AGCIEOC96/TolSSSII

AGCI Session II: Characterizing and Communicating Scientific Uncertainty

Session Chairs: Dr. Richard H. Moss and Dr. Stephen H. Schneider

July 31 to August 8, 1996

Social Costs of Climate Change

Richard S. J. Tol

Institute for Environmental Studies, Vrije Universiteit

Amsterdam, The Netherlands

Tol discussed Chapter 6 of Working Group III, on the social costs of climate change. It is a great challenge to make a comprehensive assessment of the impacts of climate change from an economic point of view. Economists try to overcome the problem of the overwhelming amount of information on the diversity of the potential impacts of climate change by assigning values to each impact, adding them, and then presenting the sum in monetary terms or as a percentage of GDP. Advantages of such an approach are that (1) it delivers a single indicator of climate change impact, and (2) it makes the impacts of climate change directly comparable to the costs of emission control, and can thus lead to an informed notion of an appropriate level of spending on emissions control.

Controversies

Regarding the work of Tol's group, an enormous controversy arose over the subject of the valuation of a human statistical life, whether such a thing is possible, and whether or not it should differentiate between income groups. The controversy took the shape of personal attacks on lead authors in the popular press regarding alleged racism and blasphemy; it reached its summit in a short occupation of the London offices of the convening lead author's headquarters. The background of this controversy stems from (1) valid concerns about equity and justice aspects of valuation, particularly regarding human mortality; (2) less valid concerns of physical scientists who thought that the damage estimates were too low (the value of a statistical life is one of the most influential parameters in total damage estimates); and (3) the fact that the controversy gave some people the opportunity to shed a nasty light on the IPCC and try to break down at least one of its Working Groups.

An enormous controversy arose over the subject of the valuation of a human statistical life, whether such a thing is possible, and whether or not it should differentiate between income groups.

Uncertainties

No assessment of uncertainties with regard to the social costs of climate change has been made as yet, but generally, the uncertainties are believed to be large. The chapter presents a range of best guess estimates. On global aggregate, impacts are expected to amount to 1 - 2 percent of GDP in a doubled CO2 climate, on an annual basis. Regionally, this varies from slightly positive impacts for the former USSR and, some believe, China, to mildly negative impacts in OECD countries, to as much as a 10 percent negative impact in Africa. The ranges do not reflect uncertainty, however; rather, they reflect different analysts' best guesses.

These conclusions are based on only five studies; two others were ignored, a third was unheard of at that time, and a fourth appeared too late to be taken up. Tol says it is remarkable that the referees did not notice the omission of these four other studies, which are not unknown in certain circles. Three of the five authors of the studies which were used were involved in the writing of the chapter and this may reflect certain biases. All of the five original analysts also drew heavily on the same set of underlying estimates in their work. The already wide range of best guesses mentioned above largely reflects difference in interpretation of the same "empirical" material. This, plus the small amount of information available and the small number of analysts interpreting it, leads Tol to believe that the uncertainty is quite large (but unknown). Nordhaus' 1994 poll of experts supports this notion (see Session Synthesis Essay, The Nordhaus Study in this report).

Most of the studies on which the chapter relies are based on one or two case studies which are then extrapolated to a whole nation or even to the whole world. For example, all the heat stress effect estimates were based on New York data. Agricultural impacts were based on 60 case studies spread over the world and used to obtain a global aggregate. Each of the case studies has its own uncertainty, and extrapolation brings additional uncertainties to the analysis. It is difficult to capture extrapolative uncertainties, but attempts should be made. The best strategy to reduce uncertainties is to expand the number of case studies. Tol believes that, in general, the chapter conveys a message that knowledge is stronger and uncertainties smaller than is actually justified.

The ranges of economic impacts estimates do not reflect uncertainty; rather, they reflect different analysts' best guesses.

Personalities

The main body of the first draft was prepared by one individual. This was revised and extended by another individual. These two (of seven) authors thus had the largest influence on the actual wording of the chapter, and they are known to be somewhat more pessimistic about the impacts of climate change than most mainstream economists.

Outliers

Regarding outliers, the group ignored two studies of comprehensive damage estimates, one because it was somewhat outdated and in the "gray" literature, and the second for the same reasons and also because it took two debatable views: one is that human statistical lives were valued equally and at the OECD level, and the second is that it included the impacts of potential increases in hunger and famine (whereas the five studies used in the report did not). The reasons for not considering hunger in the chapter (at least, quantitatively) include the fact that it would have dominated impact estimates if it was taken into account while the authors did not find the estimates of the increase of hunger attributed to climate change to be reliable. Agronomists have not reached beyond the stage of food availability; most people feel food accessibility is at least as large a driver of hunger, Tol explained.

