A man who was walking down a street late at night found another fellow on his hands and knees beneath a streetlight. "Looking for something?" the passer by asked. ''Yes, " replied the other, ''I dropped a diamond ring about a block away." ''Well then,'' the passer by sensibly inquired, "why are you looking for it here?" "Because," answered the fellow on the ground, the light is much better here.''

WHEN the United States government first began to regulate industry, the regulations were generally economic&emdash;antitrust laws, for instance&emdash;and they offered society tangible benefits in return for tangible costs. Although these trade-offs did not always allocate costs and benefits equitably, they could usually be quantified in dollars and cents.

However, the new regulations that proliferated in the 1970s&emdash;especially in environmental protection, occupational health, and consumer safety&emdash;provide benefits that we have only begun to learn how to measure. Regulators must confront scientific ambiguities about dose-response levels, latency in carcinogens, extrapolation from animal tests, and synergistic effects. Moreover, these new regulations deal with qualities outside the realm of economics: he value of life and health.

Instead of accepting such ambiguities as inherent to social policy making, many people insist that we force regulations into a single cost-benefit mold. For example, during the presidential campaign, Murray Weidenbaum, now head of the Council of Economic Advisers, stated that President Reagan would require all agencies to make decisions on the "basis of dollars and cents."

Cost-benefit analysis is valuable in regulatory decision making, but unless we recognize its shortcomings, we are likely to force a superficial quantification of issues that cannot wholly be grasped by the reassuringly precise embrace of numbers. Like the loser of the diamond ring, we shall be guided by a bright light in the wrong place, and the result will be not only bad cost benefit analysis but bad decisions.

Antipollution devices such as filters, stack-gas scrubbers, and electronic precipitators are tangible items that must be built or bought, and we can precisely determine the investment and operating costs of each. Yet the record of industry and government in estimating costs is very spotty.

For example, last year I asked my staff at the U.S. Environmental Protection Agency to compare estimated costs for controlling pollution in six industries: automobiles, iron and steel, petroleum refining, pulp and paper, utilities generally, and electric utilities in particular. Although EPA's estimates were lower than industry's for five of the six categories, both EPA and industry tended to overestimate rather than underestimate compliance costs.

For 1974 through 1977, both EPA and the petroleum refining industry estimated that pollution control would cost $1.4 billion, but the actual cost was between $550 and $750 million. EPA estimated that iron and steel plants would have to spend $830 million during those three years for pollution control and industry's figure was $1.6 billion, but the actual figures ranged from $470 to $630 million. In contrast, the pulp and paper companies came up with $1.1 billion as compared with EPA's $1.6, and the actual figure turned out to be $1.4 billion. In the six categories, three of EPA's estimates and two of industry's proved closer to actual compliance costs.

Not in the Public Interest

Even if these figures were prepared in good faith, some were wrong by 50 percent, and the worst was off by a factor of four. If we have this margin of error in conscientiously prepared cost estimates, how much inaccuracy can we expect in estimates that are purposely inflated?

For example, in 1978, the Occupational Safety and Health Administration proposed a standard to control beryllium dust and fumes in plants employing 25,000 workers. Beryllium, a known carcinogen, is used in the production of nuclear weapons. Former Energy Secretary James Schlesinger protested that the cost of compliance-&emdash; estimated at $150 million by industry&emdash; would close down plants vital to national security. Three weeks later, Department of Energy officials conceded that the actual cost would be at most $4.6 million.

In another case, a chemical industry estimate of the cost of complying with a proposed vinyl-chloride standard proved to be 200 times actual costs. And a Senate investigation found that a shoulder harness in a 1968 car cost only $5, not $25 as the auto industry reported. The reported total cost of compliance differed from the actual cost by $100 million. Such exaggerations play havoc with any cost-benefit balancing.

Estimates of benefits are even more uncertain. For example, the typical latency period for cancer is between 15 and 40 years, so that even if we were to begin regulating a newly discovered carcinogen tomorrow, we could not demonstrate reduced cancer incidence for at least 15 years. Cause and effect would be difficult to prove even then, requiring highly detailed medical records and exposure data on perhaps 100,000 people as they pass through a succession of jobs and areas of residence. The benefits estimate would be further complicated by the fact that most forms of cancer do not bear the clear imprint of a single chemical. An excess of bladder tumors among workers in the dye industry was reported in the early 1 930s. but because the workers were exposed to a number of compounds, the specific carcinogen-&emdash;benzidine&emdash;was not identified until 1965. Our analysis would also have to contend with the fact that individuals and ethnic groups differ in susceptibility to cancer.

