E T H I O P I A CO U N T RY ST U DY 68650 i Economics of Adaptation to Climate Change ETHIOPIA ii E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E EACC Publications and Reports 1. Economics of Adaptation to Climate Change: Synthesis Report 2. Economics of Adaptation to Climate Change: Social Synthesis Report 3. The Cost to Developing Countries of Adapting to Climate Change: New Methods and Estimates Country Case Studies: 1. Bangladesh: Economics of Adaptation to Climate Change 2. Bolivia: Adaptation to Climate Change: Vulnerability Assessment and Economic Aspects 3. Ethiopia : Economics of Adaptation to Climate Change 4. Ghana: Economics of Adaptation to Climate Change 5. Mozambique: Economics of Adaptation to Climate Change 6. Samoa: Economics of Adaptation to Climate Change 7. Vietnam: Economics of Adaptation to Climate Change Discussion Papers: 1. Economics of Adaptation to Extreme Weather Events in Developing Countries 2. The Costs of Adapting to Climate Change for Infrastructure 3. Adaptation of Forests to Climate Change 4. Costs of Agriculture Adaptation to Climate Change 5. Cost of Adapting Fisheries to Climate Change 6. Costs of Adaptation Related to Industrial and Municipal Water Supply and Riverine Flood Protection 7. Economics of Adaptation to Climate Change-Ecosystem Services 8. Modeling the Impact of Climate Change on Global Hydrology and Water Availability 9. Climate Change Scenarios and Climate Data 10. Economics of Coastal Zone Adaptation to Climate Change 11. Costs of Adapting to Climate Change for Human Health in Developing Countries 12. Social Dimensions of Adaptation to Climate Change in Bangladesh 13. Social Dimensions of Adaptation to Climate Change in Bolivia 14. Social Dimensions of Adaptation to Climate Change in Ethiopia 15. Social Dimensions of Adaptation to Climate Change in Ghana 16. Social Dimensions of Adaptation to Climate Change in Mozambique 17. Social Dimensions of Adaptation to Climate Change in Vietnam 18. Participatory Scenario Development Approaches for Identifying Pro-Poor Adaptation Options 19. Participatory Scenario Development Approaches for Pro-Poor Adaptation: Capacity Development Manual E T H I O P I A CO U N T RY ST U DY iii Economics of Adaptation to Climate Change ETH IOPIA Ministry of Foreign Affairs Government of the Netherlands iv E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E © 2010 The World Bank Group 1818 H Street, NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved. This volume is a product of the World Bank Group. The World Bank Group does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank Group concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank Group encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone 978-750-8400; fax 978-750-4470; Internet: www.copyright.com. All images © The World Bank Photo Library, except Pages xii, 3, 30, 62 and 92 © Shutterstock Pages 77 and 86 © iStockphoto E T H I O P I A CO U N T RY ST U DY v Contents Abbreviations and Acronyms x Acknowledgments xiii Executive Summary xv Key Concepts xv Ethiopia’s Vulnerability to Climate Variability and Change xvi Impacts xvi Adaptation xix Recommendations xxii 1 Introduction 1 Background 1 Scope and Limitations 1 2 Country Background 5 Basic Features 5 Current Growth and Poverty Reduction Policies 5 Climate and Vulnerability to Climate Change 6 Modeling Approach 6 Approach to Climate Model Uncertainty 8 Crop Model Description 10 Livestock Model Description 11 Drought Model Description 15 Road Transport Model Description 15 Flood Cost Model for Road Infrastructure 18 IMPEND Model Description 20 WEAP Model Description 21 CGE Model Description 24 Dynamic CGE models 25 Social Analysis Approach 27 Key Assumptions and Limitations 27 vi E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E 3 Impacts 31 Agriculture 31 Drought Expenditures 39 Road Transport 40 Hydropower 46 Vulnerability: A Participatory Analysis 52 Economywide Impacts 54 4 Adaptation Options 63 Agriculture 63 Road Transport 65 Hydropower 69 Summary of Sector-Level Adaptation Costs 70 Adaptation: Economywide Analysis 72 Adaptation: Priorities at the Community Level 84 5 Recommendations 89 Shorter Term (up to 2015) 89 Medium to Long Term 91 Bibliographic references 95 Annexes (available on line at www.worldbank.org/eacc) Annex 1: CLIRUN-II Rainfall Runoff model Description Annex 2: CliCrop model Annex 3. Hydropower: IMPEND Model Annex 4: WEAP21 Model Description Annex 5: CGE Model: additional data and information Annex 6: WEAP Intersectoral Analysis Data and Methods Annex 7: Livestock Model Transfer Approach and Assumptions Annex 8: Drought Model Details Tables Table 1 Adaptation Costs (annual average, 2010–50, US$ billions) xix Table 2 Adaptation Costs and Residual Damage (annual average, 2010-2050), US$ billions xxi Table 3 GCM Scenarios for Ethiopia Country Track Study 9 Table 4 Predicted Change in the Probability of Selecting Each Animal as the Primary Animal Type for the Farm 13 Table 5 Predicted Change in Net Income per Animal 13 E T H I O P I A CO U N T RY ST U DY vii Table 6 Predicted Change in Number of Animals per Farm 13 Table 7 Predicted Change in Expected Income 14 Table 8 Dose-response Descriptions for Maintenance Costs 16 Table 9 Five Agroecological Zones 27 Table 10 Base Classified and Urban Road Networks, 2006 (km) 41 Table 11 Unit Maintenance Cost Rates (US$) 41 Table 12 Cumulative Climate Change Impact on Paved and Unpaved Roads Based on Increased Maintenance Costs for the Four Climate GCMs (US$ Million) 43 Table 13 Time-Series inputs to CGE model for transport costs 43 Table 14 Summary of Impacts Costs on Roads (US$ Million, annual average) 46 Table 15 Planned Power Generation Projects Included in the Analysis 46 Table 16 Total Hydropower Production under Different Demand and Climate Scenarios (Million MWH, 2010–40) 50 Table 17 Unmet Demand for Irrigation (Million m3, annual average by decade) 51 Table 18 Scenarios 55 Table 19 Statistics on Year-to-Year Growth Rates of Household Consumption 61 Table 20 Existing Irrigation Schemes (2005/06) and Irrigation Potential 64 Table 21 Summary of Adaptation Costs in Agriculture (annual average 2010–50, US$ million) 65 Table 22 Summary of Adaptation Costs for Roads (annual average 2010–50, US$ million) 67 Table 23 Summary of Adaptation Costs in the Road Sector (annual average over the 2010–40 period, US$ million) 69 Table 24 Summary of Adaptation Costs, all sectors analyzed (annual average 2010–50, US$ million) 71 Table 25 Total Direct and Indirect Adaptation Costs ($ billions) 74 Table 26 Foreign Savings ($ billions) 75 Table 27 Net Benefits and Adaptation Project Costs, US$ billions 79 Table 28 Adaptation Costs and Residual Damage (annual average, 2010-2050), US$ billions 80 Figures Figure 1 Deviations of GDP from Base Scenario xvii Figure 2 Agricultural Year-to-Year Growth Rates: Standard Deviations xviii Figure 3 Regional GDP, Deviation from Base, Wet2 xviii Figure 4 Net Present Value (NPV) of Absorption Differences xx Figure 5 Standard Deviation of Year-to-Year Agriculture GDP Growth Rates, without and with adaptation xxi Figure 6 Benefit/ cost ratio of Upgrading Road Standards xxiv Figure 7 Economic Growth and Climate xxvi Figure 8 Flow Chart of Model Sequencing 7 Figure 9 CGM Projection on Daily Precipitation Intensity 10 Figure 10 Precipitation Changes for IPCC A1B Scenario 10 Figure 11 Flood Damage Relationship 19 Figure 12 Ethiopia River Basins Modeled with WEAP 22 viii E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 13 Example of Basin Schematic with Supply and Demand in WEAP 23 Figure 14 Overview of Ethiopia WEAP Model 23 Figure 15 Agroecological Zones in Ethiopia 31 Figure 16 Range of Percent Yield Deviations from no-CC Base (2006-2050) for Selected Crops 33 Figure 17 Biophysical Component: Ratio of Future Livestock Expected Incomes to Expected Incomes under Mean Baseline Conditions, Ethiopia Base Scenario, 2001–50 34 Figure 18 Biophysical Component: Ratio of Future Livestock Net Revenues to Net Revenues under Mean Baseline Conditions, Ethiopia Dry (on left) and Wet (on right) Scenarios, 2001–50 35 Figure 19 Biophysical Component: Ratio of Future Livestock Net Revenues to Net Revenues under Mean Baseline Conditions, Global Dry (on left) and Wet (on right) Scenarios, 2001–50 35 Figure 20 Feed Component: Ratio of Millet Yields to Mean Baseline Yields, Ethiopia Base Scenario, 2001–50 36 Figure 21 Feed Component: Ratio of Millet Yields to Mean Baseline Yields, Ethiopia Dry (on left) and Wet (on right) Scenarios, 2001–50 37 Figure 22 Feed Component: Ratio of Millet Yields to Mean Baseline Yields, Global Dry (on left) and Wet (on right) Scenarios, 2001–50 37 Figure 23 Combined Components: Mean of Biophysical and Feed Vectors, Ethiopia Base Scenario, 2001–50 38 Figure 24 Combined Components: Mean of Biophysical and Feed Vectors, Ethiopia Dry (on left) and Wet (on right) Scenarios, 2001–50 39 Figure 25 Combined Components: Mean of Biophysical and Feed Vectors, Global Dry (on left) and Wet (on right) Scenarios, 2001–50 39 Figure 26 Mean Annual Projected Ethiopian Government Expenditures in the 2000s through 2040s (Millions of 2010 $) 40 Figure 27 Average Decade Costs for each of the Four GCMs for Maintaining Gravel and Earth Roads due to Climate Change Increase in Precipitation 42 Figure 28 Average Decade Costs for each of the four GCMs for Maintaining Paved Roads due to Climate Change Increase in Precipitation and Temperature 42 Figure 29 Base Maintenance Costs to Repair Flood Damage Based on Projected Floods From Historic Climate Patterns 44 Figure 30 Total Road Capital % Loss Based on Historic Projected Flood Events 44 Figure 31 Annual Average Maintenance Cost for Existing Paved and Unpaved Roads per Decade With No Adaptation 45 Figure 32 Hydropower Generation under two Change Scenarios, 2008–50 47 Figure 33 Total Hydropower Production in the 21 Ethiopia River Basins, Assuming Growing M&I Demands and Irrigation to 3.7 Million ha, 2001–50 48 Figure 34 Mean Decadal Changes in Hydropower Production Given Increasing M&I and Irrigation Demands, Relative to a No-Demand Scenario 49 Figure 35 Mean Decadal Changes in Crop Yields Given Increasing M&I and Irrigation Demands 51 Figure 36 Welfare Loss from CC Scenarios 57 Figure 37 Deviation of GDP from Base Scenario 58 Figure 38 Standard Deviation of Agricultural Year-to-Year Growth Rates 58 E T H I O P I A CO U N T RY ST U DY ix Figure 39 Agricultural GDP, Deviation from Base 59 Figure 40 Regional GDP, Deviation from Base, Wet2 60 Figure 41 Regional GDP, Deviation from Base, Dry2 60 Figure 42 Export Share of Electricity Production 61 Figure 43 Total Adaptation Cost for New Roads (average annual cost per decade) 69 Figure 44 Benefit/Cost Ratio of Upgrading Road Standards 70 Figure 45 Cumulative Discounted Cost of Hydropower for Two Climate Change Scenarios 71 Figure 46 Net Present Value (NPV) of Absorption Differences 73 Figure 47 Average Deviation of GDP from Base Run (%) 74 Figure 48 Standard Deviation of Year-to-Year Agriculture GDP Growth Rates 75 Figure 49 Direct and Indirect Adaptation Costs ($ billions) 76 Figure 50 Foreign Saving, Differences from Base Run Values ($ billions) 76 Figure 51 GFCF, Ratio to Base Run, Wet Scenarios 78 Figure 52 GFCF, Ratio to Base Run, Dry Scenarios 78 Figure 53 Residual Damage Costs ($ billions) 80 Figure 54 Discounted Differences in Absorption from Baseline (2010-2050, Wet2 scenario) 81 Figure 55 Household Consumption: Difference from Base, CC Shocks 82 Figure 56 Household Consumption, Difference from Base, Wet2 CC Shocks and Adaptation 82 Figure 57 Household Consumption, Difference from Base, Dry2 CC Shocks and Adaptation 83 Figure 58 Coefficient of Variation of Year-to-Year Growth Rates, Household Consumption 83 Figure 59 Economic