Additional difficulties confronted this study. The published literature and available unpublished literature on OSHA is surprisingly brief. No reasonably broad, much less definitive studies of OSHA's performance have been conducted. Moreover, gaps and inconsistencies in the data base make even the most careful of studies somewhat inconclusive. Thus recommendations of necessity rely heavily on logic, theory, and indicative data. There are no controlled experiments to offer guidance, no substantial data sets from which unambiguous inferences can be drawn.

Although hesitant to advance conclusions and recommendations, we felt it essential to proceed beyond a mere review of OSHA's present status. At present, no one is truly happy with the Agency. But the dissatisfactions of various parties are quite different, and indeed are often in conflict. Concrete proposals may help sort out the differences, as well as reveal more clearly our point of view.

Some critics, particularly those concerned with the legislative mandate of OSHA, assert that it has neglected its major responsibilities. Other critics suggest that the Agency has been inefficient in carrying out its mission. Finally, a number of economists, who appreciate the accomplishments of free markets, and industrial leaders, who see OSHA as another costly, perhaps pernicious, Government intrusion into their activities, feel that the entire conception of a Government Agency regulating health and safety within the workplace is misguided. We believe that much disagreement among these groups would be eliminated, and indeed a sensible middle ground established, if a careful record of the accomplishments and the costs of OSHA could be tallied.

OCCUPATIONAL SAFETY AND HEALTH:

MARKET PERFORMANCE AND THE GOVERNMENT'S ROLE

SAFETY AND HEALTH DETERMINATION IN PERFECTLY

COMPETITIVE MARKETS

Policymaking in the United States, a society that is supposedly dedicated to free enterprise, should start with a recognition of the significant accomplishments of competitive free markets. The government should only consider stepping in when competitive conditions cannot be maintained-when, as the saying goes, the market fails. Some understanding of the appropriate regulatory role for government is gained by considering the hypothetical situation where the market functions ideally.2

If perfectly competitive markets prevail, the health and safety. characteristics of employment will be determined by the decisions of individuals on both sides of the market, workers and employers, with regard to the packages of wages and OSH conditions that they will accept and offer.

The Worker's Choice: Wage Premiums as Accompaniments to Risk The worker is confronted with a number of different employment opportunities. The free flow of information assumed in the competi

2 More extended and formal treatments of OSH determination in competitive markets may be found in the literature. See, e.g., W. Y. Oi "On the Economics of Industrial Safety." Law and Contemporary Problems, 38 (Summer-Autumn, 1974), 669-699. For an entertaining, but also instructive, "Safety Fable" on this same subject see R. S. Smith, The Occupational Safety and Health Act (American Enterprise Institute, Washington, D.C., 1976), pp. 20-26.

tive model assures that he shares all available knowledge about occupational risks. Furthermore, no one else is concerned with his choice of employment: there is no other-regarding behavior, nor will any outside party gain or lose resources depending on the worker's choice. There is no role for the government or other OSH-promoting agency in this hypothetical world. The worker will make a well-informed choice among his opportunities, trading levels of safety and health associated with alternative positions against their wages.

The Employer's Choice: Wage Costs Versus OSH Expenditures

Will the worker's choice lead to efficiency for society as a whole? Yes, if employers also follow the dictates of the competitive model. The wage premiums that must be paid to workers accepting healththreatening employment are the instruments that guide them to appropriate decisions. Employers weigh the financial costs they incur when they impose significant safety or health risks on their workers against the costs of reducing those risks. The optimal level for such risks in any workplace is set where the savings in wages from providing a more healthful workplace would just not cover the additional costs of achieving it. (Thus it is the same level that the workers themselves would set if they owned the enterprise, leaving aside the fact that their trade-off for safety and wages might change with newfound wealth.) This is the optimal level of safety, for it minimizes the sum of the two classes of costs we would like to avoid: expected accident costs plus the costs of preventing accidents.

3

The competitive market achieves its efficient outcomes by bartering wage premiums against incurred risks. Thus, at the competitive equilibrium we would expect to find that among individuals who are otherwise alike, those who assume greater risks will be receiving higher incomes. The economics of markets for risky forms of employment devotes great attention to the existence and magnitudes of these wage premiums.

Empirical Evidence

Although recognition in the economics literature of the role of wage premiums for hazardous work dates back at least to the 18th century. writings of Adam Smith, empirical studies to measure the actual magnitude of those premiums have appeared only recently. The most obvious manifestation of such premiums is in the hazard pay provisions of many contracts. A study by the Bureau of Labor Statistics (BLS) of 503 major collective bargaining agreements, both before and after enactment of the OSHAct, found that about 29 percent rereferred to such hazard pay.5 Most hazard-pay clauses were found in the construction industry, and the most common danger referred to was that of falling. Where the contracts set a specific amount of compensation for hazardous duty, the amount specified most often was 25 cents per hour.

