The first type of issue would take the evaluation towards an overall loss in the region of 13-15% (using the 10.9% figure of Table 6.1 and scaling up by one-quarter or more for distribution). The second type of issue would lead to taking a weighted average somewhere between this figure (13 or 14%) and 20%. The weights would depend on crude judgements about likelihoods of different kinds of probability distributions, on judgements about the severity of losses in this context, and on the basic degree of cautiousness on the part of the policy-maker. Together, they would make up the "aversion to ambiguity" discussed in Chapter 2 and the Appendix.
This discussion points to areas for further work in the context of this particular model: distribution within a generation and explaining different distributional judgements. Of course, there is much more to do in terms of considering different economic models – we have investigated just one – and exploring different probability distributions.
Conclusion
This Chapter has presented global cost estimates of the losses from "business as usual" climate change. They have been expressed in terms of their equivalent permanent percentage loss in consumption. They are averages over time and risk and can be compared with percentage costs, similarly averaged over time, of mitigation – that is the subject of Part III of this Review. In the final chapter of that part, we include a discussion of how much of the losses estimated in this Chapter could be saved by mitigation. The loss estimates of this Chapter should be viewed as complementary to the discussions of the scale of the separate impacts on consumption, health, and environment that were presented in Chapters 3 to 5.
What have we learned from this exercise? Notwithstanding the limitations inherent in formal integrated models, there can be no doubt that the economic risks of a "business as usual" approach are very severe – and probably more severe than suggested by past models. Relying on the scientific knowledge that informed the IPCC"s TAR, the cost of BAU climate change over the next two centuries is equivalent to a loss of at least 5% of global per-capita consumption, now and forever. More worrying still, when the model incorporates non-market impacts and more recent scientific findings on natural feedbacks, this total average cost is pushed to 14.4%.
Cost estimates would increase still further if the model incorporated other important omitted effects. First, the welfare calculations fail to take into account distributional impacts, even though these impacts are potentially very important: poorer countries are likely to suffer the largest impacts. Second, there may be greater risks to the climate from dynamic feedbacks and from heightened climate sensitivity beyond those included here. If these are included, the total cost would be likely to be around 20% of current per-capita consumption, now and forever.
Further, there are potentially worrying "social contingent" impacts such as migration and conflict, which have not been quantified explicitly here. If the world"s physical geography is changed, so too will be its human geography.
Finally, we must close with the warning about over-literal interpretation of these results with which we began this chapter. The estimates have arisen from an attempt to add two things to the previous literature on IAM models. The first is use of recent scientific estimates of probabilities and the second is putting these probabilities to work using the economics of risk and uncertainty. The most worrying possible impacts are also among the most uncertain, given that so little is known about the risks of very high temperatures and potential dynamic instability. The exercise allows us to see what the implications of the risks, as we currently understand them, might be. The answer is that they would imply very large estimates of potential losses from climate change. They give an indication of the stakes involved in making policy on climate change. The analysis of this Chapter shows the inevitable difficulties of all these models in extrapolating over very long periods of time. We therefore urge the reader to avoid an over-literal interpretation of these results. Nevertheless, we think that they illustrate a very important point: the risks involved in a 'business as usual' approach to climate change are very large.
Notes: 1 All changes in global mean temperature are expressed relative to pre-industrial levels (1750 – 1850). A temperature rise of 1°C represents the range 0.5 – 1.5°C, a temperature rise of 2°C represents the range 1.5 – 2.5°C etc.
2 "Extreme events" occur when a climate variable (e.g. temperature or rainfall) exceeds a particular threshold, e.g. two standard deviations from the mean.
3 In looking at the effects on crop yields of severe weather during the Little Ice Age, Prof Martin Parry (1978) argued that the frequency of extreme events would change dramatically as a result of even a small change in the mean climate and that the probability of two successive extremes is even more sensitive to small changes in the mean. Often a single extreme event is easy to withstand, but a second in succession could be far more devastating. In a follow-up paper, Tom Wigley (1985) demonstrated these effects on extremes mathematically.
4 Based on a technical paper prepared for the Stern Review by Challinor et al. (2006b) 5 This will also depend on efficiency of use as well.
6 Special Issue of Global Environmental Change, Volume 14, April 2004 – further details on the new analysis are available from Warren et al. (2006). Risk and uncertainty are often used interchangeably, but in a formal sense, risk covers situations when the probabilities are known and uncertainty when the probabilities are not known.
7 See, for example, Arnell (2006a) 8 IPCC (2000) 9 In 1990 US $ 10 Problematic as based on Market Exchange Rates 11 For example, many integrated assessment models – details in Chapter 7 27 Based on the work of Falkenmark et al. 1989, water availability per person per year is the most frequently used measure of water resource availability. The UN has widely adopted this measure, for which data are readily available. The next most frequently used measure is the ratio of withdrawals to availability, but this requires reliable estimates of actual and, most crucially, future withdrawals.
28 Based on work of Gleick (1996). Actual usage varies considerably, depending on water availability, price, and cultural preferences (domestic consumption in UK is around 170 L per person per day; in large parts of Africa it is less than 20 L per person per day).
29 FAO World Agriculture report (Bruinsma 2003 ed.) 30 Plants also develop faster at warmer temperatures such that the duration from seedling emergence to crop harvest becomes shorter as temperatures warm, allowing less time for plant growth. This effect varies with both species and cultivar. With appropriate selection of cultivar, effective use of the extended growing season can be made.