Growth and Climate 92 Boxes Box 1 Climate Scenarios in the Global and Ethiopia Track of the EACC 9 Box 2 Modeling Impacts on Paved and Unpaved Roads 17 Box 3 Computable General Equilibrium Models (CGEs) 25 Box 4 Government Activities in the Agriculture Sector 64 Box 5 Design Strategy Adaptation for paved and unpaved roads 66 x E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Acronyms AR4 IPCC Fourth Assessment Report ITCZ Inter-Tropical Conversion Zone BAU Business-as-usual MDGs Millennium Development Goals CAADP Comprehensive Africa Agriculture NCAR National Center for Atmospheric Development Program Research CGE Computable general equilibrium NAPA National Adaptation Plans of Action CO2 Carbon dioxide NCCAS National Climate Change Adaptation CMI Climate moisture index Strategy CRU Climate Research Unit NGO Nongovernmental organization CSIRO Commonwealth Scientific and Indus- ODA Official development assistance trial Organisation PaMs Policies and measures EACC Economics of Adaptation to Climate PET Potential evapotranspiration Change Ppm Parts per million ENSO El Niño-Southern Oscillation R&D Research and development GCM General circulation model SRES Special Report on Emissions GDP Gross domestic product Scenarios GHG Greenhouse gases SSA Sub-Saharan Africa GIS Geographical information system SST Sea surface temperature HDI Human Development Index TAR Third Assessment Report IFPRI International Food Policy Index UNDP United Nations Development IMPACT International Model for Policy Analy- Programme sis of Agricultural Commodities and UNFCCC United Nations Framework Conven- Trade tion on Climate Change IPCC Intergovernmental Panel on Climate VFS Vulnerability and food security Change E T H I O P I A CO U N T RY ST U DY xi CURRENCY EQUIVALENTS (Exchange Rate Effective) Currency Units = Birr US$1.00 = 11.7 Birr in 2009 (annual average) Note: Unless otherwise noted, all dollars are U.S. dollars. FISCAL YEAR (FY) July 8 – July 7 WEIGHT AND MEASURES Metric System xii E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY xiii Acknowledgments The report was prepared by a team coordinated on The team is grateful for the comments received the World Bank side by Raffaello Cervigni. Prin- from participants to the initial consultation cipal authors were Kenneth Strzepek, Sherman workshop held in Addis Ababa in November Robinson, Len Wright and Raffaello Cervigni; 2009, as well as the final workshop in Addis additional experts on the team were Paul Block, Ababa in October 2010. These included: Assefa Brent Boehlert, Paul Chinowsky, Chas Fant, Wil- Tofu, Kumelachew Yeshitela, Sisay Nune, liam Farmer, Alyssa McCluskey, Michelle Mini- Abdissa Megersa, Abera Tola, Ahmed Kha- hane, Niko Strzepek, and Gete Zeleke. lid Eldaw, Alessandra Tisot, Amare Kebede, Amare Worku, Amelie Blume, Angus Friday, The World Bank team comprised, in alphabetical Belay Simane, Berhanu Adenew, Berhanu order, Aziz Bouzaher (former team leader), Marie Ayalew, Berhanu Kassa, Daniel Danano, Daniel Bernadette Darang, Susmita Dasgupta, Edward Tewodros, Dessalegen Mesfin, Dubale Admasu, Dwumfour, Achim Fock, Francesca Fusaro, Anne Gelila Woodeneh, Getu Zegye, Girma Balcha, Kuriakose, Stephen Ling, Sergio Margulis (team Hailu Tefera, Hiwot Michael, Ian Campbell, leader of the overall EACC study), Stephen Jeremy Webb, Mahary Maasho, Mahlet Eyassu, Mink, Kiran Pandey (Coordinator EACC coun- Mersha Argaw, Mesfin Mengistu, Michael Cat, try studies) , Dawit Tadesse, and Fang Xu. Robert Moges Worku, Negusu Aklilu, Peter Mwa- Livernash provided editorial services, Jim Cant- nakatwe, Praveen Wignarajah, Seleshi Geta- rell contributed editorial input and coordinated hun, Shewaye Deribe, Tamiru Sebsibe, Tesfaye production, and Hugo Mansilla provided edito- Alemu, Tewolde Egziabher, Tibebu Tadesse, rial and production support. The peer reviewers Valdemar Holmegen,Wondosen Sentayehu, were John Nash, Vahid Alavian, Michael Jacob- Wondwossen Mekuria, Wolt Soer, Yacob sen, and Roger Gorham. Mulugeta, Zenebe Gegziabher and Zerihun Ambaye. From the government of Ethiopia, policy guidance was provided by Dr. Tewolde Birhan Gebre Egziab- Financial support from DFID, the governments her Yohannes (Director General of the Environmen- of the Netherlands and Switzerland, and the tal Protection Authority), Ato Dessalegen Mesfin Trust Fund for Environmentally and Socially (Deputy Director General, EPA); and Dr. Abera Sustainable Development (TFESSD) is gratefully Deresa (State Minister, Ministry of Agriculture). acknowledged. xiv E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY xv Executive Summary This report is part of a broader study, the Eco- change of the economy in the absence of climate nomics of Adaptation to Climate Change change that can be used as a basis for comparison (EACC), which has two objectives: (a) to develop a with various climate change scenarios. global estimate of adaptation costs for informing international climate negotiations; and (b) to help For Ethiopia, a baseline, no-climate-change sce- decision makers in developing countries assess the nario reflecting current government plans and risks posed by climate change and design nationa priorities and spanning the 2010–2050 time hori- strategies for adapting to it. zon, was established in consultation with govern- ment officials at a workshop in November 2009. In addition to a “global track� (World Bank The baseline includes an ambitious investment 2010), where multicountry databases were used to program in dams, hydropower development, irri- generate aggregate estimates at a global scale, the gation, water management, and road building. EACC project includes a series of country-level studies, where national data were disaggregated Impacts are evaluated as the deviation of the to more local and sector levels, helping to under- variables of interest—economic welfare, sec- stand adaptation from a bottom-up perspective. tor development objectives, and so on—from the baseline trajectory. Adaptation is defined as a set of actions intended to reduce or eliminate Key Concepts the deviation from the baseline development path caused by climate change. In accordance with the broader EACC method- ology, climate change impacts and adaptation The impacts of climate change, and the merits strategies were defined in relation to a baseline of adaptation strategies, depend on future cli- (no-climate change) development trajectory, mate outcomes. These are typically derived from designed as a plausible representation of how global circulation models (GCMs) and are uncer- Ethiopia’s economy might evolve in the period tain, both because the processes are inherently 2010–50 on the basis of historical trends and stochastic and because the GCM models differ current government plans. The baseline is not a in how they represent those processes. To cap- forecast, but instead provides a counterfactual—a ture these uncertainties, this study utilizes the two reasonable trajectory for growth and structural “extreme� GCMs used in the global track of the xvi E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E EACC (labeled Wet1 and Dry1), as well as two rainfall intensity, there would be an increased additional models that are better suited to repre- chance of severe episodic flooding caused by sent climate model uncertainty in the specific case storm runoff in highland areas. of Ethiopia (labeled here Wet2 and Dry2). The Wet1 and Dry1 are used to ensure consistency with the EACC global track; but the Ethiopia Dry Impacts (Dry2) and the Ethiopia Wet (Wet2) capture more adequately the range of variation of climate out- The analysis focuses on three main channels of comes specific to Ethiopia. climatic vulnerability that already affect the Ethi- opian economy and are likely to be of major sig- nificance under the climate of the future. These Ethiopia’s Vulnerability channels include (1) agriculture, which accounted to Climate Variability and for 47 percent of Ethiopian GDP in 2006 and is highly sensitive to seasonal variations in tem- Change perature and moisture; (2) roads, the backbone of the country’s transport system, which are often Ethiopia is heavily dependent on rainfed agricul- hit by large floods, causing serious infrastructure ture. Its geographical location and topography—in damage and disruptions to supply chains; and (3) combination with low adaptive capacity—entail dams, which provide hydropower and irrigation a high vulnerability to the impacts of climate and are affected by large precipitation swings. change. Historically the country has been prone to extreme weather variability. Rainfall is highly Changes in precipitation and temperature from erratic, most rain falls with high intensity, and the four GCMs were used to estimate (a) changes there is a high degree of variability in both time in yields for major crops and impacts on livestock, and space. Since the early 1980s, the country has (b) flow into hydropower generation facilities and suffered seven major droughts—five of which the consequent changes in power generation, (c) have led to famines—in addition to dozens of the impact of flooding on roads; (d) the effects of local droughts. Major floods also occurred in dif- more frequent droughts on government expendi- ferent parts of the country in 1988, 1993, 1994, ture on vulnerability and food security (VFS); and 1995, 1996, and 2006. (e) the loss of irrigation and hydropower due to conflicts among competing demands. The analy- Climate projections obtained from the GCMs sis assesses deviations in GDP and other variables referred to above suggest an increase in rainfall from the no-climate-change baseline growth path variability with a rising frequency of both severe for the four climate change scenarios mentioned flooding and droughts due to global warming. above, which are intended to capture the range The Dry2 scenario shows reductions in average of possible variability in Ethiopia: Dry1, Dry2; annual rainfall over 2045–55 of (a) 10–25 per- Wet1, and Wet2. cent in the central highlands, (b) 0–10 percent in the south, and (c) more than 25 percent in the For the baseline (no-climate-change scenario), north of the country. The Wet2 scenario shows the analysis uses historical monthly climate data increases in average annual rainfall of (a) 10–25 and projects the historical pattern into the future. percent in the south and central highlands, and For the climate change scenarios, stochastic rep- (b) more than 25 percent in most of the rest of resentations of weather variability in each global the country. If the Wet2 scenario is accompanied circulation model are superimposed on the base- by an increase in the variability of short-duration line to capture the variability of the future. The E T H I O P I A CO U N T RY ST U DY xvii scenarios include projections of extreme weather nearly 8 percent lower than in the baseline. While events such as droughts and floods. these are not forecasts of future climate impacts, they highlight the high degree of vulnerability of Economy-wide impacts of climate change were Ethiopian agriculture and infrastructure to the assessed using a computable general equilib- climate shocks of the future. rium (CGE) model. The results of the model- ing suggest that the GDP losses are significant, Climate change brings about increased weather but diverse across scenarios. Under the Dry2 variability, which translates into large swings in scenario, losses are large (6 to 10 percent) and the growth rate of agricultural GDP, illustrated regularly distributed during the time horizon con- in Figure 2 by the increase in standard devia- sidered. In contrast, in the case of the Wet 2 sce- tion compared to the baseline. While the simple nario, the loss of GDP is quite substantial in