AdaptCost Briefing Paper 1: Integrated Assessment Models – Africa results Key Messages 1. Estimates of the economic costs of adaptation require investigation of several lines of evidence. These range from case studies of projects and plans through to global scale assessment. Each of these evidence lines brings insight into a complex area, where we have relatively little information. 2. Many of the cited estimates of the economics of climate change are derived from global integrated assessment models (IAMs). These provide aggregated estimates, assessing costs in a single iterative framework. However, to make analysis manageable, they involve assumptions and simplifications, and they have been the subject of considerable debate. The AdaptCost project has commissioned two leading global IAMs, the FUND and PAGE models, to provide results for Africa. 3. The first important conclusion from the model runs is that the economic costs of climate change in Africa are likely to be significantly higher in relative terms than in other world regions and that they could be significant even in the short-term. The models indicate that the central net economic costs of climate change could be equivalent to 1.5 - 3% of GDP each year by 2030 in Africa. These costs include market and non-market sectors and are subject to assumptions and uncertainties. 4. The FUND model estimates that under a business as usual scenario, net economic costs could be equivalent to 2.7% of GDP each year in Africa by 2025 (around $40 billion/year [undiscounted]). The model reports large costs from water resources, health impacts, and energy costs for cooling, but some potential benefits from agriculture. Note that positives and negatives are combined in the results. It shows relatively similar levels of economic costs in future years through to 2100. A separate analysis with the FUND national model has provided net estimates for each African country. This shows strong distributional patterns of economic costs by country (and region) and strong increases in economic costs in all regions over future years. 5. The PAGE model estimates that economic costs could be equivalent to around 2% of GDP each year in Africa by 2040 (central value, market and non-market sectors, A2 scenario), with a 5-95% range of 0.4% to 4%. The model shows economic costs rises rapidly in future years, so that by 2100, they could be equivalent to 10% of GDP each year, i.e. too high for a sustainable economy. 6. The PAGE model has been run with adaptation included, which reduces costs significantly. Under the A2 scenario, the (mean) economic costs of climate change drop from 0.8% of GDP in 2020 and 1.7% in 2040 (no adaptation) to 0.5% and 1.1% respectively (with adaptation included). The benefits of adaptation are far in excess of estimated costs, with mean benefits of 17 billion in Africa in 2020 compared to mean costs of $4.5 billion and increase significantly in future years. However, the model shows high residual costs in Africa even with adaptation, estimating that these are around 50% for market sectors and around 70% of total impacts. 7. Finally, the PAGE model has been run with a mitigation scenario for a 2 degrees (450 ppm) scenario. This shows very reduced costs from climate change in later years (post 2040), such that annual economic costs of climate change to Africa are limited to just over 2% of GDP by 2100 – this is dramatically below the business as usual scenario. Adaptation reduces these economic costs further, leaving low residual impacts. 8. The runs shows that in the absence of mitigation, economic costs from climate change in Africa could be extremely large, even in the short-term. Impacts will be unevenly distributed across countries, and between sectors. Moreover, high economic costs are likely even with adaptation. Mitigation reduces longer-term impacts significantly, but there will still be significant adaptation needs and residual economic costs. Adaptation needs are similar in early years in all scenarios, due to the change already locked into the system. 9. All these estimates are indicative only. They provide some insights on signs, orders of magnitude, and patterns of effects. They can provide information on the potential economic costs, which are useful in the context of adaptation financing requirements. However, the models reflect only a partial coverage of the effects of climate change, and do not capture extreme events (including flooding and droughts), crosssectoral links and socially contingent effects, or the cumulative effects on adaptive capacity, all of which may be important for Africa. 1 Background Discussion of the Models There are relatively few estimates of the economic costs of climate change, and the costs and benefits of adaptation. Of the estimates that do exist, many derive from a set of aggregated economic models known as Integrated Assessment Models, or IAMs. What are Integrated Assessment Models? Integrated assessment is a generic term used to describe the integration of different models or methods within a single analysis. However, the term is often used in a specific sense for the integration of a number of climate change impact sectors within a single analytical model. There are a large number of IAMs (see Dickinson, 2007iv). These are used for different applications, investigating different aspects, and work at different geographical scales, but as the focus in the AdaptCost is at the African (continental) scale, the most relevant are the global integrated assessment models. These models provide a way to value the economic costs of climate change over time, at a