31 Previous crop studies use a quadratic functional form, where yields are increasing in temperature up to an "optimal" level when further temperature increases become harmful (for example Mendelsohn et al. 1994). A crucial implicit assumption behind the quadratic functional form is symmetry around the optimum: temperature deviations above and below the "optimal" level give equivalent yield reductions. However, recent studies (e.g. Schlenker and Roberts 2006) have shown that the relationship is highly asymmetric, where temperature increases above the "optimal" level are much more harmful than comparable deviations below it.
This has strong implications for climate change, as continued temperature increases can result in accelerating yield reductions.
32 Evidence reviewed in Slingo et al. (2005); Ciais et al. (2005) 33 The impacts depend crucially on the distribution of warming over land (Chapter 1). In general, higher latitudes and continental regions will experience temperature increases significantly greater than the global average. For a global average warming of around 4°C, the oceans and coasts generally warm by around 3°C, the mid-latitudes warm by more than 5°C and the poles by around 8°C.
34 Warren et al. (2006) have prepared this analysis, based on the original work of Parry et al. (2004). More detail on method and assumptions are set out in Box 3.4. Production declines less than yields with increasing temperature because more land area at higher latitudes becomes more suitable for agriculture.
35 New analysis by Long et al. (2006) showed that the high-end estimates (25 – 30%) were largely based on studies of crops grown in greenhouses or field chambers, while analysis of studies of crops grown in near-field conditions suggest that the benefits of carbon dioxide may be significantly less, e.g. no more than half.
36 The optimum temperature for crop growth is typically around 25 – 30°C, while the lethal temperature is usually around 40°C.
37 Other staple crops in Africa (millet and sorghum) are also relatively unresponsive to the carbon fertilisation effect. They all show a small positive response because they require less water to grow.
38 Types of adaptation discussed by Parry et al. (2005) 39 For example Fischer et al. (2005) 40 These effects are not yet routinely considered in climate models or impacts studies (Betts 2005).
41 According to Parry et al. (2004) people at risk of hunger are defined as the population with an income insufficient either to produce or procure their food requirements, estimated by FAO based on energy requirements deduced from an understanding of human physiology (1.2 – 1.4 times basal metabolic rate as minimum maintenance requirement to avoid undernourishment).
42 Links between changes in income and mortality are explored in Chapter 5.
43 Warren et al. (2006) have prepared these results, based on the original analysis of Parry et al. (2004) (more details in Box 3.6). These figures assume future socio-economic development, but no carbon fertilisation effect. There is likely to be some positive effect of rising levels of carbon dioxide (if not as much as assumed by most studies).
44 Turley et al. (2006) – Ocean pH has changed by 0.1 pH unit over the last 200yrs. As pH is on a log scale, this corresponds to a 30% increase in the hydrogen ion concentration, the main component of acidity.
45 Royal Society (2005) – a drop of 0.15 pH units corresponds to a 40% increase in the hydrogen ion concentration, the main component of acidity. A drop of 0.3 pH units corresponds to a doubling of hydrogen ion concentration.
46 Comprehensively reviewed by Patz et al. (2005) 47 Average life expectancy at birth has increased by 20 years since the 1960s. But in parts of Africa life expectancy has fallen in the past 20 years because of the HIV/AIDS pandemic (McMichael et al. 2004).
48 De et al. (2005) 49 See Tol (2002) for indicative figures for different OECD regions 50 Based on detailed analysis by McMichael et al. (2004), using existing quantitative studies of climate-health relationships and the UK Hadley Centre GCM (business as usual emissions) to estimate relative changes in a range of climate-sensitive outcomes, including diarrhoea, malaria, dengue fever and malnutrition. Changes in heat- and cold-related deaths were not included in the aggregate estimates of mortality. Climate change contributes 2% to today"s climate disease burden (6.8 million deaths annually) and 0.3% to today"s total global disease burden.
51 Projections from Patz et al. (2005) 52 See, for example, Tol (2002) and Bosello et al. (2006) 53 As described earlier, today 800 million people are at risk of hunger and around 4 million of those die from malnutrition each year. Once temperatures increase by 3°C, 200 – 600 million additional people could be at risk (with little carbon fertilisation effect), suggesting 1 – 3 million more dying each year from malnutrition, assuming that the ratio of risk of hunger to mortality from malnutrition remains the same. This ratio will of course change with income status – see Chapter 4 for more detail.
54 The impacts on human development mediated through changes in income are explored in Chapter 4.
55 Calculations from Warren et al. (2006) based on research from Tanser et al. (2003), using one of only two models which has been validated directly to account for the observed effect of climate variables on vector and parasite population biology. They assume no increase in population size in the future or change in vulnerability (through effective treatment/prophylaxis). While this assumption of no change in control efforts is not realistic, the results illustrate the potential scale of the problem. The study used the Hadley Centre climate model to estimate regional temperature and rainfall patterns; other models produce different rainfall patterns and therefore may result in different regional patterns for malaria.
56 Calculations from Warren et al. (2006) based on research from Van Lieshout et al. (2004), who take into account future population projections and used the Hadley Centre climate model. Similar to Tanser et al. (2003), they use the Hadley Centre model and find an increase in malaria exposure in Sub-Saharan Africa, but with slightly fewer people affected (50 million rather than 80 million for a 4°C temperature rise) because of different assumptions about rainfall thresholds.
57 Malaria in Africa is particularly difficult to control because of the large numbers of mosquitoes spreading the disease, their effectiveness as transmitting the disease, and increasing drug resistance problems. Alternatives can be very effective, but are often much more expensive (WHO 2005).
58 Hales et al. (2002) used a vector-specific model coupled to outputs of two climate models. Their estimates take account of projected population growth to the 2080s, but not any control measures.
