleading antitrust concerns, most committee members were actually worried about putting U.S. companies at a competitive disadvantage. But popular perceptions to the contrary, there are several reasons why safety standards are unlikely to be infused with anticompetitive purposes. First, price fixing schemes tend to involve the shape or size of a product, not its safety features. The former are necessary to standardize prices, the latter are not.^[27]

More important, anticompetitive motives are most likely to be a problem in interpreting or applying safety standards, not in writing them. Anticompetitive implications are the clearest in applied situations, where outside scrutiny is minimal compared to standards-writing. The most celebrated case in the field, American Society of Mechanical Engineers v. Hydrolevel Corp., involved a conspiracy between two volunteers on a small subcommittee that helps issue thousands of annual interpretations to the Boiler and Pressure Vessel Code.^[28] In 1988, the Supreme Court declined to apply the "governmental action" immunity and upheld a verdict against a producer of steel conduit that had "packed" an NFPA meeting in order to keep polyvinyl chloride from becoming an accepted substitute under the National Electric Code.

There is somewhat less here than meets the eye. There has never been a successful antitrust suit alleging that a safety standard itself, as opposed to an interpretation, violated antitrust laws. The recent Supreme Court cases involved unusual circumstances and are unlikely to have widespread effect on private standards-setting, although they suggest a trend that might be counterproductive. American Society of Mechanical Engineers v. Hydrolevel Corp. involved an interpretation of the Boiler

and Pressure Vessel Code, something that focuses directly on a single product, not on the formation of general standards. Allied Tube and Conduit Corp. v. Indian Head involved a decision on the floor of an NFPA meeting, not in a committee or subcommittee, where almost all policy decisions are actually made. Morever, as a result of changes prompted in part by these cases, standards-setters are providing fewer "interpretations" and the decisive "floor vote" is a thing of the past at NFPA.^[29]

This leaves vast regions of standards-setting largely unaffected by antitrust considerations. In fact, the law tolerates a great deal of private activity, some of it questionable by almost any measure. By applying the so-called rule of reason to cases involving standards-setting proper, courts have sanctioned several private standards acknowledged to stifle technological innovation. Absent a showing of bad faith, there is almost always an "objective" reason for a standard that can pass judicial muster.

Private responses to antitrust law often seem out of proportion to the actual requirements of the law. Antitrust law is a popular excuse for almost every negative decision in the private sector. Since standards exclude things by definition, there is always an intuitively appealing argument that proposed requirements might illegally exclude something from the market. The law is not actually this restrictive. Even standards of obviously questionable scientific merit have been treated with great deference by courts and agencies.^[30] "We give the benefit of the doubt to the standards setter," according to a Federal Trade Commission official.^[31] The makers of low-tolerance metal chimneys or high-clearance woodstoves would have a valid antitrust claim only if they could prove the standards to be "manifestly anti-competitive and unreasonable."^[32]

Antitrust law is nevertheless influential in standards-setting, according to participants. Fear of the antitrust law, well founded or not, apparently helps explain the reluctance of private organizations to ban specific product types or push technology. A representative of the Gas Appliance Manufacturers Association argues that the Z21.11.2 committee could not require the ODS device, which was manufactured solely by a European manufacturer, without raising "serious antitrust problems." UL, beset by a controversy about the temperature tolerance of metal chimneys, argues that the antitrust law is one of the reasons it cannot upgrade the standard.^[33] Similarly, after the NFPA general membership agreed in an unusual floor vote to ban woodstoves requiring more than thirty-six-inch clearances, the Standards Council overturned

the action, citing antitrust concerns.^[34] These concerns are both widespread and overstated.

Liability Law

Unlike antitrust law, liability law is seldom cited as an official reason for doing anything in the private sector. Yet the case studies suggest that liability is a more significant factor than antitrust law in explaining regulatory behavior. Liability concerns help explain why so many provisions in NFPA 61B are vague and why, in both 61B and 408, NFPA pushes the distinction between "requirements" and "appendices." More generally, liability law seems to explain the reluctance of the private sector to (1) recognize or address issues of consumer misuse, (2) embrace certain new technologies, and (3) include work rules or operational controls in safety standards.

