Neither of these policies contains much real guidance for deciding when private standards are "adequate," and neither speaks to the institutional differences between public and private approaches. Instead, agencies are left to devise their own ad hoc policies. Most seem uninterested in developing any kind of general standards policy. A report on the implementation of the OMB circular notes the sentiment that many agencies "have not yet heard of the circular, or not done anything about it, or don't really care."^[6] The Interagency Committee on Standards Policy, charged with implementing the circular, virtually disbanded in 1987 owing to a lack of interest among participating agencies.

In lieu of a general policy, the most popular way of evaluating the adequacy of private standards, when agencies have to decide, is to scrutinize compliance rates. Standards that do not engender "high enough" compliance are considered suspect. This perspective provides two omnipresent justifications for public regulation. First, public standards have the force of law. Unlike "voluntary" standards, they command the authority of the state. Second, at the federal level agencies can enforce standards nationwide. This is an important advantage if problems are national in scope or if states might otherwise engage in undesirable competition to lower regulatory requirements. Business interests sometimes support federal regulation as a method of eliminating conflicting state or local requirements, which are often taken directly from private standards.

The perceived advantages of the mandatory and national nature of certain government standards, more than anything specific about their content, underpin many public policies concerning private standards. In neither of the consumer product cases was the CPSC trying to improve the content of private standards so much as the enforcement. The real hope for the woodstove labeling rule was that it would increase certification to UL 1482. The CPSC standard depended heavily on UL for its content. The oxygen depletion sensor was already an "optional requirement" in Z21.11.2. The CPSC simply sought to make it mandatory, again leaving technical specifications and details to the private sector. These do not seem to be isolated examples of the government seeking to combine its enforcement power with the standards written by the private sector. HUD turned NFPA's standard for mobile homes into a federal regulation (much to the dismay of NFPA), and OSHA did likewise with hundreds of ANSI standards shortly after the agency was created. This enforcement-based strategy is popular because the public agency has a standard to work with, rather than starting from scratch. The government acts more as an editor than as an author, taking existing provisions and (sometimes) tinkering with them. Starting with a private standard is not only more convenient for government, it also provides the built-in credibility that distinguishes first guesses from second ones. The importance of such credibility cannot be understated in the case of "trans-scientific" issues.^[7] What a government agency may not be able to justify alone can attain instant credibility with the right private sector credentials. UL decisions are practically accepted as gospel. How many provisions of the Uniform Building Code, for example, are ever challenged on safety grounds?

Although persuasive in some contexts, the enforcement rationale does not always provide a sound reason for government intervention. First, the cases demonstrate that the marginal benefits of making private standards mandatory are often overestimated. Public standards are only as good as the incremental benefit they add to existing private ones. The size of this increment is no problem if the private standard generates very few benefits (as in the case of NFPA 61B) and the government version is both stronger and better enforced. More commonly, however, compliance with so-called voluntary standards is practically mandated through a combination of local codes and other incentives. Before government took any action, at least 50 percent of the airlines complied with the FAA's advisory circular, upward of 85 percent of the woodstove market was in compliance with UL 1482, and close to 100 percent of the unvented gas space heater market was in compliance with the "requirements" in Z21.11.2. In the latter two cases, the marginal benefit of perfect compliance would be minuscule. Both the FAA and the CPSC made the mistake of calculating the benefits of their own proposals without subtracting the existing benefits of private standards. In the case of woodstoves, the marginal benefit of the CPSC standard was very close to zero. Ironically, the CPSC justified its standard under a

statute intended to force deference to the private sector "whenever appropriate." Beneath the irony, these cases reveal a major shortcoming in basing standards policy on compliance rates.

A second problem with current efforts is that nationwide standards are not always better than regional ones. In the case of gas space heaters, some states and localities wanted to take a stricter approach than the CPSC by banning the product altogether. Their petitions for exemption from preemption were the undoing of the CPSC standard. The topic of regulatory federalism is just emerging, and there is no comprehensive theory about when the federal government should preempt state efforts.^[8] But the CPSC's experience in both cases suggests federal intervention was the wrong approach.

