POLICYFORUM

ECONOMICS

An optimal economic response to climate Climate Change: Risk, Ethics, change requires consideration of discounting and the Stern Review and value judgments. Nicholas Stern and Chris Taylor

ny thorough analysis of policy on cli- They cannot substitute for the detailed risk (3) developed the standard social welfare dis- mate change must examine scien- and cost analysis of key effects. counting formula r = ηg + ρ, where r is the Atific, economic, and political issues Our sensitivity analysis shows that our consumption discount rate, η is the elasticity and many other relationships and structures main conclusions—that the costs of strong of the social benefits attained (also called the and must have ethics at its heart. In a Policy action are less than the costs of the damage social marginal utility), g is per-capita con- Forum in this issue of Science, Nordhaus (1) avoided by that action—are robust to a range sumption growth rate, and ρ is the time dis- suggests that our results as described in the of assumptions. These assumptions concern count rate (also called the pure rate of time Stern Review (2) stem almost entirely from (i) model structure and inputs (including preference). The equation arises from com- ethical judgments. This is not correct. In addi- population, structure of the damage func- paring the social value of a bit of consump- tion to revisiting the ethics, we tion in the future with a unit now and asking also incorporated the latest sci- how it falls over time, the definition of a ence, which tells us that, for a discount rate. given change in atmospheric con- Traditionally, the discount rate has been centration, the worst impacts now applied to policies and projects involving appear more likely. Further, the small changes with direct benefits and costs

science also now gives us a better over less than one generation (say a few on February 2, 2008 understanding of probabilities, so decades at most), which means that people we could incorporate explicit risk are feeling the impact of their decisions in analysis, largely overlooked in their own lives. However, climate change is previous studies. It is risk plus an intergenerational policy issue, and thus, ethics that drive our results. we must see ρ as a parameter capturing dis- The most direct way to look at crimination by date of birth. For example, the problem of constructing an applying a 2% pure time discounting rate economic response to climate (ρ = 2) gives half the ethical weight to some- change is to look at the individual one born in 2008 relative to someone born in www.sciencemag.org impacts of climate change along- 1973. Surely, many would find this difficult side the cost of reducing emis- to justify. sions and then to ask whether it is In addition, the discounting formula worth paying for mitigation. described above depends on the path of However, we do not have the kind of informa- tion, aversion to irreversible consequences, future growth in consumption. Climate tion that would enable formally attaching future conditions, and the rise in price of change involves potentially very large

numbers to all consequences, weighting environmental goods relative to consump- changes and can reduce future growth in Downloaded from them, and adding them all up with any plausi- tion goods) and (ii) value judgments (atti- consumption, so the discount rate applied in bility. Thus, economists attempt aggregations tudes to risk and inequality, the extent to a world with climate change will be less of impacts and costs using very simplified which future generations matter, and intra- than that in a world without, all else being aggregate modeling and, in the process, throw generational income distribution and/or equal. Moreover, this logic can be extended away much that is of fundamental importance regional equity weighting). so that the uncertainty around climate to a balanced judgment. Some credible assumptions about the impacts is taken into account. For every The central estimate of mitigation costs rate at which climate change will result in possible scenario of future climate change, for stabilizing emissions below 550 ppm damage would lead to cost estimates that there will be a specific average discount

CO2 equivalent is 1% of gross domestic are much higher; our modeling approach rate, depending on the growth rate of con- product (GDP) per annum (2). The basic has been cautious. Some modelers are very sumption in that scenario (4). Thus, to speak question is thus whether it is worth paying optimistic about economic growth and of “the discount rate” is misguided. 1% of GDP to avoid the additional risks of social rates of return for the next centuries. Using η = 1 implies that a given social higher emissions. The modeling in the Stern However, they appear to overlook that such benefit will be valued more highly by a factor Review is valuable in identifying some key rapid growth is likely to lead to greater of five for someone with one-fifth the drivers of costs and benefits in terms of eco- emissions and, hence, the more rapid onset resources of someone else. Some commenta- nomic modeling approaches, scientific vari- of climate change. tors have suggested that higher values should ables, and ethical considerations. However, The ethical approach adopted in our be used. Using η = 2 would mean that an extra excessive focus on the narrow aspects of analysis focuses on the ethics of allocation benefit to the person who is poorer by a factor these simplistic models distorts and often between richer and poorer people and be- of five would have a value 25 times that to a

PHOTO CREDITS (MAIN): GETTY IMAGES; (INSET) CORBIS exaggerates their role in policy decisions. tween those born at different times. Ramsey richer person. In a transfer from the richer

www.sciencemag.org SCIENCE VOL 317 13 JULY 2007 203 Published by AAAS POLICYFORUM

