Desperately seeking numbers: global warming, species loss, and the use and abuse of quantification in climate change policy analysis.

• Author: Johnston, Jason Scott
• Position: Response to article by Wayne Hsiung and Cass Sunstein in this issue, p. 1697

Wayne Hsiung and Cass Sunstein's article, Climate Change and Animals, (1) has two objectives. The first is to present the argument that, in addition to being concerned about species loss due to climate change, people should (and at least some do) care about the suffering that climate change will inflict upon animals. As the authors only sketch this argument, (2) I focus here on their primary objective, which is to derive a dollar value for the expected loss to Americans from species extinctions caused by climate change. This analysis is easy to summarize: Hsiung and Sunstein take a number representing the number of species projected to be lost from climate change and then multiply this number by the estimated dollar loss per species. Under this basic methodology, when all kinds of value generated by species are included, they estimate that the global-warming-induced species loss would cost Americans between $162 and $399 billion per year, or 1.4% to 3.5% of GDP. (3)

Even in the big-numbers world of the costs and benefits of climate change, this is a very big number. Indeed, Hsiung and Sunstein admit that their estimated cost of species lost due to climate change would alone justify the United States in incurring the $125 billion that Nordhaus estimated in 2001 as the cost of U.S. compliance with the Kyoto Protocol. (4) This number is intended to remedy what Hsiung and Sunstein call the "virtual[] ignor[ance]" in previous economic analysis of the deleterious effect of climate change on nonhuman species. (5) It is a number that is clearly intended to alert people to the potentially devastating impact of global warming on many species around the world.

But it is not a number to be taken seriously. It is a number based upon a methodology that reflects fundamental errors and misconceptions about economics, a partial and misleading selection from the biological literature on climate change and species loss, and deep confusion over the distinction between government spending and private values.

Essentially, they multiply n, an estimated number of vertebrate species, by p, an estimated probability of species extinction due to climate change, and then multiply this figure by v, an estimate of the total per-species value to all Americans. If we let L denote the expected total loss due to climate-change-induced species extinction, then what Hsiung and Sunstein are supplying is a number for:

L = npv.

I have no quarrel with their estimated number n of vertebrate species (about 57,000). (6) Biologists are quite certain about the number of vertebrate species, and the number n that Hsiung and Sunstein use is in any event the number of known vertebrate species, so n, if anything, underestimates the total number of such species that now exist.

For the probability of extinction, p, the authors use a recent estimate provided by a team of biologists. (7) The biological literature on climate change and species loss is much richer than this, however, and carries at least two main lessons. One is that techniques for generating quantitative estimates of the global probability of species loss from climate change are especially new and controversial developments in biology. The second and more important lesson is that there are many other results in the biological literature on the impact of climate change on biodiversity species loss that are not controversial at all, and that carry concrete lessons for climate change policies. I conclude my analysis by discussing these issues. The bulk of my discussion, however, is focused on the methodologies that Hsiung and Sunstein use to come up with v, their estimate of the dollar value aggregate U.S. valuation of a species. As I shall argue, these methods are deeply flawed in a variety of ways. After briefly summarizing the way that their numbers are generated, I focus below on three major steps that they take.

1. THE SEARCH FOR A NUMBER, ANY NUMBER: A QUICK OVERVIEW OF HSIUNG AND SUNSTEIN ON THE SOCIAL COST OF SPECIES (OR IS IT ANIMAL?) LOSS DUE TO GLOBAL WARMING

   Before delving into a more detailed critique of the methods that Hsiung and Sunstein use to come up with aggregate willingness to pay for species preservation, it is important that readers understand that in this critique, I do not assess a straw man, but the actual analysis presented in their article. Hence, I begin by briefly surveying the remarkable number of twists and turns that Hsiung and Sunstein take in arriving at their final number for the aggregate value to Americans of the expected species loss due to climate change.