Given these problems, it would probably be best if government minimized the use of the "enforcement rationale." What is needed instead is a standards policy that builds on the comparative advantages of public standards-setting. Two major components of such a policy are discussed below: (1) identifying uniquely public objectives and policy instruments and (2) defining niches for public standards that complement private ones.

Defining a Role for Public Standards

Promoting "Public" Values

All the attention paid to enforcement and compliance issues obscures a more basic reason for government to intervene in the world of private standards. There are some things the private sector simply does not want to do. These lines between public and private are not readily obvious, but their contours are suggested by the regulatory philosophies sketched in chapter 7. Certain approaches to safety regulation are favored almost exclusively by the public sector. The government may call them protective; the private sector considers them overzealous or paternalistic. In either case, government should not waste time waiting for private standards-setters to implement this part of the public agenda. Rather, public agencies should identify and pursue the strategies and goals that set them apart from the private sector. Paternalistic regulation and "technology-forcing" stand out from the case studies. Concerns about consumer misuse distinguish the CPSC from both UL and AGA Labs. A willingness to engage in technology-forcing distinguishes the public standards for space heaters and aircraft smoke detectors from the private ones. And both NFPA standards underscore the reluctance of the private sector to regulate operating procedures, maintenance schedules, training routines, and other "managerial" issues.

Government should proceed with caution, however, since the case studies also suggest that some strategies disfavored by the private sector do not always work well in practice. Paternalistic regulation can be self-defeating or ineffective. To have been worth the effort, the CPSC woodstove rule would have had to have been dramatically more effective than any previously studied information strategy. Similarly, the desirability of operational controls depends on the quality of enforcement. It remains to be seen whether OSHA inspectors will apply the "action level" for housekeeping in a flexible but firm fashion. The agency certainly has a reputation for unreasonable enforcement.

The verdict on technology-forcing is at least partially in the government's favor. The CPSC apparently advanced the introduction of safer technology for gas space heaters. A trade association representative familiar with gas appliances admits that there is no evidence of the technological problems industry predicted while refusing to adopt the "unproven" technology. The jury is still out on smoke detectors for the passenger compartments of commercial aircraft. Whether the FAA rule will spur the development of better detectors remains to be seen. The FAA rule was written in ignorance of several matters of importance, however, and the gas space heater was successful primarily because the CPSC had the support of the National Bureau of Standards. This suggests that technology-forcing should only be undertaken when an agency has strong technical support.

Whether government should engage in either of these strategies is ultimately a question of values. Whatever the answer, it would be an improvement if these matters were highlighted and debated rather than concentration being put on contrived issues such as those raised in the woodstove proceedings.^[9] The CPSC has never given any real content to its mandate to protect against "unreasonable risk of injury." As with most administrative agencies with broad mandates, the agency has left its regulatory mission rather vague. The issue of public policy toward private standards provides an ideal opportunity for facing these value questions directly.

Niches for Public Standards

An interactive standards policy should also be based on identifying niches for public standards. This strategy targets issues or areas that private standards do not cover. Some oppose this effort at interactive regulation, fearing that public standards will "drive out" private ones. That has happened several times on a programmatic level. The emergence of federal automobile safety standards displaced the safety program of the Society of Automotive Engineers. (The SAE still writes test methods that are referred to in government standards, however.) Similarly, the creation of the Civil Aeronautics Board ended UL's involvement in airplane certification. But the argument does not provide a valid reason for avoiding overlap in individual standards. Public and private standards-setters can and do operate in the same policy space. They sometimes even cooperate. UL works with the Food and Drug Administration in regulating microwave ovens and with the Coast Guard in regulating marine safety equipment. But even when the two sectors compete—attempting, as they did in all four case studies, to regulate the same subject independently—the efforts of one sector do not necessarily detract from those of the other. In none of the cases did the public standard "drive out" its private counterpart, although both UL and AGA Labs argued that they would. Identifying areas where public standards would best complement private ones is difficult because there are so many private standards. Nevertheless, the case studies suggest three possible niches.

