Imbalancing act.

ENVIRONMENTAL regulations cost Americans roughly $150 billion a year, and we don't seem to be getting our money's worth. A study released last summer by the Harvard Center for Risk Analysis estimated that the Environmental Protection Agency's actions cost more than $7.6 million for each year of life saved.

In response to such profligacy, many critics of the EPA argue that the agency should be forced to weigh the costs and benefits of regulations before implementing them. Newt Gingrich's Contract with America includes legislation that would impose such a requirement. But reformers who put their faith in cost--benefit analysis should take a look at the EPA's recent report on the impact of a nationwide smoking ban. They will see how easy it is to justify a regulation when you get to decide which costs and benefits to consider and how to measure them. The report suggests that, unless Congress establishes effective procedures for challenging the EPA's findings, cost--benefit analysis may simply become a cover for regulation as usual.

The EPA prepared its assessment of the Smoke-Free Environment Act, which would ban smoking in almost all buildings open to the public, at the request of the bill's author, Representative Henry Waxman (D., Calif.), then chairman of the Subcommittee on Health and the Environment. The report concluded that the ban could save $39 billion to $72 billion a year, mainly by preventing deaths due to secondhand-smoke exposure. Specifically, it projected that Waxman's bill would mean "an average reduction of 7,000 to 12,900 premature deaths per year over the first 50 years, and approximately 7,500 to 13,000 thereafter."

These figures are based partly on the EPA's estimate that secondhand smoke causes about 3,000 lung-cancer deaths a year. There are reasons to question the EPA's conclusion that secondhand smoke causes lung cancer (see "Just How Bad Is Secondhand Smoke?," NR, May 16). But even if we accept that judgment, the death estimate is very soft. In its December 1992 report on secondhand smoke, the EPA conceded that "based on statistical variations, estimates as low as 400 and as high as 7,000 are possible." And even that range relies on several controversial assumptions, including the notion that there is no safe level of secondhand smoke.

The numbers for heart disease, which by the EPA's reckoning would account for the vast majority of premature deaths avoided and money saved through a smoking ban, are even shakier. The EPA took no position on the alleged link between secondhand smoke and heart disease in its 1992 risk assessment. After noting this, the cost--benefit analysis cited a 1991 review by the National Institute for Occupational Safety and Health (NIOSH), which "concluded that the evidence suggested a possible link between exposure to ETS [environmental tobacco smoke] and an increased risk for heart disease of nonsmokers." (Emphasis added.)

There is good reason for such tentative language. In a review of the literature publihshed in the January 1, 1992, Journal of the American Medical Association, Kyle Steenland of NIOSH estimated that 35,000 to 40,000 people died of heart disease each year during the early 1980s because of exposure to secondhand smoke. But Steenland qualified this "crude estimate" with several important caveats. "A number of assumptions are involved in estimating the heart disease mortality due to ETS, adding an unfortunate level of uncertainty," he wrote.

Steenland called misclassification of former smokers as nonsmokers "potentially a serious problem" and noted that "due to the relatively slight increased risk of heart disease for passive smokers and the many factors known to affect heart disease, the possibility of uncontrolled confounding as a cause for the increased risk cannot be ruled out." (That is, the observed association between ETS and heart disease could in fact be due to other variables, such as diet and exercise.) Steenland added that his estimate was based on the assumption that any level of ETS poses some risk; if there is a threshold below which ETS has no effect, the number would be much lower.

This uncertainty did not faze the EPA. Noting that "reported estimates for the effect of ETS on heart disease fall in the range of 32,000 to 40,000 heart-disease deaths per year," the agency used the lower end of the range as a "high estimate." It cut this number in half to obtain a "low estimate." Then it made "an arbitrary additional adjustment of 75 per cent," yielding a range of 12,000 to 24,000 deaths a year. The report said such "conservative adjustments" are appropriate because "these estimates are substantial and because EPA ... did not specifically address heart disease in its risk assessment." But the size of the estimates is irrelevant; the question is whether they're accurate. If they are, there is no need for "adjustments." If they're not, it's hardly "conservative" to take unreliable numbers and reduce them by an arbitrary amount, especially since it's not clear that ETS causes heart disease at all.

The EPA also broke new ground in estimating the smoking ban's impact on asthma. Its 1992 risk assessment said "the evidence is suggestive, but not conclusive, of a causal association" between secondhand smoke and asthma. Testifying before Waxman's subcommittee in July 1993, Morton Lippmann, chairman of the Science Advisory Board committee that reviewed the risk assessment, said "the data were insufficient to come to firm conclusions on causality" for "induction of asthma in previously asymptomatic children." Yet the cost--benefit analysis attributed billions of dollars in savings to the prevention of asthma.