Turning first to misuse: incorporating misuse considerations into standards can nullify the one effective defense to product liability suits that product certifiers and manufacturers still retain in some states—the misuse defense.^[35] This defense works only if the plaintiff's use of the product was unforeseeable to the manufacturer. Accordingly, the more misuse is acknowledged in standards, the more manufacturers and product certifiers might be held responsible for it (no matter how impossible the task). In other words, because of liability law it is not in UL's interest to think too much about potential consumer misuse.

There is a similar reluctance to adopt certain new technologies. The failure of a new safety device might leave a manufacturer or certifier in a worse legal position than if it had not addressed the problem to begin with. The argument, advanced to explain the slow diffusion of antilock braking technology, appears to explain the reluctance to require the ODS for gas space heaters. Committee members admit privately that data on failure rates of the device in Europe were considered legal dynamite.^[36] Even if it significantly reduced the number of carbon monoxide deaths, as expected, manufacturers worried that the occasional injury it failed to prevent—as a result of equipment failure or poor maintenance—would be seen as the fault of the manufacturer, not the consumer.

Standards-setters also worry that "upgrading" a standard leaves existing products more vulnerable to product liability suits. Improvements can be cited in some jurisdictions as evidence that the old standard was

inadequate. When, as in the case of UL's metal chimney standard, hundreds of thousands of products manufactured under the existing standard are still in use (and will be for many years to come), there is considerable reluctance to "upgrade" the standard. This reason, admitted privately by UL officials, makes much more sense than the stated concern about excluding low-tolerance chimneys from the market.

Finally, liability concerns help explain why the operational controls were placed in the appendices of NFPA 408 and 61B. Technical and engineering standards are much easier to control. Third-party certification assures compliance with many product standards. Compliance with other provisions in NFPA 408 and 61B involves fairly discrete acts—carrying a certain type and number of fire extinguishers, attaching devices to converyor belts in grain elevators, using specified building materials. Operational controls, on the other hand, require unfailing implementation. They are dependent on the goodwill and reliability of employees. They are also where corners are likely to be cut in the face of various economic pressures. Either way, liability fears apparently keep the private sector from mandating operational controls.

The Burden of Justification

The most important legal influence on public regulation appears to be the specter of judicial review. There is no analogy in the private sector. Except for violations of antitrust law, there is no possible basis for challenging the substance of private standards in court (and the antitrust law, as already explained, gives substantial deference to private safety standards in the absence of proven bad motives). With the CPSC, standards are not just reviewable by court, the statute allowing such review is intended to encourage closer judicial scrutiny than under the traditional, more deferential approach.^[37] The agency has not fared well in the courts. A standard for swimming pool slides was rejected because the CPSC "provided little evidence that warning signs would benefit consumers."^[38] The court rejected arguments based on "common sense" and faulted the agency for not having tested the effectiveness of the required signs. That same year, another court, rejecting portions of a safety standard for matchbook covers, practically demanded that the agency justify all of its decisions with cost-benefit analyses.^[39]

This sort of judicial review creates a burden of justification on public agencies that does not hinder their private counterparts. "We cannot go to court on engineering judgment," notes a CPSC staff member with

dismay. This suggests a major difference between public and private standards-writing. Public agencies cannot "get away with" what forms the basis for most private standards: engineering judgment, common sense, and educated guesses. The burden of justification probably helps explain why the government agencies shied away from technical issues in the case studies, seizing issues for which technical proof is least significant, such as warning labels and effective dates.

The final three chapters assess the policy implications of these observations and explanations. Chapter 10 aggregates the institutional considerations, setting forth the comparative institutional advantages of each system. Chapter 11 examines the most common policy prescription for improving standards-setting: changing administrative procedures. Chapter 12 considers interactive strategies and alternative policy instruments that are also worthy of consideration but are often overlooked in discussions of standards policy.

Ten
Comparative Institutional Advantages

Conclusions about specific safety standards are like snapshots—in order to be understood, they must be placed in context. The analysis thus far has been particularized and static, concentrating on the dynamics and outcomes of specific cases. This approach is suggestive but fragmentary. It points out specific differences between public and private approaches but says little about them as systems . This mirrors one of the main shortcomings with most discussions of "standards policy"—they amount to little more than empty calls for case-by-case decisionmaking. The problem is that policymakers have little understanding of the institutional differences between the public and private sectors. The goal of this chapter is to transform the observations of this study into some general conclusions about the comparative institutional advantages of public and private standards-setting. This discussion will be followed (in the final two chapters) by an analysis of specific proposals for exploiting and improving these institutional features.