Orphan Standards . One promising niche for public standards is similar to that of "orphan drugs"—efficacious drugs that private companies decline to produce.^[10] The most common examples of orphan drugs are treatments for rare diseases, where the number afflicted (the potential market) is too small to justify the expense of producing the drug. A similar condition obtains with safety standards: market forces limit the development of possibly useful private standards. The case studies suggest two types of market failure in the market for safety standards. First, as with orphan drugs, some standards are not written because the development costs are too high, given the anticipated income. There is such a limited market for, say, aviation fire extinguishers that the cost of developing special requirements for the aircraft environment are not worthwhile. Yet a standard of this type would probably do more to improve aviation fire safety than the specific actions taken by either the FAA or the NFPA. UL officials are quick to point out that some of its standards are "charitable." The standard for septic tanks, for example, apparently generates far less income than it costs. Yet there are undoubtedly other areas where standards would help promote the public interest but do not currently exist for economic reasons. Second, there is a related type of market failure involving chronic hazards. The cost of developing standards to control truly longterm hazards is simply prohibitive. The private sector readily admits its failure to address such safety problems. Engineers at AGA Labs agree that the hazards of long-term exposure to combustants from space heaters should be studied, and possibly regulated, by the government. An enlightened government standards policy would seek to identify and address these gaps in private standards.

Comparative Standards and Information . Another possible niche for public action involves comparative standards and information. Economists argue that optimal levels of safety can be achieved when consumers select their own levels of risk.^[11] That only works, however, if comparative safety information is available. A wider range of choices will also be available to the public if graded standards are favored over "pass-fail" ones. An assistant attorney general in the Carter administration cautioned: "A single quality level standard if set too high may unjustifiably exclude substitute products from the marketplace. Pass-fail standards may also blunt the incentive for further product development and innovation."^[12]

Standards that facilitate safety comparisons are also desirable when information on causal links is weak. For example, there is significant uncertainty about the connection between high-absorbency tampons and toxic shock syndrome. Given this uncertainty, a standard that allows consumers to compare absorbency rates and apply their own notion of risk aversion seems preferable to one that imposes a single, arguably unjustifiable, level of risk.

Private action of this type is highly unlikely given the competitive implications of comparative safety standards. Private standards are built on common interests. Comparative standards are antithetical to these common interests. UL assiduously avoids rating safety. Instead it maintains the myth that products fall neatly into two categories, safe and unsafe. After extensive negotiations, the major tampon manufacturers were unable to agree on a test method for rating absorbency. Each method was considered favorable to a particular brand, raising the ire of various competitors.^[13] In sum, the private sector is generally unwilling to develop standards that distinguish levels of performance. To the extent that this strategy has a place in safety regulation, it is located in the public sector. Several comparative standards have been adopted by the federal government in the automotive field: octane ratings, fuel economy ratings, crash ratings, and tire treadwear ratings. Not all these standards have been successful, but this type of effort is worth considering and probably has a place in any enlightened standards policy.

Integration and Coordination . Finally, there is a role for public standards to cut across private regulatory boundaries. The narrow focus of private standards-setting tends to eliminate safety judgments across products or programs. Neither AGA Labs nor UL, for example, considers the relative safety of gas appliances versus electric appliances. AGA concentrates on the former, UL on the latter. There is no equivalent to the OMB in private standards-setting, trying to ensure that the next regulatory efforts are undertaken where they will do the most good. The government could provide a valuable service in some areas by considering together the universe of products or processes regulated separately by the private sector. An overall policy of, say, banning certain propane appliances in favor of natural gas and electricity might create more benefits than individual standards aimed at making all three somewhat safer.