59 Reviewed in Epstein and Mills (2005) 60 More detail in Chapter 1 61 See Chapter 6 for discussion of implications for global trade 62 Ali (2000) 63 Munich Re (2005) 64 Increased storm intensity could cause similar impacts and will exacerbate the effects of sea level rise – these effects are not included in the impact estimates provided here (see Chapter 6).
65 Warren et al. (2006) have prepared these results, based on the original analysis of Nicholls (2004), Nicholls and Tol (2006) and Nicholls and Lowe (2006) for impacts of sea level rise on populations in 2080s with and without climate change. More details on method are set out in Figure 3.8. "Average annual people flooded" refers to the average annual number of people who experience episodic flooding by storm surge, including the influence of any coastal protection. In some low-lying areas without protection, 66 International Federation of Red Cross and Red Crescent Societies (2001) 67 Myers and Kent (1995) 68 Emanuel (1987) 69 In fact Nordhaus (2006) found that economic damages from hurricanes rise as the ninth power of maximum wind-speed, perhaps as a result of threshold effects, such as water overtopping storm levees.
70 Munich Re (2006) 71 Nelson (2003) 72 Root et al. (2005); Parmesan and Yohe (2003) 73 Arctic Climate Impacts Assessment (2004) 74 Pounds et al. (2006) 75 Ricketts et al. (2005) 76 For example, fast-growing tropical tree species show greater growth enhancements with increased carbon dioxide concentrations than slower-growing species and could gain a dominant competitive advantage in tropical forests in the future (Körner 2004).
77 Reviewed in detail in Hare (2006). These figures are likely to underestimate the impacts of climate change, because many of the most severe effects are likely to come from interactions with factors not taken into account in these calculations, including land use change and habitat fragmentation/loss, spread of invasive species, new pests and diseases, and loss of pollinators. In addition, ecosystem assessments rarely consider the rate of temperature change. It is likely that rates of change exceeding 0.05 – 0.1°C per decade (regional temperature) are more than most ecosystems can withstand, because species cannot migrate polewards fast enough (further details in Warren 2006).
78 According to the Arctic Climate Impacts Assessment (2004), Arctic ecosystems will be strongly affected by climate change as temperatures here are rising at close to double the global average.
79 Thomas et al. (2004a) – these (and subsequent) estimates of extinction risk are based on calculations of decreases in the availability of areas with suitable climate conditions for species into the future. As suitable areas to support a certain level of biodiversity disappear, species become "committed to extinction" when the average rate of recruitment of adults into the population is less than the average rate of adult mortality. There is likely to be a lag in response depending on the life span of the species in question – short-lived species rapidly disappear from an area while long-lived species can survive as adults for several years. There is a great deal of uncertainty inherent in such estimates of extinction risk (Pearson and Dawson 2003) and alternative modelling approaches have been shown to yield different estimates (Thuiller et al. 2004, Pearson et al. 2006). However, other studies looking at climate suitability also predict high levels of extinction, for example McClean et al. (2005) predict that 25 – 40% of African plan species will lose all suitable climate area with 3°C of warming globally.
80 Coral bleaching describes the process that occurs when the tiny brightly coloured organisms that feed the main coral (through photosynthesis) leave the skeleton because they become heat-stressed. Bleached corals have significantly higher rates of mortality.
81 Donner et al. (2005) 82 Leemans and Eichkout (2004) 83 Malcolm et al. (2006) 84 This effect has been found with the Hadley Centre model (Cox et al. 2000) and several other climate models (Scholze et al. 2006).
85 Visser and Both (2005); Both et al. (2006) report declines of 90% in pied flycatcher populations in the Netherlands in areas where caterpillar numbers have been peaking two weeks earlier due to warming, which means there is little food when the flycatcher eggs hatch.
86 For example, Rial et al. (2004) 87 As set out in a Pentagon commissioned report by Schwartz and Randall (2004) 88 Schellnhuber (2006) 89 Nicholls et al. (2004) 90 Kumar et al. (2006) 91 Gupta et al. (2003) 92 Collins and the CMIP Modelling Groups (2005) 93 Defra (2005) 94 Challinor et al. (2006a) 95 Reviewed in detail in a report prepared for the Stern Review by Challinor et al. (2006b) 96 This is a result from Arnell (2006b), who superimposed rainfall and temperature changes from past extreme monsoon years (average over five driest and five wettest years) on today"s mean summer climate to understand consequences for water availability.
97 Described in detail in Munich Re (2006) 1 The physical effects of climate change are predicted to become progressively more significant by the 2050s with a 2 to 3°C warming, as explained in Chapter 3.
2 IPCC (2001). The classification of sensitivity is similar to susceptibility to climate change, the degree to which a system is open, liable, or sensitive to climate stimuli.
3 Intra-annual variability refers to rainfall concentrated in a single season, whilst interannual variability refers to large differences in the annual total of rainfall. The latter may be driven by phenomena such as the El Nino/Southern Oscillation (ENSO) or longer-term climate shifts such as those that caused the ongoing drought in the African Sahel. Brown and Lall (2006) 4 De et al (2005) 5 Challinor et al (2006). The scale of losses in the agricultural sector is indicated by the fact that this sector contributed just over one fifth of GDP at the time.
6 Nordhaus (2006). Approximately 20% of the difference in per capita output between tropical Africa and two industrial regions is attributed to geography according to Nordhaus" model and analysis.
7 Sachs (2001a) 8 Mendelsohn et al (2006) 9 World Bank (2006a) using 2004 data 10 ILO (2005). The employment figures are given as a share of total employment, 2005.