the means of annual growth rates are similar across 2040–49 decade because of the costs of coping the scenarios, high variability leads to significant with damage caused by extreme weather events, welfare losses. A priority for adaptation invest- especially floods, from the 2030 decade onward. ment is therefore to reduce income variation and The 10-year average GDP for the final decade is the related welfare losses. Figure 1 DEVIATIONS OF GDP FROM BASE SCENARIO 2015 2025 2035 2045 0.00 -2.00 WET 2 PERCENT DEVIATION FROM BASE -4.00 WET 1 -6.00 DRY 1 DRY 2 -8.00 -10.00 -12.00 xviii E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 2 AGRICULTURAL YEAR-TO-YEAR GROwTH RATES: STANDARD DEVIATIONS 7.00 6.00 5.00 PERCENTAGE POINTS 4.00 3.00 2.00 1.00 0.00 BASE DRY2 WET2 Figure 3 REGIONAL GDP, DEVIATION FROM BASE, wET2 40.00 30.00 rrR1 PERCENT DEVIATION FROM BASE 20.00 rrR2 rrR3 10.00 rrR4 0.00 rrR5 2015 2025 2035 2045 rrURB -10.00 -20.00 Notes: Regions R1 to R5 and Urban. E T H I O P I A CO U N T RY ST U DY xix Variability in agricultural income tends to affect the poor more, with values of the standard devia- Adaptation tion on average some 10 percent higher than for the non-poor, under both wet and dry scenarios. The investment program included in the no-cli- mate-change baseline established in consultation Finally, as shown in Figure 3 for the Wet2 sce- with the government is likely to enhance Ethio- nario, climate change impacts are likely to vary pia’s resilience to climate change. However, addi- significantly across regions. tional efforts are required to attenuate climate change impacts. Adaptation strategies were there- The arid lowland zone 5 (R5) derives substantial fore identified as additions to—or modifications benefits from the increase in total rainfall, which of—current government programs. supports livestock, while relative losses are concen- trated in the cereals-based highlands zone 2 (R2) More specifically, adaptation in agriculture and in urban areas. The latter reflects the down- included increasing irrigated cropland and invest- stream consequences of flooding and weather vari- ing in agricultural research and development. In ability. The dry scenarios have reverse impacts, with the transport sector, adaptation options included the arid lowlands and livestock suffering greatly. increasing the share of paved and hardened roads, as well as “soft� measures such as changes In addition to analysis of the three priority sectors in transportation operation and maintenance, (agriculture, roads, and hydropower), the study development of new design standards that con- also analyzed potential conflicts under climate sider projected climate changes, transfer of rel- change in the use of water across sectors. A water evant transportation technology to stakeholders, planning model was used to evaluate the potential and the enhancement of transportation safety interactions among growing municipal and indus- measures. In the hydropower sector, adaptation trial (M&I), irrigation, and hydropower demands policies included altering the scale and timing under climate change. The model evaluates these of planned projects, as well as constraining total intersectoral effects between 2001 and 2050 and downstream flow and irrigation flow. generates time series of impacts to irrigated agri- cultural yields and hydropower generation under These strategies were first assessed on a sector-by each of the climate scenarios. sector basis. When the full set of economywide linkages is taken into account, direct plus indirect The result of the intersectoral analysis indicates adaptation costs increase significantly, as indi- that hydropower production is impacted by irriga- cated in Table 1. tion and M&I withdrawals. Under the Dry2 sce- nario, if priority is given to agricultural demands, Table 1 ADAPTATION COSTS (ANNUAL AVERAGE, 2010–50, there is a loss of hydropower capacity equivalent US$ BILLIONS) to 100 percent of the 2000 installed capacity and Scenario Direct, Indirect Total direct 10 percent of the hydropower capacity planed by sector costs and level costs indirect the government for the period 2011–15. costs Wet 1 0.19 0.60 0.79 If, on the other hand, priority is given to hydro- Dry 1 0.17 0.77 0.94 power, up to a billion cubic meters of water Wet 2 0.16 2.30 2.46 might be taken away from irrigated agriculture. Dry 2 0.26 2.55 2.81 That would cause a 30–40 percent yield drop in an area of some 250,000 hectares that would be forced to revert to rainfed conditions. xx E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E To evaluate its welfare implications, the adapta- at relatively low cost; and (c) that adaptation low- tion strategy was analyzed in a CGE framework ers income variability. by comparing a no-climate change baseline— reflecting existing development plans—with As shown in Figure 4, adaptation greatly reduces climate change scenarios reflecting adaptation the welfare loss due to climate change (measured investments. The main findings are that adapta- here by the difference from the baseline of total tion (a) reduces, but does not eliminate, welfare absorption –GDP plus imports minus exports, dis- losses; (b) that such welfare gains can be achieved counted over the 40-year time horizon). Figure 4 NET PRESENT VALUE (NPV) OF ABSORPTION DIFFERENCES 0.0 WET 2 DRY 2 WET 1 DRY 1 -1.0 -2.0 RATIO �%� TO NVP OF BASE GDP -3.0 -4.0 -5.0 -6.0 -7.0 -8.0 -9.0 -10.0 NO ADAPTATION ADAPTATION Note: NPV of Absorption, Difference from Base (% of NPV of GDP). Absorption is defined as GDP, plus imports minus exports E T H I O P I A CO U N T RY ST U DY xxi Figure 5 STANDARD DEVIATION OF YEAR-TO-YEAR AGRICULTURE GDP GROwTH RATES, wITHOUT AND wITH ADAPTATION 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 NO WITH NO WITH ADAPTATION ADAPTATION ADAPTATION ADAPTATION BASELINE WET 2 DRY 2 The (undiscounted) welfare benefits of the adap- Table 2 ADAPTATION COSTS AND RESIDUAL DAMAGE (ANNUAL AVERAGE, tation strategy are significantly larger than the 2010-2050), US$ BILLIONS project-level costs of implementing it, resulting Scenario Adaptation Residual Total in benefit/cost ratios ranging from 5 to over 13. costs damage Finally, adaptation restores the variability of agri- Wet2 2.45 1.52 3.97 culture GDP growth close to the baseline scenario Wet1 0.79 0.43 1.22 (Figure 5). Dry1 0.94 0.81 1.75 Dry2 2.81 3.03 5.84 While the benefits of adaptation investments are significant, they do not fully offset the nega- tive impact of the climate change scenarios. Two The second approach is to include an additional options were explored to close the “welfare gap� labor-upgrading program in the adaptation strat- caused by climate change. egy. In this scenario, 0.1 percent of rural unskilled labor is assumed to be transferred to the urban The first is to estimate the “residual damage region, with additional upgrading so that all the costs� as the transfer (in $) that would be required urban labor categories, skilled and unskilled, to completely offset the loss of absorption from grow uniformly faster than in the base run. When climate change shock, after implementing adap- tested under the Wet2 scenario, an adaptation tation investments. Closing the “welfare gap� strategy—including such a labor-upgrading pro- through residual compensation would entail gram—appears to be able to more than offset the mobilizing significant resources compared to negative impacts of climate change. direct project-level adaptation costs. xxii E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E While no information was available within the level, similar options were identified, along with analysis’ time frame to properly estimate the cost a focus on early warning systems and flood con- of the skill upgrading program, this finding points trol measures, agricultural technology, finance to the significant potential benefits of accelerat- and market development, renewable energy, and ing the diversification of the economy away from urban planning. The adaptation options iden- highly climate sensitive sectors, such as agricul- tified at the local and national levels generally ture. In the Wet2 scenario, for any value of the aligned with the natural resource and agriculture program cost below $1.5 billion/ year, a devel- focus in the NAPA, which also identifies needed opment strategy including a skill upgrading pro- investments in crop insurance, wetlands protec- gram like the one considered here would appear tion, carbon livelihoods, agroforestry and anti- to be preferable to the residual compensation malaria initiatives. approach. ADAPtAtion PrioritieS: loCAl-level Recommendations PerSPeCtiveS The findings of this analysis suggest that impacts Land and water management are central con- of climate change will be quite significant, par- cerns in Ethiopia, which is subject to extremes ticularly as Ethiopia approaches the middle of of drought and floods. Vulnerable groups iden- the century. While the magnitude of the impacts tified through community discussions included remains considerable—irrespective of whether asset-poor households with very limited means the climate of the future will be wetter or drier— of coping with climate hazards, the expanding several important adaptation decisions are sensi- group of rural landless who lack income opportu- tive to what climate is expected. nities, the urban poor living in flood-prone areas of cities, and the elderly and the sick due to their Given the large uncertainty on future climate limited adaptive capacity. Women and children outcomes, the approach to enhance Ethiopia’s left behind as male adults migrate for employ- climate resilience should be couched in terms ment during drought-related production failures of a gradual, adaptive, and learning paradigm. were also identified as vulnerable during and after Such an approach could be articulated for both extreme events. Other vulnerable groups identi- the shorter term—including the implementation fied included communities living on already-de- of the Growth and Transformation Plan (GTP) graded lands, and pastoral communities who face recently issued by the government—and for the severe conflicts over natural resources (especially longer term. access to land for herd mobility) with agricultural- ists and the state. Shorter term (uP to 2015) Local participatory scenario development (PSD) By and large, the Growth and Transformation workshops identified soil and forest rehabilitation, Plan supports a number of actions that, by boost- irrigation and water harvesting, improved agri- ing growth, will contribute to the enhancement cultural techniques and drought-resistant variet- of Ethiopia’s resilience to climatic shocks. Robust ies, education, and land use rights for pastoralists growth based on infrastructure investment is as adaptation preferences. Regional development likely to be the first line of defense against climate and the need for structural shifts toward service change impacts. Relatively small deviations from and industry sectors to improve employment out- the ambitious investment targets set forth by the comes were also raised as issues. At the national government for roads, dams, hydropower, water E T H I O P I A CO U N T RY ST U DY xxiii management, and irrigation would significantly roAD infrAStruCture increase longer-term vulnerability to climate change and thus make adaptation costlier. The GTP aims to expand the coverage and enhance the quality of infrastructure: “Focus will However, there are a number of additional be given to the development of roads, railways, issues the government could consider to further energy, telecommunication, irrigation, drinking enhance the contribution of GTP to Ethiopia’s water and sanitation, and basic infrastructure climate resilience—and thus, ultimately, to the developments … With regard to roads, rural ability of the country to support sustained, longer roads will be constructed on all regions and all term growth. rural kebeles will be connected (through) stan- dardized all-weather roads with main highways.