A Preliminary Summary of Contested Assumptions
Cost Benefit Analysis and Climate Change
Appendices:
If governments agree to slow the pace of global warming during the next decade, it will largely be due to the efforts of the Intergovernmental Panel on Climate Change (IPCC). The IPCC was established in 1988 by the United Nations Environment Programme (UNEP) and the World Meteorological Organisation (WMO) to assess the science of climate change in order to provide a basis for international and national policy- making The IPPC's First Assessment Report (1990) defined cuts in greenhouse gas emissions of between 60% and 80% as immediately necessary to stabilise greenhouse gas concentrations in the atmosphere with a view to halting global warming.
Since 1990 the IPCC has been preparing a Second Assessment Report (SAR) which it hopes to publish by the end of this year. The report is authored by three working groups.
Working Group I is reviewing the science of how the earth's climate system functions and how this might change as a result of human activities.
Working Group II is assessing published work on the health and other effects of climate change and on the measures which could be adopted in sectors such as agriculture, energy production, industry and transportation to minimise those effects.
Working Group III is preparing a technical assessment of the state of knowledge of the "socio economics of climate change mitigation" and "other cross-cutting issues", a phrase which was intended to signal a full sociological assessment of the issues at hand.
Working Groups I and II are well advanced with their reports, drafts of which have been circulated for comment in academic circles and in part on the Internet. There have been no major disagreements about these drafts' content and conclusions. The draft report by Working Group III (WG3), however, ran into severe criticism when its section on the "Social Costs" of climate change was discussed at a WG3 meeting in Geneva in July 1995.
In this paper we outline the concerns raised about WG3's social costs assessment and recalculate those social costs in the light of these criticisms.
The difficulties of placing a monetary value on the damage which is likely to be caused by global warming are legion. The costs are long-term, highly uncertain and in some cases unknowable in advance, even in principle. For many types of damage such as species extinction, the assignment of a monetary value makes little sense, and some economists go part way to acknowledging this by distinguishing between 'tangible' and 'intangible' costs.
In spite of this, the WG3 team for the "Social Costs" of climate change attempted to put a cost figure on the damage global warming might do, basing their estimates largely the work of Fankhauser (1) and Tol, (2) - both members of the group - who built on earlier work by two other members of the group - Cline (3) and Pearce (4) - together with that by Nordhaus (5) and Titus (6).
The team's summary assessment of the global damages consequent on climate change is that monetary losses will equal to 1.5% to 2% of Gross World Product (GWP). (B) This is an estimate for a single, unspecified, year - the year when CO2 equivalent © concentrations will have doubled. They assume that this doubling will happen in around 2050 or 2060. (D)
the global economy will have progressed from the present to the year 2050 on a "business-as-usual" path;
global mean temperature will have risen by the "mean" figure of
2.5 C by that year,
it is useful to give policy-makers a "snap-shot" of that single year's damages, ie one divorced from a cumulative assessment of damages for the period between the present and 2050.
This figure of 1.5% to 2% of GWP is significantly lower than that reached by some other analysts - most notably Hohmeyer and Gaertner (7) in their 1992 report to the European Union. Their study estimated accumulated damage costs of potentially $900 trillion by 2030; that is, well beyond 100% of GWP by that year and therefore up to two orders of magnitude greater than the figures reported in the WG3 draft.
WG3 also estimated regional damage costs as being equivalent to 1% to 1.5% of GNP in OECD countries and between 2% and 9% of GNP in countries outside the OECD. These regional losses were derived exclusively from the work of Fankhauser and Tol.
In our view, both the global and regional ranges of damage figures currently drafted in WG3 contain errors, are unjustified and should be replaced. Using Fankhauser's raw damage figures as the starting point for developing our arguments, we conclude that the expected extent of global damage for the year 2050 as a result of warming is highly uncertain but probably lies in a range between 12% and 130% of GWP. Within this, for the OECD region, the range is from 0.6% to 17% of Gross Regional Product (GRP), while for the Rest of World (RoW) (those countries outside the OECD) it is from 25% to 250% of GRP. This represents accumulated losses between 1990 and 2050 of between $50 and $600 trillion. We consider even these estimates are on the low side, as we have made many conservative assumptions and made only very limited allowance for surprises.
