Discounting Human Lives

One of the missions of the U.S. Environmental Protection Agency (EPA) is to protect human health by reducing human exposure to pollution. Some EPA regulations—those controlling chemical plant accidents, for example—yield immediate health benefits. Others will save lives—that is, prevent premature deaths—but not for some time. Examples include the cleanup of Superfund sites to prevent groundwater contamination and the banning of asbestos in a variety of uses. Both actions reduce the public's exposure to cancer-causing substances; but because the cancer cases avoided would not have occurred until many years after exposure, the actions save lives in the future rather than today.

This creates a problem in evaluating the benefits of environmental regulations. In counting the number of lives saved by a regulation, should a life saved 25 years from now be equivalent to a life saved today? According to EPA, the answer is yes. In evaluating regulations to ban asbestos under the Toxic Substances Control Act, for example, EPA simply added up the number of lives that would be saved, regardless of when they would be saved.

However, analysts at the Office of Management and Budget and others have criticized EPA for not figuring the benefits of these regulations in the same way that it calculated costs. Instead of merely adding up the dollar costs of the ban, regardless of when they occurred, the agency valued a dollar ten years from now at the amount one would have to put in the bank today to yield $1 in ten years—the so-called present discounted value of the dollar. If money in the bank earns interest at 5 percent, this amount is only 61 cents. EPA thus discounted the costs of the asbestos ban but not the benefits.

Over the last three years, researchers at Resources for the Future (RFF) and the University of Maryland have investigated how members of the public feel about the discounting of human lives. At the heart of this investigation was the following question: Does the public feel that a life saved in the future is equivalent to a life saved today, or does it feel that a life saved in the future counts less than a life saved today?

Far from arcane, the subject of discounting lives arose in a recent court ruling. On October 18, 1991, the U.S. Court of Appeals for the Fifth Circuit overturned EPA's regulations of a variety of asbestos-containing products. Among other reasons for its decision, the court cited the agency's unwillingness to discount the lives saved in the same way it discounted the future costs of the regulations (see Corrosion Proof Fittings v. EPA).

To shed light on the issue of discounting lives, the RFF and University of Maryland researchers surveyed approximately 2,600 households—1,600 in the state of Maryland and the Washington, D.C., metropolitan area, and 1,000 nationwide. In a telephone interview, each household was told that the government could fund only one pollution control program and was asked to choose between a program that would save X lives today and another program that would save Y lives in the future. From the answers to these questions, the researchers were able to infer the number of lives that must be saved in the future to make people as happy as would saving one life today. The larger this number, the more heavily people discount lives saved in the future.

Discounting lives saved in the future

As part of the Maryland Poll, a public opinion survey conducted by the University of Maryland Survey Research Center in November and December of 1990, approximately 1,000 households in Maryland were asked to choose between two hypothetical life-saving programs based on the following information:

Each year some people in the United States may die as a result of exposure to certain kinds of pollutants. Unless there are programs to control this pollution, 100 people will die this year from pollution and 200 people will die T years from now. The government has to choose between two new programs to control this pollution. The two programs cost the same, but there is only enough money for one.

Program A will save 100 lives now.

Program B will save Y lives T years from now.

Which program would you choose?

In the national survey of 1,000 households, the RFF and University of Maryland researchers explored the relationship between peoples' discount rates for money and their discount rates for saving lives. Of interest was whether, on average, these two discount rates are the same and whether individuals who have high discount rates for money also have high discount rates for saving lives, reflecting, perhaps, a general orientation toward the present.

In the national survey, people were asked to choose between receiving $10,000 today and receiving a larger sum in either 5 or 10 years. Their choices implied an average monetary discount rate of 20 percent for a 5-year time horizon and 10 percent for a 10-year time horizon. These rates, while higher than market interest rates at the time of the survey, are close to monetary discount rates inferred in other studies from purchases of energy-saving appliances and from reenlistment bonuses paid to members of the military. More important, they are approximately equal to the discount rates for saving lives (see table, p. 3).

The researchers also found that people with high monetary discount rates have high discount rates for saving lives. Those people who chose to receive $10,000 today rather than a larger sum in the future were for the most part the same people who preferred the present-oriented life-saving program to the future-oriented program.

Factors affecting discount rates

A question that arises in interpreting answers to queries about choices among life-saving programs is whether these answers reflect pure altruism or incorporate some selfish concerns. Did respondents to the above surveys discount lives saved in the future because they or their families are unlikely to benefit from future-oriented life-saving programs or because they do not feel as close a kinship with persons living in the future as they do with persons alive today?

One way of answering this question is to ask the respondents whether they considered how life-saving programs would affect them personally. In the national survey, they were asked whether, in making their choices, they had considered the effect that these programs would have on them or their families. Forty percent said they had; however, statistically, this consideration did not increase the chance that they would choose the present-oriented program.

An indirect way of investigating whether responses reflect selfish concerns is to see if the variation in responses can be explained by the respondent's age or by whether or not he or she has children. If older people are less likely to benefit from future programs, and responses are partly selfish, then older people should have higher discount rates than younger people. This should also be true of people with young children. When confronted with a choice between a program that saves lives today and a program that saves lives in 25 years, a person with young children should be more likely to choose the former—all else being equal—if he or she is more concerned about protecting his or her children when they are children than about protecting them when they are adults.

Statistical analysis of survey responses confirmed these hypotheses. The older the respondent was, the higher was his or her discount rate for saving lives. Having children who were under the age of 18 and who lived at home raised the discount rate for saving lives 25 years or more years in the future, but had no statistically significant effect on the discount rate for saving lives in 5 or 10 years. This is consistent with the hypothesis that 5 years from now a person's children will still be children and just as deserving of protection as they are now, but 25 years from now they will be adults and able to take care of themselves.

The only other demographic variable that consistently affected discount rates for saving lives was race. Black people had significantly higher discount rates than other races, as has been found when it is money, rather than lives, that is discounted. Because the analysis controlled for—that is, held constant—education and income variables, the race variable may have reflected cultural factors or the fact that Black people have shorter life expectancies.

It is perhaps surprising that income and education had no effect on the discount rate. Further reflection, however, suggests that there is no reason why low-income persons, who have been found to discount monetary rewards more heavily than high-income persons, should discount lives at a higher rate.

Caveats

The above findings should be regarded as preliminary for several reasons. First, relatively brief telephone interviews are an imperfect vehicle for eliciting preferences with regard to choices as difficult as those presented. Second, some evidence suggests that the order in which questions are asked has a slight effect on responses and thus on the calculation of implicit discount rates. Third, as reported above, some fraction of the respondents took into consideration the fact that the present-oriented program could protect them personally while programs with time horizons of 50 years or more were unlikely to do so. Since the responses of these people reflect selfish concerns, some people might question the validity of using the discount rates inferred from the RFF study to make social decisions.

In spite of these caveats, however, the overwhelming majority of those questioned attached a lower priority—sometimes much lower—to lives saved in the future, even when the time horizon was quite short (5 or 10 years). For example, for a program that would save lives 25 years in the future to be preferred to a program that saves lives immediately, it had to save at least six times as many lives. If borne out by additional research, this finding would have important implications for the evaluation of many programs to regulate health and safety. In view of the resources being devoted to these programs, such research appears to be worth undertaking.