11 For example, the Central African Republic derives more than 50% of its export earnings from cotton alone (1997/99). Commission for Africa (2005) 12 Natural medicines, for example, are often the only source of medicine for poor people and can help reduce national costs of supplying medical provisions in developing countries. The ratio of traditional healers to western-trained doctors is approximately 150:1 in some African countries for example. UNEP-WCMC (2006) 13 Vedeld et al (2004). This effect on the Amazon has been found with the Hadley Centre model, as reported in Cox et al. (2000), and several other climate models (Scholze et al. 2006) as discussed in Chapter 3.
14 World Bank (2003b) 15 Stern et al (2005) 16 World Population Prospects (2004); and World Urbanization Prospects (2005).
17 Hewawasam (2002) 18 For example, proximity and economies of scale enable cost-effective and efficient targeting and provision of basic infrastructure and services.
19 Approximately 72% of Africa"s urban inhabitants now live in slums and squatter settlements for example (Commission for Africa, 2005) 20 WHO (2005). Poverty impacts a person"s standard of living, the environmental conditions in which they live, and their ability to meet basic needs such as food, housing and health care that in turn affects their level of nutrition.
21 One of the MDGs is to halve, between 1990 and 2015, the proportion of people who suffer from hunger. In 2002 there were 815 million hungry people in the developing world, 9 million less than in 1990. (UN, 2005) 22 Irrigation plays an important role in improving returns from land, with studies identifying an increase in cropping intensity of 30% with the use of irrigation (Commission for Africa, 2005). Similarly, effective water management enables water to be stored for multiple uses, increases the reliability of water services, reduces peak flows and increases off-peak flows, and reduces the risk of water-related shocks and damage (World Bank, 2006b).
23 World Bank (2006c) 24 Brown and Lall (2006) 25 International Commission on Irrigation and Drainage (2005) 26 Roy (2006) 27 An estimated 2.5 billion low income people globally do not have access to bank accounts, with less than 20% of people in many African countries having access (compared to 90-95% of people in the developed world) (CGAP, 2004). Poor people are typically constrained by their lack of collateral to offer lenders, unclear property rights, insufficient information to enable lenders to judge credit risk, volatile incomes, and lack of financial literacy, among other things.
28 The incomes of poor people will become less predictable, making them less able to guarantee the returns that are needed to pay back loans, while insurers will face higher risks and losses making them even less willing to cover those most in need.
29 IMF (2003) 30 Freeman et al (2002) 31 For example, a recent study from the Hadley Centre shows that the proportion of land experiencing extreme droughts is predicted to increase from 3% today to 30% for a warming of around 4°C, and severe droughts at any one time will increase from 10% today to 40% (discussed in Chapters 1 and 3).
32 Data extracted from Munich Re (2004). These figures are calculated on the basis of the occurrence and consequences of "great natural disasters". This definition is in line with that used by the United Nations and includes those events that over-stretch the ability of the affected regions to help themselves. As a rule, this is the case when there are thousands of fatalities, when hundreds of thousands of people are made homeless or when the overall losses and/or insured losses reach exceptional orders of magnitude. While increases in wealth and population growth account for a proportion of this increase, it cannot explain it all (see Chapter 5 for more details). The losses are given in constant 2003 values.
33 The true cost of disasters for developing countries is often undervalued. Much of the data on the costs of natural disasters is compiled by reinsurance companies and focused on economic losses rather than livelihood losses, and is unlikely to capture the effect of slow-onset and small-scale disasters and the impact these have on households. Furthermore, the assessments typically do not capture the cumulative economic losses as they are based on snapshots in time. Benson and Clay (2004) 34 IMF (2003) 35 Warren et al (2006) 36 Huq and Reid (2005) 37 Alderman et al (2003) 38 Wagstaff and van Doorslaer (2003) 39 These results were estimated after controlling for initial poverty, economic policy, tropical location and life expectancy (using different time frames). Sachs and Gallup (2001) 40 Carter et al (2004) 41 Vos et al (1999) 42 IMF (2003). This was largely due to the higher price of food that had to be imported following a drought induced reduction in agricultural output, as described in Box 4.2, coupled with an increase in inflation to 46%.
43 Commission for Africa (2005) 44 Strzepek et al (2001) 45 Cited in Nkomo et al (2006) 46 This refers to a minimum asset threshold beyond which people are unable to build up their productive assets, educate their children and improve their economic position over time. Carter et al (2005) 47 Dercon (2003). Households with an average livestock holding in Tanzania were found to allocate 20% less of their land to sweet potatoes than a household with no liquid assets, with the return per adult of the wealthiest group being 25% higher for the crop portfolio compared to the poorest quintile.
48 Hicks (1993) 49 A household survey in eight peasant associations in Ethiopia found that distressed sales of livestock following the drought in 1999 sold for less than 50% of the normal price. Carter et al (2004) 50 People can be pushed below a critical nutritional level whereby no productive activity is possible, with little scope for recovery given dependence on their own labour following the loss or depletion of their physical assets. Dasgupta and Ray (1986) 51 Hoddinott (2004) 52 IMF (2003); World Bank (2006a) 53 Benson and Clay (2004). Similarly the climatically less severe 1994/95 drought involved costs of US$1 billion in cereal losses (due to higher prices in a tighter international cereal market).
54 ODI (2005) 55 IMF (2003) 56 Information based largely on Challinor et al (2006). See also Roy (2006) 57 As ever it is difficult to attribute an outside event to climate change but the evidence is strong that the severity of such events is likely to increase.
58 Challinor et al (2006). 70% was the maximum reduction in yield that came from the study, in northern regions. Reductions in the 30-60% range were found over much of India. Strictly speaking these results are for groundnut only, although many annual crops are expected to behave similarly. The study was based on an SRES A2 scenario. The values assume no adaptation.