To address the range of temperatures which may plausibly obtain in 2050, we have made assumptions about how damage costs vary with temperature change. Clearly such variation will not be linear and we have assumed an S- shaped relationship, so that costs rise very slowly with the first increment of temperature change and approach a limiting value at temperature rises above 30 C. We do not consider that the present state of knowledge justifies building a more complex model. Details are given in Appendix A.
The gulf between our figures and those in WG3's current draft report can be explained in large part by our having employed different assumptions and methods to those used by WG3. The areas of dispute are summarised below and then in more detail in subsequent sections.
WG3 assumes "Willingness-To-Pay" (WTP) as an acceptable method of assessing damages costs. We argue that "Willingness- To-Accept Compensation" (WTAC) is a more sensible method.
WG3 compute damages for the single year of CO2 doubling, that is the year in which global mean temperatures will be 2.5 C higher than pre-industrial. We argue that it is most useful to policy makers to focus the assessment on a particular year and the period leading up to that year. This is more useful than focusing on the "moving target" of when CO2 doubling may or may not occur. We suggest that 2050 should be used, a date within the range expected by the IPCC. However, we also argue that by 2050, various factors may well have increased CO2 and equivalent greenhouse gas concentrations in the atmosphere to more than double the pre-industrial levels and that global mean temperature is consequently likely to be higher than the stated 2.5C.
WG3 authors calculated regional GNP losses by dividing damages corrected for "Purchasing Power Parity" (PPP) by GDP figures which have not be corrected for PPP. We argue that this procedure is arithmetically wrong and also now seen to be wrong. It results in another seriously misleading factor in the representation of the proportional damages in and between different regions of the world. Results based on this procedure must not be published by the IPCC, and regional losses must be recalculated using sound methods.
WG3 assumes that neither "climate sensitivity" , (E) "feedbacks" or other uncertainties need be portrayed in its bottom-line results. We argue that the IPCC WG3 must reflect the full range of "uncertainties" and "sensitivity" in the bottom-line figures it publishes in its final report, and in its Summary for Policy-Makers (SPM).
WG3 also assumes that in key areas where there are uncertainties over the complexity of imminent warming factors (such as positive feedbacks and sulphate aerosol removal) that these can be given a value of zero in the assessed damages. We argue that they realistically cannot and that they must be represented by numbers greater than zero.
Deaths due to malaria and malnutrition have unrealistically been omitted from the WG3 draft assessment. We argue that these must be assessed and included in the report.
What follows sets out these argument in more detail.
Working Group Three's damage estimates are based on the "Willingness- To-Pay" (WTP) method of assessing damage costs. WTP leads to discriminatory differential estimates in cost rates between the OECD and the rest of the world, most notably differential estimates of the value of a "statistical life". It would have been more correct to use the "Willingness- To-Accept Compensation" (WTAC) method.
"Willingness to Accept Compensation" is regarded as the "conceptually correct" (8) procedure in Cost-Benefit Analysis - that is, it assesses costs in terms of what losers are willing to accept as compensation for any inflicted disbenefit. Willingness to Pay (WTP) is appropriate only for benefits. By describing potential payments for the avoidance of climate-change damage costs as "benefits", the WG3 authors give dubious plausibility to the use of WTP. In reality, however, there will be in a broad view no benefits from climate change, only different kinds of costs or disbenefits borne by different groups of people. (F)
WTAC naturally results in very much higher damage costs than WTP, since the amount that people are willing to accept as compensation for major losses is not constrained by their income and - most people being poor - is many times greater than what they are willing and able to pay to prevent undesirable impacts on their lives. The use of WTP also leads naturally to the adoption of differential 'statistical' life evaluation, sometimes known as "Values of Statistical Lives" (VOSLs). This has been the subject of much heated debate. We state here our position.
There is an extensive literature on whether it is admissible to give human life a monetary value, and, if admissible, what value life has. Some reject the idea out of hand. Nonetheless, in certain industries, it has become an accepted management tool. A good overview from the perspective of the oil industry can be found in Fleishman (9) who concludes that a valuation in the range of o500,000 to o5 million is appropriate (approximately $750,000 to $7.5 million).