   Hsiung and Sunstein begin by noting that they cannot do the first best valuation analysis, which apparently would be to elicit from species how much they would need to be paid to consent to their extinction. They lament that such "[a] contingent valuation study based on the preferences of animals would be infeasible. Polar bears do not have money, and they cannot tell us how much they care about Arctic sea ice." (8) And so Hsiung and Sunstein ultimately concede that "any monetary valuation of animals will inevitably be made by human beings." (9)

   Having conceded that we are stuck with human valuation of nonhuman species, Hsiung and Sunstein next concede that people might not be willing to spend as much per species to save, say, one million species, as they would to save one or two species, and that people are willing to spend much more to save some species (so-called "charismatic megafauna," such as the bald eagle) than others. These problems they simply assume away. (10) They then concede that much of the species loss caused by climate change might happen a long time from now. But rather than discounting those losses to a present value, Hsiung and Sunstein simply assert that existing survey evidence on how much people are willing to pay to save species already presumes that all that people are getting is a reduction in the probability of loss, which is pretty much the same thing as discounting for possible future species loss due to climate change. (11) Hence, they conclude that it is acceptable to use survey data on willingness to pay to save particular species as measuring willingness to pay to reduce the chances of global species loss due to climate change. (12) Further, they argue, whatever numbers they use will be seriously biased downward because extinction rates ignore the death and suffering of individual animals (13) and do not "account for the expected costs of extreme weather events or abrupt climate change." (14) In summary, because some people care about the suffering that species will undergo if the global climate warms and becomes stormier, it is acceptable to apply economic estimates of willingness to pay to protect particular species to all the species that are expected to become extinct due to climate change, and to act as if all of those species are going to become extinct today.

   Hsiung and Sunstein are now ready to derive their estimate of species value. First, they get a number for what economists call the use value of species--the value that people get from using wild species in various ways (e.g., as pollinators, for pest control, and as sources of chemical compounds used in pharmaceuticals). They initially attempt to set this value by taking a recent estimate of the global value of biological services, (15) and multiplying it by the U.S. share of world GDP (remember that what they derive is aggregate American value from species that will be lost due to climate change). But Hsiung and Sunstein say that this first number underestimates the true value of "non-human life" because it fails to include the value of ecosystem services such as erosion control that "are of mixed biological and nonbiological origin." (16) So, they use another study that does include all ecosystem services to which "biological sources" contribute. Taking this study's number and multiplying by the U.S. share of world GDP, they come up with a lost use value from species to become extinct due to climate change of between $58 and $144 billion annually, or 0.6% to 1.4% of GDP. (17) In sum, Hsiung and Sunstein take estimates of the global use value of biodiversity or biological services as a measure of the global value of species, and assume that the United States gets a fraction of this use value equal to its share of world GDP.

   The authors now come to the final stop in their derivation of aggregate species value: the assessment of what economists call existence or nonuse value. Existence value is the value that people get from a species' existence even if they never actually incur any real expenses to see or visit the species, and even if they never get any concrete economic benefit from the species' existence. There is, obviously, no market measure for existence value. Instead, to measure existence value of various natural resources, including species, economists have developed a technique known as contingent valuation. Contingent valuation is a survey in which samples of people are asked how much they would be willing to pay for measures taken to protect or preserve a species.

   Hsiung and Sunstein did not conduct their own contingent valuation study, but instead use two meta-analyses (studies of studies) of contingent valuation studies concerning willingness to pay for species preservation. These studies, surveyed by the meta-analyses used by Hsiung and Sunstein, valued partially overlapping sets of ten and seventeen threatened species, (18) but when put on per-household bases, the two meta-studies generate roughly the same willingness to pay for species protection--about $12 per species. (19) Multiplying once again by the number of Americans and the number of vertebrate species that are expected to become extinct due to climate change, Hsiung and Sunstein's contingent valuation studies give them an annual U.S. loss from species extinction at between $29 and $71 trillion, anywhere from three to seven...