global and continental scale (e.g. Africa), in a single iterative framework, and to look at the potential costs and benefits of adaptation, all within a consistent economic framework. They are therefore an extremely important part of the overall evidence base on the economics of climate change in Africa. Most of the recent policy estimates have focused on a few key models. The results of these models from a large part of the economic evidence base on climate change, as reported in the AR4 2007 chapter 20 WGIIi. They were used to provide the headline economic costs in the Stern review (2006) ii, i.e. that ‘BAU climate change will reduce welfare by an amount equivalent to a reduction in consumption per head of between 5 and 20%’. The FUND model (Tolv); The PAGE model (Hopevi); The DICE/RICE/AD-RICE models (Nordhausvii). The MERGE model (viii). There are other models which use the functional impact forms in the models above. This includes a new generation of global IAMs that are based on computable general equilibrium models (CGEMs), as well as more detailed energy / mitigation models. They were also used in the IIED/Grantham study (2009iii) to estimate the global costs and benefits of adaptation, which reported that adaptation measures had large (mean) benefit:cost ratios of about 60 under a business-as-usual A2 scenario, and about 20 in the aggressive abatement ‘450ppm’ scenario. However, their widespread use has also led to considerable debate about the models and results. How do these models work? The global IAMs combine the scientific and economic aspects of climate change within a single, iterative analytical framework. They typically include an energy / economy / emissions module, a climate module and an impact/valuation module. The models have an additional element where climate impacts feed back to the socioeconomic module thereby linking emissions, climate modelling, climate change impacts and the economy. This briefing note reviews the methods and estimates in this area. It is one of a series of briefing notes from the UNEP sponsored AdaptCost project1, which is investigating the economic costs of adaptation in Africa. It provides results from a number of IAM modelling runs which have been commissioned by the project to provide specific output for Africa. The briefing note does not necessarily represent the official views of the sponsors or the host organisations. To make analysis of economic costs manageable, they use simplified analysis of climate projections - rather than full-scale climate models - and simplified impact relationships. The models do 1 The AdaptCost project benefits from related projects funded by DFID and DANIDA (Economics of Climate Change in East Africa) and the EC (ClimateCost). 2 not undertake physical impact assessment per se, but instead link changes in climate to economic values, using relationships between climate change and economic damage. Some models (such as the PAGE model) use generic categories of damages to do this, working with the broad categories of market and non-market sectors. Other models (such as the FUND model) use reduced form equations that work at a sectoral level, so that economic costs are estimated for agriculture, human health, sea level rise, etc. Some of the models can also look at the economic costs of climate change with and without adaptation, estimating the costs and benefits of adaptation (PAGE) – or a mix of mitigation and adaptation policies (AD-RICE). The models have a very strong focus on uncertainty analysis, and a number of them (PAGE, FUND) include Monte Carlo analysis, a risk modelling technique that presents both the range, as well as the expected value, of collective risks: designed for when there are many variables with significant uncertainties, or when the implications of uncertainty (and the impact on a decision) cannot be adequately captured through sensitivity analysis. The models also have mitigation modules, with some based on top-down cost curve information, and others including some bottom-up detail. Note that the wider list of IAM models includes a much greater diversity on the mitigation side (including the CGEM models above). Some also consider adaptation, discussed below. A key benefit of these models is that they provide information and outputs that cannot be generated by other approaches, such as the SCC, or the PV of future economic costs. They also produce overall context and ‘headline’ values, putting climate policy in an economic context. What uses do these models have? The global IAM can be used to address a number of policy questions. In relation to the potential economic costs of GHG emissions and climate change this includes: Aggregate economic costs, e.g. at the global or continental scale, presented as $ values for any future year and as the equivalent % costs of gross domestic product (GDP) in that year. This can be undertaken for a business as usual scenario, or for a specific scenario such as stabilisation of CO2 equivalents at 450 CO2 ppmv. Note the models also estimate economic costs with temperature changes. The total economic costs over time – for each year - aggregated and expressed in present day prices (present values). The marginal global economic costs caused by one tonne of carbon emitted, often known as the Social Cost of Carbon (SCC), defined as the net present value of climate change impacts over the next 100 years (or longer) of one additional tonne of carbon emitted to the atmosphere today. The models can be extremely useful in providing information on the drivers of economic costs, including the influence of different parameters