However, a different outlier was included (perhaps because the author of this study was on the lead author team). That study involved the marginal benefits of GHG emission abatement. The "marginal benefit" of a tonne (metric ton) of carbon abatement is the difference in damage from emitting one tonne of carbon less than otherwise would have been emitted. The range of marginal damage estimates reported was $5 per tonne up to $125 per tonne, but most estimates were concentrated between $10 and $25 per tonne. The $125 figure comes from using a pessimistic estimate of the impact of climate change, expressing this as a function of the magnitude of climate, projecting it far into the future with no change in society's ability to adapt, and using a very low discount rate. This figure was an outlier and should have been excluded (at least, from the policymakers summary), Tol believes, but was not.

The range of marginal damage estimates reported was $5 per tonne up to $125 per tonne, but most estimates were concentrated between $10 and $25 per tonne.

Valuation

The primary aim of the social cost chapter was to derive a comprehensive, monetary estimate of all impacts of climate change. The estimated monetary impact is the estimated impact multiplied by its estimated price. A monetary analysis is easy to do if there is a market impact, land loss due to sea level rise for example, because the price is more or less known, but much harder to do for things not traded in markets. There are basically three methods to value goods or services not traded in markets. First, is the travel cost method. People are willing to spend a certain amount to go to a particular place and that is one way of measuring the value of that place. Second, is the Hedonic method. A house on a beautiful site is worth more than the same house in an industrial area; the price difference is an indication of the worth of that beautiful site. Both of these methods are subject to measurement and value uncertainty, however, and do not capture total value. A third method is to directly interview people as to their preferences. People are asked, "Would you be willing to spend a certain amount to preserve a certain environmental good?" This presents other difficulties as most people do not have clearly expressed preferences in their minds.

Most importantly, all three of these methods are highly context specific. There are strong theoretical and empirical indications that we cannot take a measure of a non-market item from one place or one issue and use it to predict the value of a similar item in another place for another issue, but that is exactly what was done for this social cost assessment. Thus, the uncertainty is large but, again, unknown and perhaps unknowable, Tol says.

The most difficult aspect of this valuation is that monetary values are always income dependent. A poor person would be willing to spend less in absolute terms than would a rich person. This brings us back to the problem of the value of a statistical life. Based on income differences, the authors ended up saying that an American is "worth" $1.5 million while a person from a low-income country such as India or China is "worth" $150,000, which led to the controversies mentioned earlier. There was some discussion around this issue, with various suggestions from the group about using value units other than dollars for human lives. Tol pointed again to the usefulness of being able to value all things in the same units in order to enable comparison of costs and benefits.

Equity

Economics is a science based on the presumption that there is a strong nation state which clearly defines and enforces property rights and which also guarantees some form of compensation between losers and winners of certain policies. The general approach taken to issues such as the valuation of lives is that one takes an average value within one nation. This approach tacitly assumes compensation between citizens. At the international level, however, property rights are not clearly defined or enforced and there are no compensation rules. Thus it makes sense to differentiate values between countries. Otherwise, inconsistencies between national policies on global issues and national policies on domestic issues would arise. If one would use an OECD value globally, it would require developing countries to put more money into climate change abatement than into basic health care. This, of course, would not make sense, Tol explains, whereas differentiated values do.

Another issue that arises in the context of valuation is responsibility for climate change. Though most of the historical accumulation of GHGs is due to OECD nations, developing countries are more vulnerable to the impacts. Why should damages be ascribed to developing countries when they are not, in fact, responsible? To some this means that higher impacts should be ascribed to the OECD countries, making them responsible for paying for the damages to developing countries which the OECD countries caused. In addition, people value things differently depending on perceived property rights and responsibilities. It makes a great difference whether an increase in mortality risk is voluntarily, or is imposed by someone else, particularly if that someone else reaps the benefits and is much wealthier. In the latter case, theoretical and empirical results support the intuitive notion that the value of a statistical life is (much) higher.

Though most of the historical accumulation of GHGs is due to OECD nations, developing countries are more vulnerable to the impacts.

On the whole, despite the considerable attention the chapter received, the comments in both the expert and government reviews were not very numerous or very substantive. In response to the concern that comments on the chapter may not have been taken seriously and adequately addressed by the authors, Tol says that the text was revised to reflect concerns raised but the bottom line estimates were not. The justification for this was a strict interpretation of IPCC rules directing the authors to reflect what was in the published learned literature, and most of the concerns raised were not in the published learned literature (note that the interpretation may have been so strict exactly for this purpose) or in the review.

Tol points out that economic impact assessments are so rudimentary that it comes down to discussing numbers rather than the methods for arriving at them.

In addition, Tol points out that economic impact assessments are so rudimentary that it comes down to discussing numbers rather than the methods for arriving at them. Therefore, concerns and caveats could not be calculated into the values, at least not in a manner that the strict interpretation of the IPCC rules would allow the authors' team. In addition, tensions rose so high that the authors' team was not inclined to address everyone's concerns. Others point out that virtually nothing has been spent on thinking systematically about these issues compared to what has been spent on the physical science of climate change, and that consequently, the quality of the work should not be judged on the same basis.

Forward to Next Section//Back to Table of Contents//AGCI Homepage//Comments: agcimail@agci.org