The concept of "statistical" life has been introduced into the debate not because person A is being asked how much he or she is willing to pay or to accept for himself/herself or for person B to be definitely killed, but because of attempts to place a value on how to much to pay or accept for a relatively low probability - normally less than 1% - of any particular individual being killed. To do this, one essentially values the life at, say, $1.5 million, and multiplies by the (low) probability of an individual dying as well as by the total population size involved. If the probability of an individual being killed reaches a sufficiently high level, the whole process of valuation is rejected and the life is effectively regarded as having infinite value. According to Fleishman, there is little agreement as to how great a risk is acceptable in this sense, because it all depends on society's perception of the value of the risk-creating activity.
Major problems arise when one life is valued at more than another, as is done by Fankhauser and Tol. Following Hohmeyer and Gaertner, we argue that no differentiation by nationality, race or gender should be adopted, on grounds both of straight forward ethics and of practical international politics. This is regardless of whether the life is "statistical" or not. If differential values arise logically from a theory such as WTP, that merely demonstrates the inapplicability of the theory.
The ethical argument suggests a method of valuation based on how much someone is willing to pay can only be used as an input to some kind of averaging process. The highest value we might consider is Tol's OECD value of $3 million, the lowest Fankhauser's world average of $350,000. Advocates of differential statistical life evaluation seem to think that because the risk of death is being costed rather than the certainty of death, the equity argument is nullified. We disagree strongly. In addition, using WTP, they find a single global value unrealistic. Thinking in terms of WTAC, however, makes such a value quite plausible, provided that an OECD-derived value is used. (10) Following Fankhauser, we use $1.5 million.
Differential discount rates by region have also been advanced to make the "present value" - that is, the "discounted" value of future lives - different. This too is unethical and unacceptable. The "present value" of a Chinese life in 2050 must be treated as the same as the "present value" of an American life in 2050.
Parity-Unit-Damage-Valuation (PUDV)
If one accepts the equal life valuation argument above, the next step is to extend the same principle to the rates for valuing all the other kinds of damage costs. In Hohmeyer and Gaertner analysis, this was done explicitly for agricultural land values and implicitly for most other impacts. The case for doing so is presented below. While we feel the case is strong, it admittedly lacks the absolutely imperative character of equal life valuation. It can be justified prescriptively or descriptively.
For every identified cause of damage, a lower figure is given by Fankhauser for the impact on the Third World. To take but one example: the loss of a hectare of Chinese wetland is assessed as bearing a cost of just 10% of that of an OECD hectare. One of the stated reasons for differentiation, in this case, is the assigning of a much higher value to the loss of recreational use in the OECD than in the South. We find this ethically indefensible. Once wetland has gone, it has probably gone for many decades or centuries, if not for ever. Why should the future Chinese be assumed not to need wetlands as much as future Americans, whether for recreation or for livelihoods? Clearly from an ethical standpoint, one country's hectare of wetland should be treated as worth the same as any other country's, and similarly for all the other damage categories. (We list the categories in Appendix C).
This leads to the question of whether to value all hectares of wetland, and other resources at risk, at a rate calculated on the basis of first-world damage costs, or on some global average basis. We argue that the former could be considered the appropriate basis on the following grounds:
Fankhauser uses a methodology which effectively assumes that in the year 2050 the international breakdown of world GWP will be the same as it was in 1988. By definition, this means that the existing average income disparity between the OECD and the LDCs will remain unaltered. Others, including Nordhaus (12) and Greenpeace, (13) posit a significant degree of income convergence between the OECD and the LDCs. Such convergence is also a widely-shared policy goal. On this latter view, damage valuation, even on a WTP basis, would likewise converge and adoption of the current OECD values as a world average for 2050 becomes more plausible.
Fankhauser's damage costs are calculated for a single year - which could be 2050 or 2060 - when it is assumed that the levels of CO2 and equivalents will have doubled in the atmosphere and global mean temperature will be 2.5 C above pre-industrial. Using this date as a reference point, he then expresses the damages in 1988 monetary values, thus giving a "snapshot" of potential future damage costs due to global warming for one year only.
"The time of CO2 doubling" has become an accepted benchmark for discussions in the climate-change field; apparently for reasons of ease of computation and comparison. However, this approach de-emphasises those factors, both natural and anthropogenic, that might well speed up or retard the time of doubling. It also diverts the focus of attention from the much more serious longer-term hazards, as was noted by Cline. (14)
The IPCC in its reference scenarios IS92 a,b,e and f forecast dates of doubling between 2050 and 2075. (15) Fankhauser has assumed that the instant of doubling would probably be around 2050 to 2060; Cline and Hohmeyer and Gaertner assumed around 2030. The latest results from the Hadley Centre (16) forecast a 0.2 C (approximately) per decade rise in temperature, reaching 1.8 C above pre-industrial levels by 2050, the end date of their published charts. Extrapolating from these figures would suggest 2085 as the date by which CO2 levels will have doubled, with 2.5 C the most likely temperature rise due to CO2 doubling. However the Hadley Centre forecasts that doubling will be reached at 2050 assuming there is no further increase in sulphur emissions.