There are no widely accepted criteria for comparing regulatory regimes. In order to provide a rounded view of the comparative advantages of public and private standards-setting, three approaches are taken below. The first examines the costs associated with decisionmaking, an approach favored by those who believe that more substantive measures are either impossible or simply not worthwhile. The second examines information inputs, an approach favored by those partial to rational decisionmaking models. The third is an amalgam of the re-

maining factors stressed by Lester Lave and others.^[1] It adopts an evolutionary perspective, comparing how the systems change and adapt over time.

Decisional Costs

One way to compare standards-setting systems is by the cost of the enterprise, what economists call the "transaction costs." Standards-setting takes time and money, and some systems are likely to run more smoothly and inexpensively than others. Of course, transaction costs say nothing about substance. Perhaps more expensive standards are better in other respects. Given the troubling uncertainties about the costs and benefits of safety regulation, however, it is often impossible to evaluate standards substantively. If having a standard, any standard, is preferable to not having one, then the system that produces standards most quickly and inexpensively should generally be favored. But the difference between public and private decisional costs, as explained below, is more complicated than is often imagined.

Government rulemaking is often criticized for being time-consuming and expensive. Government agencies can be notoriously slow. It took the CPSC two years to decide whether to grant the Banner petition and another three years to develop a relatively simple woodstove labeling rule. The gas space heater proceeding was no quicker. OSHA's grain elevator standard was almost ten years in the making. Only the FAA stands in notable exception, having adopted the aviation fire safety standard in less than a year—a dubious distinction, considering its content.

But the case studies also demonstrate that government has no monopoly on slowness. NFPA proceedings can drag on for years. "Decisionmaking by town meeting" is how one committee member describes the process. It took NFPA much longer than the FAA to revise its standard for aircraft fire extinguishers. The same problem apparently plagues ASTM, ASME, and similar organizations. And there is reason to think that they will become slower in the future. As measured by the number of appeals to the NFPA Standards Council and the ANSI Board of Standards Review, contentiousness is on the rise. Moreover, standards-setting is expensive, and groups such as NFPA do not always see a sufficient return on the cost (in staff time alone) of meeting frequently to revise them. Again, there is an important distinction between standards developed in conjunction with product certification and most

other private standards. Testing labs have a business interest in assuring that product standards are readily available when new products are submitted for approval. Standards developed for certification purposes are written in very short order, often in a matter of weeks or months. The resulting "desk standards" at UL have been criticized as violative of due process, but they are to be credited with remarkable timeliness. When private standards-setters lack this economic incentive, or when the cost of developing standards gets too high relative to the income generated by product testing, their efforts are likely to be as slow as those of government.^[2]

It is impossible to compare the total administrative cost of the two systems, because so many of the costs on the private side are decentralized. Practically everyone on the NFPA and AGA/ANSI committees is sponsored by his or her employer or trade association. Undoubtedly, the total of these costs is still lower on the private side because the total number of person-days involved in any single standard is so much less.^[3] UL keeps close track of the costs and revenues associated with its standards, but the organization is unwilling to reveal these figures. However, private standards written for certification purposes are apparently cheaper than government standards, at least from the point of view of the standards-setting organization.^[4] In short, the private sector appears to have the advantage in terms of money and, to a lesser degree, time—particularly in connection with product testing.

The verdict on adversariness is less clear. Although private standards-setting is often considered less adversarial than government regulation, the case studies suggest that adversariness may be more issue-dependent than sector-dependent. In the woodstove and aviation safety cases, there was no more adversariness on the government side than there was on the private. In the grain elevator and space heater cases, the antagonism that marked the government proceedings was also present in the private sector—it was just less open and publicized. There were strong disagreements within AGA about the space heater. Even stronger disagreements lurk behind NFPA 61B; the National Grain and Feed Association opposes practically everything about it. The observed tranquility in the grain elevator proceedings came at the expense of not addressing the most important safety issue: housekeeping. The private sector has one important advantage, however. The scope of private standards is so broad that there is no stigma attached to particular standards-setting activities. When the CPSC recently took an interest in metal chimneys, the industry felt it had been unfairly singled out. No-

body has ever felt that way about a UL standard. This is not to say that the private sector does not "miss" some issues. It does. Certain types of issues seem to fall through the cracks in private standards-setting, some warranting government attention. This is one of the ways (discussed in chapter 12) in which government standards-setting could profitably complement its private counterpart.