The case studies also include examples of assorted "holes" in private standards that could possibly be filled by public action. For years, NFPA 211, the installation code for chimneys, fireplaces, and woodstoves, required certified wall pass-through systems, even though none were actually available. A similar situation obtained with spark arresters and fire-resistant floor coverings. NFPA standards mandated a certified product that did not exist. Installation codes can also fall out of touch with reality. NFPA 211 calls for chimney specifications that few contractors ever follow. The portions of the National Electric Code incorporated into NFPA 61B are not really based on the conditions encountered in grain elevators. So, too, NFPA 408 is built on a rating system for fire extinguishers that has nothing to do with the aviation environment. The causes and possible remedies for these dysfunctional aspects of private standards are not clear. Private standards-setters admit that the relationship between product standards and installation codes is arcane and puzzling. Public scrutiny of the connections between private standards might provide insights and improvements.

Alternative Policy Instruments and Institutions

Developing public standards is not the only way for government to seek what safety standards have to offer. Government can also attempt to influence the development of private standards. Since private standards cover literally thousands of subjects unlikely to be regulated by government in the foreseeable future, this strategy may have a much greater effect on public safety than government could ever hope to achieve through the small number of standards that survive the political process and the courts. Unfortunately, standards policy has been almost silent on other methods of improving private standards-setting.

Safety standards interact with a host of other policy instruments and influences. Harter and Eads observe that liability law, workers compensation law, information, and wage differentials are among the array of instruments that affect workers' health and safety.^[14] However these instruments and institutions are evaluated on their own account, they affect standards-setting and hold the potential for improving it indirectly. Harter and Eads emphasize the importance of taking these instruments and institutions into account when analyzing a specific policy space: "Policy instruments coexist and interact. Modifying one does not represent the adjustment of an isolated instrument, but a shift in the constellation of instruments and institutions affecting private behavior."^[15]

It is beyond the scope of this study to examine the role of these "external" forces independent of their effect on safety standards, although they should certainly be considered in any analysis of specific standards. OMB raised such issues in the grain elevator proceedings, appropriately questioning whether the force of liability law and insurance provided sufficient incentives for safety. But, as Harter and Eads suggest, changing these policy instruments will probably alter the standards-setting system. These need not be mere side effects. They might form the basis for useful, purposeful changes in standards-setting. The case studies suggest many important linkages between these policy instruments and standards-setting. Several policy instruments and institutions whose relevance to standards-setting was suggested by the case studies are discussed below.

Reforming Liability Law

Liability law is probably the most widely recognized policy instrument that affects private standards-setting. It was an important influence in several of the case studies. The debate over "tort reform" encapsulates the familiar arguments about whether the direct effects of liability law hamper public safety objectives.^[16] The case studies add to that debate by suggesting that liability law has undesirable effects on standardssetting. At least two reforms seem in order. Both would help ensure that improvements in standards are encouraged, rather than discouraged, by the law. These two changes would probably have a larger impact on private standards-setting than any aspect of the various "standards policies" offered in the past ten years.

First, improvements made in a standard should not be admissible in cases concerning mishaps prior to the improvement . Although there is some support for this position in the law, "the growing tendency," according to UL's general counsel, "is to admit into evidence post-accident revisions to standards."^[17] This can only discourage improvements in standards. UL would be more willing to upgrade its standard for metal chimneys if doing so did not create the implication that thousands of existing chimneys are, by UL's apparent admission, inadequate. Similar concerns were apparent when the NFPA's Agricultural Dusts committee added provisions for motion-detection devices to the appendix of the standard in 1973. These provisions might have been made mandatory but for the implication that facilities without such devices were below the generally accepted industry practice. Protecting standards-setters and businesses against liability for making improvements would certainly advance public safety objectives.