� AgriCulture Modeling results show that existing infrastruc- ture design standards (level of prevention against The GTP purports to “continue the ongoing extreme events, e.g. local and regional flooding) effort of improving agriculture productivity in a are inadequate to address current climate variabil- sustainable manner so as to ensure its place of ity and will impact economic growth rates in the the engine of growth.� The analysis of this report near- and mid-term. Results from climate change indicates that under future climates many regions analyses show that this issue is likely to become of Ethiopia will face decreases in agricultural worse in the mid to long term. The government production. This suggests that agricultural pro- should consider enhancing infrastructure design duction as an engine of growth is vulnerable to standards as soon as possible. climate change and climate variability. While the more pronounced effects on crops and livestock Even under current climate, the direct bene- are likely to materialize in later decades, efforts to fits—in terms of increased lifetime—of roads enhance the resilience to climate shocks of crop designed following higher standards outweigh the yields and livestock production should be stepped corresponding costs in a discounted benefit/cost up as soon as possible, particularly on account of analysis. The case for improved design standards the lead time needed to strengthen research sys- is even stronger under climate change, irrespec- tems and to transfer and adapt findings from the tive of climate outcomes: the benefit/cost ratio lab to the field. of adopting higher design standards is 17 percent to 75 percent higher than in the baseline under Investments in improved agricultural productiv- the Wet2 scenario, and 16 to 55 percent higher ity—such as watershed management, on-farm in the Dry2 scenario (Figure 6). And in addition technology, access to extension services, trans- there are important indirect economywide ben- port, fertilizers and improved seed varieties, and efits: a more climate-resilient road network can climate and weather forecasting—will enhance avoid costly disruptions of communications links the resilience of agriculture both to droughts and and supply chains that increased flood frequency to waterlogging caused by floods. National and might bring about. local actions will need to be supported by inter- national efforts (e.g. through the CGIAR system) The GTP includes ambitious targets for upgrad- to develop climate resilient agricultural technolo- ing the road network, including 70,000 km of gies, given the global public good nature of these all-weather, Woreda (locally administrative unit) innovations. managed roads. Unpaved roads only have a 5-year design span until resurfacing and they are very susceptible to flooding damage, which has xxiv E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 6 BENEFIT/ COST RATIO OF UPGRADING ROAD STANDARDS WET 2 SCENARIO DRY 2 SCENARIO 5.00 4.00 3.50 4.00 3.00 3.00 2.50 2.00 2.00 1.50 1.00 1.00 0.50 0.00 0.00 2020 2030 2040 2050 2020 2030 2040 2050 NO CLIMATE CHANGE NO CLIMATE CHANGE WITH CLIMATE CHANGE WITH CLIMATE CHANGE very large indirect, economywide costs on supply wide planning purposes, although not for plant- chains, health and education services, etc. The level design and operation), provides support, government may want to consider a more detailed from a climate change perspective, to the GTP economic analysis of the road expansion targets targets. The projects likely to go online in the next to determine if building fewer but more climate 5 years have very low risk of being impacted by resilient roads is preferable to building a larger climate change. number of roads, which are likely to be more vul- nerable to climate shocks. The case for the for- While in the longer term (see below) hydropower mer option seems compelling under the current development will become increasingly more cli- climate, and will become even sounder under the mate sensitive, projects in the current pipeline climate of the future. In addition, should interna- are likely to be less vulnerable to shocks, as the tional climate finance resources become available overlap between their life span and the time when in the future for Ethiopia (from the Copenhagen stronger climate change effects will materialize is Green Fund or other mechanisms), the govern- relatively limited. Some climate change scenarios ment might consider utilizing these resources for actually project an increase in Ethiopian runoff, enhancing the climate resilience of the road net- resulting in larger volumes of hydropower gen- work expansion plans. eration, and thus making the case for investment in hydropower stronger. energy In the nearer term, the economics of hydropower Current water resources and Ethiopian topog- investments will be influenced less by climate, and raphy indicate an overall potential of more than more, on the demand side, by the evolution of 30,000 megawatts in economically viable hydro- domestic and external markets (Regional African power generation capacity. The GTP approach is power grids). A sustained expansion of national to focus on “the development of water and wind and foreign demand for power will be key to sup- energy options to fulfill the energy demand of the port the expansion of Ethiopia’s hydropower country,� with targets for hydropower of 6,000 to sector, which in turn will be vital to support the 8,000 MW in additional generation capacity. The country’s accelerated economic growth. hydropower analyses of this report (conducted at the monthly scale, which is adequate for sector- E T H I O P I A CO U N T RY ST U DY xxv In the short run, expansion of hydropower gen- meDium to long term eration should be accelerated as a way to support growth and to facilitate the transition of the econ- As Ethiopia looks into the next stages of devel- omy from being highly agriculture-dependent to opment—starting with preparation of the next having a broader productive base in industry and growth plan, which will follow the GTP 2011– services. Given the vulnerability of the agricul- 15)—it might want to evaluate more closely the tural sector to current climate shocks (let alone implications of climate change for its overall those to be expected in the future), strengthen- policies and infrastructure development pro- ing of the electricity sector, and in particular the grams. Early planning for the more severe cli- promotion of regional and Africa-wide power mate impacts of mid-century is desirable, so as to grids to receive Ethiopia’s excess power, should avoid locking the country into a climate-vulner- be a priority in the investment strategy. Strength- able development trajectory, particularly when it ened hydropower development can both increase comes to economic processes with a high degree near-term economic growth and make the energy of inertia, or investment decisions concerning system more climate resilient, since more reser- infrastructures with a long life span. voir storage distributed over the country provides more reliability and protection from regional Due to the uncertainty of future climate, a risk- droughts. based investment planning approach should be adopted. Robust decision-making principles are needed to minimize the “regrets� of climate- sensitive decisions. As climate shocks become more frequent and severe, the opportunity costs xxvi E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E of capital invested in projects and programs that employment, fiscal revenues, capital formation, are viable only under a limited set of climate out- the drain on government expenditure, aid flows comes becomes too large. In developing a climate to support disaster relief, and so on. risk management approach to support long-term development, some key considerations include Under climate change, renewed efforts will be the following. necessary to buffer the economy from more frequent and/or severe climate shocks. These mACroeConomiC mAnAgement include strengthening social safety nets, access to relief funds, drought early warning systems, crop Historically the Ethiopian economy has been insurance programs, grain banks, and strengthen- vulnerable to climate fluctuations (Figure 7). The ing infrastructure design. analysis of this report shows that climate variabil- ity will increase under all scenarios. Since agricul- Promote DiverSifiCAtion ACroSS ture (the economy’s most climate sensitive sector) SeCtorS of inCome AnD emPloyment is likely to remain for some time one of Ethiopia’s main engines of growth, climate-induced shocks In the longer term, however, accelerated diversi- will continue to be a threat to macroeconomic fication of income and employment sources away stability, because of the impacts on income, from climate-sensitive sectors such as agriculture Figure 7 ECONOMIC GROwTH AND CLIMATE 80 25 20 60 15 40 10 20 5 -0 % 0 -5 2000 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 -20 -10 RAINFALL VARIABILITY -15 -40 GDP GROWTH -20 -60 AG GDP GROWTH -25 -80 -30 Source: De Jong, The World Bank (2005) E T H I O P I A CO U N T RY ST U DY xxvii is likely to become increasingly important under are crucial to help identify which climate change a more erratic climate. It be should explored in path Ethiopia is actually on, and to provide inputs closer detail, particularly because it holds prom- to the adaptive management process for resource ise to be a cost-effective way to eliminate residual management. Better data on hydrometeorologi- welfare damage caused by climate change. cal processes, and stronger capacity to analyze and model them, is key to making more informed The government may want to look into ways to decisions on issues such as the number of hydro- accelerate the absorption of the rural labor force power plants, the design of individual plants, and into non-agricultural activities, including skills- the operation of the grid. upgrading programs and encouragement of growth poles around medium-size municipalities. ProACtively ADDreSS ConfliCtS in wAter uSeS evAluAte the ClimAte reSilienCe of lArge infrAStruCture ProjeCtS Under “dry� future climate scenarios, compe- tition among users of water—municipal and As we move toward mid-century, the range of industrial consumption, hydropower generation, possible climate futures broadens to encompass and irrigation—might become more acute, par- markedly different “wet� and “dry� scenarios. ticularly in certain river basins. The availability This has implications for the optimal timing of of water to downstream riparian countries might dams and other investments in water infrastruc- also be affected. ture, which is likely to be quite sensitive to climate outcomes. Large projects of this type should be Given the significant pay-off in addressing inter- subject—on account of the large capital outlays nal and transboundary conflicts on water use involved—to careful climate-robustness tests. before they arise, the government might want to consider investments in river basin planning To adequately inform the design of subsequent systems and institutional arrangements that can generations of water infrastructure projects, facilitate information sharing, dialogue, and dis- investments in enhancing national hydrometeo- pute resolution. rological services, data collection, and analysis xxviii O NE E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 1 Introduction Background