Institutional Knowledge

Harold Wilensky coined the term "organizational intelligence" to describe the institutional ability to process and utilize information.^[5] The concept seems particularly relevant to standards-setting, since the activity is so information-intensive. Three kinds of information contribute to "organizational intelligence" in setting safety standards: technical know-how, information on real-world experience, and applied research and development. The comparative strengths and weaknesses of the two standards-setting systems are striking. Private institutions are most intelligent in the first respect, public institutions in the second and third.

Technical Know-How

The most basic information utilized in standards-setting is know-how; that is, elementary knowledge about how a product or process works. The private sector is not only, as attorney David Swankin points out, "where the bodies are" (tens of thousands of people participate in private standards-setting); it is also where practical and technical knowledge often originates. Private standards-setters have a marked advantage in this regard over their public counterparts. They usually have a working knowledge of technical terms and basic engineering considerations, and they understand the practical implications of commercial use. One of the most active participants in revising NFPA 408 supervised Factory Mutual's study of hand-held fire extinguishers (conducted under contract for the FAA). Similarly, engineers for producers of space heaters and component parts participated on the Z21.11.2 subcommittee.

The personnel of public agencies, on the other hand, seldom have technical backgrounds or previous experience with the products or processes they regulate. Few are engineers. Many are lawyers or former compliance officers. In 1985, three of the CPSC commissioners were lawyers; none were engineers. A similar imbalance exists on the staff.

Neither regulators nor rulemaking staff members accumulate much technical experience over time, because public agencies do not specialize to the same extent as private standards-setters. Even at the FAA, which has a narrow mandate compared to most regulatory agencies, a rulemaking staff member might work on fire extinguishers one day, fire-blocking seat cushions the next, and tactile aisle markers thereafter. A staff member at the CPSC might work on woodstoves, gas furnaces, or a host of other "fire and thermal burn" hazards.

The result in all four public cases was ignorance and confusion over basic facts. Limited know-how led to longer and often more adversarial public proceedings. Woodstoves, described by a former CPSC commissioner as a "simple" issue, and airplane smoke detectors, described by Congressman Mineta as "straightforward," both involved issues that exceeded the technical capabilities of the respective regulatory agencies. The CPSC spent two years trying to decide whether to grant the Banner petition.

In the process of acquiring knowledge, government agencies often lose credibility and are put on the defensive. It took the CPSC years to differentiate in its injury statistics between the various types of space heaters; some say that many CPSC investigators still don't understand the difference between vented and unvented equipment.^[6] The CPSC staff also did not understand the special problems posed by fireplace inserts until very late in the proceedings on wood and coal-burning stoves. The commissioners had an even worse understanding of the technical issues. Analysts at the FAA lacked a basic understanding of the different types of Halon extinguishers and the special training needed for their use. OSHA's proposal for smaller grates on grain-loading pits was apparently born of ignorance about the operating effect of grate size on grain throughput and about the availability of other options for removing hazardous debris. Lacking basic knowledge, public standards-setters also tend to gloss over difficult technical questions, concentrating instead on more accessible, but less important, issues. The FAA plowed ahead in ignorance on smoke detectors. "They may be a good idea," noted an engineer at Factory Mutual, "but the feds are jumping in without the technical background to do the right thing. They were too slow to recognize the issue, and now they are moving too fast." They knew there were serious technical questions about smoke detectors and fire extinguishers, but they chose to ignore them. In the woodstove and gas space heater proceedings, the CPSC left the significant decisions to industry, concentrating instead on issues that

could not be resolved through technical knowledge—such as the wording of warning labels and the effective date of regulations—even though technical questions, particularly about creosote, seemed to have greater safety implications.

In short, how standards-setters approach the task appears to be partly a function of their technical knowledge and capabilities. Lacking specific expertise, government agencies try to avoid technical issues, concentrating on issues where the agency is at less of a disadvantage. When technical issues are unavoidable, however, limited technical knowledge tends to reduce the agency's credibility and to result in longer, more contentious proceedings.