Second, in assessing the reasonableness of the risk attendant to a product or process, the law should consider overall effects, not just specific incidents. In other words, the law should not impose liability when overall social benefits clearly outweigh specific adverse effects. This argument has been advanced in connection with innovation by individual firms such as drug companies.^[18] There are implications for standards-setters as well. The most persuasive reason why AGA did not require the oxygen depletion sensor on space heaters without prodding by the CPSC—a reason never committed to paper in either the public or private proceedings—is that companies were afraid that an occasional failure of the device would create new liabilities. Similar fears have slowed the introduction of other technologies, from vaccines to antilock brakes, that would unequivocally increase overall levels of safety. The perverse result is that standards-setters rationally avoid some actions that would make the world safer because, unfortunately, these actions might also increase exposure to liability.

Loosening Antitrust Law

Though not thought of as an instrument of public safety, antitrust law is often considered a proper tool for controlling private standards setting. Strengthening the influence of antitrust law is always a popular proposal. (Who wants to defend "trusts" or "combinations"?) Such proposals are particularly popular because more substantive suggestions for controlling private standards-setting are lacking. In the area of standards and certification, the FTC houses some of the strongest advocates of increased antitrust scrutiny. These staff attorneys place their hope in stricter antitrust enforcement largely because the proposal to regulate standards-setting by rule was abandoned by the commission under President Reagan.

There are two reasons for restraining this enthusiasm for controlling private standards-setting through antitrust law. First, reducing one type of error often increases another type. Stricter antitrust enforcement would probably eliminate some of the worst standards currently in use, but it would probably also eliminate some of the good ones. This study suggests that producers are increasingly crying "antitrust" to intimidate standards-setters from taking socially desirable actions. Stories about antitrust law inhibiting otherwise desirable actions are almost as common in the private sector as those about liability law. Certain metal chimney producers raised this argument with UL and were effective in keeping UL from upgrading the standard, even though research results from NBS support such a move. Antitrust considerations were also instrumental in keeping AGA Labs from adopting ODS technology that was produced by only one company. Increasing the strength of antitrust law may well decrease the propensity to upgrade private standards.

The second reason for restraint is that antitrust law threatens to heap undesirable administrative costs on the private sector. That is practically the intention of those pushing the due process argument. As two members of the Supreme Court realized in the Indian Head case, "insisting that organizations like NFPA conduct themselves like courts of law will have perverse effects."^[19] One effect is that due process itself can be exploited for anticompetitive reasons. The number of appeals to NFPA's Standards Council has increased to the point where it threatens to overburden the system. Higher administrative costs are likely to result in fewer standards and less frequent updating. The antitrust influence also threatens to burden the private sector with perhaps the worst aspect of public regulation: the judicial second-guessing that follows on the heels on most standards. The current level of discussion about antitrust and standards-setting is uninspired, relying on contrived distinctions that classify NFPA as "commercial" and the CPSC as "political."^[20] At the very least, antitrust doctrine should be reexamined i n a manner that recognizes the relevant institutional similarities and differences between public and private standards-setting. It should also be recognized that proposals to strengthen antitrust enforcement might actually be counterproductive, eliminating more desirable standards than undesirable ones. As Maitland argues, there are good reasons to permit more collective action by business organizations.^[21] Whether a loosening of sorts is possible without allowing an undue amount of undesirable activity is worth examining.

Public Information Systems

Perhaps the most promising possibility for government action suggested by this study is an information strategy. The government has a clear comparative advantage in generating two types of information particularly important to setting safety standards: applied research and feedback on real-world experience. The private sector is lacking in its capacity to generate such information. Government overcomes the "free rider" problem that plagues private efforts to produce so-called public goods. Government is also the only avenue for collecting vital information such as medical records that are otherwise protected by privacy laws.

The potential for influencing private standards-setting with information about real-world experience is significant. The private sector has demonstrated its willingness to alter standards in light of relevant information. A former CPSC voluntary standards coordinator cites numerous examples where the private sector responded to public information. Liability law also helps provide an incentive for the private sector to respond to injury information. Unfortunately, it also provides a countervailing incentive against collecting it. What you know can hurt you. In any case, the incentive to respond depends on the quality of information. Industry feels no need to respond to many of the existing CPSC data because they are so clearly inadequate.