Under the country track, impacts of climate change and adaptation costs are being assessed The Economics of Adaptation to Climate by sector, but only for the major economic sectors Change (EACC) has two specific analytical objec- in each case study country. Differently from the tives. The first is to develop a “global� estimate global analysis, though, vulnerability assessments of adaptation costs to inform the international and participatory scenario workshops are being community’s efforts to help those developing used to highlight the impact of climate change countries most vulnerable to climate change to on vulnerable groups and to identify adaptation meet adaptation costs. The second objective is strategies that can benefit these groups. Further- to help decision makers in developing countries more, macroeconomic analyses using CGE mod- to better understand and assess the risks posed eling are being used to integrate the sector level by climate change and to better design strategies analyses and to identify cross-sectoral effects, such to adapt to it. as relative price changes. The EACC study comprises a global track to meet the first study objective and a case study track to Scope and Limitations meet the second one. The country track comprises seven countries: Ethiopia, Mozambique, Ghana, The purpose of this study is to assist the govern- Bangladesh, Vietnam, Bolivia, and Samoa. ment of Ethiopia in its efforts to understand the Under the global track, adaptation costs for all potential economic impacts of climate change developing countries are estimated by major and to develop sound policies and investments economic sectors using country-level data sets in response to such impacts. Adaptation options that have global coverage. Sectors covered are and their costs are estimated and compared with agriculture, forestry, fisheries, infrastructure, the costs of inaction. The analysis of impacts and water resources, coastal zones, health, and eco- adaptation focuses on three sectors considered: system services. Cost implications of changes in agriculture, road infrastructure, and hydropower. the frequency of extreme weather events are also In addition, an evaluation of the economywide considered, including the implications for social repercussions of the sector-wise impacts is also protection programs. included. 2 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E The study methodology and a first set of results conducted at the subnational level (the five agro- were discussed at a workshop with government ecological regions discussed in chapter 3), but fur- officials and other stakeholders in Addis Ababa ther spatial disaggregation would be desirable for in November 2009. This report presents a set of future finer-level analysis on specific sectors such findings that take into account comments and sug- as agriculture. gestions made at the meeting, particularly on the definition of a development baseline consistent In addition, the report’s scope is limited in terms with the government’s priorities and plans. This of the adaptation options considered. The analy- report also incorporates new work on livestock, sis focused on a relatively limited numbers of irrigation, road infrastructure, and tradeoffs in options in each of the three sectors considered. water use between hydropower and irrigation. The choice was guided by availability of data, and by the possibility of utilizing straightforward The study does not have the ambition to cover and broadly accepted methodologies. More work the full range of impacts that climate change would be required to evaluate a fuller range of might have on Ethiopia’s economic and social adaptation options. This would best be done development. Given resources and time con- through follow-up studies undertaken at the level straints, it focused on three sectors (agriculture, of individual sectors, rather than at the multisec- roads, and hydropower) that play a strategic role toral, economywide level of the present study. in the country’s current economic structure and in its future development prospects. However, it is The report is structured as follows. Background important to recognize that climate change might information related to climate vulnerability spe- have important impacts in other areas not cov- cific to Ethiopia is provided in chapter 2. Meth- ered by this report; for example, the provision of ods used to assess the economics of adaptation ecosystem services (e.g. by forests or wetlands), or are described in chapter 3. Estimated impacts of the effects of climate change on human health. climate change are presented in chapter 4. Adap- As a result, the estimates of impacts are likely tation options, including economywide costs, are to be a conservative, lower-bound approxima- outlined in chapter 5, and conclusions are pro- tion of the fuller spectrum of effects that climate vided in chapter 6. change might bring about. The analysis has been E T H I O P I A CO U N T RY ST U DY 3 4 T wO E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 5 Country Background Basic Features strategic framework known as Plan for Acceler- ated and Sustained Development to End Poverty With around 75 million inhabitants, Ethiopia is (PASDEP). The third strategy plan, recently pub- the second most populous country in Sub-Saha- lished, is the “Growth and Transformation Plan- ran Africa (SSA). Despite rapid economic growth ning (GTP) 2011–2015.� over the past five years, per capita income—$255 in 2006/07 at current prices—remains well Policies related to human development, rural below the region’s average. The 2006/07 share development, food security, and capacity build- of agriculture in GDP is 46 percent, while indus- ing that were a priority in the SDPRP were aug- try accounts for 13 percent and services for 41 mented under the PASDEP. In addition to these, percent (CSA 2008). The main export commod- new strategic directions with emphasis on com- ity is coffee with a share of 35.7 percent in total mercialization of agriculture and an emerging merchandise exports in 2006/07, followed by oil urban agenda are also pursued as a means to seeds (15.8 percent), gold (8.2 percent), chat (7.8 mitigate the challenges faced by the agriculture percent), leather and leather goods (7.5 percent), sector and the overall economy. and pulses (5.9 percent) (IMF 2008). Ethiopia is a net importer of wheat; petrol, coal, and gas are The government strategy in agriculture empha- only imported. sizes a major effort to support the intensification of marketable farm products by both small and large farmers. To help jump-start this process, a Current Growth and Poverty range of public investments have been identified Reduction Policies and are being implemented. The major invest- ments include the construction of farm-to-market roads and area irrigation through multipurpose The Ethiopian government prepared and adopted dams. Other related services include measures to three successive poverty reduction strategy pro- improve land tenure security, reforms to improve grams. The first was the Sustainable Development the availability of fertilizer and improved seeds, and Poverty Reduction Program (SDPRP). that and specialized extension services for differenti- covered three years from 2000/01 to 2003/04. ated agricultural zones and types of commercial The second is a five-year (2005–10) guiding agriculture (Ahmed et al. 2009). 6 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Growth and Transformation Planning (2011–2015), between climate change variations and overall recently issued by the Ministry of Finance and economic performance. Economic Development, outlines growth strat- egies planned for the next 5 years. The agricul- Ethiopia is historically prone to extreme weather tural sector is identified as a key driver of growth. events. Rainfall in Ethiopia is highly erratic, As this sector is particularly sensitive to climate, and most rain falls intensively, often as convec- there is an opportunity to consider the implica- tive storms, with very high rainfall intensity and tions of the EACC findings with respect to the extreme spatial and temporal variability. Since the GTP (Chapter 5). early 1980s, the country has suffered seven major droughts, five of which led to famines in addition to dozens of local droughts (Diao and Pratt 2007). Climate and Vulnerability to Survey data show that between 1999 and 2004 Climate Change more than half of all households in the country experienced at least one major drought shock (UNDP 2007). Major floods occurred in different Around 45 percent of Ethiopia consists of a high parts of the country in 1988, 1993, 1994, 1995, plateau with mountain ranges divided by the East 1996, and 2006 (ICPAC 2007). African Rift Valley. Almost 90 percent of the population resides in these highland regions with It is important to highlight that climate change is elevations greater than 1,500m above sea level. projected to take place over the course of the next The surrounding lowlands (<1,500m) are mostly century. In accordance with the approach fol- populated by pastoralists. Ethiopia’s varied topog- lowed by the broader EACC study, this report will raphy has traditionally been associated with three only consider the implications of climate change main climatic zones: (a) the warm, semiarid Kolla, to 2050 (by decadal increment), even though cli- <1,500m above sea level; (b) the Woinadega, a cool, mate change is expected to be most severe toward subhumid temperate zone, 1,500–2,400m above the end of the century. The long time frame con- sea level; and (c) the cool and humid Dega, mostly sidered—40 years into the future—means that >2,400m above sea level. As the population dynamic processes are important. Two distinct increased and agricultural activities expanded, approaches to the analysis were used: quantitative two more were added at the extreme ends of the modeling for biophysical and economic assess- nation’s climatic conditions: the hot, arid Bereha, ments, and participatory social analysis. and the cold and moist Wurch. Ethiopia is heavily dependent on rainfed agricul- Modeling Approach ture. Its geographical location and topography in combination with low adaptive capacity entail a The economywide modeling component of the high vulnerability to adverse impacts of climate EACC Ethiopia case study involves linking a change. Regional projections of climate models dynamic multisectoral and multiregional com- not only predict a substantial rise in mean tem- putable general equilibrium model (CGE) with peratures over the 21st century, but also suggest a range of sectoral climate change impact mod- an increase in rainfall variability with a rising fre- els that generate quantitative estimates of effects quency of both extreme flooding and droughts on water systems, agriculture, hydro-energy, and due to global warming. The agricultural sector road transport infrastructure (Figure 8). also affects performance in other sectors of the economy. Hence, there is a strong observable link E T H I O P I A CO U N T RY ST U DY 7 Figure 8 FLOw CHART OF MODEL SEqUENCING Location GCM GENERAL CIRCULATION MODEL TEMPERATURE PRECIPITATION Surface Slope CliRun CLIMATE RUNOFF TEMPERATURE PRECIPITATION TEMPERATURE RAINFALL RUNOFF Soil Composition Reserve Specifications Crop Type Discount Rate IMPEND INVESTMENT MODEL FOR CliCrop PLANNING ETHIOPIAN CLIMATE CROP AND NILE DEVELOPMENT WATER RESOURCE ALLOCATIONS IRRIGATION DEMAND CROP YIELD Reservoir Specifications River Basin Management Municipal and Industrial Demand WEAP WATER EVALUATION AND PLANNING RESOURCE ACCOUNTING Discount Rate CGE COMPUTABLE GENERAL EQUILIBRIUM 8 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Global circulation models (GCMs) lie at