Defending the decision to include heart disease and asthma, David Mudarri, the analyst who prepared the report, argues that the standards for a cost--benefit analysis are looser than the standards for a risk assessment. He also notes that the analysis was not intended to justify EPA regulations. But it was intended to justify federal legislation imposing sweeping restrictions throughout the country. Given the clear policy implications, the EPA cannot disavow responsibility simply because it does not have the authority to regulate indoor air.

Deaths to Dollars

TO translate its estimate of the smoking ban's health effects into dollars, the EPA used data from surveys that ask people how much they are willing to pay to avoid premature death or disease. This technique is widely used to assess environmental regulations, and it has a certain intuitive appeal. But it's not clear how meaningful such "willingness to pay" figures are. The answers are entirely hypothetical, since they do not consider ability to pay, an essential component of demand. "There are no constraints," observes Lynn Scarlett, vice president for research at the Reason Foundation. "In the absence of transactions, you've got funny money."

We can get a better idea of the value that people attach to avoiding the risk of exposure to secondhand smoke if we look at how they actually behave. Taking into account the possible health effects (as well as the annoyance), people are willing to dine in restaurants, fly in airplanes, and work in offices where smoking is permitted. They accept the costs in exchange for the benefits they get: food, transportation, employment. Indeed, as long as individuals are free to make such decisions for themselves, it's hard to see how a government-imposed ban can be justified on economic grounds.

Death and disease estimates account for nearly all the savings projected by the EPA, but there are several other alleged benefits of a smoking ban, all of them questionable. For example, the EPA claimed that employers would save $4 billion to $8 billion a year on office cleaning and maintenance. Mr. Mudarri concedes "there is not a lot of foundation" for that estimate: just one study of a single business, plus one or two anecdotal reports suggesting savings on the order of 10 per cent. Furthermore, as George Mason University economist Robert B. Tollison noted in congressional testimony last May, the EPA's numbers "defy common sense. If $8 billion (or even $4 billion) in employee-maintenance savings are readily available, why have American managers not captured those savings, transferring them directly to the bottom line?"

The same observation applies to projections of reduced absenteeism, lower insurance costs, or any other financial benefit that might result from banning smoking. If anyone cares about the costs associated with allowing smoking in the workplace, employers do. There is no reason to believe that the Federal Government is in a better position to assess the merits of a no-smoking policy than the business people who have to live with the consequences.

Where Are the Costs?

BOLD in estimating the benefits of a smoking ban, the EPA was suddenly reticent when it came to assessing the costs. It assumed, for example, that forcing employees outside for smoking breaks would not have a significant impact on productivity "because taking occasional breaks is already a normal part of the workday for most persons." But surely a smoking ban would tend to increase the frequency and duration of smokers' breaks. The EPA could have considered the experience of businesses in jurisdictions that have already banned smoking in office buildings. Judging from all the smokers standing on the sidewalks, there must be some work that's not getting done.

The analysis also excluded costs imposed on smokers by the ban, including inconvenience and forgone pleasure. Mr. Mudarri notes that it would be hard to put a price tag on such considerations. But a good place to start would be an estimate of the impact that a smoking ban would have on cigarette purchases. Depending on what smokers use the money for instead, the loss could be quite substantial.

If the EPA had included such costs, or measured the value of reduced disease differently, or picked a lower number for heart-disease deaths, or decided that the link between ETS and heart disease was too speculative, the outcome of the analysis could have been dramatically different. The EPA might have arrived at a much lower net-benefit figure; it might even have found that the costs of a ban would outweigh the benefits.

Clearly, the outcome of a cost--benefit analysis hinges on which factors are considered and how they are measured. Those parameters are apt to be influenced, consciously or not, by the policy preferences of the people doing the analysis. In this case, a powerful congressman with authority over the EPA asked the agency to weigh the costs and benefits of his own legislation. Given the inherent flexibility of cost--benefit analysis, it's hardly surprising that the agency was able to tell him what he wanted to hear. If the EPA is charged with assessing its own proposed regulations, the tendency to cook the books may be even greater.

The Contract with America addresses this problem in two ways. Under the proposed Job Creation and Wage Enhancement Act, which requires the EPA to weigh the costs and benefits of new regulations, each analysis has to undergo peer review by a panel of independent scientists. Furthermore, individuals "adversely affected" by the EPA's failure to conduct a proper analysis can sue the agency (and recover lawyers' fees if they win). Another idea that deserves consideration is a proposal by Fred Smith, president of the Competitive Enterprise Institute, to create an office within the EPA charged with criticizing proposed regulations.

Whatever the specifics, Congress should include safeguards in any legislation dealing with cost--benefit analysis. Otherwise, the benefit of this reform could approach zero.