CHAPTER

25

Better Evaluation of Life-Saving Environmental Regulations

by Maureen L. Cropper

Mr. President, I recommend that you issue a new executive order requiring that cost-effectiveness analyses be conducted for all major health and safety regulations, using a uniform set of protocols. Agencies such as the U.S. Environmental Protection Agency, the Food and Drug Administration, and the Department of Transportation, which issue regulations whose primary benefits take the form of reductions in illness, injury, and premature mortality, would be required to calculate the cost per unit of health benefit (such as cost per quality-adjusted life year saved) using identical methodologies. Such an executive order would facilitate a comparison of the cost-effectiveness of health and safety regulations both within and across agencies, and it would serve as a complement to existing benefit–cost analyses of health and safety regulations. Several reasons exist for requiring cost-effectiveness analysis as an adjunct to benefit–cost analysis: it avoids monetization of health benefits, which can be controversial; it provides an alternative method of aggregating health benefits that emphasizes life years saved, rather than private willingness to pay; and the cost per quality-adjusted life year saved, which is commonly employed in the public health arena, provides a simple, transparent way of comparing the efficiency of various health and safety regulations designed to achieve similar objectives. In the long run, expansion of the analytical framework to include cost-effectiveness analysis should lead to more efficient regulation and to greater public acceptance of regulatory decisions.

Background

Executive Order 12866 currently requires that regulatory impact analyses be conducted for all economically significant regulations—those regulations with costs over $100 million per year. The Office of Management and Budget (OMB) suggests that benefit–cost analysis (BCA) should be a primary tool for regulatory impact analysis, and it is the tool that the U.S. Environmental Protection Agency uses in evaluating environmental regulations. In the case of air quality and drinking-water regulations, whose primary benefits are improvements in human health, a BCA calculates the expected reduction in illness and premature mortality associated with the regulation and monetizes these health benefits. Subtracting regulatory costs from monetized benefits provides an estimate of the net benefits of the regulation.

Within EPA, a BCA can inform regulation in three ways:

In practice, the most prominent use of BCA is to examine whether a given regulation passes the benefit–cost test. Recent air quality regulations, such as the 1990 Clean Air Act Amendments, the Tier II Emissions Standards, and the Heavy Duty Engine/Diesel Fuel Rule, have been estimated to yield extremely large net benefits. Indeed, OMB (2003) estimates that four air quality regulations accounted for more than half of the net benefits of 107 health and safety regulations issued between 1992 and 2002. This suggests a second, interagency, use of BCA: comparing net benefits of regulations across agencies suggests where future regulatory efforts might be directed—namely, toward agencies whose regulations have yielded large net benefits.

How would current practice in evaluating environmental regulations be altered by requiring cost-effectiveness analysis (CEA) as a supplement to BCA? Under CEA, reductions in cases of illness and premature mortality would be aggregated by calculating the number of equivalent life years gained, rather than by using weights that reflect what people would pay to avoid illness and reduce risk of death, as is the case with BCA. To illustrate, if an air quality regulation were estimated to save the lives of 100 75-year-olds, this would be translated into the equivalent number of life years saved. If the regulation were estimated to prevent 100 cases of chronic bronchitis (with a given age of onset), this would be translated into an equivalent number of life years saved by determining the relationship between a year of life with chronic bronchitis and a year of life in good health. If a year with chronic bronchitis were equivalent to two-thirds of a year in good health, avoiding a year of chronic bronchitis would save one-third of a life year. Requiring a “reference case” CEA would mean using a uniform set of such weights to aggregate health effects.

Rationale for a Reference Case CEA

CEA health and safety regulations using a uniform set of protocols should be required, rather than continuing to rely on BCA, for at least three reasons. The first stems from difficulties in empirically implementing BCA: specifically, problems exist in obtaining reliable monetary values of the health benefits in a BCA, especially for reductions in risk of death. These problems are avoided by CEA, which does not entail monetization of health outcomes. Second, CEA presents an alternative approach to calculating benefits. BCA is based on the premise that health improvements should be valued by what people would pay to obtain them—an approach that, if applied consistently, means that the value of saving the life of a rich man counts more than saving the life of a poor man. In contrast, CEA counts a year of life saved the same regardless of the economic status of the individual whose life is extended, although the value may depend on the individual’s health status. Third, CEA facilitates making comparisons of the cost of health and safety regulations both within and across agencies, which should promote transparency in decisionmaking.

