the same resource expenditures if we enforced standard A more vigorously and abandoned standard B. All parties could be made better off if regulatory efforts were adjusted so that low incremental cost means of increasing OSH levels were pursued more vigorously and high cost measures relaxed, so that the marginal costs of promoting OSH were made constant.

This argument is less compelling, although only slightly so, if we look across industries or classes of workers. Most would agree that we should not spend $5 million to save one life, say through safety railings, while sacrificing another, perhaps through exposure to cotton dust, to save $50,000. We should formulate a regulatory system that attempts to achieve the greatest OSH gain for the levels of expenditure it entails, what economists would call a cost-effective system.

Significant gains could be achieved if OSH-promoting efforts were rationalized so that resources were directed where they were most productive. In the hypothetical example above, for instance, 99 additional lives would be saved if $5 million were redirected from safety railing expenditures to efforts to reduce cotton dust exposure. If this process of resource shifting continued, always seeking the area where the greatest OSH gain could be achieved for a given level of expense, a cost-effective outcome would be reached. One of its significant characteristics would be that the cost of securing additional OSH would be the same in all areas of expenditure. This condition, familiar from economics, is that for an efficient outcome the marginal cost of achieving a particular benefit must be everywhere the same.

We conjecture that the gains in ŎSH would be much greater if we were to rationalize our present expenditures to be cost-effective than if we were to increase all present areas of expenditure by say 10 percent. The pursuit of a cost-effective outcome would yield an additional benefit: It would automatically generate information on how much additional increases in the level of OSH would cost. This information, presently unavailable, would be made accessible to the political process. Congress and the executive branch would have relevant information for decisionmaking. A major policy conclusion of this analysis is that the interventions of OSHA should be rationalized so that they follow the principle of cost-effectiveness.

Equity

Three classes of observations introduce equity considerations into the discussion of appropriate policy for OSH: (1) health and safety threats vary in a recognizable manner from occupation to occupation and industry to industry; (2) those who suffer the highest OSH risks are differentially those with low earning opportunities; 17 (3) those who suffer serious losses of health, though perhaps receiving lost wages and medical expenses, are rarely compensated sufficiently to make them whole. Moreover, some forms of OSH-promoting interventions inhibit market forces which generate higher wage levels as accompaniments to OSH risks.

Proponents of government programs for OSH frequently express the belief that every American worker has the right to a safe and

17 This does not imply that wage premiums are not being paid for assuming risks, since individuals differ substantiallly in marketable labor skills. High skill people, in most cases, are "paid" in both higher OSH levels and higher wages.

healthful workplace.18 They may argue further that it is an inalienable right, one not for sale. No humane society should offer employment that imposes significant risks on its workers' health. Individuals presently working in high risk industries, they might observe, have little opportunity to switch. Indeed, it is increasingly difficult now to attract people to industries, such as coal mining and asbestos, that are known to be risky. The kernel of this argument is that workers, particularly those disadvantaged in other ways, should not be forced to run significant health risks merely to earn a minimal standard of living.

Those who support market outcomes would point out in response that premiums are being received for the risks that are being run. Moreover, the government, by prohibiting certain classes of activities, might end up hurting those it intends to help. (Which groups will be hurt and which will benefit in any case will depend on supply and demand elasticities for labor.) Why has this argument failed to convince the proponents who view safe and healthful workplaces as a right? First, there is the special nature of health, a nonmarket good provided at least initially by nature. Second, discomfort arises from observing the correlation between occupational risk and the distribution of income. Losses of life and limb draw attention to income inequalities that would otherwise be fuzzier in our consciousness.19 But the better remedy is to provide more income to poor people, not to deny them opportunities.

It might help to get our thinking straight on some of these issues if we considered some hypothetical questions. What would happen if we banned all hazardous employment? Would our feelings about market determined OSH conditions differ if we had more substantial programs to transfer income? Given such programs, would risks diminish? Would equity become a less compelling argument in relation to OSHA and OSH promotion? Many areas of social policy are exceedingly inefficient places to bring about the redistribution of income or, more generally since we are including health in our discussion, of welfare. Occupational safety and health seems to be one of them, particularly if, as some suspect, some of its efforts work to the disadvantage of low-opportunity individuals, those who may welcome additional income even at the expense of additional risk.