on the overall numbers. They can often provide important underlying information that would not be derived by other means. As an example, the FUND model has identified that the economic costs of air conditioning to address increased cooling demand could be one of the most important sectoral effects in global costs (see Downing et al, 2005ix) – yet this area is often omitted in national bottom-up assessments. Similar, the PAGE model has demonstrated the economic importance of major discontinuities, even for events post 2100 when discounted. These models are generally applied at the global scale, though they work by aggregating outputs from a continental / large-scale regional level. While their use has predominantly been to look at global estimates, they have also been used for regional and national assessments, with examples in the US and Caribbean (SEI, 2008x) and in South East Asia (AdB, 2008xi). The models can be used to look at costs and benefits of a baseline vs. mitigation scenarios, and can determine the optimized level of mitigation using cost-benefit analysis, though such assessments have been controversial. 3 What are the advantages and disadvantages of these models? The integrated models important advantages. have a number of They represent multi-sectoral and/or multiimpact inter-linkages in a quantitative way. They allow analysis of a very large number of possible scenarios very quickly. They provide high flexibility and can work with a variety of metrics, including the social cost of carbon, estimates in future years, GDP % losses, present values, and even cost-benefit analysis, which cannot be derived from bottom-up impact assessments. assumptions about how steeply damages increase as temperatures rise. However, these studies may also have overlooked positive impacts of climate change and not adequately accounted for how development could reduce impacts of climate change. Key Strengths Key Weaknesses They provide estimates over very long timescale and post 2100 – for example PAGE extends to 2200 and FUND to 2300. They factor in uncertainty, with several models working with Monte Carlo analysis. The main disadvantages of these integrated assessment models is that they are technically complex to construct, that they often cover a limited number of impacts and linkages, and they are often considered “black boxes” in the sense that model parameterisation and structure may not be entirely transparent, documented or independently validated. These models – and their strengths and weaknesses - have been the subject of considerable debate, not least due to their high profile use in the Stern Review. Their strengths and weaknesses are outlined below. Useful headline values. Total damages (in given time period, including GDP loss) and cumulative present value as well as Social Cost of Carbon values. Detailed uncertainty analysis (e.g. Monte Carlo assessment) Extend to longer time periods (>2100) No representation of physical impacts and partial coverage of effects due to use of reduced form impact functions. Coarse spatial resolution and high level of aggregation. Do not capture adaptation, or only in a highly theoretical and aggregated form. Partial coverage of potential effects, e.g. basing analysis on average temperature but not the effects of extreme weather events or major discontinuities. Omission of many categories of effects. Poor coverage of existing vulnerability to climate variability and short-term analysis. Based on existing literature, thus limited in terms of transferability at a regional or national level, and not validated at a local scale. The review concluded that the current generation of aggregate estimates may understate the true cost of climate change because they tend to fully capture extreme weather events, major climate discontinuities and socially contingent effects (see also next section); to underestimate the compounding effect of multiple stresses; and to ignore the costs of transition, learning and institutional development The models have been the subject of several reviews and inter-comparisons. Warren et al (2006xii) reviewed the main four global IAMs listed earlier, and concluded that most were based on literature from 2000 and earlier, and thus that they may omit some predictions of climate impacts (e.g. from AR4) that have become more pessimistic. The review also identified issues in relation to the calibration of the models based on US literature, the higher possibility of discontinuities at lower temperatures, and the treatment of adaptation. It also highlighted that many of the IAMs rely on aggregated damage functions with a smooth functional form, and make The issue of the damage functions has reemerged more recently (see Ackerman et al, 2009xiii), with concerns that the functions may not perform well at higher temperature levels. Note, however, that since this time all the models have been updated, with the emergence of a 4 newer version of DICE, the forthcoming PAGE2009 model, and multiple updates to the FUND model. Moreover, many of the criticisms of the models reflect underlying gaps in the evidence base: there is simply a lack of information on the economics of climate change, and as this improves, so will integrated assessment. weighting has a significantly bearing on the results. Use of an approach that includes equity weights increases the values, but there have been concerns raised over their use (e.g. see Anthoff et al, 2009xiv). This is a key issue for Africa. The consideration of uncertainty and risk aversion, and related to this, whether best guess / central values are used (and whether mean