59 Challinor et al (2006) 60 Information based largely on Nkomo et al (2006) 61 The regions at risk of climate change were identified by looking at the possibility of losses in length of growing period that was used as an integrator of changing temperatures and rainfall to 2050. This was projected by downscaling the outputs from several coupled Atmosphere-Ocean General Circulation Models for four different scenarios of the future using the SRES scenarios of the IPCC. Several different combinations of GCM and scenario were used. The vulnerability indicator was derived from the weighted sum of the following four components: 1) public health expenditure and food security issues; 2) human diseases and governance; 3) Human Poverty Index and internal renewable water resources; and 4) market access and soil degradation. (Thornton et al, 2006) 62 Cited in Warren et al (2006) based on the original analysis of Parry et al. (2004). These figures assume future socio-economic development, but no carbon fertilisation effect, as discussed in Chapter 3.
63 McClean et al (2005). This is estimated using the Hadley Centre third generation coupled ocean-atmosphere General Circulation Model.
64 van Lieshout et al (2004) 65 Republic of South Africa (2000) cited in Nkomo et al (2006) 66 Strzepek et al (2001) 67 Gambia (2003) and Republic of Kenya (2002) cited in Nkomo et al (2006) 68 Information based on Nagy et al (2006) 69 El Nino-Southern Oscillation events (as discussed in Chapter 1).
70 Jones and Thornton (2003), cited in Nagy et al (2006) 71 Information based on Erda and Ji (2006) 72 Tang Guoping et al (2000) 73 NBSC (2005) 74 Warren et al (2006) 75 Warren et al (2006) 76 Increased agricultural productivity has been identified as a key factor in reducing poverty and inequality. This is based on work undertaken by Bourguignon and Morrisson (1998) using data from a broad sample of developing countries in the early 1970s and mid 1980s. Evidence from Zambia, for example, suggests that an extra US$1.5 of income is generated in other businesses for every $1 of farm income. Hazel and Hojjati (1995). Similarly, Block and Timer (1994) estimated an agricultural multiplier in Kenya of 1.64 versus a non-agricultural multiplier of 1.23 in Kenya.
77 Cited in Roy (2006) 78 World Bank (2006c) 79 World Bank (2006c). The model shows growth projections dropping 38% when historical levels of hydrological variability are assumed, relative to the same model"s results when average annual rainfall is assumed in all years. Hydrological variability included drought, floods and normal variability of 20% around the mean.
80 This model picks up the aggregate impacts of climate change on a range of market sectors such as agriculture. The estimates used in this analysis are based on the impact of climate change on market sectors. PAGE2002 allows examination of either market impacts only (as used here to ensure no double counting of poverty impacts) or market plus non-market impacts. These estimates and further details on the PAGE2002 model are given in Chapter 6.
81 The baseline-climate-change scenario is based largely on scientific evidence in the Third Assessment Report of the IPCC, in which global mean temperature increases to 3.9°C in 2100 (see Chapter 6 for more detail).
82 Using the IPCC A2 SRES baseline 83 In the high-climate-change scenario, global mean temperature increases to 4.3°C in 2100. The high-climate- scenario is designed to explore the impacts that may be seen if the level of temperature change is pushed to higher levels through positive feedbacks in the climate system, as suggested by recent studies (see Chapter 1 and Chapter 6 for more detail).
84 Other factors – such as changes in income distribution – that may also affect poverty levels or child mortality are assumed to be constant.
85 Ravallion (2001) 86 Kraay (2005) 87 World Bank (2000) 88 The formulae express the level of poverty as a function of the poverty line, average household income and the distribution of income. The $2 poverty line is used throughout.
89 This figure is obtained from a cross-country regression of rates of growth in mean household expenditure per capita on GDP per capita. Ravaillion (2003) 90 It is important to note that income alone does not determine health outcomes, efficient public programmes and access to education for women are also important factors, for example. Furthermore, the way in which GDP per capita changes (for example if there is a change in the distribution of income that coincides with the change in national income) can affect the impact it has on health.
91 Analysis demonstrates the health effects today of slowing or negative per capita growth. For example, in 1990, over 900,000 infant deaths would have been prevented had developing countries been able to maintain the same rate of growth in the 1980s as in the period 1960-80 (assuming an elasticity of -0.4), rather than the slow or negative growth they in fact experienced. The effects were particularly significant in African and Latin America, where growth was lower by 2.5% on average (Pritchett and Summers, 1993).
92 The elasticity is assumed to be a constant across countries and over time, consistent with econometric evidence (such as Kakwani (1993)). However, the average elasticity of child mortality with respect to GDP over a period of time will typically not be the same as the actual elasticity that applies on a year-to-year basis, even if the latter is assumed constant, because of compounding.
93 Myers (2005) 94 Nicholls (1995) and Anwar (2000/2001) 95 Barnett and Adger (2003) 96Warren et al. (2006) analysing data from Nicholls (2004), Nicholls and Tol (2006) and Nicholls and Lowe (2006). This is calculated on the basis of the number of people that are exposed each year to storm surge elevation that has a one in a thousand year chance of occurring. These odds and the numbers explored could be rising rapidly. This has already been demonstrated in the case of heat waves in Southern Europe where the chance of having a summer as hot as in 2003 that in the past would be expected to occur once every 1000 years, will be commonplace by the middle of the century due to climate change, as discussed in Chapter 5.
