at present. It should be stressed, however, that whether difficult tradeoffs involving resources and health are dealt with explicitly and analytically, or are ignored, they are still made every time a decision is reached affecting human health. Where occupational health decisions. are concerned, we believe that a stronger emphasis on analytic approaches is needed.

Given the difficulties, both analytic and political, of valuing lives and health, we believe that the most promising approach for setting standards is to employ cost-effectiveness analysis. For each standard proposed, several different levels should be considered, including the possibility of no standard at all. Dollar costs should be estimated as accurately as possible for each possible level. The benefits associated with each level should also be assessed, measured perhaps in such terms as the number of disabling injuries or illnesses prevented, not in dollars. In many cases, the available data will permit only the crudest estimates encompassing wide ranges of uncertainty. Still, we would argue, if the data available are too weak to make even crude estimates of the benefits, they also are too weak to establish standards following any procedure.

Once cost and effectiveness estimates are assembled, at least rough answers can be provided to the question: As a standard is tightened progressively, how much does it cost per additional life saved (or illness/injury prevented) to achieve the higher level of protection? 193 This procedure would not by any means determine the actual level chosen. The very process, however, would force OSHA to examine the consequences of its standards more closely and would also provide useful information for interested parties, including other executive agencies and the Congress. Perhaps more importantly, it would highlight inconsistencies in different areas; it might show, for example, that at current levels of stringency one standard costs $5 million at the margin per expected life saved, while another could be tightened at a cost of only $5,000 per expected life, thereby yielding 1,000 times the OSH gain for its cost impositions. In such a case, by loosening the first standard and tightening the second, it would be possible both to increase longevity and to free resources for other uses.

If costs are an important consideration in setting standards, the level of protection afforded will vary significantly across and even within industries. In general, the lower the cost of compliance, the more stringent the standard should be. For example, it is obviously more expensive for a metal stamping firm to come into compliance with OSHA noise standards than it is for an electronics firm to do so. OSHA should consider setting standards at different levels for differ

183 Typically, as a standard is tightened the cost of achieving a fixed increment of health protection rises, sometimes quite rapidly. Consider a proposed standard that will prevent 100 injuries at a cost of $200,000. If OSHA were adopting policies that implicitly valued Injuries at $3,000, and if this level of standard were the only option, it would seem well justified. since it prevents an injury at an average cost of $2,000. But other levels may be feasible. Suppose that a slightly relaxed standard would cost only $150,000 and would prevent 95 injuries. In that case, the marginal cost of preventing the last five injuries is substantial. To be precise, it is ($200,000-$150.000)/100-95)= $10.000 per injury. If that $50.000 could be directed to more beneficial purposes elsewhere. including achieving more for the promotion of OSH, the slightly relaxed standard should be adopted.

The cost-effective setting of standards requires that the marginal cost of preventing an injury (or saving a life) by tightening a standard must be the same for all standards.

ent industries, or at least allowing a limited number of industries with very high costs relative to benefits to meet less stringent standards.194 The costs of compliance will also vary significantly from firm to firm within individual industries. Clearly it is impractical to set standards on a firm-by-firm basis, although flexibility in setting abatement periods for cited firms may achieve that goal partially. Significant cost savings could be achieved through the use of "grandfather clauses" in those cases where the costs of compliance are much higher with older plants or equipment than they are with new ones. Assurances that new standards would generally exempt existing plants and equipment would reduce the problems that firms now encounter in trying to anticipate future OSHA standards when undertaking major capital expenditures. For example, Harvard University is currently planning to construct some new laboratories. In addition to making sure that the laboratories will comply with existing OSHA requirements, administrators must also try to guess what additional standards will be imposed during the buildings' lifetimes.195

Relaxation of OSHA's hostility to the use of personal protective devices could also generate very significant resource savings with little or no sacrifice in worker health and safety. OSHA has correctly pointed out the danger that workers may not use their protective equipment. This argument lends indirect support for incentives-based approaches to OSH, for with them firms will have strong inducements to get workers to keep protective equipment in place.

Enforcement

Given the limited size of its compliance staff, OSHA is able to inspect only a tiny fraction of the workplaces under its jurisdiction each year. At current rates, the average employer will see a Federal OSHA inspector roughly twice a century. Thus the effectiveness of OSHA enforcement strategies will be a critical factor in determining its degree of success in improving OSH.