Problems in Monetizing Health Benefits

The net benefits of environmental regulations, especially regarding air quality and drinking-water quality, are extremely sensitive to the value attached to reductions in premature mortality. For example, in BCAs of air quality regulations, the value of reductions in premature mortality typically constitute 90 percent of the monetized benefits of the regulation. Controversies surrounding the value of a statistical life (VSL) have led to controversies in the use of BCA. The VSL is the value of reductions in the risk of dying that, when added up, will save one life. It is the sum of what people are willing to pay for these risk reductions. For example, a regulation that reduces risk of death by 1 in 10,000 for each of 10,000 people will, on average, save one of these lives. If each person would pay $500 for this risk reduction, the VSL is $5 million.

At least two concerns surround estimates of the VSL. First, although economists have made progress in estimating the value of reductions in mortality risk (for example, by examining compensating wage differentials in the labor market), empirical estimates of the VSL remain very fragile—that is, they are very sensitive to the specification of the models used to estimate them. Indeed, in a 2003 study commissioned by EPA to examine the robustness of compensating wage differentials, Dan Black and his coauthors¹ reported that “collectively, these findings lead us to have severe doubts about the usefulness of existing estimates [of compensating wage differentials] to guide public policy.”

A second concern is that estimates of the VSL depend primarily on studies of prime-aged workers—men in their forties—but they are used to value reductions in risk of death for much older persons; typically, half of the lives extended by an air quality regulation are those of persons 75 years of age and older, whose remaining life expectancy is much lower than that of workers in compensating wage studies. According to BCA, the lives of older persons should be valued by what these persons are willing to pay for reductions in their risk of death; however, few reliable estimates exist of the willingness to pay (WTP) of older persons. Simulations of standard economic models (the life-cycle consumption-savings model) suggest that the WTP for mortality risk reduction eventually should decline with age.

Reductions in mortality risk are not the only health effect that is difficult to value. Table 25-1, which summarizes the benefits of a recent air quality regulation, the Heavy Duty Engine/Diesel Fuel Rule, lists the other health effects that usually are quantified in the BCA of an air quality regulation: chronic bronchitis, asthma attacks, and hospital admissions for respiratory illness and cardiovascular disease. The striking fact that this table illustrates is that these serious illnesses count for very little relative to avoided premature mortality. This is in part due to difficulties in estimating what people would pay to reduce their risk of contracting chronic bronchitis or suffering a heart attack. As a result, a hospital admission for cardiovascular disease is valued by the reduction in medical costs associated with the episode; the value of avoided pain and discomfort are ignored. This has the unintended effect of diminishing the importance of reducing serious illness relative to reducing premature mortality. In the table, the 5,500 avoided cases of chronic bronchitis, each with an average age of onset of 45, count less than 4 percent of the value of extending the lives of 8,300 persons with an average age of 75. A CEA of the same rule gives a very different picture: the reduction in quality-adjusted life years (QALYs) provided by 5,500 avoided cases of chronic bronchitis is more than 40 percent of the QALYs associated with avoided mortality.²

Table 25-1 U.S. EPA Primary Estimate of the Annual Quantified and Monetized Benefits Associated with Improved Air Quality Resulting from the HD Engine/Diesel Fuel Rule in 2030

Source: Federal Register, Vol. 66, No. 12, Thursday, January 18, 2001, p. 5105.