individual to the poorer one, how much (future generations) have no say in the cal- term costs but expect greater reductions in would you be prepared to lose in the process culus, and those who feature in the market the future as they place a larger weight on and still regard it as a beneficial transfer? In less prominently (the young and the poor) consumption now over the effects on future the case of η = 2, as long as less than 96% is have less influence on the behaviors that generations (thus perpetuating the delay for lost, it would be seen as beneficial and, for are being observed. significant reductions). η= 1, less than 80%. Although it is a tenable The issue of ethics should be tackled We have argued strongly for an assess- ethical position, those who argue for η as high directly and explicitly through discussion (7). ment of policy on climate change to be as 2 should be advocating very strong redis- No discussion of the appropriateness of based on a disaggregated approach to con- tribution policies. particular value judgments can be decisive. sequences—looking at different dimen- In the case of η = 3 in Nordhaus’example, Alternative ethical approaches should be sions, places, and times—and a broad ethi- over 99% could be lost and a transfer would explored: Within the narrow confines of the cal approach. Nevertheless, our modeling still be beneficial. Does he advocate huge modeling, sensitivity analysis does this. sensitivity analysis demonstrates that the increases in transfers from rich to poor in the There is also scope for further work attempt- treatment of risk and uncertainty and the current generation? ing to disentangle the roles of risk aversion extent to which the model responds to A value of unity for η is quite commonly and inequality aversion that are conflated progress in the scientific literature, are of invoked, but higher values of ρ are some- (via η) in this modeling. Furthermore, we roughly similar importance in shaping times used in cost-benefit analysis. Indeed, should go beyond the narrow framework of damage estimates as our approach to ethics there are a number of reasons why a smaller- social welfare functions to consider other and discounting. It is these three factors scale project such as a new road or railway ethical approaches, including those involving that explain higher damage estimates than may not be as valuable—or relevant at all— rights and sustainability. those in the previous literature. in several years time as circumstances We note briefly that Nordhaus misrepre- Given the centrality of risk, scientific change. However, avoiding the impacts of sents the Stern Review on the subject of advance, and ethics, in our view, the question climate change (the value of a stable climate, taxes. He argues that we propose a tax of should really be why, with some important human life, and ecosystems) is likely to con- $85 per ton of carbon dioxide, which exemptions, did the previous literature pay

tinue to be relevant as long as the planet and equates to $312 per ton of carbon. This was inadequate attention to these issues? on February 2, 2008 its people exist. our estimate of the marginal environmental There was much structural caution in our Further, as people become richer and cost of each extra carbon emission (the approach. We left out many risks that are environmental goods become scarcer, it “social cost of carbon,” hereafter SCC) likely to be important, for example, the pos- seems likely that, rather than fall, their value under business-as-usual, with no policies sibility of strong disruption of carbon cycles will rise very rapidly, which was an issue to reduce emissions. To identify this with a by changes to oceans and forests. It is possi- raised in chapter 2 of our review and has been recommended tax makes two mistakes. ble that risks and damages are higher than investigated in later analyses (5). And the First, any estimate of the SCC is path- we estimated. But one thing is clear: how- flow-stock nature of greenhouse gas accu- dependent. In chapter 13 of the Stern ever unpleasant the damages from climate mulation, plus the powerful impact of cli- Review, we justify our proposed policy change are likely to appear in the future, any www.sciencemag.org mate change, will render many conse- goal of stabilizing emissions between 450 disregard for the future, simply because it is

quences irreversible. Thus, investing else- and 550 ppm CO2 equivalent. In this range, in the future, will suppress action to address where and using the resources to compensate we estimate the social cost of carbon to be climate change. for any later environmental damage may be between $25 per ton of carbon dioxide (450 References and Notes very cost-ineffective. ppm) and $30 per ton (550 ppm), and so the 1. W. Nordhaus, Science 317, 201 (2007). Many of the comments on the review have proposed package of policies should be 2. N. Stern, The Economics of Climate Change: The Stern

suggested that the ethical side of the modeling broadly consistent with this range. Second, Review (Cambridge Univ. Press, Cambridge, 2006). Downloaded from should be consistent with observable market in distorted and uncertain economies, any 3. F. P. Ramsey, Econ. J. 38, 543 (1928). 4. In the review’s modeling, the g in the discount rate is behavior. As discussed by Hepburn (6), there tax should be different from an SCC (8). specific to the growth path in each of the thousands of are many reasons for thinking that market Stabilization between 450 and 550 ppm is model runs in the Monte Carlo analysis of aggregated- rates and other approaches that illustrate equivalent to reductions of around 25 to impact cost estimates. 5. T. Sterner, U. M. Persson, “An even sterner review: observable market behavior cannot be seen as 70% in 2050. Nordhaus claims erroneously Introducing relative prices into the discounting debate,” reflections of an ethical response to the issues that the review suggests reductions on this Working draft, May 2007; www.hgu.gu.se/files/ at hand. There is no real economic market that scale over the next two decades. The Stern nationalekonomi/personal/thomas%20sterner/b88.pdf reveals our ethical decisions on how we Review is also clear that prices should 6. C. Hepburn, “The economics and ethics of Stern dis- counting,” presentation at the workshop the Economics should act together on environmental issues in increase over time, although perhaps not as of Climate Change, 9 March 2007, University of the very long term. sharply as Nordhaus suggests. Birmingham, Birmingham, UK; www.economics.bham.ac. Most long-term capital markets are The ethical approach in Nordhaus’mod- uk/maddison/Cameron%20Hepburn%20Presentation.pdf 7. N. Stern, "Value judgments, welfare weights and dis- very thin and imperfect. Choices that eling helps drive the initial low level of counting," Paper B of "After the Stern Review: Reflec- reflect current individual personal alloca- action and the steepness of his policy ramp. tions and responses," 12 February 2007, Working draft tions of resource may be different from As future generations have a lower weight of paper published on Stern Review Web site; www.sternreview.org.uk collective values and from what individu- they are expected to shoulder the burden of 8. N. Stern, "The case for action to reduce the risks of als may prefer in their capacity as citizens. greater mitigation costs. This could be a climate change," Paper A of "After the Stern Review: Individuals will have a different attitude to source of dynamic inconsistency, because Reflections and responses," working draft of paper risk because they have a higher probability future generations will be faced with the published on Stern Review Web site, 12 February 2007; www.sternreview.org.uk. of demise in that year than society. Those same challenge and, if they take the same who do not feature in the market place approach, will also seek to minimize short- 10.1126/science.1142920

204 13 JULY 2007 VOL 317 SCIENCE www.sciencemag.org Published by AAAS