The coming months will be filled with debate over the merits of environmental regulation, headed by the controversial reauthorization of the Clean Air Act. Changes in many of the act's specific provisions will be proposed, but affecting all of them will be the proposal to require that henceforth clean air controls must first be subjected to a formal cost-benefit analysis showing that they provide benefits of greater monetary value than the cost of achieving them. Cost-benefit analysis is not, however, the hoped-for answer that will automatically produce more sinsible and effective government regulation.

Such accounting required, first of all, that dollar values be assigned to intangible benefits. For examble, the medical costs incurred by exposure to unhealthy air can be measured, as can lost wages and lowered productivity, but pain and suffering, psychological trauma and premture death are far harder, if not impossible, to gauge. Attaching specific values to aesthetic improvements is almost as difficult.

Another complicating factor is that costs and benefits arise at different times, with the costs generally arising immediately and the benefits spread over years of generations. Traditional economic methods for treating such problems do not work well. Cost-benefit analyzers must also make assumptions about the value society attaches to lowering a particular risk, yet society's choices in this realm tend to be irrational and unpredictable. Americans are willing to pay a great deal of attention to some risks (a nuclear accident) and very little to lower the risk of a more familiar but much more certain danger (cigarette smoking). Finally, cost-benefit analysis cannot accommodate the fact that costs and benefits are usually felt by different people.

Though costs are usually easier to measure than benefits, they also raise enormous uncertainties. Unpredictable technological innovation frequently invalidates the most careful estimates of the costs of meeting new controls. Productivity may be lowered by the diversion of scarce research funds, or raised by new processes or equipment developed to meet environmental requirements.

Putting all these uncertainties together means that the final answer is likely to mean little or nothing. Applying hard numbers to soft assumptions does not hide the fact that the method is highly subjective -- not the hard-headed objective analysis it might seem. The Clean Air Act, for example, is the most exhaustively studied of all environmental laws, yet recent, reputable estimates of its annual benefits differ by tens of billions of dollars. Its estimated costs are equally vague.

Eventually, cost-benefit analysis should develop into a more useful technique, but at this point it is too crude to be made the sole basis for deciding on a particular regulatory goal. The most likely outcome of such a requirement will be years of expensive delay while courts struggle to decide among wildly conflicting estimates. For the present, the provisions of the Clean Air Act need to be individually examined and weakened or strengthened on the basis of all of the best available evidence.