97 Warren et al. (2006) based on the original analysis of Parry et al. (2004).
98 Warren et al. (2006) based on the original analysis of Arnell (2004) for the 2080s.
99 Miguel et al (2004), Collier and Hoeffler (2002), Hendrix and Glaser (2005) and Levy et al (2005) 100 University for Peace Africa Programme (2005) 101 For example, there are 20 plans in place to build large dams along the Niger River alone.
102 Niasse (2005) 103 Ethiopia, the Sudan, Egypt, Kenya, Uganda, Burundi, Tanzania, Rwanda, the Democratic Republic of Congo and Eritrea.
104 Strzepek et al (2001). Whilst there is general agreement regarding an increase in temperature with climate change that will lead to greater losses to evaporation, there is more uncertainty regarding the direction and magnitude of future changes in rainfall. This is due to large differences in climate model rainfall predictions.
105 Meier and Bond (2005) 106 AIACC (2005) 107 Niasse (2005) 108 Christian Aid (2006) 109 Tanzler et al (2002) 110 ODI (2005) 111 This takes into account farm size, inputs, hours worked etc. This is drawing on evidence from Malaysia, Ghana and Peru Information drawn from Birdsall (1992) 112 Chew and Ramdas (2005) 113 Chew and Ramdas (2005) 1 Tol et al. (2004) set out these arguments in some detail and with great clarity.
2 Projections for changes in rainfall patterns in developed countries are generally more reliable than those in developing countries (due to their higher latitude location).
3 Schröter et al. (2006) and Arnell (2004) 4 Hayhoe et al. (2006) 5 Preston and Jones (2006) 6 Using a general equilibrium model for the USA, Jorgenson et al. (2005) found that agriculture contributed 70 – 80% of the changes in GDP driven by climate change (more details later in chapter). This work did not include the costs of extreme weather, particularly infrastructure damage from hurricanes and storms.
7 Mendelsohn et al. (1994); see also Schlenker et al. (2005) for a recent critique of this work 8 COPA COGECA (2003) 9 Warren et al. (2006) have prepared these results, based on the original analysis of Prof Nigel Arnell (University of Southampton). Energy requirements are expressed as Heating Degree Days and Cooling Degree Days (more detail in Table 5.1).
10 MICE (2005) 11 Cayan et al. (2006) 12 Department of Health (2003) study for the UK found an increase in heat-related mortality by 2,000 and decrease in cold-related mortality by 20,000 by the 2050s using the Hadley Centre climate model.
13 Benson et al. (2000) report on studies in five US cities in the Mid-Atlantic region (Baltimore, Greensboro, Philadelphia, Pittsburgh and Washington DC) and find a net increase in temperature-related mortality of up to two- to three-fold by 2050 (using outputs from three global climate models). These cities see larger increases in summer heat-related mortality than some other cities in the USA.
14 Hamilton et al. (2005) 15 Preston and Jones (2006) 16 Suggested by Pew Center study by Jorgenson et al. (2005) 17 All impacts in the Arctic are clearly and comprehensively set out in the Arctic Climate Impacts Assessment (2004) 18 Nick Rowley and Josh Dowse of KINESIS Consulting, Sydney, Australia http://www.kinesis.net.au 19 Environment Agency (2006), McGregor et al. (2006) 20 O"Brien et al. (2006) 21 Environment Agency (2003) 22 Brookings Institution (2005) 23 Described by low frequency but high impact events (e.g. more than two standard deviations from the mean) 24 Hallegatte et al. (2006) define the "economic amplification ratio" as the ratio of the overall production losses from the disaster to its direct losses.
25 2005 prices for total losses (insured and uninsured) – analysis of data from Swiss Re and Munich Re in Mills (2005) and Epstein and Mills (2005); Munich Re (2006) 26 Muir-Wood et al. (2006) 27 Based on simple extrapolation through to the 2050s. The lower bound assumes a constant 2% increase in costs of extreme weather over and above changes in wealth and inflation. The upper band assumes that the rate of increase will increase by 1% each decade, starting at 2% today, 3% in 2015, 4% in 2025, 5% in 2035, and 6% in 2045. These values are likely underestimates: (1) they exclude "small-scale" events which have large aggregate costs, (2) they exclude data for some regions (Africa and South America), (3) they fail to capture many of the indirect economic costs, such as the impacts on oil prices arising from damages to energy infrastructure, and (4) they do not adjust for the reductions in losses that would have otherwise occurred without disaster mitigation efforts that have reduced vulnerability.
28 Stott et al. (2004) 29 Recent papers from Nordhaus (2006) and the Association of British Insurers (2005a) examined consequences of increased hurricane wind-speeds of 6% on loss damages, keeping socio-economic conditions and prices constant. Several climate models predict a 6% increase in storm intensity for a doubling of CO2 concentrations (close to a 3°C temperature rise). The insurance study used existing industry catastrophe loss models validated with historic events to predict future losses. The extreme event costs are defined from an event with a 0.4% chance of occurring (1 in 250 year loss).
30 Heck et al. (2006) 31 UK Government Foresight Programme (2004) calculations for flooding from rivers, the sea and flash-flooding in urban areas. Prof Jim Hall at the University of Newcastle has provided some additional analysis. Assumes no change in flood management policies.
32 Research from the Association of British Insurers (2005a) extrapolated from a UK-based study of flood losses that assumed no change in flood management policies beyond existing programme. Some of the increased cost is driven by economic growth of the century and greater absolute wealth in physical assets.
33 Preston and Jones (2006) 34 Hayhoe et al. (2006) 35 London Climate Change Partnership (2004) 36 Association of British Insurers (2004) estimates that subsidence costs to buildings could double by the middle of the century to £600 million (2004 prices).