These sulphur emissions come mainly from power stations, and we suggest that it is only prudent to make the stronger assumption that they decrease, rather than merely fail to increase. There are already international agreements to cut back on these emissions to check acid rain, and such action is quite likely to intensify. We suggest therefore, that it as advisable for climate change impact planning to expect that the existing aerosol cooling effect will in fact be further reduced.
Thus the fashion for concentrating on a time of CO2 doubling of about 2050 or 2060 seriously misleads the debate. On current trends, there is a real risk that CO2 concentrations may double much earlier. With rigorous policy measures, CO2 doubling could perhaps be avoided.
To face this very considerable policy challenge of averting climate change, what policy makers need to know is not just the range of best-guess damage estimates for the year of CO2 doubling from a group of Cost/Benefit Analysts. What policy makers need to know is what the range of accumulated damage is that is likely to occur across a firmly defined period of time. In other words, 2050 is only a suitable calendar reference point for policy makers, if the intention is to assess damages up to and including that point, recognising the non-linearity of climate change in its evolution to this point (and beyond) and the consequently vast unpredictability of damages within this time-frame.
The present "snap-shot" of 1.5% to 2.0% of GWP possibly being adopted into the Social Costs chapter of IPCC WG3 is spuriously precise and more generally, raises questions about the appropriateness of cost-benefit- analysis (CBA) as a policy tool for making decisions about climate change (see below).
The distribution of the cost estimates between the OECD and the rest of the world is unsound. Crucially, the method adopted by Fankhauser and Tol for calculating these estimates expressed as percents of GDP likely to occur in the LDCs, is based on what we and many others see as a basic arithmetical error. This error has a substantial influence on the present distributional results in the Social Costs assessment.
Because the Gross National Product (GDP) of individual countries is measured in the country's own currency, international comparisons require the use of a set of conversion factors. The set used universally, until very recently, was the Trading Exchange Rates (TERs). (G) This rates an Indian rupee at the number of dollars that it can buy on the international money exchanges. However, the TER typically fails to reflect, by a wide margin, the local purchasing power of that rupee.
For an average basket of goods and services, the bulk of which are produced locally, most LDC currencies are worth double the TER values. Some are worth five times more. So in the last few years tables have been published and adopted by among others the World Bank and the IMF giving Purchasing Power Parity (PPP) values for countries' currencies and for their GDPs.
These tables were perhaps not available to Nordhaus, Titus and Cline when they did their pioneering costings of climate change some years ago. They were, however, available to WG3 and, according to Fankhauser, Tol and Pearce, the damage costs - at least for the LDCs - are indeed corrected for PPP.
The arithmetic mistake then arises when PPP-corrected damage costs for the non-OECD countries are divided by their uncorrected TER GDP totals to deduce the percentage of GRP losses which are quoted. Although we have been told that this is the procedure adopted, it is nowhere explained in the text, though there is a footnote now in Summary for Policy-Makers (SPM) which refers to this. The effect of the erroneous arithmetic is to give quotable LDC damage percentages of GRP up to five times higher than they should have been. This gives a false credibility to the WTP-based assessment where in the figures currently quoted in the draft in billions of dollars are $180 for the OECD and $89 for the Rest of World (ROW). When the arithmetic is done correctly, the LDC percentage losses as a whole are approximately halved. (H) In our judgement it would be wholly inappropriate for IPCC to agree to the publication - in its name - of data which is derived from a method which is known - and admitted - to be wrong. (I)
Some potential positive feedback effects (including several identified by another IPCC working group, Working Group I - see Appendix C) were not taken into account in the literature reviewed by WG3, largely because they were not represented in most or all of the climate models. We argue that an allowance for the feedback mechanisms identified in WG1 must be made. In addition, the effects of removing certain pollutants from the atmosphere must also be taken into account. For example, an additional warming effect will occur if human-made sulphate aerosols are no longer present in the atmosphere in their present concentrations as a result of necessary efforts to curtail acid rain. Sulphate aerosols have a cooling effect and thus represent latent, committed warming, which will become actual very quickly once they are removed from the atmosphere. The effect of these aerosols is now being built into global circulation models; but this was not done in sufficient detail in the earlier models on which Fankhauser's and the others' cost estimates were based, because the WG1 report (17) quantifying the effect was only published recently. This could also bring forward the time of CO2 and equivalent doubling.