Some of these premiums may be in a form other than higher monetary wages, such as pleasant working conditions or more extensive health and disability benefits. A. Smith, The Wealth of Nations (Modern Library, New York, 1937).

U.S. Bureau of Labor Statistics, "Major Collective Bargaining Agreements in Selected Industries: Safety and Health Provisions Before and After the Occupational Safety and Health Act of 1970." Report prepared by the Office of Wages and Industrial Relations for the Office of Occupational Health Statistics, Bureau of Labor Statistics, Washington, D.C., February 1975.

Ibid., ch. VIII.

In most cases, the additional wages paid for hazardous work are not explicitly identified in collective bargaining agreements. Walter Oi cites an unpublished study by K. Gordon, who found that wage differentials across U.S. Class I railroads were approximately equal to the differences in lost wages due to injuries. Richard Thaler and Sherwin Rosen used insurance company data on the risk of mortality in various occupations to calculate the "value of saving a life." Using multiple regression, a technique that holds other factors equal when estimating the effect of a particular variable, they found premiums for risks to life which implied a valuation of between $176,000 and $260,000 per expected life lost (i.e., a premium between $176 and $260 for exposure to a .001 risk of death). An alternative (log-linear) specification of their model yielded somewhat lower estimates, with a range of $136,000 to $189,000 using the mean values. Their estimates suggest that unionized workers receive higher premiums for risky work.8

Robert Smith also employed regression techniques to estimate wage premiums, using data tabulated by industry rather than by occupation. He found that the premiums for non-fatal injuries were insignificant, but that the premiums for the risk of death were very large, approximately $1.5 to $2.6 million per expected life lost." In contrast, James Chelius, who used data on individual firms in thirteen states, found (with a four equation simultaneous equation model) no evidence of wage premiums in higher-risk industries.10

In the studies cited above, the assumption is made, at least implicitly, that workers are aware of the risks that they face in different industries or occupations. W. Kip Viscusi has approached the problem of estimating the impact of death and injury rates somewhat differently, employing survey data on workers' own perceptions of the riskiness of their jobs. Using multiple regression techniques to control for worker and job characteristics, he found that workers who reported that their jobs exposed them to dangerous or unhealthy conditions received on average about $375 more per year than those who did not. It should be noted that over one-half (52.2 percent) of the blue-collar workers surveyed claimed to be exposed to dangerous or unhealthy conditions. Viscusi also estimated the same equations using the BLS injury rate instead of workers' perceptions of their risks. He found that workers exposed to the average manufacturing injury rate received about $420 annually for incurring that risk.11 Viscusi's results, consonant with those of Thaler and Rosen, suggest that workers in risky occupations are able to command higher wage premiums for risks if they are unionized.

Viscusi also shows that quit rates are substantially higher for hazardous employment.12 Although alternative explanations relating to changing preferences or skill levels are possible, these differential quit rates seem to suggest that workers learn about hazards on the job. To

7 W. Y. Oi, "On the Economics of Industrial Safety," p. 693.

8 R. Thaler and S. Rosen, "The Value of Saving a Life," paper presented at NBER Conference on Income and Wealth, Household Production and Consumption, Washington, D.C., November 30, 1973.

R. Smith, The OSHAct, Appendix B.

10 J. R. Chelius, "The Control of Industrial Accidents: Economic Theory and Empirical Evidence." Law and Contemporary Problems, 38 (1974), 700-729.

11 W. K. Viscusi, "Employment Hazards: An Investigation of Market Performance," Ph. D. dissertation, Harvard University, 1976, ch. 12.

12 Ibid., ch. 11.

the extent that government intervention might provide information that would improve initial perceptions, or change risks so that a large number of workers would not be drawn initially into hazardous employment, it might prove beneficial, particularly if worker mobility is limited.18

The studies discussed above suggest that as economic theory predicts, market forces do lead to higher wages for workers in risky professions. This implies that workers are at least to some extent cognizant of the occupational risks they run. It also implies that if such risks were to be eliminated, say through government regulatory efforts, those workers would likely suffer a loss of income. What these studies do not tell us is how accurately workers assess the heightened risks they run or how fully they are compensated. The studies are aggregative and may not distinguish, for example, among workers at widely varying risks within the same occupation or industry. The studies also do not tell us how workers get sorted into different occupations, whether, for instance, workers presently in safe occupations would demand a roughly equal or much more substantial premium to incur higher risks. Information on this issue might affect the implications for equity of the finding that wage premiums accompany hazardous employment.