Perhaps the most influential government organization involved in collecting information is the National Transportation Safety Board. The NTSB has no decisionmaking power, but its recommendations carry considerable weight. The NTSB was split off from the FAA "so that the same person  was not both promulgating civil air regulations and investigating the accidents they might cause."^[22] Avoiding this "conflict of interest," whether real or imaginary, seems to have bolstered the NTSB's clout. The FAA adopts many NTSB recommendations in short order. Others are eventually adopted under pressure from members of Congress who use these authoritative recommendations to register support for the popular cause of aviation safety. "Without the NTSB," muses a lobbyist for the Association of Flight Attendants, "the FAA could turn all of these issues into technical mush." Devoting a specific organization to the collection and analysis of injury data would probably improve standards-setting in other areas as well.

Government could also exert a more positive influence on private standards-setting by doing more applied research. Most private standards-setting organizations would gladly replace various "engineering judgments" with decisions based more on science—so long as someone else shoulders the research costs. Voluntary associations such as NFPA do almost no applied research; they do not have the resources. Testing labs occasionally conduct such research, but seldom in connection with a single standard. Perhaps the best policy for government is to increase funding for the National Bureau of Standards. Not only does NBS have the necessary technical capability and reputation to conduct useful research and development, but it offers the right institutional setting for developing an intelligent research strategy. The cases demonstrate that prolonged NBS involvement can help improve private standards. The agency combined its technical knowledge and reputation to bring about beneficial changes in NFPA 211 and ANSI/AGA Z21.11.2. "Engineering needs a loyal opposition," argues Edwin Layton in a prominent book on the profession.^[23] Engineers face issues that are technical and political, in an organizational context that profoundly shapes personal incentives. They can best be kept in check by other engineers with the technical knowledge and institutional independence to challenge them. The National Bureau of Standards (now the National Institute of Standards and Technology) fits the bill, although its budget is still minuscule compared to its potential agenda.^[24]

Engineering Schools, Insurance Companies, and Other Influential Institutions

There are a host of institutions at the periphery of this study that obviously play an influential role in the achievement of public safety objectives. These institutions interact with private standards-setting in ways that are not well understood and deserve more attention. The rest of this chapter contains speculation about the policy implications of the influence of three "external" institutions on private standards-setting.

The discussion is organized in decreasing order of speculativeness. That is, the institutions most peripheral to the case studies, about which only the most tentative statements are possible, are discussed first.

Engineering Schools . One important institutional influence on standards-setting comes through engineering education. The influence is diffuse and difficult to measure but nevertheless important in shaping the ethics that predominate in the private sector. Engineering education has been criticized for being too narrow in focus.^[25] This study does not examine the details of engineering education, but it supports the conclusion that revisions might improve private standards-setting. A greater educational emphasis on economic trade-offs and human factors might improve the reasonableness of standards in two ways, respectively making them less demanding when the benefits are low and more demanding when human factors account for significant injuries.

Building Code Organizations . By design, this study also concentrates on "paired" public and private cases. In reality, however, private standards typically come in connected pairs or interlocking webs. The tendency of the private sector to separate product standards from installation and use codes is exemplified by the division of labor between UL and NFPA. The symbiotic relationship between these organizations is somewhat mysterious, and at times problematic. Matters are complicated further by the process through which these standards most often become law: municipal building codes. Building codes draw on the efforts of NFPA and UL, as well as those of various regional building code conferences—organizations that draft model building codes. The importance of the linkages between all these standards is clear. How to improve building code enforcement and achieve better coordination between installation codes and product standards is not. Perhaps the recently abandoned federal housing code deserves reconsideration. Possibly the government should institute a mechanism for bringing together those who write installation codes and product standards. Without improvements in the enforcement of building codes or the integration of design and use standards, it appears likely that some improvements in private standards will ultimately prove ineffective.