Differences in Ethical Perspectives

This leads to the second advantage of CEA as an adjunct to BCA: it provides a different method of aggregating health benefits, doing so by considering healthy time gained rather than private willingness to pay. CEA weights the value of reductions in risk of death by the expected number of life years gained rather than by the dollar amount that an individual would pay for the risk reduction. This implies that saving the life of a 75-year-old, with a remaining life expectancy of 11 years, will be given one-third the weight of saving the life of a 45-year-old, with a remaining life expectancy of 34 years. (This assumes that future life years saved are not discounted, which would reduce the 3:1 ratio.) Current BCAs assign equal weights to saving the two lives. Which approach is correct? The answer is that both approaches have validity and are consistent with individual preferences, depending on how choices are presented to people. When individuals are placed in the role of social decisionmakers and asked how they would allocate medical research dollars to save the lives of young people or of older people, people of all ages—especially those over 65—value young people relative to older persons even more than in proportion to the ratio of their remaining life expectancies.³

Is this social perspective valid? In the medical community, a reason frequently given for using CEA to evaluate health programs is that the costs of life extension in this country, under Medicare and Medicaid, are indeed social costs. They are borne in large part by taxpayers rather than by the individuals receiving treatment. Parallels exist in the case of environmental programs. Extending the life of an 85-year-old by reducing air pollution will, in addition to benefiting the individual, likely impose social costs if U.S. taxpayers are paying for the medical and nursing care this person receives. Why should a public health perspective not be examined as well as a private one?

Transparency in Decisionmaking

The third reason for requiring CEA is to promote greater transparency and efficiency in regulation. Listing the cost per QALY, as well as the number of QALYs, saved by a regulation provides a means of comparing regulations that is easy to understand. This would be a useful adjunct to current regulatory impact analyses within agencies and, if cost-effectiveness ratios were calculated using the same methodology, a convenient way of making comparisons across agencies. The point here is not that you should rely exclusively on CEA to make regulatory decisions, but that it provides useful information to supplement existing BCAs and studies of the distribution of benefits and costs of regulation.

Why not make such comparisons using BCA? Currently, cross-agency comparisons of regulatory effectiveness are difficult both because some agencies have rejected BCA as a method of analysis, and because agencies that do perform BCAs often use different values to monetize health endpoints. For example, the Food and Drug Administration regularly employs CEA rather than BCA for its regulatory impact analyses. The Department of Transportation conducts BCAs but uses a lower VSL ($3 million) than does EPA. You could require that BCAs be performed by all agencies using a common methodology; however, such a requirement would have to deal with the problems that arise in attaching a dollar value to health benefits. One advantage of CEA is that it avoids the problem of monetization.

The use of CEA also would help inform decisionmaking within agencies. In evaluating air quality regulations within EPA, CEA would provide useful information on the cost-effectiveness of air quality regulations from a health perspective. Currently, BCAs of air quality regulations differ widely in the extent to which nonhealth benefits are quantified. When a regulation yields large health benefits, less attention may be paid to quantifying nonhealth benefits. It would be useful to know, from a health perspective, whether the Heavy Duty Engine/Diesel Fuel Rule is a “better buy” than the Tier II Emissions Standards. A CEA—computed solely using health benefits—would answer this question. This does not imply that you should ignore the nonhealth benefits of either rule, but that it would be useful to compare the rules on an equal footing. The rules also could be compared taking into account the nonhealth benefits that have been quantified by subtracting the value of these benefits from costs before dividing by the number of QALYs saved.

Conclusions

A 2003 guidance by the OMB⁴ suggests that CEAs should be conducted for health and safety regulations whenever possible. OMB’s guidance, however, stops short of requiring a CEA, or of requiring that a set of uniform procedures be used by all agencies in conducting a CEA. I believe that a reference case CEA would be a useful adjunct to existing regulatory impact analyses conducted by agencies that issue health and safety regulations, including EPA. It would avoid the controversies surrounding monetization of health benefits, provide a public health perspective on health valuation, and facilitate comparing the cost of regulations that deliver health benefits, both within and across agencies.

How, exactly, should your administration use the results of a CEA in conjunction with BCAs? Currently, estimates of the net benefits of regulations provide a useful input into the regulatory process, but they are not used in a rote fashion; regulations may be issued even if their net benefits are small (or even negative), if compelling equity or other considerations exist. The same is true of CEA. No one would suggest that a regulation with a cost per quality-adjusted life year saved in excess of some bright line be rejected solely on the grounds of having a high cost-effectiveness ratio. Efficiency in regulation is, however, an important goal, and providing alternate efficiency measures should ultimately improve the policy process.

M.L.C.

Notes