Information about Real-World Experience

Obtaining feedback on real-world incidents is the second form of institutional knowledge essential to setting safety standards. Without information about the type and frequency of accidents, it is almost impossible to spot trends or even identify some hazard scenarios. As a UL vice president puts it, "The proof of the pudding is in the field evidence." The case studies suggest, however, that such evidence is rarely generated by the private sector. Government agencies, although far from ideal, have much better information systems than their private counterparts.

Private information sources are largely anecdotal and play only a negligible role in shaping safety standards. UL, for example, has institutionalized contacts with building inspectors in order to learn about problems "in the field." But a UL engineer acknowledges that the information is of limited use. UL also has a clipping service that collects newspaper stories on product-related injuries. The information is always sketchy and often inaccurate. How the injury actually came about is unlikely to be described in any detail; brand names are rarely mentioned, let alone model numbers; and a vented heater with clogged vents might be described as unvented. The AGA also has a national reporting system—dubbed the Gas Appliance Information Network (GAIN)—which relies on voluntary reporting from gas utilities. The system is much less impressive than its name. An earlier version was nicknamed N-FLOP by staff members at AGA.^[7] GAIN is also a flop. Reporting is scattered and, according to an AGA staff member, only one report was forwarded to a standards-writing committee in 1985.

NFPA makes a greater effort to collect injury information, but with only slightly more success. It, too, has a clipping service as a supplement

to reports received from local fire departments. This was the only available data base on grain elevator fires when the NAS began its study. Unfortunately, the information was of questionable reliability. NFPA also sponsors comprehensive investigations of major fire incidents, such as the MGM Grand Hotel fire and the Air Canada incident. These investigations can improve standards-setting by providing better information on specific hazard scenarios, particularly if they are not duplicative of government efforts (as they were with the Air Canada fire). Unfortunately, the investigative function of NFPA is generally limited and plays little role in standards-setting. The 408 committee drafted its requirements before the Air Canada fire, and the NFPA investigation did not result in any changes. Committee minutes do not indicate whether the 61B committee ever evaluated NFPA's survey of grain elevator explosions. Only anecdotal information was discussed during committee deliberations at the July 1985 meetings.

There are two reasons why the private sector collects so little useful injury information: information has the quality of a public good, and it often carries worrisome liability implications. Public goods have value to those beyond the immediate purchaser. National defense is the classic example of a good that benefits all, whether or not they pay for it. A similar phenomenon affects the collection of injury information. Many private organizations would benefit from reliable national information on consumer product injuries, but the cost of any given organization collecting such information is prohibitively high. Since private standards-setting organizations are decentralized, there is also no easy mechanism for spreading the cost to all those who would benefit from a national information system.

Injury information also carries threatening legal implications. Lawsuits are rarely discussed openly in standards-setting committees, although Eads and Reuter report that such discussion might occur off the record.^[8] Most firms keep records of consumer complaints involving allegations of injury, but it is unlikely they would share such information, given the adverse effect it could have in court. Even trade associations—according to staff members at the CPSC, the Outdoor Power Equipment Institute, and the Gas Appliance Manufacturers Association—generally do not receive this kind of information from their own members. NFPA's fire investigators are also inhibited by liability concerns, sometimes stopping short of certain conclusions or recommendations because the organization does not want to become embroiled in the litigation that inevitably follows the kinds of disasters they investigate.

Public information systems, though obviously flawed, are vastly better than private ones. The public sector does much more than clip newspapers. The CPSC has several information systems to provide feedback on consumer product safety. The agency collects injury data daily from hospital emergency rooms around the country and, through its field offices, conducts hundreds of in-depth investigations each year on selected hazards.^[9] The hospital data provide the basis for national injury estimates; the accident investigation reports provide details of specific hazard scenarios. Similarly, the FAA maintains an extensive computerized file of Service Difficulty Reports, and the NTSB investigates all serious airplane accidents. In both the grain elevator and aviation fire safety cases, the government paid particular attention to accident investigations. OSHA deferred drafting its grain elevator standard so that the NAS could collect explosion data and conduct in-depth studies. Many of the recommendations from the NAS study were incorporated into the OSHA proposal.