the in the next phase of work. Reservoir flow calcu- beginning of the modeling chain. These mod- lated in WEAP may change previous inputs into els take as inputs quantities of greenhouse gases IMPEND, thus requiring the net benefits to be emitted and produce climate outcomes through recalculated and their implications re-modeled time and across space as a function of these emis- in WEAP. The following subsections describe sions and initial conditions. A series of hydrologi- the sector models CliCrop, a road algorithm, cal and crop models are the next elements in the IMPEND, and CGE. chain. These models take climate outcomes and convert them to natural outcomes on the ground. Outcomes of particular interest are crop yields, Approach to Climate Model which temperature and rainfall can influence Uncertainty substantially, and hydrological flow within river basins, including the incidence of floods. Historic and future climate inputs specific to Ethi- The flow of information through the integrated opia and its river basins—such as monthly tem- river basin and water resource model is generally perature and precipitation—are used to drive the linear, as shown above in Figure 8. Climate data river basin and water resource model and crop is entered into CliRun and CliCrop in order to models outlined below. Historic inputs have been produce streamflow runoff estimates and crop gathered using data from the Climate Research irrigation demand estimates. Unit (CRU) on global monthly precipitation and temperature. Information on future climate has CliRun is a two-layer, one-dimensional infiltration been taken from four general circulation models and runoff estimation tool that uses historic run- (GCMs), forced with different CO2 emission sce- off as a means to estimate soil characteristics. A narios to represent the total possible variability in 0.5° by 0.5° historic global runoff database gener- precipitation. ated by the Global Runoff Data Center (GRDC) and measured historic runoff is used to calibrate There were four climate change scenarios used CliRun. CliCrop is a generic crop model described for this analysis. The scenarios were selected to in chapter 2. Inflows calculated using CliRun are span the range of possible climatic change as passed to IMPEND, where storage capacity and measured by the Climate Moisture Index (CMI). irrigation flows are optimized to maximize net The CMI is an aggregate measure of annual benefits. The outputs from IMPEND, along with water availability. It provides for an integrated the irrigation demands estimated from CliCrop, measure of the impact of climate change on are then passed to the Water Evaluation and soil moisture and runoff from changes in both Planning System (WEAP), where water storage temperature and precipitation. Since CMI is an and hydropower potential are modeled based on annual measure, it does not account for seasonal their interaction with the climate and socioeco- changes and the potential for increased flood- nomics of the river basins being modeled. ing due to changes in daily and monthly scale precipitation processes in the midst of annually Finally, this information is passed to the CGE, drier climate. Two scenarios, the Dry1 and Wet1, where the economic implications of the modeled come from the EACC global track (Box 1). They data are assessed. Within the river basin model are the wettest and driest scenarios measuring there is, however, one interaction with the poten- the changes in CMI over all the land area of the tial for nonlinearity. The interaction between globe. These scenarios were examined to provide IMPEND and WEAP is an iterative process for linkage and comparison of the EACC global depending on the scenario and will be completed and country-study tracks. E T H I O P I A CO U N T RY ST U DY 9 box 1 CLIMATE SCENARIOS IN THE GLOBAL For Ethiopia both of the test scenarios show mod- AND ETHIOPIA TRACK OF THE EACC est temperature increase and very small changes in annual precipitation. Dry2 and Wet2 are the In the EACC global track study, two mod- els—the National Center for Atmospheric extremes scenarios of CMI change evaluated Research (NCAR) CCSM3 and Common- only over the area of Ethiopia (Table 3). Over wealth Scientific and Industrial Research Ethiopia, the Dry2 scenario shows reductions in Organization (CSIRO) Mk3.0 models—with average annual rainfall over 2045–55 greater than SRES A2 emission forcings were used to 15 percent. The Wet2 scenario show increases in model climate change for the analysis of average annual rainfall close to 25 percent over most sectors. These models capture a full most of the rest of the country. spread of model predictions to represent Table 3 GCM SCENARIOS FOR ETHIOPIA inherent uncertainty, and they report specific climate variables—minimum and maximum COUNTRY TRACK STUDY CMI Devia- temperature changes—needed for sector Scenario GCM SRES tion* (%) analyses. Though the model predictions do ncar_ not diverge much for projected tempera- Wet 1: Global Wet ccsm3_0 A2 10 ture increases by 2050—both projecting csiro_ Dry 1: Global Dry mk3_0 A2 -5 increases of approximately 2°C above pre- ncar_ industrial levels—they vary substantially for Wet 2: Ethiopia Wet pcm1 A1b +23 precipitation changes. Among the models Dry 2: Ethiopia Dry ipsl_cm4 B1 -15 reporting minimum and maximum tempera- ture changes, NCAR was the wettest sce- Note: *The Climate Moisture Index (CMI) depends on average annual precipitation and average annual potential evapotranspira- nario and CSIRO the driest scenario globally tion (PET). If PET is greater than precipitation, the climate is con- sidered to be dry, whereas if precipitation is greater than PET, the based on the climate moisture index (CMI). climate is moist. Calculated as CMI = (P/PET)-1 {when PET>P} and CMI = 1-(PET/P) {when P>PET}, a CMI of –1 is very arid and a CMI of +1 is very humid. As a ratio of two depth measurements, CMI In line with the global track, the climate is dimensionless. Average annual PET is a parameter that reflects the amount of water lost via evaporation or transpiration (water projections from these two GCMs are used consumed by vegetation) during a typical year for a given area if to generate the “global wet� and “global sufficient water were available at all times. Average annual evapo- transpiration (ET) is a measure of the amount of water lost to the dry� scenarios for the Ethiopia country- atmosphere from the surface of soils and plants through the com- track study, referred to as Wet1 and Dry1. bined processes of evaporation and transpiration during the year (measured in mm/yr). Evapotranspiration, which is both connected In addition, the climate projections from the to and limited by the physical environment, is a measure that quan- tifies the available water in a region. Potential evapotranspiration two GCM/SRES combinations with the low- is a calculated parameter that represents the maximum rate of ET est and highest climate moisture index for possible for an area completely covered by vegetation with ade- quate moisture available at all times. PET is dependent on several Ethiopia were used to generate “Ethiopia variables, including temperature, humidity, solar radiation, and wind velocity. If ample water is available, ET should be equal to PET. dry� and “Ethiopia wet� scenarios. Precipi- tation and temperature data acquired from these simulations are used to estimate the interPreting ClimAte SCenArioS availability of water at a sub-basin scale. Historical climate data for each basin have IPCC reports that while in terms of annual been gathered using available precipitation precipitation in Ethiopia, some climate models and temperature data when available, along indicate increases and some decreases, all mod- with the Climate Research Unit’s 0.5° by 0.5° els suggest increases in precipitation intensity at global historical precipitation and tempera- the daily and weekly scale (Figure 9). This implies ture database. more flooding even in scenarios that suggest more drought. Both increased flooding and increased drought are projected by the same scenarios. 10 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Another important aspect of the IPCC climate change scenarios is that they are generated by Crop Model Description dynamic models that are transient over the cen- tury. While temperature mostly increases mono- CliCrop is a generic crop model used to calculate tonically, precipitation does not. Figure 10 shows the effect of variations in CO2 daily precipitation a set of climate change precipitation projections patterns caused by climate change on crop yields averaged over East Africa. These show that there and irrigation water demand. The model was is a great deal of variability from year to year. developed in response to the available crop mod- Some scenarios even switch from increasing to els that use monthly average rainfall and tem- decreasing precipitation, or from decreasing to perature to produce crop outputs. These monthly increasing, over different periods of the 21st Cen- models do not capture the effects of changes in tury. These changes can lead to climate change precipitation patterns, which greatly impact crop impacts that vary by decade within a less transient production. For example, most of the IPCC multi-decadal trend of reduction or increase in GCMs predict that total annual precipitation will precipitation. decrease in Africa, but rain will be more intense Figure 9 CGM PROjECTION ON DAILY PRECIPITATION INTENSITY Figure 10 PRECIPITATION CHANGES FOR IPCC A1B SCENARIO E T H I O P I A CO U N T RY ST U DY 11 and therefore less frequent. Currently CliCrop is able to produce predicted changes in crop yields Livestock Model Description due to climate change for both rainfed and irri- gated agriculture, as well as changes in irrigation This analysis evaluates the effect of changing cli- demand. Since it was developed to study effects mate conditions on livestock productivity, relying of agriculture on a global or continent scale, it on a hybrid approach that has two components: a is a generic crop model. CliCrop can be used at “biophysical� component that considers the effect a variety of scales, from field level to agroeco- of temperature on expected livestock incomes, logical zone. It was developed to measure climate and a “feed� component that incorporates the change impacts on water stress and examine field- effect of changing availability of livestock feed. level adaptation options, including mulching and (Ideally, a livestock process model would be avail- water harvesting. able that was capable of explicitly analyzing the effect of changing climate conditions on livestock The inputs into CliCrop are CO2 concentration, productivity, and the resulting adaptive responses temperature, and precipitation soil parameters of livestock farmers. Because no such model is (field capacity, wilting point, saturated hydraulic publicly available, this analysis relies on the hybrid conductivity, and saturation capacity); historic approach described above.) The biophysical com- yields for each crop by province; crop distribution ponent relies on the results of a structural Ricard- by province; and current irrigation distribution ian model of African livestock developed by Seo estimates by crop. The weather inputs into Cli- and Mendelsohn (2006).1 This model measures Crop for future scenarios are extracted directly the interaction between temperature and livestock from the four GCMs referred to above. The daily and considers the adaptive responses of farmers distributions of the precipitation and temperature by evaluating which species are selected, the num- are derived from the NASA POWER data set for ber of animals per farm, and the net revenue per both the baseline and the future scenarios. All of animal under changes in climate. the soil parameters required are extracted from the FAO soils database. The current analysis transfers the findings from Seo and Mendelsohn (2006) to the Ethiopia-spe- The output of CliCrop is used as input to the cific context. The feed component of the hybrid computable general equilibrium model (CGE) as model uses the outputs of CliCrop to identify shocks/stressors caused by the predicted weather how climate change may affect the yields of mil- changes from the GCMs. The CGE model let, a primary source of livestock feed in Ethiopia. includes details about Ethiopia’s agricultural crop To evaluate the overall effects of climate change and livestock commodities, as well as capital, on livestock productivity, the biophysical and feed land, and labor inputs. The CGE model is used components are combined to form vectors of to study and evaluate impacts of climate change changes in expected livestock productivity under adaption strategies in the agriculture sector and the baseline and four climate scenarios between consequently in the other sectors of the economy. 