37 Nicholls and Klein (2003) 38 Association of British Insurers (2005b) 39 As set out in a Pentagon commissioned report by Schwartz and Randall (2004) 40 A complete collapse of the Thermohaline Circulation is considered to be unlikely (but still plausible) this century (Chapter 1).
41 Vellinga and Wood (2002) 42 Salmon and Weston (2006) 43 Mills (2005) 44 Mills and Lecomte (2006) provide many examples of increasing prices or withdrawing cover in the US. For example, reinsurance prices have increased by 200% in some parts of the US. Commercial customers are also being affected by the availability and affordability of insurance. Allstate insurance dropped 16,000 commercial customers in Florida in 2005, and some commercial businesses in the Gulf of Mexico are unable to find insurance at any price.
45 Crichton (2006) found that today in the UK one-third of small and medium-sized businesses had any form of business interruption cover against extreme weather.
46 "Extreme" is defined by an insurers risk appetite and regulatory requirements.
47 Heck et al. (2006) 48 The fundamental drivers of past, current and future world migration are clearly set out by Hatton and Williamson (2002).
49 See, for example, Woodham-Smith (1991) 50 Brooks et al. (2005) 51 Shiva (2002) describes several examples of conflict within a nation or between nations that has been exacerbated by tensions over construction of dams to manage water availability. Every river in India has become a site of major, irreconcilable water conflicts, including the Sutlej, Yamuna, Ganges, Krishna and Kaveri Rivers. The Tigris and Euphrates Rivers, the major water bodies sustaining agriculture for thousands of years in Turkey, Syria and Iraq have led to several major clashes among the three countries.
The Nile, the longest river in the world, is shared by ten African countries and is another complicated site of water conflict, particularly following construction of the Aswan Dam.
52 For example, hurricane damages scale as the cube of windspeed (or more), which itself increases exponentially with ocean temperatures.
1 Cline (1992) 2 Ethical perspectives other than those embodied in the models below – such as the approaches based on rights and liberties, intergenerational responsibilities, and environmental stewardship discussed in Chapter 2 – also point towards focusing on the costs of climate change in terms of income/consumption, health, and environment.
3 Pearce et al. (1996) 4 Mendelsohn et al. (1998) 5 Tol (2002) 6 Nordhaus and Boyer (2000) 7 The European result is driven in large part by Europe"s expected willingness to pay to reduce the risk of a catastrophic event such as a significant weakening of the Atlantic thermohaline circulation – part of which keeps Western Europe warmer than its latitude would otherwise imply.
8 Stern (1977), Pearce and Ulph (1999) 9 Equity weights should reflect the choice of social welfare function – sometimes called the "objective" function. This aggregates the consumption of individuals over space and time, reflecting judgements about the value of consumption enjoyed by individuals in different regions at different times (see the Appendix to Chapter 2). Here we focus on how this weighting should be carried out across regions within the present generation when considering the aggregation of small changes. The first step in calculating a weighted average change is to calculate the proportional impact of climate change on the representative individual in each region. If the utility function for an individual has constant marginal utility, the proportional impacts on per capita consumption can then be aggregated to give the proportional impact on overall social welfare by weighting them by the share of each individual"s consumption in total consumption. At the regional level, this means weighting the impact on the representative individual by the region"s share in global consumption (i.e. regional per-capita consumption multiplied by regional population, as a share of total global consumption). With a utility function given by the log of individual consumption, the proportional impacts on individuals should simply be added up; thus, at the regional level, the proportional impact on the representative consumer is weighted by the region"s population.
10 There are several reasons why the "Mendelsohn" model estimates the lowest cost of climate change. Adaptation is likely to be one, its omission of non-market impacts and the risk of catastrophe another.
11 That is, they estimate the relationship between production in their five market sectors and climate based on how production varies across current world climates, and control for other important determining factors.
12 See Hitz and Smith (2004) 13 Watkiss et al. (2005) 14 Warren et al. (2006) 15 Hallegatte and Hourcade (2005) and Chapter 4.
16 Although this depends on how rapidly costs increase in proportion to temperature.
17 Nordhaus and Boyer (2000) 18 Nordhaus and Boyer (2000) 19 Because the Nordhaus and Boyer model simplifies the economy to one sector, it ignores the possibility that productivity will increase if production is shifted from low productivity/highly climate-sensitive sectors to high productivity/low sensitivity sectors.
But a multi-sector study for the USA (Jorgensen et al., 2005) indicates that such processes are negligible, at least in that region.
20 Fankhauser and Tol (2003) 21 Warren et al. (2006) 22 Hope (2003) 23 We follow PAGE2002 in referring to "GDP" but, as remarked above, it is preferable to think of a broader income concept in interpreting some of the results.
24 Tol (2005) 25 Nordhaus and Boyer (2000) 26 IPCC (2001) 27 For example, the central value is based on Gedney et al. (2004) assuming 4.5°C temperature rise in 2100 28 IPCC (2001) 29 Nordhaus and Boyer (2000) 30 Tol (2002) 31 In these calculations, we assume that some fixed proportion of income is saved for future consumption. A more sophisticated model would vary the rate of saving as a result of prospects for future consumption, as determined by the model itself.
32 As in Nordhaus and Boyer (2000) 33 We are not considering here the discounting of extra units of consumption in the future because consumption itself may be higher then.
34 Proposed by Mirrlees and Stern (1972) 35 Formally, the change in the BGE is a natural commodity measure of welfare that expresses changes in future consumption due to policy in terms of the percentage increase in consumption (along a steady-state growth path), now and forever, that is equal to the changes that are forecast to follow from the policy change being examined. In a one-sector growth model with natural growth a and consumption C at time t, we want to calibrate welfare from the path [C(t)]. If this is equivalent, in welfare terms, to the balanced growth path yielding consumption ?eat, then ? is the BGE of [C(t)].