The threat of fines for noncompliance has been an empty one; the average fines are so low relative to the costs of compliance that for most firms delaying action until after receiving a citation is rational. One possibility would be to raise fines significantly in order to make delay less attractive. It would seem, however, with the current low probability of inspection, that fines would have to be increased many times over in order to obtain much impact. Current experience with serious violations suggests that very high fines may make inspectors less willing to cite violations. Large fines for relatively minor violations might appear capricious, further damaging OSHA's image, both with business and the public. Thus a major increase in the size of fines levied appears to be undesirable. Small fines have little incentive value, and are perceived by many businessmen as further evidence of OSHA's petty harassment. Indeed, the fiscal year 1977 Appropriations Act eliminates small fines in most cases. 196

194 In an interview with the authors, Dr. Corn suggested that OSHA take account of cost variations by allowing high-cost firms longer compliance periods. While we welcome this consideration of costs, we would prefer to see economic costs enter into the standard-setting process as well.

125 Personal communications from Harvard University administrators.

196 First-instance penalties for nonserious violations are not permitted unless there are 10 or more violations. Bolle memorandum.

OSHA's general worst-first philosophy for allocating inspections is a reasonable one, although the particular plan now in effect suffers several shortcomings, as discussed earlier. As more data become available, it should become possible to target inspections more accurately. For example, firms could be selected for inspection on the basis of their individual injury records, as reported to the BLS. The danger of such a policy, however, would be that it would provide an incentive for firms to underreport. Future statistical studies, similar to DiPietro's work summarized earlier, might show differences in the effectiveness of inspections across industries, thus providing another basis for setting priorities.

OSHA's new National Emphasis Program (NEP) for foundries, which is intended to serve as a model for similar efforts in other industries, appears to incorporate a number of attractive features: industry participation through the major trade association has been encouraged, apparently with some success; inspectors have received special training to enable them to detect the hazards likely to be found in foundries; and inspections will cover health as well as safety hazards. Perhaps most welcome, OSHA has announced its intention of incorporating a strong evaluative component in each of the NEP's. 197 If these inspections, involving a much greater input of resources than usual, cannot be shown to have a significant impact on occupational injuries and illnesses, it will provide strong support for the view that OSHA's approach to promoting OSH through the promulgation and enforcement of standards is misguided. Any evaluation of the NEP's should also include a careful accounting of the costs involved. Uninspected firms and industries

Under present OSHA procedures, and under any reform that seems likely, a vast proportion of American workplaces will go uninspected. Indeed, given the worst-first approach for general inspections, most firms that will not be inspected can predict that fact. If proper calculations are made, the deterrent effect of OSHA inspections and fines will vanish. Feelings of anxiety or annoyance about potential inspections should also dissipate.

A variety of responses to this situation are possible. We might propose that a small proportion of inspections be directed at firms or industries with strong safety records, where except for deterrent effects, inspections must be expected to have low productivity. But since the deterrent effects of potential fines are already minimal, it is hard to see how moving from a zero to, say, a .0001 probability of inspection would change matters. The major argument for maintaining some finite probability of inspection for every workplace would be its symbolic value: OSHA cares about all workers; no firm is immune to OSHA; your workplace may be inspected; no, OSHA is not picking just on the high-risk firms. Though we see the argument for appearances and equity, a more vital objective is that OSHA direct its resources in the most productive directions, and we would not urge that random, very low probability inspections be targeted on low priority firms. Should it be acknowledged that such inspections will not be made? Probably yes. The effects on deterrence will be trivial. More

197 The National Emphasis Program: A View from the Top," Job Safety and Health, 4 (April 1976), p. 35.

over, employers who know they will not be inspected should look on OSHA with diminished apprehension and animosity.

The basic problem remains. How do we feel when a regulatory effort with mandatory standards is only imposed on a small portion of the population? Four options are open:

(1) We could accept this as the natural outcome of a cost-effective application of OSHA's resources.

(2) OSHA could attempt to recruit sufficient inspectors to have some reasonable probability of inspecting all workplaces.

(3) We could dramatically raise fines so that even a low probability of inspection would generate some deterrent effect.

(4) We could find some other way of promoting OSH in low-priority firms.