Insurance Companies . Insurance companies are also key actors affecting public safety. Not only do they spread risk, but, through insurance underwriting, they determine which risks are insurable and at what rate. If premiums accurately reflect risk, then insurance can help internalize externalities and improve public safety. Several major standards-setting organizations, including UL and NFPA, trace their origins to the insurance industry. Safety standards help insurers in two ways: (1) they provide convenient underwriting criteria, and (2) they promote general loss control. But what about the reverse influence—that is, how insurance companies affect standards-setting? Two possible influences are indicated by the case studies. Both seem rather weak, however, suggesting the potential for improvement.

The first way insurers affect standards-setting is with information. Insurance companies identify problems and manage information. In theory, they should be able to provide standards-setters with helpful data on the size and nature of various hazards because insurers pool information on claims through institutions such as the Insurance Services Organization. In practice, however, insurers apparently contribute little useful information to standards-setting efforts. An insurance company representative testifying in favor of the CPSC woodstove rule was embarrassed by questions about the actual magnitude of the problem. In an analysis of residential fire insurance, David Hemenway concludes that "the insurance industry has not been especially helpful in providing useful [loss control] data." He suggests several policies that might improve the situation, but all require further study."^[26]

Second, as one of the primary users of safety standards, insurers are potential watchdogs for the quality of these standards. Unlike many participants in standards-setting whose interests are vested in particular products, insurance companies have a purer motive: reducing losses. Insurers certainly make use of these standards. Product liability insurers require compliance with UL and AGA standards. Some grain elevator insurers require compliance with NFPA 61B. But the case studies suggest that insurers are not very demanding users of private standards-setters. They are seldom involved enough in the details of particular hazards to know a good standard when they see one. Insurers play almost no role in the NFPA committees on aviation fire safety, even though individual fires can be catastrophic. The grain elevator standard is sadly lacking in critical areas of grain elevator safety, but reinsurers are apparently unaware that more effective standards for safety are possible. Moreover, in neither the grain elevator case nor the woodstove case does it appear that the premiums charged by insurance companies reflect differences in risk. Airline  

officials also contend that their premiums have nothing to do with their loss control efforts.

This study suggests that insurance companies could improve the quality of private standards if they were more demanding. A small reinsurance company was apparently instrumental in getting UL to change its requirements for metal chimneys. But the low priority that many insurance companies seem to place on loss control has been noted by Hemenway.^[27] How to maximize the potential of insurance as a regulator of risk has long been a puzzle in the area of workers compensation. Somehow, the critical role of insurers as assessors and regulators of risk should be more widely recognized so that these issues are addressed in various areas of public safety.

Conclusions

Both public and private standards-setting are more complicated than is often imagined. The popular characterization of the public sector as slow and rigid is sometimes true, but there are notable exceptions. The FAA moved quickly in adopting standards for fire extinguishers and smoke detectors; OSHA adopted flexible and reasonable standards for grain elevator safety. Similarly, private standards are not always weak and "watered down." Three of the four private standards in this study provided meaningful, if not stringent, safety regulation.

The relevant questions for public policy are also more complicated than is generally recognized. The choice is rarely between public and private standards. Rather, the challenge is recognizing ways in which the two sectors can interact beneficially. This study identifies the comparative institutional advantages of public and private standards-setting and suggests several ways to take advantage of them. One approach is to emphasize "public" values. In other words, the public sector should emphasize certain strategies shunned by the private sector. Second, government should identify niches where public standards are likely to complement private ones. A particularly promising strategy is filling "holes" in private standards. Finally, the importance of alternative policy instruments must not be overlooked. Standards-setting in both sectors is affected by a variety of "external" factors that are subject to influence by government action. Several of these factors hold the promise of improving standards-setting in ways that could never be mandated directly. Altering the product liability law and improving the education of engineers could both lead to significant improvements in standards-setting. Recognizing these subtle influences and pursuing strategies that build on the complex interaction of public and private safety standards is the key to an intelligent and productive standards policy.