Also, the uncertainty described by IPCC Working Group 1 for the temperature rise to be expected from a given CO2 increase - normally referred to as the "climate sensitivity" - is seriously underplayed in the present WG3 results. The crucial summaries and tables ignore it. In addition, there are many other significant sources of uncertainty - indeed every factor under consideration is uncertain. Nonetheless, the costings are presented as point estimates, with no quantifiable indication by error bars, confidence intervals or otherwise of the range of uncertainty that accompanies them, although the text stresses the uncertainty qualitatively. It is stressed in the text of the SAR that the estimate of 1.5% to 2.0% of GWP is not an uncertainty estimate but simply a range, comprising the "best guesses" of the various authors. This distinction is likely to be lost on many readers and policymakers.
Every aspect of potential climate change impacts is beset with uncertainty. We feel it is of the greatest importance to represent this adequately within any summary results. There are different types of uncertainty, which can be classified as follows:
For example, economic growth rates; CO2, SOx and other emissions; population growth rates. In our own analysis we assume that economic growth and CO2 emissions follow a trajectory along the lines of the IPCC's IS92a scenario. CO2 emissions are very closely linked to economic growth, as so much economic activity is dependent on fossil fuels. However, sulphur emissions come from point sources and are therefore separately controllable. And we consider it is now important to explicitly take account of that, independently of IS92a.
There is no guarantee current carbon sinks will continue to absorb, as they do now, about half of worldwide CO2 emissions. There is also great uncertainty about sources, sinks and atmospheric concentration changes in the minor greenhouse gases such as methane and nitrous oxide. We do not address these points here, though there is certainly scope for unpleasant surprises.
There is uncertainty about the impact of increased CO2 (and other greenhouse gas) concentrations in the atmosphere on the climate.
This is the climate sensitivity, identified by the IPCC in 1990 and confirmed by their 1992 and 1994 reports. It is the proposition that the global average warming to be expected from CO2 and equivalent doubling is most likely to be 2.5 C, but might be between 1.5 C and 4.5 C.
Since the effects of sulphate emissions are localised, they are much harder for climatologists to model. Only recently, in 1995 publications by the Hadley research Centre and others, have they have been quantified in any useful way.
Several positive feedback mechanisms are likely to exist which could mean that, once temperatures begin to rise, factors will come into play beyond those which have been explicitly modelled and this will cause temperature to rise faster than the GCMs (General Circulation Models) predict. These factors bring forward in time the expected instant of CO2 doubling and therefore bring forward the time of the expected temperature rise or damages. And they do increase the damages we should expect at our chosen time of 2050, and the damages to be expected per tonne of emitted CO2. Just because they are not well understood or quantified does not mean that the positive feedbacks should be ignored, as the almost universal focus on costs at the time of doubling has ensured. (J)
Finally, there is uncertainty in the magnitude of each of the damage categories identified by Fankhauser. Most significantly, the damage costs are extremely sensitive to the surmised death rate, which has been predicted largely on the basis of a study by Kalkstein (18) into the effects of a 4 C rise on the inhabitants of fifteen US cities, and a series of extrapolations.
We combine these uncertainties using simple statistical methods, also explained in Appendix A. The main effect of the excessive simplicity in our statistics is likely to be to lead us to state incorrectly low combined uncertainty values, due to the assumptions of independence which we make.
Fankhauser does not assume that the dozen or so damage categories he uses (see Appendix C) are a reliable guide to all the untoward impacts of climate change. Nevertheless that is how his work and that of WG3 may well be interpreted. Many areas of expected damage are omitted - i.e. costed at zero - due to inadequate knowledge. For example, he only costs deaths due to heat stress and storms, not to disease or other indirect effects, though the text of Chapter 6 of the SAR asserts that indirect health effects "could far exceed direct effects". The very considerable "costs of acclimatisation" are not obviously quantified even though they are identified. In particular cost estimates were given in 1992 by Hohmeyer and Gaertner for the increased incidence of malnutrition and of malaria which far exceeded the direct costs.