2001 and 2050. Details of the livestock modeling The output of CliCrop will also be used in the approach are presented in Annex 8. WEAP model used to calculate the changes in irrigation demand on the reservoir water supply. 1 The Ricardian approach examines how crop production varies in regions of different climates and then infers the effect of climate from these differences (Mendelsohn et al. 1994). The approach explicitly embeds farm adaptations as found in the pooled (time series and cross-sectional) data. Using these data, it is possible to forecast how climate changes affect profits and production in future years. 12 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E BioPhySiCAl ComPonent In all three tables, beef cattle and chickens are more sensitive to changes in climate than dairy Although the direct effects of heat stress on live- cattle, goats, and sheep. In prior studies, beef stock have not been studied extensively, warm- cattle have been found to experience increases in ing is expected to alter the feed intake, mortality, mortality, reduced reproduction and feed intake, growth, reproduction, maintenance, and produc- and other negative effects as temperatures rise tion of animals. Collectively, these effects are (Adams et al. 1999). Butt et al. (2005) found that expected to have a negative impact on livestock small ruminants (such as goats and sheep) are productivity (Thornton et al. 2009). Below, the more resilient to rising temperatures than beef approach and findings of the Seo and Mendel- cattle. Chickens are particularly vulnerable to sohn (2006) study are reviewed, and the transfer climate change because they can only tolerate methodology is explained. narrow ranges of temperatures beyond which reproduction and growth are negatively affected. Seo and Mendelsohn rely on a survey of over Further, increases in temperature caused by cli- 5,000 livestock farmers in ten African countries, mate change can be exacerbated within enclosed one of which was Ethiopia. In this data set, the poultry housing systems. variation in livestock productivity and expected incomes in different regions demonstrates a clear Seo and Mendelsohn (2006) predict that rela- relationship to regional climate, which provides a tive to the baseline, the probability of choosing mechanism (i.e., through spatial analogue) to statis- beef cattle and chickens will decline with rising tically analyze how climate change may affect live- temperatures, but that the probability of select- stock incomes across Africa.2 The authors develop ing dairy cattle, goats, and sheep will increase. a three-equation farm-level model. The first equa- Table 5 shows that predicted changes in expected tion predicts the probability of selecting each live- income per animal are most dramatic for beef stock type as the primary animal for the farm, the cattle and chickens, which fall 32 percent and 48 second predicts the net income of each animal, percent, respectively, with an increase of 5.0°C. and the final equation predicts the number of ani- Finally, Table 6 indicates that rising temperatures mals on each farm. Farm net revenues are the sum reduce the predicted number of beef cattle and product of these three outputs; that is, the prob- chickens on each farm, but will increase the num- ability of selecting each type of animal multiplied ber of other livestock types. by the number of animals and then the expected income per animal, summed across animal types. SM uses this model to evaluate fixed changes in temperature of 2.5°C and 5.0°C from the base- line. The resulting predicted changes in the prob- ability of selecting an animal, expected income per animal, and the total number of animals under these changing climate conditions are pre- sented in Tables 4 through 6. 2 Because the raw data from this survey were not available, it was not possible to compare the climatic conditions observed in the Seo and Mendelsohn (2006) survey to the conditions in Ethiopia. If conditions are more attuned to the warmer and more variable climates of western Africa, the impacts estimated here may be overstated. E T H I O P I A CO U N T RY ST U DY 13 Table 4 PREDICTED CHANGE IN THE PROBABILITY OF SELECTING EACH ANIMAL AS THE PRIMARY ANIMAL TYPE FOR THE FARM (%) Beef cattle Dairy cattle Goats Sheep Chickens Baseline probability 11.8 23.1 23.4 19.4 22.3 Increase temp 2.5C -1.7 +0.4 +0.8 +3.3 -2.8 Increase temp 5C -3.8 +2.1 0.0 +8.7 -7.0 Table 5 PREDICTED CHANGE IN NET INCOME (US$) PER ANIMAL Beef cattle Dairy cattle Goats Sheep Chickens Baseline income 145.54 132.09 6.49 11.77 1.14 Increase temp 2.5C -27.80 -3.40 -0.81 -2.55 -0.34 Increase temp 5C -47.09 -21.36 -0.54 -3.49 -0.55 Table 6 PREDICTED CHANGE IN NUMBER OF ANIMALS PER FARM (ANIMALS/HOUSEHOLD) Beef cattle Dairy cattle Goats Sheep Chickens Baseline number 63.47 23.84 15.36 34.05 790.09 Increase temp 2.5C -9.00 1.84 1.45 0.35 -112.61 Increase temp 5C -18.96 2.88 2.14 3.20 -183.84 14 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Table 7 PREDICTED CHANGE IN ExPECTED INCOME* Bootstrap** lower Bootstrap upper Mean (US$/farm) % Change 95% 95% (US$/farm) (US$/farm) Expected income 3,023 Increase temp 2.5C -964 -31.90 -1,077 -722 Increase temp 5C -2,083 -68.89 -2,452 -1,631 Note: *SM’s model does not include any information on livestock prices, only data on per animal net income. As a result, there is no interac- tion between supply of cattle and market prices in the model. In a model that considered market prices, as livestock numbers fell, prices would likely rise, thus lessening the economic impact of rising temperatures in the model.. ** Because effects on income are the product of three separate statistical processes, SM were unable to estimate variance parametrically (i.e., from a well-defined probability distribution). Instead, they used bootstrap methods, which involved resampling from the underlying observed data and then re-running their three-stage model to generate an approximating distribution. This approximating distribution provided the 95 percent confidence interval reported here. Table 5 combines the above effects into predicted feeD ComPonent changes (within a 95 percent confidence interval) in expected farm-level income. Seo and Mendel- The feed component of the model reflects the sohn predict a reduction in expected income of availability of feed for livestock under changing 32 percent and 69 percent resulting from a 2.5°C climate conditions. Rather than transferring the and 5°C increase in mean temperatures. Overall, results of a study, the feed vectors are ratios of the although both the probability of selection and the projected millet yields from 2001 to 2050 under number of animals per farm increases for certain the baseline and four climate change scenarios types of livestock, the negative responses to cli- within each AEZ, relative to the mean millet mate overwhelm these positive effects. In addi- yields under the baseline scenario (25 vectors in tion to the fact that the predicted change in net total). As a result, the feed ratio can be greater income per animal declines universally across or less than one. The millet yield projections are animal types, the animal types that are most developed using CliCrop, which produces projec- affected by changes in temperature (beef cattle tions of both irrigated and rainfed yields for the and chickens) collectively make up over 60 per- 2001 to 2050 period; this analysis uses the rainfed cent of baseline TLUs and are predicted to gen- millet yields. erate net incomes that are over 40 percent lower with a 2.5°C increase in mean temperatures. ComBining the BioPhySiCAl AnD feeD For comparison, dairy cattle generate the largest ComPonentS increases in incomes under the 2.5°C rise at only 7 percent. Next, the vectors from the biophysical and feed components are combined in a weighted aver- To generate vectors of projected changes in age where each vector is given equal weight; that expected income from livestock within each AEZ, is, a simple average. If additional information the analysis uses the framework and findings of becomes available on the relative importance of Seo and Mendelsohn coupled with country- these two factors in determining livestock pro- specific livestock data and climate projections. ductivity, these weights could be adjusted accord- Details of this transfer are in Annex 8. ingly. The final product is 25 vectors—sets of five AEZ vectors for the baseline and each of the four E T H I O P I A CO U N T RY ST U DY 15 climate scenarios—that consider both the bio- physical and feed effects of climate on livestock Road Transport Model productivity. The findings of the livestock analysis Description are presented in the livestock results in chapter 4. The stressor-response methodology used in this Drought Model Description report is based on the concept that exogenous fac- tors, or stressors, have a direct affect on and sub- Periodic drought in Ethiopia causes severe reduc- sequent response by infrastructure materials used tions in food availability, causing government in roads.5 In the context of climate change and expenditures on food aid and emergency drought infrastructure in this section, the exogenous fac- relief to swell during these periods. In recent tors are the individual results of climate change, years, the Ethiopian government has maintained including changes to precipitation levels and tem- records of expenditures on vulnerability and food peratures. Therefore, a stressor-response value is security (VFS), which have typically increased the quantitative impact that a specific stressor has extreme droughts (e.g., 1999–2000 and 2003– on a specific infrastructure element. For example, 04).3 Using a reduced-form statistical model, this an increase in precipitation level is going to have analysis estimates the relationship between cli- a specific quantitative impact on the life span of mate drivers and Ethiopian VFS expenditures, an unpaved road, based on the change in pre- and develops nationwide projections of those cipitation. (The quantitative estimates presented expenditures from 2001 to 2050 under the base here, as well as the analysis of the related policy and four climate scenarios.4 implications, consolidate the findings of COWI (2009)). The reduced form statistical model relies on historical expenditure data on VFS, historical Variations across infrastructure type in the rela- climate data in drought-prone regions, and a tionships between climate and life span reflects, dummy variable reflecting a significant increase among other factors, differences in the materials in VFS