Applied Research and Development

Applied research is the third type of knowledge essential to a standards-setting system. It is the only method short of actual experience for determining whether new technologies are actually feasible and reasonably effective. Such questions were prominent in all four case studies. The oxygen depletion sensor had a track record in Europe, and the question was whether the device would be as reliable with American fuels.^[10] Similarly, questions were raised about the effectiveness of using household smoke detectors in airplanes. In the grain elevator case, there was significant controversy over whether pneumatic dust control could achieve airbone dust levels below the lower explosive limit. And with woodstoves there were a surprising number of technical questions, including the extent to which catalytic combustors could minimize creosote production.

Some standards-setters in both sectors have the in-house capability to conduct such research. The FAA Technical Center is well respected by industry. So is the National Bureau of Standards, which conducts applied research under statutory agreements with several agencies, including the CPSC. On the private side, UL and AGA conduct applied research both for in-house purposes and under contract. NFPA also supports a limited fire safety research effort.

The case studies suggest, however, that government does a better job of generating the kind of applied research that can inform standards-setting. The NBS conducted numerous helpful studies on wood-burning appliances. Its study on wall pass-through systems, a major source of fires related to woodstoves, filled a gaping hole in the private standard. Experiments conducted at the NBS also convinced the CPSC of the reliability of the oxygen depletion sensor—something that AGA Labs was reluctant to admit. The FAA funds extensive research into aviation safety, including the investigation of hand-held fire extinguishers, prompting the president of a major airline to declare at a recent Flight Safety Foundation conference that "airlines and manufacturers rely on the government to do development and testing."^[11] There were only scattered instances of private research efforts connected with the case studies. Two airlines and a major airframe manufacturer conducted tests on Halon fire extinguishers. The other major research effort, considered a public relations ploy by some, was undertaken by the National Grain and Feed Association. In the woodstove case, however, the lack of private research was notable. Catalytic combustors, a possible method of reducing creosote formation, were not (and have not) been considered by UL because, as a UL engineer said, "Nobody [in the industry] wanted to spend the money."

Research is expensive and private organizations generally do not do it unless someone else pays. Some private groups simply have no resources. "ASTM is just a building with rooms and secretaries," quips a former CPSC commissioner. It relies on its members to bring to the standards-setting process any information about relevant research. NFPA does much the same. Committees do not have budgets; nor does NFPA at large conduct applied research to support its standards-setting activities.

Product certifiers are also reluctant to undertake research aimed at improving standards. This research can have the quality of a public good. If UL improves a standard through applied research, other testing labs might be able to capitalize on the expenditure. Since UL must ultimately pay for such research efforts through its certification fees, the organization is not likely to conduct research in those areas, like woodstoves, where it shares the certification market with other labs. According to a UL spokesman, however, woodstoves are unusual in this respect. "We are often a monopoly for all practical purposes."

The funding of public research depends on the politics of the budgetary process, but there is certainly the potential (realized in several of

the cases) to fund projects not likely to be done privately. Many research budgets have been cut on the public side, particularly at the CPSC, but there are also indications that funding might actually increase for the FAA. The need for more public research and information collection is discussed in the final chapter.

The Evolutionary Perspective

More important than any single institutional feature is how an organization changes over time. Aaron Wildavsky has stressed the importance of "resilient" approaches to safety regulation.^[12] Others have emphasized the advantages of "flexibility" and "responsiveness."^[13] These concepts place standards-setting in an evolutionary context. What matters is not so much how individual cases are resolved as how results change over time. A regime that generates standards that are considered "too lenient" (at the time of adoption) might actually be more desirable than one that generates standards closer to the social optimum (at the time of adoption) depending on when the respective standards are adopted and how they are adjusted over time .

The case studies suggest that there are significant evolutionary differences between public and private standards-setters, differences that indicate several previously unrecognized advantages of the private sector. In short, private standards-setting is prospective and ongoing, while public efforts are usually corrective and singular. Private standards-setters tend to intervene relatively early in the life cycle of an issue, adjusting the standard subsequently over time. Public standards-setters, by contrast, are likely to get involved later, often after a major disaster, adopting a "one-shot" standard without the benefit of subsequent adjustments. The evolution of standards-setting systems can vary in two important ways: (1) in the timing of interventions and (2) in the nature of any subsequent adjustments.