As reported in New Scientist (13th May 1995), recent research by scientists at the Tropical Vegetation Monitoring Unit of the European Commission's Joint Research Centre at Ispra, Italy strongly supports the view that malaria will spread far beyond its present range. Widespread debilitation and increased mortality would result across much of the densely inhabited northern temperate zones whose populations have no natural immunity.
Hohmeyer and Gaertner have suggested 10 million extra cases of malaria worldwide by the time of doubling, (which they expect in 2030). We translate this to be a rate of 500,000 extra cases per year. In the absence of evidence to hand, we split this between the OECD and the LDCs in proportion to their population. Hopefully better estimates will become available shortly. To estimate a suitable WTAC-based cost, we asked a small sample of UK citizens unconnected with GCI or other environmental group what lump-sum compensation they would be willing to accept for the increased risk of malaria and received replies ranging from GBP 5,000 to GBP 1,000,000, with the most often chosen value being GBP 50,000 and the median somewhat higher. To be conservative, we have used the figure of GBP 50,000 (i.e. $75,000). At 5% of the value ascribed to a life this seems consistent. In addition, Hohmeyer and Gaertner suggest 0.5% mortality is likely (of the 10 million total cases, not of the 500,000 annual increase), that is another 50,000 deaths per year.
Another extra cost which we feel it is important to incorporate is an estimate of the cost of forced migration to the migrant. Tol does include such a cost in his work - at a rate of three times the migrant's average annual income - but Fankhauser does not, costing migration only insofar as it induces costs in the host nation. We use a rather smaller figure than for malaria, $50,000 or 3% of the value of a life. This is approximately consistent with Tol for OECD countries.
Finally we add in the largest cost identified by Hohmeyer and Gaertner - death through malnutrition, a factor not quantified by Fankhauser. Hohmeyer and Gaertner forecast at least 10 million deaths per year - a very high number but only a doubling, according to them, of the present level. Remaining conservative, we use half this figure as our best guess, so that the high end of our forecast range will be their figure of 10 million.
We do not claim that these extra damage categories are all-there will be others which are even harder to quantify or have simply not been thought of: remember that no-one forecast polar ozone holes when the debate on CFCs and ozone depletion was starting. So this means that our estimate, like all others, is more of a lower bound than a forecast.
Our re-analysis of data on costs is set out in Table A. We show the effect of our differing assumptions on Fankhauser's costings in a series of stages, represented by the columns of the table. We also show the costs in PPP (1988) US dollars and as proportions of GRP (K) and of GWP. (L)
Column F of Table A gives our estimates of damage costs, which range up to over 130% of GWP, many times higher than the costs estimated by Fankhauser. The discrepancy results from the extra cost categories (based largely on Hohmeyer and Gaertner's work) which we have taken into account, in particular malnutrition-related deaths. Even if these deaths are not incorporated into the calculations, however, our damage costs range up to 37% of GWP (see Column E). For the LDC region, high-end costs go up to over 250% of GRP reflecting the high impact of deaths costed at OECD rates. At the low end of our overall uncertainty range, on the other hand, global costs could be as little as 1.3% of GWP without the allowance for malnutrition-related deaths or 12.5% with it.