The critical question for the equity discussion is: If workers wish to make a well-informed choice to assume a higher level of risk in return for greater wages, and if there are no externalities, does society have the right or obligation to interfere with that decision? If the answer is no-and ours is the government's appropriate interventionary role to promote occupational health and safety is better defined. The government should look for situations where the market is, or is likely to be, performing inadequately, and where regulatory intervention can be expected to improve matters.

18 For example, Rep. Phillip Burton, a member of the House Education and Labor Committee when it reported on the OSH Act, stated in his "Separate and Concurring Views": "This bill represents a long-overdue significant recognition that working men and women need Federal assistance to secure their inalienable right to earn their living free from the ravages of job-caused death, disease and injury." [Italic supplied], as quoted in Bureau - of National Affairs. The Job Safety and Health Act of 1970 (Bureau of National Affairs, Washington, D.C., 1971), p. 199.

19 For a fuller discussion of these issues, see Zeckhauser, "Procedures for Valuing Lives."

MARKET IMPERFECTIONS

The issues of equity discussed above arise even when the competitive market functions perfectly, that is, efficiently. However, the efficiency of market outcomes is called into question by several imperfections. Our discussion focusses on three issues: (1) imperfect information, (2) institutional rigidities, and (3) externalities.

Imperfect information

Risks to health and safety are difficult to estimate. Given the importance of the area, we might expect that there would be extensive data series relating occupational conditions to health and safety conditions. There are not. No individual would have the incentive to tabulate these data himself, for he would have to bear all of the costs, but would reap only a small portion of the benefits.20 It would not be efficient for a private business to do so; to recoup its costs, it would have to charge recipients more than the marginal costs of providing the service. Moreover, it would have to make resource-wasting, though profits-conserving, efforts to prevent noncustomers from securing this information. In some instances a national union which represents the vast majority of workers concerned with particular OSH findings might allocate an efficient level of resources to generate information on OSH.21 In general, however, given the public good aspects of OSH information, we must expect the Federal Government to support efforts to collect information on OSH, both statistical information from the field and experimental information from the laboratory. Some forms of OSH-related statistical information would not be worth collecting even if the benefits to all citizens were taken into account. Conceivably, then, the government has been efficient by not seeking to provide useful information on the relation between OSH experience and various hazards. The evidence suggests, however, that the predominant reason is that no division within the government has taken as its responsibility the need to generate both a consistent data base and studies drawing upon it that could inform worker, employer, and government decisions. The Federal Government can play a useful role by gathering and processing information on OSH risks, and making its results known to all at a charge no higher than the cost of distribution.

22

Even if relevant information on OSH were provided, workers might have a difficult time making informed decisions. Experimental and empirical evidence suggests that individuals make poor decisions where low probability events are concerned.23 In their decisionmaking they may find it difficult to distinguish operationally between a .0001 probability and a .00001 probability. There may be enough poor assessors or processors of probabilities to fill the riskiest professions

20 W. K. Viscusi's research suggests that workers do draw inferences based on the accident experiences of their fellow workers. In 1970, the United Rubber Workers negotiated contracts specifying that the firms involved would contribute $.005 per worker per hour to a fund for research on the OSH risks facing members of the union.

2.There is the danger, of course, that once such information is generated there will be irresistible pressure for the government to undertake poorly considered interventions. For experimental evidence, see A. Tversky and D. Kahneman, "Judgement Under Uncertainty: Heuristics and Biases," Science, 185 (1974), 1124-1131. For an analysis of survey data on decisions to purchase flood and earthquake insurance, see H. Kunreuther, "Limited Knowledge and Insurance Protection," Public Policy, 24 (1976), 227–262.

with those who by chance underestimate OSH risks.24 (The inference from this line of reasoning is that if workers understood the risks they faced they would not assume them for the wage premiums now paid.)

OSHA standards can be seen as a device for disseminating information. In that case, though, the Government has gone one step further than gathering information on the magnitudes of risks associated with various conditions. It has, in theory at least, processed the information and chosen the optimal response, which it identifies as a standard. The potential difficulty with this approach is that the Government will not have available information pertinent to particular workplaces. A standard can not recognize firm-to-firm differences in the costs of meeting a standard, or in risks associated with particular conditions-for example, skill level of workers or frequency with which a machine is used-or differences among workers in their willingness to accept particular risks in return for wage premiums.

As difficult as information on risks of accidents may be for an individual to gather and process, the information on health risks is many times less tractable. Losses in health may not show up for many years. Moreover, precise attribution of an incident of health loss to employment may be exceedingly difficult if the disease can have other causal factors. For example, even the most observant workers would probably fail to notice that working with substance X raises the risk of lung cancer 20 years later by 2 percent.