or median values are cited as the central estimate). Both the mean and the median have been used as a measure of central tendency. However, when uncertainty is considered, the models show right skewed distributions, thus the mean is greater than the median and studies which consider uncertainty often derive higher values. A key issue here regards climate sensitivity, i.e. the equilibrium warming expected with a doubling of CO2 concentrations, and the risk of low probability high consequence events. However, a key point for Africa is that these IAMs are global models. None have been developed in Africa to capture particular development futures or specific impacts. They have not been validated in Africa with detailed case studies, noting that very few such case studies exist in Africa. Key issues of debate with the models The outputs from these models have been the subject of considerable debate, particularly following their use in the Stern review (2006) and the critiques that followed. This focuses on a number of specific factors (Watkiss and Downing, 2008), notably; Each of these factors can significantly influence the values from the models, leading to a range that approaches several orders of magnitude. For instance the IPCC noted plausible estimates of the social cost of carbon from $1 to 1000/tC (Yohe et al, 2007). There is no consensus on which parameter values or ranges to use, though there are many alternative views expressed. The discount rate. The choice of discount rate has a very large effect on any values, because most impacts of climate change occur in the future. Higher discount rates lead to lower values. The choice of discount rate was the main controversy in the Stern review, though as noted above, this is an issue with the economics of climate change, not just IAMs – though the longer time horizons in these models make this issue more important. The choice of specific parameters can explain the difference between recent Government Social Cost of Carbon estimates in the UK (Stern, 2006) and the wider literature (such as Tol, 2007xv; Yohe, 2007xvi, Nordhaus, 2006xvii), even though both use much the same models and damage functions. The Stern Review adopted a global rather than a national perspective, with substantial aversion to risk, and consideration of intertemporal equity. This translated to the use of a very low discount rate (0.1% PRTP), use of mean values and higher climate sensitivity based on more recent evidence. The study time period and time horizon. Aggregate models suggest that net impacts of climate change may be modest in the short term, but strongly negative for more severe climate change in the future. The time horizon chosen, even with discounting, can therefore substantially affect the economic costs of climate change. The models can address this through the long time-scales, but this increases uncertainty. The approach to weighting impacts in different regions (equity weighting), when aggregating global values. Most adverse impacts in the medium term occur in developing countries, and so the decision on whether to use compensation values or adjust estimates using a form of distributional or equity There is also an issue with the coverage of the models. Climate change includes many types of climatic parameters, which in turn affect many sectors (market and non-market) in different ways. A key issue is therefore over the coverage (or completeness) of any valuation of climate change. It is clear that different estimates of the 5 valuation of greenhouse gas emissions are based on different types of climate effects, and include different impacts across varying sectors. Watkiss and Downing (2008xviii) framed this issue of coverage in a matrix, see figure below. Market Projection e.g. temperature and sea level rise Bounded e.g. precipitation and extremes Major change e.g. major tipping points All models include Model coverage is partial or missing PAGE/DICE include Non-Market All models include in deriving long-term global omissions are more important. None None None None these It is also stressed that the missing categories are likely to include both positive and negative effects. However, there is a general view that the missing effects are likely to have net damages, which could be potentially very large. A clear research priority is to investigate the missing elements of the risk matrix – to fill the evidence gaps. An initial inventory of adaptation estimates indicates a similar early concentration on categories that are relatively easy to quantify. Socially contingent Some initial coverage values, A final issue here centres on the use of the models in cost-benefit analysis. CBA for climate change relies critically on the assumption that marginal costs and benefits, as well as absolute costs and benefits, are known and finite. This is not necessarily the case and has been highlighted in work by Weitzman (2009xx), which concludes that including plausible, if unknown, probabilities of catastrophic climate change (and so called fat tails), leads to radically different conclusions for policy from the conventional advice coming out of a standard economic analysis and formalised CBA, as the conventional approach essentially ignores this kind of potential for disasters. Figure 1. Coverage of the global IAMs Mapping the IAMs onto a matrix of coverage of climate parameters, and coverage of categories of impacts/valuation shows that most of the model outputs focus on the top left area reflecting market damages from predictable events such as temperature and SLR. There is some coverage of non-market damages, but the models