The Timing of Regulatory Interventions

Turning first to intervention, the case studies suggest a major difference between the public and private sectors. The private sector appears to intervene relatively early in the life cycle of an issue, often in anticipation of problems rather than in direct response to them. Private standards often flow directly from engineering decisions. They are usually written in anticipation rather than in response to accidents. NFPA 408 and 61B were in place long before the first serious in-flight fire or the

well-publicized series of grain elevator explosions in 1977–78. UL and AGA initiate the development of standards as products are submitted for certification, which is almost always before they are marketed. The implications of this difference depend critically on how (and whether) these standards are adjusted over time.

In contrast, government interventions appear to occur relatively late in the life cycle of an issue and are usually reactive. Grain elevator safety was considered a problem by fire prevention engineers and insurance companies in the 1920s. OSHA did not get actively involved in the issue until after the disastrous Christmas of 1977. NFPA formally considered questions about aviation fire extinguishers more than ten years before the FAA first issued its advisory circular, and almost thirty years before it adopted a formal standard on the matter. Government standards also tend to respond to specific problems identified through real-world experience. For every serious airplane accident, there is a new FAA rule. This means that decisions are often made under pressure and with emotion, as in the months following the Air Canada fire. The pressure to "do something" becomes so overhwelming that valid concerns, such as whether household smoke detectors will work reliably on airplanes, too easily get brushed aside. Conversely, once "something" is done, there is little or no pressure to follow up on the action and reassess its wisdom in light of subsequent field experience.

Private standards-setters are also generally more comprehensive in defining "the problem," although the grain elevator case demonstrates that there are certain problems that only the government is likely to address. Private standards are more likely to be conceived as a total package, covering most aspects of the product or process from design and performance to labeling and production. In contrast, both of the CPSC standards addressed a single issue or single solution. So, too, with the FAA. Whether comprehensiveness is an advantage depends on the relationship between private and public standards. If the two were mutually exclusive, the private approach would probably be preferable. "It is generally better to regulate a range of problems satisfactorily than to do one in great detail," admitted a former CPSC commissioner who often favored government regulation.

Adapting Standards over Time

Over time, private standards-setting appears to be more flexible and adaptable than government regulation. Private standards are continu-

ally being revised. Many government standards are one-time interventions. Public standards rarely evolve the way private ones do. The "regulatory ratchet" characterizes some government standards, while the mechanisms for revising standards remain unused in others.^[14]

In none of the public sector case studies was standards-setting seen as an evolutionary process. There was little expectation that a standard, once adopted, would be altered in the foreseeable future. To the contrary, attaining closure was an important goal in its own right—one that took OSHA almost nine years in grain elevator safety. The FAA brushed aside technical questions about smoke detectors in order to avoid going through the requirements of "notice and comment" rulemaking all over again. In the gas space heater proceeding, the prospect of "gearing up" for a new rulemaking proceeding seemed so onerous to one CPSC commissioner that when the commission decided to revoke the space heater rule—in light of good evidence that it was no longer needed—she objected strenuously on the grounds that "restoration of the protection previously provided by the mandatory standard would require initiation anew of the entire time-consuming, resource-intensive process" (original emphasis).^[15]

Whether or not their perception of the rulemaking process is accurate, government agencies seem to operate on the principle that "it's now or never." As a result, public standards are often confined in scope, and they tend to stay fixed in their original form. The CPSC, for example, had reason to believe that gas space heaters might pose a chronic nitrogen dioxide hazard. But lacking an ongoing process for standards-writing, it saw the choice as either delaying the ODS rule (possibly for years) or not addressing this hazard at all. Revisions are always possible on the public side, but the process is ad hoc and seldom used. Several problems with the woodstove labeling rule have come to light since its adoption, but the idea of amending the rule has not been seriously considered. According to CPSC staff members, the effort required to enact the rule in the first place is the reason amendments are not seriously being contemplated. Instead, the agency has tried to amend the rule informally by working directly with the testing labs.^[16] Some of the advantages of a formal adjustment process might be captured through such informal efforts, but on the whole it seems likely that the public sector suffers significantly from the lack of revisions.