The difficulty of computing benefits for environmental and safety regulations is illustrated by a study prepared for a Senate committee and released in March 1980. In . situations where a few hazards could be . clearly identified and regulated, the benefits computations were found to be quite precise. For example, a safety standard for infants' cribs has reduced injuries by 44 percent since 1974. Automobile safety controls, including seat belts, saved 28,000 lives between 1966 and 1974.

But in the case of air pollution, where we are dealing with dozens of compounds of varying concentrations in different places, there is considerably less precision. Calculations of dollar benefits--based mainly on reductions in health costs, worker sick leave, property damage, and impact on real estate values&emdash;ranged from $5 billion to $58 billion annually. Benefits from auto pollution controls alone were estimated at between $2.5 and $10 billion a year.

Because of such imprecision, the authors of the Senate study concluded that "a strict cost-benefit approach in federal regulation does not appear advisable. Certainly every appropriate effort should be made to more fully identify the projected consequences of proposed regulatory action. But the findings of this report strongly suggest that placing too great a reliance on such analyses would not be in the public interest."

Environmental regulation also produces unanticipated benefits, stimulating innovation in dozens of industrial processes. Rather than simply hanging filters at the ends of smokestacks, industrial managers are learning to reduce the amount of waste generated in the first place.

For example, a pollution-reducing program inaugurated by the 3M Co. in 1975 saved the company $17.4 million in three years. Through an ingenious manipulation of sulfur trioxide emissions, the Long Island Lighting Co. saved $2 million in fuel costs and recovered 362 tons of vanadium, which it sold for $1.2 million. And a 1977 Census Bureau survey estimated the value of energy and materials recovered as a result of all U.S. pollution-control measures at more than $950 million.

Even if we learn to calculate both costs and benefits within acceptable margins, we will always confront the greatest single obstacle--that of placing of an economic value on human life and health.

One of the most widely used methods calculates deferred future earnings&emdash;how much income a worker loses as a result of injury or death. This approach places little economic value on the lives of the poor or old and none at all on the lives of the unemployed. The Washington Post quoted this statement from a major oil company's protest against new clean-air standards in Montana: "Some of the people who will die from air pollution are unemployed, and therefore have no economic value." Victims of occupationally induced cancer may receive the best medical treatment and financial compensation based on some multiple of lost income, but this will not replace their health, mobility, enjoyment of the present, and anticipation of the future.

The Right Yardstick

The fundamental problem with cost-benefit analysis may not involve computational difficulties but rather our extraordinary emphasis on economics in social decision making. Interestingly, when someone offered to endow the first chair in economics at Oxford in 1825, a provost of the university resisted the gift because he did not want to admit into the curriculum a science "so prone to usurp the rest." And even the donor himself expected the university to keep the new study "in its proper place."

Economics is a useful tool for comparing the worth of unlike things when a society must decide how to distribute its limited resources. However, this balancing begins to fail when we try to reduce all goods to their monetary value. We would not attempt to measure color or temperature with a yardstick, nor would we accept decibel level as a measure of good music.

But this does not mean that we should throw up our hands at any decision involving human life and health. Life is priceless to an individual human being, but society must make cold-eyed calculations about how much it can afford to preserve a human life. The value of human life implied by social investment has ranged from as high as $158 million for coke-oven emission standards to as low as $10,000 to preserve the life of an unknown individual in a public nursing home.

Cost-benefit analysis, no matter how precise, cannot replace social policy judgments, but it can help policy makers see the dimensions of their decisions: Which of a number of hazards presents the greatest threat to society? Which should be dealt with first, and which can be postponed? How should limited resources be used to achieve the greatest benefit?

If, because of scientific ambiguity or the subtleties of balancing life against economics, we must look for prudent social decisions in dark places, we must still search. We cannot substitute the mechanical business of piling numbers on either side of a balance for the agonizing process of making a fallible human judgment. Such judgments are both the dilemma and the glory of any society that aspires to be just.

Back...