do not generally include consideration of all factors associated with ecosystem services, nor do they capture all the effects associated with bounded elements such as extreme events. Consideration of adaptation Within the global IAMs, adaptation is also aggregated and partial, though this reflects the underlying information available on the costs and benefits of adaptation. There is far less explicit coverage in the models, and also wide differences between models. Where adaptation is included, it is based on an aggregated approach without any technological, institutional or spatial detail. Some models do include some potential to consider major catastrophic events, i.e. so called tipping points (Lenton et al, 2008 xix), but tend to focus on economic considerations, i.e. omitting the full implications of ecosystem services and in particular socially contingent effects, i.e. large scale dynamics related to human values and equity that are poorly represented in cost values, e.g. from regional conflict, famine, poverty. The main IAMs adopt different approaches. In FUND, adaptation happens optimally in the damage function specification, such that damage functions are residual damage and adaptation costs, though this is more explicit in some sectors, such as coasts. In PAGE, the adaptation analysis is run separately, using parameterized functions to represent how much adaptation can reduce impacts. Adaptation affects the rate and level of temperature change at which an onset of impacts begins, and can reduce the severity of these impacts once they begin. It can also assess the It is highlighted that these gaps also exist for all other approaches, e.g. bottom-up assessments; however, because of the uses of these IAMs, e.g. 6 costs and benefits, though this is not run as an optimisation calculation. Within the AD-DICE model, the analysis disaggregates the damage function into adaptation costs and residual damages. While adaptation and mitigation are not modelled as substitutes, the model can select a preferred combination of mitigation and adaptation in response to climate impacts. standard monetary valuation methods, particularly benefit transfer. It has a time period through to 2300, and has 16 world regions. The FUND model runs commissioned for AdaptCostxxii estimate that climate change could lead to economic costs equivalent to 2.7% of GDP each year in Africa by 2025 (central value, including market and non-market sectors). The values are shown below, along with the splits for north and sub Saharan Africa. These approaches highlight an essential difference in the assessment of adaptation. The IAMs assume adaptation reduces projected impacts, that is, they adopt a simple predict-andprovide rational decision model. Conversely, most adaptation case studies (but not all) evaluate adaptation as a socio-institutional process, based on decision models such as act-learn-then act again (as portrayed in Klein et al. 2007xxi). Economic costs of climate change as % of GDP/year in 2025 2025 The assessment of ‘actual’ adaptation includes costs that anticipate risk but might not reduce actual impacts (akin to risk aversion), maladaptation, where actions increase impacts (e.g., displaced flooding risk) and costs not directly associated with additional climate change but essential to build competence and capacity to adapt. These processes are the foundation of the adaptation signature approach (see separate briefing note). Running the Models for Africa As part of the AdaptCost project, the project has commissioned two of the leading IAM runs for Africa, the FUND and PAGE models. The results are summarised below. 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% North Africa Sub Saharan Africa All Africa of GDP. as a Faction of Climate Costs 2. Annual Figure 2. Figure Annual costs as a Change % of GDP in 2025 Source FUND model Source FUND model The FUND Model: Africa Results In absolute terms, the costs predicted are extremely high, with estimated net costs in 2025 (undiscounted) of $10 billion for North Africa and almost $30 billion for Sub-Saharan Africa, and thus a total of $40 billion overall (values are presented as estimated model outputs, and are not discounted). FUND (The Climate Framework for Uncertainty, Negotiation, and Distribution model), is an IAM which couples demographics, economy, technology, carbon cycle, climate, and climate change impacts. It includes: sea level rise (including adaptation), energy consumption, agriculture, forestry, water resources, cardiovascular and respiratory diseases, malaria, dengue fever, schistosomiasis, diarrhoea and ecosystems, at varying levels of detail. The model includes reduced forms of more complex models. It values impacts using These economic costs include both market and non-market sectors. The economic costs are dominated by energy (cooling of buildings), water resources and health (as negative effects) but also include agriculture, where there may be positive effects (assuming adaptation). 7 An additional more disaggregated version of the model has also been run, the FUND national model, for Africa, with the results shown in the figure belowxxiii – with strong orange and red colours showing higher levels of equivalent GDP loss in 2030 – and through to 2050 and 2100. This shows that the economic costs of climate change have strong differences across countries and regions in Africa. Note this contrasts with the main FUND model shows much more modest future changes post 2030, with economic costs rising in North Africa (equivalent to 1.6% of GDP each year in 2025, to 3.4% in 2100), but with very little change in SubSaharan Africa, partly due to the assumed economic growth in the region, and the greater adaptive capacity. The national model shows that the economic costs could potentially increase very significant in 2050, and especially towards 2100 (noting the very high uncertainty in such long-term predictions) with very strong increases in some countries that would have major effects for the economies of these countries. 