The revision process is institutionalized on the private side. NFPA standards are supposed to be reviewed and, if necessary, revised every five years. For product testing labs, the process is continual. There is

some merit in this attribute alone. The ongoing nature of private standards-setting may facilitate the resolution of conflicts by tilting the strategy of participating parties in the cooperative direction. Periodically revising standards is also the only way to keep up with changes in information, technology, and preferences. But whether and how standards-setters take advantage of this opportunity is of paramount importance. The evidence from the case studies is mixed. On the positive side, the "ratchet effect," described by Eugene Bardach and Robert Kagan in their study of government regulation, does not seem to characterize the private sector.^[17] Through the adjustment process, private standards are made stricter in some respects and more lenient in others. On the negative side, such flexibility can foster nonchalance. Private standards-writers are often willing, sometimes even anxious, to postpone addressing certain complex or controversial problems until the next time the standard is supposed to be revised. In 1980 the NFPA Agricultural Dusts committee decided that action on several controversial proposals should be postponed "for further study." These proposals were summarily dismissed at a meeting five years later.^[18] UL convinced members of the canvass on woodstoves to withdraw their opposition on the assurance, still unfulfilled, that the creosote problem would be handled soon. At AGA Labs, the surface temperature of space heaters has been a "continuing agenda item," without apparent effect, for over twenty years.

Comparative Summary

The difference between public and private standards-setting is not just a matter of degree, it is a matter of kind. Public efforts are not simply stricter or more lenient than their private counterparts. The systems are too different to be compared one-dimensionally (see table 8). There are important differences in how public and private standards-setters (1) resolve philosophical questions about the appropriate scope of safety regulation, (2) estimate costs and benefits, and (3) act over time and in the context of a larger regulatory program. These observations are troubling because they cast doubt on the reasons most commonly given in favor of one form of regulation over the other. Public regulation is not always stricter; private regulation is not always more reasonable. The performance of the two sectors is mixed, and there are different reasons to favor each.

The conventional wisdom about strictness and leniency is only par-

tially correct. Although public agencies appear to err systematically on the side of safety, their private counterparts do not always err in the other direction. Three of the four private cases studied are characterized by decisions that also err on the "safe side." At least one of these cases, the NFPA standard for aviation fire extinguishers, is unreasonably strict, providing reason to conclude that the private sector occasionally makes the same mistakes as the public sector. A rough estimate of overall economic efficiency suggests that, for the standards studies here, neither sector has a clear advantage. On the public side, two of the standards were unreasonably strict (the CPSC on woodstoves and the FAA on aviation fire safety), and two were within the "zone of reasonableness" (the CPSC on space heaters and OSHA on grain elevators). On the private side, two of the standards were also in the zone of reasonableness (UL and AGA), while the others were split: one too lenient (NFPA on grain elevators) and the other too strict (NFPA on aviation fire extinguishers).

But these conclusions capture only a portion of the interesting differences between the two sectors. Public and private standards-setters do not just select different outcomes. They have entirely different ways

of looking at problems. Public agencies are more paternalistic in defining "the problem" for regulation. They are also more willing than private standards-setters to select early deadlines, require unproven technologies, and regulate in a manner that interferes with traditional notions of managerial discretion.

Finally, aside from the specific standards, the two sectors vary over time and in the relationship between single standards and the standards-setting system. Sometimes private standards-writers do not have adequate information; they do not even know that a "problem" exists. At other times, they do not agree that "the problem" should be addressed. When changes are made, they are most likely the result of either government information or anecdotal evidence. Private standards are rarely unreasonable, however, in the sense of requiring something that is not generally feasible both technically and economically.

Public standards-writing, on the other hand, is reactive and rarely adaptive. Standards-writing is usually prompted by accidents or injuries. Standards are viewed as one-time corrections and not, with the exception of the FAA, as something likely to be amended in the future. Government standards are generally narrower than private ones. Although prompted by specific problems, it is questionable whether the resulting standard will address them effectively. Information about potential costs and benefits is assiduously compiled, but seems to play only an indirect role in decisionmaking. Technical issues are generally avoided in favor of "softer" issues, such as how strong a warning should be or when a regulation should take effect. But government is willing to do things the private sector will not—protect people against their own mistakes, "push" technology—and in some cases it is successful.

The remaining two chapters examine proposals for improving public and private standards-setting systems. Chapter 11 examines popular proposals for altering procedural requirements, particularly in the private sector. Chapter 12 examines more promising alternatives, including interactive strategies, niches for public standards, and methods of improving private standards-setting.