Institutional rigidities

Even if employers and employees have the appropriate information on safety and health risks, various factors may prevent them from responding as they would were OSH and employment decisions made in a competitive market. Where employment decisions are concentrated in the hands of large unions and large employers, some forces that would otherwise push toward optimal health and safety levels will be discouraged. If wages are to be set on an industry-wide basis, a single employer will have a diminished incentive to attempt to provide a safer or more healthful workplace; his wage bill will not be affected. (Any interference with market processes for wage determination, for example, minimum wage legislation, may also have this effect.) The result may be to make health and safety conditions as well as wages more of a matter for bargaining. Under such circumstances, the union finds itself in an adversary relationship with the employer with regard to the determination of safety levels. If only one element were being bargained, say the wage, we could be relatively confident that an efficient outcome would be achieved. With many issues on the table, it is quite possible that the ultimate agreed upon package will be inefficient, in the sense that both the employer and the employee would prefer an alternative package of conditions.25

24 No bias in assessment is implied here. There will also be a group of "over-estimators" which will choose safer employment than would be optimal for them. The overall distribution of risks in employment will not be the same as if all individuals could predict risks accurately.

25 As a technical point, it should be recognized that safety levels for all workers in a firm are likely to be set simultaneously. They will not be chosen individually on a worker-byworker basis as was implied in our discussion of the competitive model. The safety level in a firm may be what economists refer to as a public good. There is the possibility that the level of provision will be inefficient, either too high or too low, if the firm looks only to the preferences of the marginal worker when deciding whether to raise or lower its safety

When the government imposes a minimum safety level, the nature of the bargaining process is changed. Whether outcomes will be more or less efficient is difficult to predict.

A variety of institutional rigidities inhibit market forces that would convey financial reward to an employer who provides a safer or more healthful workplace. Insurance policies for workmen's compensation or employee health coverage are much less than fully experience rated due to the nature of the insurance industry and its regulatory structures. That is, an employer's OSH record is not sufficiently taken into account when his premiums are set. Employers thus pay much less than the resources lost when their workers get sick or injured; OSH suffers. The array of factors that limit worker mobility also works against OSH; workers will not respond fully to differentials in OSH conditions. In particular, when a productive process or workplace is discovered to be unhealthful, workers may find it costly or difficult to remove themselves.

The presence of such rigidities implies that market processes will not achieve an efficient outcome. No one particular form of government intervention is indicated, though it does suggest that measures to limit institutional rigidities should be considered.

Externalities

The competitive market could only work efficiently if no one other than the employee and his employer were concerned with the level of health or safety provided him on his job. If the welfare of others is affected, then there will be externalities to his work choice. Competitive market processes will not be efficient for they will neglect these externalities. Two types of externalities may be present: physical and financial.26 A physical externality might take the form of a worker contracting a communicable disease, or carrying a health-endangering substance outside the workplace on his clothes. Though evidence establishes the existence of this form of externality in some cases (e.g. for asbestos, lead, and kepone), the occupational health hazard that conveys a physical danger to others is clearly the exceptional case.27 Physical externalities thus cannot serve as a major justification for Government intervention in markets for occupational safety and health.

Financial externalities are generated by the institutional arrangements of our society to care for and compensate individuals who are sick or injured, including those who suffer an occupational illness or disability. Life insurance benefits, health care costs, the expense of supporting a family whose wage earner is ill or deceased are borne only in

26 Some analysts might identify what could be labelled an attitudinal externality, related to the special role our society accords to health. Health is sometimes referred to as a merit good; it is suggested that unlike clothes or televisions, society has an obligation to provide all of its citizens the resources required to achieve maximum available health. The existence of extensive government programs to subsidize health care shows the strength and persistence of this attitude.

The authors of this report, feeling strongly that individuals' preferences should be respected wherever possible, question the wisdom of our government's efforts to redirect resources towards health that would otherwise be spent for other purposes. We would argue in particular that those receiving these resources would be better off if they could spend them as they chose.

Should arguments supporting this attitudinal externality be persuasive, it would then be appropriate to measure the externality's magnitude. It may be far smaller than would be inferred from our present array of programs.

The same processes causing hazards to worker health may, of course, also release pollutants into the general environment, thus threatening the health of nearby residents. Such effects, however, are probably dealt with more appropriately by an agency concerned with environmental pollution, such as EPA, rather than by OSHA.