2050 Key 0 – 1% GDP loss 1 – 2% GDP loss 2 – 3% GDP loss 3 – 5% GDP loss 5 – 10% GDP loss >10% GDP loss 2030 2100 Figure 3. Annual Costs from Climate Change as a Faction of GDP in Africa. Source FUND national model 8 The PAGE Model: Africa Results In absolute terms the model estimates total economic costs (mean, market and non market) at $49 billion in 2020. PAGE2002 is a leading IAM. It classifies economic costs into three categories: market, non-market, and future large-scale discontinuities. Parameter values are taken from the IPCC Third Assessment Report. It is a multi-gas model with 8 world regions. Rather than only give single estimates, PAGE builds up probability distributions of results of key inputs. The PAGE model was used in the Stern review, and generated the main headline values that climate change could lead to 5 to 20% equivalent losses. The model is then run with adaptation included, shown over the page, with adaptation cost levels broadly similar to the Stern review analysis (see the investment flow briefing note). Adaptation is found to have significant economic benefits. Under a BAU A2 scenario, the economic costs of climate change drop from an annual cost equivalent of 0.8% of GDP in 2020 and 1.7% in 2040 (no adaptation) down to 0.5% and 1.1% respectively, with adaptation (mean values). The PAGE model runs commissioned by the AdaptCost project use the Stern parameter assumptionsxxiv shown in the figure below. Under an unmitigated A2 scenario, it estimates that change could lead to economic costs which would be equivalent to just under 2% per year GDP by 2040 in Africa (central mean value, including market and non-market sectors, with no adaptation). The lower value (5% of costs are less than this) from the model is 0.4% of GDP in 2040 and the upper value (95% of costs are less) from the model is 4.1% of GDP by 2040. Adaptation reduces the total economic costs in Africa from $49 billion to $32 billion in 2020 (a benefit of $17 billion. The benefits in 2040 are even larger. The costs of adaptation xxv to achieve these benefits are based on input assumptions, and are equivalent to mean costs for Africa of $4.5 billion per year from 2020 on, with a 5 to 95% range of $1.9 to 7.8 billion per year. However, there are still high residual impacts even with adaptation. 25.6% Business as usual scenario 15.6% Annual Cost, as a % of GDP 12 10 Major (catastrophic) Non-Market Market 9.6% 8.2% 8 6.1% 6 4.1% 4 3.4% 1.9% 2 1.7% 0.8% 0 2020 2040 2060 2080 2100 Equivalent Annual Cost of Climate Change in Africa, as a % of GDP PAGE model, A2 scenario (BAU) mean, 5 and 95% range, no adaptation 9 25.6% 18.2% With Adaptation 15.6% 12 11.2% Annual cost, as % of GDP Baseline 10 With adaptation 9.6% 8.2% 8 6.6% 6.1% 6 5.2% 4.1% 4.1% 4 3.4% 2.7% 1.9% 2.2% 1.7% 2 1.1% 0.8% 0.5% 0 2020 2040 2060 2080 2100 Equivalent Annual Cost of Climate Change in Africa, as a % of GDP With and without adaptation PAGE model, A2 scenario (BAU) mean, 5 and 95% range A PAGE model run also been undertaken for a 450 ppm type scenario. This would give a mean expectation of limiting average temperature change to around 2 degrees by 2100. This dramatic reduction is due to the reduction in damages in the economic and non-economic sectors, but also due to the reduced probability of large-scale discontinuities (major events). The results are shown below, comparing the baseline run against the 450 scenario – in both cases with no adaptation included. When adaptation is included, the mean annual costs are reduced even further. The PAGE model estimates that under the 450 ppm scenario, adaptation reduces down the economic costs so that residual impacts are limited to a maximum (mean) annual cost equivalent to 1.5% of GDP by 2100. Note that adaptation needs are similar in early years irrespective of the scenario, due to the climate change already locked into the system. The real benefit of the 450 ppm scenario is in limiting the potentially much higher costs in the longer term through to 2100. The PAGE model estimates that under the business as usual scenario the mean economic annual costs around 10% of GDP by 2100. However, under the 450 ppm scenario, the annual economic costs of climate change are projected to fall to annual costs equivalent to 2.3% of GDP by 2100. Note that these results are for one IAM only. There is therefore a need for global mitigation, as well as adaptation. 10 450 ppm/2 Degrees scenario Annual cost, expressed as equivalent % of GDP Annual cost, as % of GDP (no adaptation) 10 Baseline 8 450 ppm scenario 6 4 2 0 2020 2040 2060 2080 2100 Equivalent Annual Cost of Climate Change in Africa, as a % of GDP A2 business as usual vs. 450 ppm 2 degrees scenario PAGE model, mean value Key Findings and Conclusions 2. The economic costs of climate change in Africa are large, even in early periods. The IAMs have an important role in providing key messages and insight to the debate on the economics of climate change. They provide aggregated estimates and important policy metrics, though there are also considerable sources of uncertainty. The key findings of the IAM model runs for Africa are summarised below. The models provide an indication of the potential scale of the economic costs of climate change in Africa. While caution must be taken in using the values, reflecting the issues outlined above, they provide aggregated assessment of the potential economic costs. 