Volume 20, No.3, Fall 2002

Killing Us with Kindness

By Daniel K. Benjamin

Economists long have argued that life-saving regulations can be counterproductive. Regulations are costly and reduce the income we have for other things. When people are poorer, they spend less on health care and safety measures and engage in riskier behavior. (For example, they buy smaller cars and visit the doctor less often.) Hence, regulations that reduce people’s incomes cause fatalities to go up elsewhere, even as they cut them where intended.

In principle, high-cost health and safety regulations could lead to enough extra fatalities elsewhere to yield a net overall rise in mortality. To date, it has been difficult to obtain a reliable estimate of whether this has happened. Recent research (Gerdtham and Johanesson 2002) helps resolve this difficulty, revealing that any regulation costing more than about $8.4 million for each life "saved" will cause overall fatalities to rise.

The study is based on information from three massive data sets covering more than 40,000 people in Sweden aged 20-84, spanning the years 1980-1996. Unlike prior studies, the authors are able to match individuals’ wealth and income attributes with their initial and final health status, together with a host of demographic characteristics, including education and family size. With this information, they can control for other important conditions, such as the fact that people with high blood pressure or who are unemployed are more likely to die than other people of the same age and income. Similarly, they can control for the fact that married people tend to live longer and that those with children suffer even lower mortality rates. (Two is the right number of children if longevity appeals to you.)

By adjusting for these and other factors, the authors can home in precisely on the link between income and mortality, estimating the impact with an unprecedented degree of reliability. Assuming that an income loss is borne proportionately at all income levels, the authors find that a drop of about $8.4 million in the nation’s aggregate income will induce one more fatality in the economy.

The practical import is that any regulation that costs more than about $8.4 million to save one life will actually cause the overall mortality rate to rise, because the loss of income induces more than one fatality elsewhere. These findings can be used to evaluate the net impact on mortality of any health or safety regulation, or even the impact of supposedly life-saving medical interventions. To illustrate, Gerdtham and Johanesson apply their findings to health interventions in Sweden ranging from hypertension treatment to vaccinations. For fully 10 percent of these interventions, the authors conclude that the net impact is a rise in mortality.

Because the study uses only Swedish data, the authors urge caution in applying it to the regulatory experiences of other nations. (The greater homogeneity and lower income inequality of Sweden, for example, might make their work unrepresentative of other countries.) I shall throw caution to the wind, however, and ask what we might learn about the mortality effects of some U.S. regulations. Thus, I have reexamined a classic paper by John Morrall (1986) on the costs of various federal health and safety regulations.

After applying an inflationary update to make his numbers comparable to the study at hand, I discover some facts that give cause for concern. On the bright side, all three of the Federal Aviation Administration regulations studied by Morrall cost less than $8.4 million per life saved and thus arguably yield a net saving of lives. The same is true for all four of the National Highway Traffic Safety Administration rules.

The record is not so good for the Occupational Safety and Health Administration (OSHA). Indeed, the seventeen OSHA regulations studied by Morrall are about evenly divided between those cheap enough to save lives on balance and those (such as OSHA’s ethylene dibromide and formaldehyde rules) so costly that they have no doubt killed far more people than would have died in the absence of the regulations.

But the worst offender is the Environmental Protection Agency (EPA), which has an almost unblemished record of killing us with its regulations. Of the sixteen EPA regulations studied by Morrall, two probably have saved lives on balance (one regulates chloroform and related chemicals, the other restricts fugitive emissions of benzene, such as at gas pumps). Another EPA rule (regulating uranium mines) is likely a "wash," killing about as many people as it has saved.

The other thirteen EPA rules are all killers. The arsenic standard, for example, costs almost $27 million per life saved according to the official numbers. According to the Gerdtham-Johanesson analysis, this income loss itself leads to about three added fatalities elsewhere for each life saved. Thus, the standard yields a net increase in fatalities. Similarly, the EPA asbestos standard was supposed to save ten lives each year. But its cost per life saved (about $145 million) means that 170 people die each year to save those ten. With friends like that, who needs enemies?


Gerdtham, Ulf-G., and Magnus Johannesson. 2002. Do Life-Saving Regulations Save Lives? Journal of Risk and Uncertainty 24: 231-49. Morrall, John F., III. 1986. A Review of the Record. Regulation,. November/December: 25-34.

Daniel K. Benjamin is a PERC senior associate and professor of economics at Clemson University. His regular column, "Tangents-Where Research and Policy Meet," investigates policy implications of recent academic research. He can be reached at: wahoo@clemson.edu

Related Content