Bjorn Lomborg and the 500 lifesaving interventions

Originally published as "Environmental Skepticism" in the Skeptic - 2002 - Vol. 22 No. 2, page 62.

I recently read Bjorn Lomborg’s TheSkeptical Environmentalist and was sufficiently impressed by his detailed analysis that I made the effort to obtain some of his source material.
In a section called entitled “The costs of the Litany”, Lomborg gives what would appear to be a comprehensive overview of research conducted at the Harvard University Center for
Risk Analysis by Tammy Tengs and John Graham. In their initial paper, “Five-hundred Life-Saving Interventions and their Cost-effectiveness”, they analysed the comparative cost effectiveness of 587 policy initiatives in terms of their dollar cost per year of life saved. The interventions covered health care, housing regulations, transportation, occupational and environmental controls. As well as a comprehensive tabulation of the 587 interventions they also summarised the information across policy sector and government department by stating the median costs. The environmental sector performed worse with a median of $4 200 000 compared to $19 000 for health and an overall median of $42 000.

Lomborg’s presentation of the data focuses on these median costs. In a note he says “The advantage of the median is that it is less effected by very atypical (high) prices.” Yet adding the
total years of life saved and dividing by the total cost would have given more representative figures, particularly if the data set contained a lot of high valued initiatives that were of
relatively low cost, but negligible life
saving benefit.

In a subsequent paper, “The Opportunity Costs of Haphazard Social Investments in Life-saving”, the researchers jettisoned 402 of the policy initiatives because their initial estimates were derived from geographically limited or small data-sets and comparable national figures were unavailable. Lomborg, mistakenly asserts these 185 initiatives in the 1996 paper are actually implemented policies.

Tengs and Graham used linear programming to calculate various optimal policy mixes and measured them
against the policies that the US government had actually implemented.

According to their study, the status quo involved spending 21.4 billion dollars per annum, in order to save 592 000 life years. They found that by changing the policy mix and degree of implementation, a maximum of 1 230 000 years of life could be saved. They recalculated the problem with the added constraint of keeping spending in the five policy areas invariant. Surprisingly, this strategy was still able to save 1 190 000 years of life.

Tengs and Graham also considered a further subset of 134 life-saving initiatives which were proposed (or implemented) by five government agencies. Rather than use median cost per life year as a measure of departmental performance, the researchers obtained an average for each department by adding the estimated cost of implementation for the policy set and dividing by the estimated years of life saved. Although, they omitted their baseline figures from the paper, their optimised results do not reflect the extreme disparities that Lomborg suggests.

With the same departmental budgets, another 86,300 lives could have been saved at a marginal cost per life year of $1 510 000 for the Consumer Product Safety Commission, $497 000 for the Occupational Safety and Health Administration, $45 000 for the
Federal Aviation Administration, $35 200 for the Environmental Protection Agency, and $23 800 for the National Highway Safety Administration. Yet according to Lomborg “the extremely high typical cost of $7.6 million for the EPA area is fairly representative of the cost of saving life by means of toxin control”.

When Lomborg closes the passage by saying, “the Harvard study gives us an indication that, with greater concern for efficiency than with the [Environmentalist] Litany we could save 60 000 more American lives year”, he is guilty of a blatant misrepresentation.

I agree with Ian Plimer when he warns of “the misuse, by self-interested bodies, of statistics ... in supporting scare-mongering and other irrational claims”. However, I think it is unfortunate that Bjorn Lomborg has fallen into the same pattern of abuse that he seeks to expose.