1. The economic costs of climate change in Africa relative to GDP are much greater than for other world regions. The results indicate that the potential scale of economic costs could be equivalent to 1.5-3% of GDP each year in Africa by 2030. Note that these include market and non-market sectors, so they are not the loss in annual GDP, i.e. they give the full social costs of climate change. Nonetheless, even in early periods, these costs are very significant. Moreover, it is possible that these values are underestimates, because they exclude existing climate variability, extreme events (including flooding), cross-sectoral links and socially contingent effects and the cumulative effects on adaptive capacity. In the absence of mitigation, the economic costs of climate change for Africa could potentially be very large. The models show much higher economic costs in Africa than for other regions, reflecting the consensus that damages will be higher because of greater vulnerability and lower adaptive capacity. As an example, the PAGE model shows economic costs that are equivalent to almost 10% of GDP each year by 2100 for Africa. These are far higher than the world’s at 2.6%. Global mitigation does reduce these significantly. 11 This implies that adaptation financing needs will also need to be high, i.e. $billions of annual investment, even in early years, and could increase potentially very significantly in the future. It also shows that there are potentially high residual economic costs in Africa, even with adaptation. The AdaptCost Project The AdaptCost Africa project, funded by United Nations Environment Programme (UNEP) under the Climate Change – Norway Partnership, is producing a range of estimates of the financial needs for climate adaptation in Africa using different evidence lines. The study aims: To help African policymakers and the international climate change community to establish a collective target for financing adaptation in Africa. 3. Impacts will be unevenly distributed across countries, and between sectors. The models show a range of values according to the sector and location, even for some models, with winners and losers (net benefits and net gains between sectors). While the exact distribution will vary with the model, it is clear that there will be a very strong location-specific nature to future climate change impacts. To investigate estimates to adapt to climate change and improve understanding of adaptation processes. This will provide useful information for planning adaptation programmes and support decision-making by national governments and multi- and bilateral donors by allowing them to better compare projects and policies on their economic grounds. In the process, countries will also gain a better understanding of their adaptation investment requirements, and build a stronger basis for articulating their financing priorities and attracting capital. Just as there are distributional effects at the aggregate level, there will also be differences between countries, and between socio-economic groups within countries. The distributional consequences are therefore extremely important. These effects may also mask further inequalities between the formal and non-formal economy, and particularly on groups that depend on ecosystems services, because of the potentially higher effects on these. 4. The values are very dependent assumed growth and development. This briefing note was prepared by Paul Watkiss, based on model runs commissioned from Chris Hope and David Anthoff. It is stressed that the views expressed in this note do not necessarily represent the views of these contributors. The paper also benefited from comments from several reviewers. on The future damages from the models are determined by input parameters, notably the assumed rate of growth. This is particularly the case when expressing values as % GDP equivalents. Footnotes and References i Yohe, G.W., R.D. Lasco, Q.K. Ahmad, N.W. Arnell, S.J. Cohen, C. Hope, A.C. Janetos and R.T. Perez, 2007: Perspectives on climate change and sustainability. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 811-841. ii Stern . N., Peters, S., Bakhshi, V., Bowen, A., Cameron, C., Catovsky, S., Crane, D., Cruickshank, S., Dietz, S., Edmondson, N., Garbett, S., Hamid, L., Hoffman, G., Ingram, D., Jones, B., Patmore, N., Radcliffe, H., Sathiyarajah, R., Stock, M., Taylor, C., Vernon, T., Wanjie, H., and Zenghelis, D. (2006). The Economics of Climate Change. Cabinet Office – HM Treasury. Cambridge University Press. Different models give different results according to the assumed level of growth and development, and the changes that this may have on future vulnerability, reflecting the fact that development should increase adaptive capacity. However, our current understanding of (future) adaptive capacity, particularly in developing countries, is still too limited to allow firm conclusions about the direction of the estimation. 