Option 2 would be extraordinarily expensive, and would run directly counter to our proposal to channel OSHA to its most productive tasks. Option 3 seems unrealistic given the traditional relationship between fines and the magnitudes of offenses in the United States. Option 4 presents the only realistic alternative to option 1, the present system. Without detailed study, we would be loathe to propose any new regulatory effort to supplement a standards system for those firms that cannot be inspected. What follows is not a firm recommendation, but, we hope, a promising possible approach.

Inspection and compliance could be made the responsibility of the employer. He would submit a brief annual form declaring that he is in compliance. Some special procedures might be developed to secure worker input to the firm-certification procedure.

Clearly any system of this sort could turn into an administrative nightmare, with some 5 million establishments subject to OSHA jurisdiction. No more than say 20 or 30 standards should be listed on the form. They should be selected as a function of their level of health or safety threat (as documented by statistical studies), their expected cost of abatement, and their frequency of occurence. Different industries might have different forms. Clarity would be a critical element in presenting the standards.

Such a system might restore a deterrent where none now exists, promote some degree of equal treatment across workplaces, force OSHA to identify those violations which are of highest priority, and create the appearance of less harassment in the workplace. Falsification of information could be made subject to a new system of fine structures. It might also fall under existing statutes which impose criminal penalties for falsifying information on official Government documents. As part of any type of self-certification system, it would be necessary to work out abatement procedures for violations. Here too bureaucratic involvement should be minimized. Below some minimum number of violations, a firm might be given a fixed period, perhaps until the next regular filing, to abate a violation.

Compliance with a voluntary system of this sort might be much higher than with the present low-probability-of-inspection system; it could hardly be lower. Employers who persist in violations when the only threat is a small fine might be most hesitant to lie on Government forms bearing their signatures, particularly with criminal penalties as a threat. Obviously, there still would have to be some low probability of inspection to determine whether the attested levels of compliance were in fact met.

41-907 079-16

Provision of information, consultation

In the promotion of occupational safety and health, the clearest and best justified role for the Government is as a provider of information. To a limited extent, OSHA and its accompanying agencies have played this role. They have set standards in a great variety of areas. In theory at least, they should have set on the basis of careful evaluation of the data available.

OSHA's major shortcoming on information provision, we would argue, has been its basic strategy of standards, inspections, and fines. It is perceived, particularly by industry, as a policing agency, not a consulting agency. At present, Federal OSHA inspectors are not permitted to engage in onsite consultation. This prohibition may help maintain the legitimacy and deterrent capability of inspections, and may prevent unmanageable demands on manpower. However, for each of four reasons, we think it should be reviewed:

(1) It enhances rather than plays down the adversary relationship between the Government and industry in OSH promotion.

(2) It prevents the Government from providing some quite useful inputs.

(3) Less helpful offsite consultation, including such procedures as the review of blueprints, is permitted already.

(4) There is a consistency and equity issue since most State plans allow for onsite consultations.

In some nonplan States, the Federal Government has contracted with State officials to provide onsite consultations. Quite simply, the Federal Government has a clear responsibility to provide information on ways of promoting OSH; onsite consultation seems a good way. If consultation services turn out to be expensive, a marginal cost charge would be appropriate.

The Federal Government can provide information that will inform bargaining and discussion relevant to OSH between workers and employers. Whatever its regulatory format, the Government should collect and disseminate information on the types of conditions that are most threatening to safety and health. At the very least, this could lead to improvements in OSH in the vast majority of the workplaces where inspectors have never ventured, and where the deterrent effects of OSHA are minimal. For large workplaces, more likely unionized, this information could provide a helpful input to collective bargaining on work practices and working conditions. This could lead to a push to exceed OSHA specifications in some areas, and possibly worker as well as employer pressure to lower unrealistic standards.

Some observers have proposed that where an established group of workers' representatives exists, and where that group, say a national union, has sufficient resources to evaluate information on OSH, safety and health determination should be left to the bargaining process between management and workers. 198 If this aproach were followed, even to a limited extent, the Government could play a useful role by pro

18 John Dunlop. former Secretary of Labor, argues in "The Limits of Legal Compulsion." unpublished paper, Nov. 11, 1975, that "the actions of various regulatory agencies need to be brought into greater harmony with collective bargaining." He suggests that OSHA seek to encourage bargaining between labor and management over OSH conditions, possibly following the general pattern set by the Labor Department when it fulfilled its labor-protection responsibilities under the Urban Mass Transit Act by bringing together union and transit representatives to negotiate 3-year labor-protection agreements when Federal funds were used to take over failing private transit systems.