In sum, the mere existence of wage premiums for risk does not indicate whether the market is working effectively. It suggests that the labor market recognizes risks, at least partially. Substantially more elaborate studies than those conducted to date would be required to determine whether such risks were fully recognized.

If such risks were fully recognized by both employers and workers, and if preferences and institutional structures were such that a worker's health status was only his own concern, we could be assured that the outcome of competitive market processes would be efficient. Any Government intervention to promote or alter levels of OSH would by necessity reduce the welfare of some parties.14

We do not attempt in this analysis to assess the overall performance of labor markets, an analysis that might lead us to a general prescription for government intervention or nonintervention. Rather, we shall attempt to identify areas where the justifications for some government OSH-promoting efforts are weakest or strongest, and propose that resources be reallocated toward the latter. We shall also be concerned with identifying the types of government programs that seem most likely to generate efficient outcomes.

APPROPRIATE LEVELS OF SAFETY AND HEALTH

The language establishing OSHA defines its primary objective: "to assure so far as possible. . . safe and healthful working condi

13 This argument, however, should be distinguished from a similar argument that market outcomes will not be efficient: The market will not provide premium pay for exposure to unknown hazards. One example cited is vinyl chloride, which has been shown to increase the chances of developing angiosarcoma, a rare type of cancer of the liver. In that case workers were exposed to the hazard for several decades before the risks became known. The fact that vinyl chloride workers were not compensated for the extra risks they faced does not mean that the market operated inefficiently. Given the information available, market forces probably allocated resources efficiently. Until the risk was recognized, government regulation was also impossible.

14 Even if labor markets do satisfy the competitive conditions, the government may have a role as a generator of new information about the relationship between potential hazards and OSH.

tions." 15 There are at least two classes of difficulties with this charge. First, in conception it leaves aside other valued objectives, the most important of which is that appropriate value be attached to the resources of society directed to OSH promotion. Second, it lacks the degree of precision needed to make it applicable to and employable in policy formulation. No quantitative interpretation is given to levels of risk that might be regarded as safe or healthful.

How might appropriate levels of risk be defined? Zero risk levels are clearly inappropriate. In many circumstances they are technologically infeasible. In others, economic considerations make it clearly undesirable to eliminate all OSH risks. Some appropriate nonzero level of risk must be defined. This suggests that OSHA will be confronted with the need to make trade-offs, implicitly or explicitly, between OSH levels and the resources (including OSHA's resources) required to achieve them. This is precisely the type of trade-off that is made implicitly when we choose not to further lower speed limits, when we impose limits on expenditures for medical care, or fail to require fire escapes on single-family dwellings. A critical question for policy is who should be making the trade-offs in the OSH area.

Efficiency

A reasonable response to that question is that the worker himself should be the judge. He is the one whose health is being threatened. If the worker is the judge, the trade-off rate will be his willingness-topay for increased OSH levels. Moreover, if labor markets were perfect and competitive, wage premiums accompanying risk would just reflect willingness-to-pay. If labor markets are not perfect, in which case there is a rationale for government intervention, benefit-cost analyses that value OSH losses at workers' willingness-to-pay will lead to efficient outcomes.

No issue in OSH has generated more vitriolic disagreement than the question of the appropriate weighing of economic costs. The arguments against the benefit-cost approach include: This is not what Congress intended when it made OSH a matter of public policy. Life is sacred and should not be bartered for dollars. Health and bodily integrity are not traded on markets; the quasi-market measure willingnessto-pay is therefore not an appropriate indicator of value. (Even if willingness-to-pay were accepted, there would be substantial disagreement on issues of calibration from surveys or market studies, whose willingness-to-pay, before or after injury, etc.16)

Fortunately, substantial progress in the reform of OSHA's procedures can be achieved without attempting to resolve the problem of valuing lost lives or impaired health. OSHA's procedures have been formulated with minimal concern for economic resources. It should not be surprising, therefore, that there are substantial inconsistencies across areas in the cost of achieving OSH improvements. For a particular firm, compliance with either standard A or standard B may reduce the number of expected accidents by .1 per year. Yet compliance with A may cost $1,000 per year whereas compliance with B may cost 10 times that amount. We could achieve more OSH for the same workers for

15 OSHAct, sect. 2(b).

16 For a discussion of various methods for valueing lifesaving interventions, see R. Zeck. hauser, "Procedures for Valuing Lives," Public Policy, 23 (1975), 419–464.