12 iii xiv Martin Parry, Nigel Arnell, Pam Berry, David Dodman, Samuel Fankhauser, Chris Hope, Sari Kovats, Robert Nicholls, David Satterthwaite, Richard Tiffin, Tim Wheeler (2009) Assessing the Costs of Adaptation to Climate Change: A Review of the UNFCCC and Other Recent Estimates, International Institute for Environment and Development and Grantham Institute for Climate Change, London. iv Dickinson, T., 2007, The Compendium of Adaptation Models for Climate Change:First Edition, Adaptation and Impacts Research Division, Environment Canada, 52 pgs. http://www.preventionweb.net/files/2287_CompendiumofAdapt ationModelsforCC.pdf v Tol, R.S.J. (2002a) New estimates of the damage costs of climate change, Part I: Benchmark estimates. Environmental and Resource Economics 21 (1): 47-73. Tol, R.S.J. (2002b) New estimates of the damage costs of climate change, Part II: Dynamic estimates. Environmental and Resource Economics 21 (1): 135-160. vi Hope C, 2004, The marginal impact of CO2 from PAGE2002: An integrated assessment model incorporating the IPCC’s five reasons for concern, submitted to Integrated Assessment. vii Nordhaus, W.D. and Boyer, J.G. (2000) Warming the World: Economic Models of Global Warming, Cambridge: The MIT Press. Nordhaus, W. D. (2007) The challenge of global warming: Economic models and environmental policy. New Haven, Connecticut: Yale University. viii Manne, A.S., R. Mendelsohn, and R. Richels. 1995. "MERGE -- A Model for Evaluating Regional and Global Effects of GHG Reduction Policies." Energy Policy, 23(1):1734. Manne, A.S., Richels, R.G., 2004. MERGE: An integrated assessment model for global climate change. Stanford University, Stanford, CA, USA. ix Downing, T. Downing, David Anthoff, Ruth Butterfield, Megan Ceronsky, Michael Grubb, Jiehan Guo, Cameron Hepburn, Chris Hope, Alistair Hunt, Ada Li, Anil Markandya, Scott Moss, Anthony Nyong, Richard Tol, Paul Watkiss (2005). Scoping uncertainty in the social cost of carbon. Final project report. Social Cost of Carbon: A Closer Look at Uncertainty (SCCU). July 2005. Report to Defra. http://www.defra.gov.uk/environment/climatechange/carboncos t/aeat-scc.htm x SEI (2008). The Cost of Climate Change. What We’ll Pay if Global Warming Continues Unchecked. Frank Ackerman and Elizabeth A. Stanton Global Development and Environment Institute and Stockholm Environment Institute-US Center, Tufts University. May 2008. xi ADB (2009). The Economics of Climate Change in Southeast Asia: A Regional Review, Asian Development Bank, May 2009 xii Warren, R, Hope, C., Mastrandrea, M., Tol, R., Adger, N and Lorenzoni, I (2006). Spotlighting Impacts Functions In Integrated Assessment Research Report Prepared for the Stern Review on the Economics of Climate Change. September 2006. xiii Ackerman, F., Stephen J. DeCanio,S. J., Howarth, R., B., Sheeran, K. (2009). Limitations of Integrated Assessment Models of Climate Change. Forthcoming in Climatic Change. Stanton, E. A., F. Ackerman and S. Kartha (2009). "Inside the Integrated Assessment Models: Four Issues in Climate Economics." Climate and Development 1.2. Anthoff, D., C.J. Hepburn and R.S.J. Tol (2009) "Equity weighting and the marginal damage costs of climate change." Ecological Economics, 68(3): 836-849. xv Tol, R.S.J. (2006). The Stern Review Of The Economics Of Climate Change: A Comment. Economic and Social Research Institute/Hamburg, Vrije and Carnegie Mellon Universities. Octobe, 2006. xvi Yohe, G. (2006). Some thoughts on the damage estimates presented in the Stern Review—An Editorial Integrated Assessment Journal. Vol. 6, Iss. 3 (2006), Pp. 65–72. xvii Nordhaus, W.D. (2006) The Stern Review on the Economics of Climate Change. Journal of Economic Literature (forthcoming). Yale University, http://nordhaus.econ.yale.edu/stern_050307.pdf. xviii Watkiss, P and Downing, T. E (2008). The Social Cost of Carbon: Valuation Estimates and their Use in UK Policy. The Integrated Assessment Journal, Vol 8, Issue 1 (2008), xix Lenton, T. M., H. Held, E. Kriegler, J. W. Hall, W. Lucht, S. Rahmstorf and H. J. Schellnhuber (2008) Tipping elements in the Earth’s climate system, Proceedings of the National Academy of Sciences USA 105(6), 1786–1793. xx Weitzman, M. L. (2009). "On Modeling and Interpreting the Economics of Catastrophic Climate Change." The Review of Economics and Statistics 91(1): 1-19. xxi Klein, R.J.T., S. Huq, F. Denton, T.E. Downing, R.G. Richels, J.B. Robinson, F.L. Toth, 2007: Inter-relationships between adaptation and mitigation. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 745-777. xxii Run commissioned from the FUND model by David Anthoff, funded by the DFID / Danida project Economic Impacts of Climate Change in East Africa / UNEP funded AdaptCost project/ EC ClimateCost study. The run was for the business as usual scenario with the main FUND model and ‘best estimates’ and reporting of median value. The model estimates effects on agriculture, energy demand, health, forests, storms, coastal protection, water, and ecosystems. It does not include the effects of extreme events (including inland flooding), major events, or socially contingent events. The estimates include adaptation in some sectors. The main model reports Africa in two main regions, Sub-Saharan Africa and North Africa. xxiii This map was produced by Paul Watkiss, based on a FUND national model run commissioned by David Antoff and outelined above. xxiv Run commissioned from Chris Hope. The run was for an A2 scenario with 'Stern review' assumptions, without adaptation. Note the values and graph show mean estimates. The lower and upper values (and upper line on the later graph) show the 5% and 95% value. The model includes three aggregate categories of damage: market, non-market and major events (discontinuities) based on the available literature. xxv Note that for adaptation in developing countries, for market sectors, it is assumed that 50% of economic damages are eliminated by low-cost adaptation. For non-economic impacts, adaptation is assumed to remove 25% of the impact. No adaptation is assumed for discontinuity impacts. 13