36 An extrapolated version of the IPCC"s A2 scenario (IPCC, 2000), characterised by annual average GDP growth of about 1.9%.
37 Also extrapolated from the IPCC"s A2 scenario. Annual average population growth is about 0.6%.
38 In fact, the model is restricted to a subset of uneven time steps. Thus we interpolate linearly between time steps to produce an annual time series.
39 See Pearce and Ulph (1999).
40 Pearce and Ulph (1999) and Stern (1977).
41 Nordhaus and Boyer (2000).
42 Mirrlees Stern (1972) 43 Rothschild and Stiglitz (1970) 44 Atkinson (1970) 45 Nordhaus and Boyer (2000)
Successive IPCC assessments of the IAM literature can be found in Pearce et al. (1996) and Smith et al. (2001). Hitz and Smith (2004) provide a more recent summary, focussing on the nature of the relationship between rising temperatures and the cost of climate change. William Cline"s 1992 book The Economics of Global Warming and William Nordhaus and Joseph Boyer"s 2000 book Warming the World provide an important and well-structured discussion of the issues, while Hope (2005) explains Integrated Assessment Modelling in detail. Watkiss et al. (2005) is a valuable discussion of the uncertainties around estimating the monetary cost of climate change, while Warren et al. (2006) subject the damage functions in IAMs to critical scrutiny.
Atkinson, A. B. (1970): "On the measurement of inequality", Journal of Economic Theory, 2(3): 244-263 Cline, W.R. (1992): "The Economics of Global Warming". Washington, DC: Institute for International Economics Fankhauser, S. and R.S.J. Tol (2003): "On climate change and economic growth", Resource and Energy Economics 27: 1-17 Friedlingstein, P., P. Cox, R. Betts et al. (2006): 'Climate-carbon cycle feedback analysis: results from C4MIP model intercomparison', Journal of Climate, 19: 3337-3353 Gedney, N., P.M. Cox and C. Huntingford (2004): "Climate feedback from wetland methane emissions", Geophysical Research Letters 31(20): L20503.
Hallegatte, S. and J.-C. Hourcade (2006): "Why economic dynamics matters in the assessment of climate change damages: illustration on extreme events", Ecological Economics (forthcoming).
Hitz, S. and J.B. Smith (2004): "Estimating global impacts from climate change", The Benefits of Climate Change Policies, J. -C. Morlot and S. Agrawala. Paris: OECD, pp 31-82.
Hope, C. (2003): "The marginal impacts of CO2, CH4 and SF6 emissions," Judge Institute of Management Research Paper No.2003/10, Cambridge, UK, University of Cambridge, Judge Institute of Management.
Hope, C. (2005): "Integrated assessment models" in Helm, D. (ed.), Climate-change policy, Oxford: Oxford University Press, pp 77-98.
Intergovernmental Panel on Climate Change (2001): Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change [Houghton JT, Ding Y, Griggs DJ, et al. (Eds.)], Cambridge: Cambridge University Press.
Jorgenson, D.W., R.J. Goettle, B.H. Hurd et al. (2004): 'US market consequences of global climate change', Washington, DC: Pew Center on Global Climate Change.
Mendelsohn, R.O., W.N. Morrison, M.E. Schlesinger and N.G. Andronova (1998): "Country-specific market impacts of climate change", Climatic Change 45(3-4): 553-569 Mirrlees, J.A. and N.H. Stern (1972): "Fairly good plans", Journal of Economic Theory 4(2): 268-288 Murphy J.M. et al. (2004): "Quantification of modelling uncertainties in a large ensemble of climate change simulations ", Nature 430: 768 – 772 Nordhaus, W.D. and J.G. Boyer (2000): 'Warming the World: the Economics of the Greenhouse Effect', Cambridge, MA: MIT Press.
Pearce, D.W. et al. (1996): "The social costs of climate change: greenhouse damage and the benefits of control" Climate Change 1995: Economic and Social Dimensions of Climate Change, Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press pp 183-224 Pearce, D.W. and A. Ulph (1999): "A social discount rate for the United Kingdom" Economics and the Environment: Essays in Ecological Economics and Sustainable Development, D.W. Pearce, Cheltenham: Edward Elgar.
Rothschild, M.D. and J.E. Stiglitz (1970): "Increasing Risk: I. A Definition", Journal of Economic Theory, September 2, pp255-243.
Smith, J.B. et al. (2001): "Vulnerability to climate change and reasons for concern: a synthesis" Climate Change 2001: Impacts, Adaptation and Vulnerability, Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press pp 913-967.
Stern, N. (1977): "The marginal valuation of income", in Artis, M. and R. Nobay (eds.), Studies in Modern Economic Analysis, Oxford: Blackwell.
Tol, R.S.J. (2002): "Estimates of the damage costs of climate change – part II: dynamic estimates", Environmental and Resource Economics 21: 135-160 Tol, R.S.J. (2005): 'The marginal damage costs of carbon dioxide emissions: an assessment of the uncertainties', Energy Policy 33(16): 2064-2074 Warren, R. et al. (2006): 'Spotlighting Impacts Functions in Integrated Assessment Models', Norwich, Tyndall Centre for Climate Change Research Working Paper 91.
Watkiss, P. et al. (2005): 'Methodological Approaches for Using Social Cost of Carbon Estimates in Policy Assessment, Final Report', Culham: AEA Technology Environment.
Autor:
Universidad de Oxford.
| Página siguiente |