| Basis of cost estimate | A | B | C | D | E | F
|
||
|---|---|---|---|---|---|---|---|---|
| $(1988)Billions | OECD | low | $32 | $55 | $55 | $55 | $72
|
|
| medium | $181 | $181 | $325 | $325 | $325 | $387
|
||
| high | $1,100 | $1,741 | $1,741 | $1,741 | $1,916
|
|||
| LDCs | low | $16 | $27 | $58 | $221 | $2,365
|
||
| medium | $89 | $89 | $160 | $514 | $1,217 | $10,830
|
||
| high | $546 | $868 | $3,724 | $6,098 | $25,614
|
|||
| WORLD | low | $48 | $82 | $114 | $276 | $2,437
|
||
| medium | $270 | $270 | $485 | $838 | $1,542 | $11,217
|
||
| high | $1,646 | $2,609 | $5,465 | $7,839 | $27,530
|
|||
| %s Regional GNP (PPP) | OECD | low | 0.3 | 0.5 | 0.5 | 0.5 | 0.6
|
|
| medium | 1.6 | 1.6 | 2.9 | 2.9 | 2.9 | 3
|
||
| high | 10 | 15 | 15 | 15 | 17
|
|||
| LDCs | low | 0.2 | 0.3 | 0.6 | 2.3 | 24
|
||
| medium | 0.9 | 0.9 | 1.6 | 5 | 12 | 111
|
||
| high | 6 | 9 | 38 | 63 | 263
|
|||
| WORLD | low | 0.2 | 0.4 | 0.5 | 1.3 | 12
|
||
| medium | 1.3 | 1.3 | 2.3 | 4.0 | 7 | 53
|
||
| high | 8 | 12 | 26 | 37 | 131
|
|||
| %s Global GNP (PPP) | OECD | low | 0.2 | 0.3 | 0.3 | 0.3 | 0.3
|
|
| medium | 0.9 | 0.9 | 1.5 | 1.5 | 1.5 | 2
|
||
| high | 5.2 | 8 | 8 | 8 | 9
|
|||
| LDCs | low | 0.1 | 0.1 | 0.3 | 1.0 | 11
|
||
| medium | 0.4 | 0.4 | 0.8 | 2.4 | 6 | 51
|
||
| high | 2.6 | 4 | 18 | 29 | 122
|
|||
| WORLD | low | 0.2 | 0.4 | 0.5 | 1.3 | 12
|
||
| medium | 1.3 | 1.3 | 2.3 | 4 | 7 | 53
|
||
| high | 8 | 12 | 26 | 37 | 131
|
|||
The critique we have made in this paper raises wider questions about he validity of using CBA and related techniques as tools for policy making. When WG3 was restructured in 1992, its terms of reference were broad, stressing the need for the assessment to be set in the context of "Sustainable Development" and even to take account of the "cross-cutting economic and other issues". The bulk of the work of WG3 since then has however, been carried out by economists with relatively little input from other disciplines. Thus little attention is paid to the 'other issues'.
As events unfolded, the original proposal broad discussion in WG3 "Assessing the Benefits of Responses to Climate Change" was transformed into an overwhelmingly market-valuation based assessment of global GDP losses, following the earlier work of Nordhaus, Cline, Pearce, Titus, Tol and Fankhauser.
Indeed, much of WG3's effort has been in practice an attempt to apply the technique of Cost-Benefit Analysis (CBA). CBA works very well in microeconomic decision-making, and comes naturally to economists and businessmen, but is generally very unsuitable in national and international affairs. It has not featured, for example, in the fairly successful negotiations, starting at Montreal, on ozone depletion and CFCs. CBA methods are inevitably biased towards the rich, and there is a well- documented history of conflict aggravation (rather than resolution) between winners and losers assessed with it. An excellent summary of this is given by Adams. (19) Here we summarise some of the major problems with CBA, particularly with regard to the climate debate.
If not CBA, then what? The techniques of multicriteria analysis (MCA) and decision analysis, mentioned in the text of the SAR but ignored in the conclusions, might help. MCA however in practice usually, and as described in the SAR, ends up by combining the different criteria into a single weighted value, and so seems essentially equivalent to CBA. Tol uses Decision Analysis, (20) but his use of advanced statistical techniques but the conclusions of this part of his work do not appear to be reflected in the SAR. Funtowicz and Ravetz (21) call for ethics-based methods that do not rely on monetary valuations. Adams says, and we agree, that 'We are stuck with the messy and protracted process of argument, discussion, negotiation and compromise. The skills in shortest supply are not economic, but scientific and diplomatic'. In effect, CBA needs to be abandoned. Instead we need to revert to old-fashioned, if difficult, political negotiations based on a proper use of the precautionary principle and on a realistic assessment of a range of possible futures.
The authors gratefully acknowledge the assistance of many people in the production of this article, though responsibility for the contents rests solely with the authors.
Curtis Bohlen, John Adams, Chakravarthi Raghavan, Ian Douglas, Helena Paul, Peter Bunyard, Colin Price, Richard Douthwaite, Titus Alexander, Jim Berreen, Paul Ekins and Bill Hare and Alan McGowen.
In addition Sam Fankhauser and Richard Tol sent us extensive and very pertinent comments on an earlier draft, for which we were extremely grateful.
Appendices: