Few dimensions of modern life reflect a larger divergence between the perceptions that affect government policy and the available evidence than those that bear on the safety and health of the American population. As expressed by the political scientist Aaron Wildavsky, "What are Americans afraid of? Nothing much, really, except the food they eat, the water they drink, the air they breathe, the land they live on, and the energy they use." The popular press has amplified these anxieties by alarmist stories about chemicals in the soil at Love Canal and Times Beach, residues of the pesticide EDB in food and Alar in apples, and nuclear radiation from Three Mile Island. The government has responded to these anxieties by environmental regulations that now cost over 3 percent of GNP and a broad range of other safety and health regulations.
The available evidence, however, provides a very different perspective on the safety and health conditions of the American population. None of the several recent "crises" mentioned above, for example, had a significant effect on the health of the exposed populations. Chemicals in the environment account for only about 2 percent of the avoidable (nongenetic) incidents of cancer. The age-adjusted rates of cancer incidence have declined for several decades for all forms of cancer except lung cancer among women. The infant mortality rate has declined about 50 percent each twenty years since 1900. And, most generally, the average expected life of Americans at birth has increased about 60 percent since 1900, somewhat more among women and substantially more among minorities. In fact, the safety and health of the American population have increased substantially for many decades and compare favorably, with the exception of such conditions as homicide and the infant mortality rate among minority groups, with conditions in any nation.
Making sense of safety was the focus of a conference organized by the Cato Institute in March 1991. The major objectives of the conference were to understand the large differences between risk perceptions and the available evidence and, more important, to suggest changes in government safety and health programs that would increase the expected well-being of the American population. The speakers and most of the participants at the conference were risk analysts but represented a broad range of professions-economists, lawyers, public health specialists, public officials, and the occasional biologist, engineer, psychologist, physicist, political scientist, journalist, and even one bioethicist. The conference proved to be a remarkably productive dialogue among this group. There is more reason to be concerned about whether anyone else is listening.
Most of the articles in this issue of Regulation are based on the papers presented at this conference. As an editor, my most difficult task was to select those papers to be published. Most of the other papers were also quite good and merit at least a brief summary. Michael Gough, a microbiologist, summarized the evidence, from both epidemiology and animal experiments, of the effects of environmental chemicals on human health and concluded that "if we eliminate all of the carcinogens that EPA can regulate, we will see no improvement in cancer rates." Baruch Fischhoff, a psychologist, summarized the typical relations between risk perceptions and objective measures of risk. Howard Kunreuther, an economist, addressed the special case of low-probability high-consequence events. (A paper on airline safety by Richard McKenzie, an economist, was published in the Summer issue of Regulation.) William Evans, an economist, summarized the effectiveness of various automobile safety regulations and concluded that they are relatively efficient. Alan Katzenstein, a chemical engineer, reviewed the studies of the effect of environmental tobacco smoke on cancer and heart disease and concluded that the evidence is not yet sufficient to merit regulation. Lyn Weiner, a public health specialist, summarized the evidence on the effects of alcohol consumption during pregnancy on the health of the child and concluded that public attention should be focused on those with the highest rate of consumption. Alan Schwartz, a law professor, evaluated the implicit logic of product liability law and concluded that it should be replaced entirely by an improved national system of warnings and safety instructions. Margaret Maxey, a bioethicist, criticized the ethical basis for many current safety regulations. And Lester Lave, an economist, addressed the limitations on the use of benefit-cost analysis to choose safety policies. Copies of those papers are available from the authors or from the Cato Institute.
We were also blessed by three fine luncheon or dinner speeches for which papers were not prepared. Aaron Wildavsky treated us with his synthesis of economic, political, and cultural insights about safety policy; those of you who know Wildavsky will understand my comment that he was on a roll. James MacRae, the acting administrator of the Office of Information and Regulatory Affairs, summarized the limited but still valuable role of that agency. And federal judge Stephen Breyer concluded the conference with some proposed institutional reforms to improve safety policy.
The conference, including the papers published in this issue, conveyed three general themes. First, individual perceptions about risk are reasonably accurate when such perceptions are based on individuals' own experience. (The variance of individual perceptions about very low probability events, in contrast, is quite high and provides little basis for rational behavior.) Government safety warnings and alarmist news stories, however, lead people to overestimate the risks, for example, of cigarette smoking and toxic waste dumps. Moreover, individual responses to risk, with some exceptions, seem quite rational (consistent) when the individuals who bear the risks also bear the costs of risk-reducing behavior. For example, most of the studies about how much people are willing to pay to reduce risk estimate that the revealed value of a "statistical" life is in the range from $2 million to $10 million. And people who are more willing to accept some types of risks, such as cigarette smoking, also appear to be more willing to accept other types of risks, such as not wearing a seat belt. There may be some exceptions to this general conclusion. From the perspective of risk analysts, for example, more people should wear seat belts and buy earthquake or flood insurance. Even in these cases, however, individual behavior may be rational; the implicit cost of wearing a seat belt may be higher than what other people consider reasonable, and the small amount of private earthquake and flood insurance may be based on a reasonable expectation that government will bail out the affected parties.
Second, government safety and health programs and regulations, in contrast, are extraordinarily inconsistent. For example, the EPA now imposes costs of over $100 million per statistical life saved by regulating environmental carcinogens. At the same time, identifiable changes in highway design and infant health programs costing less than $1 million per life saved remained unaddressed. Even the same agencies treat different risks inconsistently. The Food and Drug Administration, for example, is extraordinarily careful not to approve an unsafe drug but seems indifferent to the lives lost by the delay in approving a safe and effective drug. Similarly the National Highway and Traffic Safety Administration has imposed progressively more costly auto safety standards for twenty-five years but seems indifferent to the lives lost due to the fuel economy standards. These comparisons, and the many others documented in these articles, lead to a conclusion that a reallocation of the same level of safety expenditures could greatly increase safety. A somewhat more controversial conclusion is that all safety programs and regulations that cost more than $10 million per life saved should be eliminated.
The third general conclusion of the conference was unstated and self-serving but still probably correct: Both the public and government officials would benefit by paying more attention to the available evidence and the recommendations or professional risk analysts.
Somewhat to my surprise and disappointment, the conference did not address one major issue. Should safety policy differ as a function of the conditions in which a person is subject to some risk? Most of the papers appeared to endorse a common cost per life saved or a common benefit-cost standard for all types of risks. Only the paper by Lester Lave suggested some reservations about a general application of a benefit-cost standard to choose safety policies, but he did not develop the implications of his reservations. For that purpose, there are three types of risks: those imposed by others without our consent, those imposed by others that are accepted by individuals as part of a contract relation, and those that are a consequence of our own behavior.
Environmental pollution is an example of the first type of risk. Risks in the workplace and home are examples of the second type of risk. And smoking, eating junk food, and not wearing a seat belt are examples of the third type of risk. There is a reasonable basis for government safety standards affecting the first type of risk, and a benefit-cost analysis is probably the best basis for setting such standard. It is much less obvious that the government should set safety standards affecting the recurrent interaction of people within firms or families; in such a case, benefit-cost analysis may be useful to advise people about safety standards within those organizations but is not a sufficient basis for setting the standards. And for my part, I strongly object to government standards affecting my choice to smoke, drink, eat, wear a seat belt or bicycle helmet, etc., where I bear the full costs of these choices. In that case, all I want from the government is succinct unbiased information about the consequences of my own choices. Our government has enough important things to do without wasting our taxes and reducing our liberties by playing Uncle Nag.