funding from 2001/02 forward because with which different types of infrastructure are of a federal special purpose grant for food secu- constructed, and the ways in which different types rity (World Bank 2008). The latter variable is of infrastructure are used; for example, buildings included in the statistical formulation to explain often provide heating and cooling. In addition, this large surge in expenditures on VFS. Detailed variation in the stressor-response relationship by information related to the drought model is con- country reflects inter-country variation in labor tained in Annex 9. and materials costs as well as terrain; for exam- ple, varying degrees of flat versus mountainous terrain. In this analysis, stressor-response factors were developed based on multiple inputs. A com- 3 The original data is reported by Ethiopian calendar year, which extends from September 11th to September 10th on the Grego- bination of material science reports, usage stud- rian calendar (e.g., September 11, 1997 to September 10, 1998 on ies, case studies, and historic data were all used to the Gregorian calendar is 1990 on the Ethiopian calendar). All dates in this analysis are reported in Gregorian terms. develop response functions for the infrastructure 4 A reduced form statistical model is a model that identifies statisti- categories. Where possible, data from material cal relationships between a dependent variable and one or more manufacturers was combined with historical data independent variables. In the present model, these independent variables are used to explain variations in VFS expenditures. Because this drought model uses observed historical data, the 5 The impact figures presented here are conservative, lower-bound results of this analysis indicate correlation between the dependent estimates of actual climate change damage. See COWI (2010) for and independent variables, but not causation. additional analysis. 16 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E to obtain an objective response function. How- new construction or an increase in maintenance ever, when these data were not available, response for existing infrastructure. As documented, this functions were extrapolated based on perfor- strategy is realized individually for the various mance data and case studies from sources such infrastructure categories. as departments of transportation or government ministries. Determining impacts. The dose-response rela- tionship between climate change and the cost To provide a contextual boundary for the func- of maintaining road networks is a central con- tion derivation, two primary climate stressors cern for climate change adaptation (Table 6). To were included; temperature and precipitation. determine the costs of climate change impact, Cost data for the general study were determined two different elements are considered; (1) costs based on both commercial cost databases and to maintain existing roads, and (2) costs to adapt specific country data where available. roads by improving the roads at regular design life intervals. The former of these can be consid- Finally, the stressor-response factors presented ered the direct impact and necessary response to below are divided into two general categories; climate change. The latter is the optional adap- (1) impacts on new construction costs, and (2) tation that can be done to minimize increased impacts on maintenance costs. New construction maintenance costs. cost factors are focused on the additional cost required to adapt the design and construction of Paved road maintenance. In determining the a new infrastructure asset, or rehabilitating the climate-change-related costs for paved roads, the asset, to changes in climate expected to occur underlying focus is to maintain the road network over the asset’s life span. Maintenance cost effects that is in place by increasing spending on mainte- are those maintenance costs—either increases or nance to retain the 20-year design life cycle. The decreases—that are anticipated to be incurred 20-year life cycle is based on the assumption that due to climate change to achieve the road’s design roads are repaved at the end of each 20-year life life span. In each of these categories, the under- cycle in a standard maintenance cycle. To deter- lying concept is to retain the design life span for mine the increased impact of climate change the structure. This premise was established as a stressors on this maintenance cycle, the impact baseline requirement in the study due to the pref- of temperature and precipitation is applied to the erence for retaining infrastructure for as long as road. These two factors are the significant factors possible rather than replacing the infrastructure for road maintenance, as precipitation impacts on a more frequent basis. Achieving this goal may both the surface and the roadbed, while tempera- require a change in the construction standard for ture impacts the asphalt pavement based on the Table 8 DOSE-RESPONSE DESCRIPTIONS FOR MAINTENANCE COSTS Road Type Precipitation Temperature Paved Roads – Existing Change in annual maintenance costs per Change in annual maintenance costs per km per 10 cm change in annual rainfall pro- km per 3°C change in maximum of monthly jected during life span relative to baseline maximum temperature projected during climate. life span. Unpaved Roads Change in annual maintenance costs per 1% Not estimated. Impact likely to be minimal. change in maximum of monthly maximum precipitation projected during life span. E T H I O P I A CO U N T RY ST U DY 17 design of the asphalt mix. Using this approach, this approach, the impact is based on potential the cost increase for the annual maintenance life-span reduction that could result from climate based on dose-response values is based on the change if maintenance practices are not adjusted concept of infrastructure life-span decrements. In to meet the increased climate stress (Box 2). box 2 MODELING IMPACTS ON PAVED AND UNPAVED ROADS PAveD roADS Modeling paved roads involves two basic steps as seen in Equation 1 below: (1) estimating the life- span decrement that would result from a unit change in climate stress, and (2) estimating the costs of avoiding this reduction in life span. For example, if a climate stressor is anticipated to reduce the life span by 2 years or 10 percent, and the cost to offset each percent of reduction is equal to a percentage of the current maintenance cost, then the total would be (10%)(current maintenance cost) to avoid decreasing the current design life span. MTERB = (LERB)(CERB) (Equation 1) where MTERB: Change in maintenance costs for existing paved roads associated with a unit change in climate stress LERB: Potential change in life span for existing paved roads associated with a unit change in climate stress CERB: Cost of preventing a given life-span decrement for existing paved roads. To estimate the reduction in life span that could result from an incremental change in climate stress (LERB), we assume that such a reduction is equal to the percent change in climate stress, scaled for the stressor’s effect on maintenance costs (Equation 2). S LERB = (SMT) (Equation 2) BaseS where S: Change in climate stress (i.e., precipitation or temperature) BaseS: Base level of climate stress with no climate change SMT = Percent of existing paved road maintenance costs associated with a given climate stressor (i.e., precipitation or temperature) unPAveD roAD CAlCulAtionS (DireCt reSPonSe methoDology) The change in unpaved road maintenance costs associated with a unit change in climate stress is estimated as a fixed percentage of baseline maintenance costs. In general terms, this approach is summarized by Equation 3. MTURR = M X BURR (Equation 3) where MTURR: Change in maintenance costs for unpaved roads associated with a unit change in climate stress M: Cost multiplier BURR: Baseline maintenance costs 18 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E The potential change in life span is dependent has established a norm in construction estimating on the change in climate stress. For precipita- that maintenance is based on initial construction tion effects, a reduction in life span is incurred by costs as a standard estimating procedure. Given existing paved roads with every 10 cm increase these two factors, we directly relate changes in in annual rainfall. For temperature, a life-span maintenance costs to specific changes in climate reduction is incurred with every 3°C change in or infrastructure design requirements using a maximum annual temperature for existing paved direct response methodology (Box 2). roads (FDOT 2009a; FEMA1998; Miradi 2004; Oregon DOT 2009; Washington DOT 2009). The stressor-response relationship discussed in Box 2 is applied as the change in maintenance The estimate of the potential reduction in life costs associated with a 1 percent change in span associated with a given change in climate maximum monthly precipitation. Research has stress reflects the contribution of that stressor demonstrated that 80 percent of unpaved road to baseline maintenance costs (SMT). For paved degradation can be attributed to precipitation, roads, precipitation-related maintenance repre- while the remaining 20 percent is due to traffic sents 4 percent of maintenance costs and temper- rates and other factors (Ramos-Scharron and ature-related maintenance represents 36 percent MacDonald 2007). Given this 80 percent attribu- (Miradi 2004). tion to precipitation, maintenance costs increase by 0.8 percent with every 1 percent increase in After estimating the potential reduction in life the maximum of the maximum monthly pre- span associated with a given climate stressor, we cipitation values projected for any given year. estimate the costs of avoiding this reduction in life Published data indicates that the baseline cost of span. To estimate these costs, we assume that the maintaining an unpaved road is approximately change in maintenance costs would be approxi- $960 per km (Cerlanek et al. 2006). Therefore, mately equal to the product of (1) the potential for every 1 percent increase in maximum precipi- percent reduction in life span (LERB), and (2) the tation, we assume a maintenance cost increase of base construction costs of the asset. Therefore, $7.70 per km. if we project a 10 percent potential reduction in life span, we estimate the change in maintenance costs as 10 percent of base construction costs. We Flood Cost Model for Road estimate base construction costs for a primary Infrastructure paved road at $500,000 per km. Unpaved road maintenance. Maintenance of flooDing CoSt imPACtS unpaved roads is primarily focused on the need to reseal the road every five years to preserve a The development of annual cost estimates for the usable driving surface and reduce the impact flooding impacts from climate change combines of erosion from precipitation. To estimate dose- life-cycle concerns, climate impacts and resis- response values for unpaved road maintenance tance, and cost estimates. Specifically, the cost of costs, we use a more direct approach for estimat- climate change impacts on roads must be exam- ing the cost impact of changes in climate stres- ined from the design phase through the main- sors. The need for this approach is based on two tenance phase and through the rehabilitation factors; (1) maintenance tasks for unpaved roads phase. This life-cycle perspective incorporates the are related closely to the initial construction tasks, complete spectrum of costs that climate change and (2) the reduced life span of an unpaved road imparts on the road infrastructure. This section E T H I O P I A CO U N T RY ST U DY 19 outlines the overall methodology and describes biased errors. Given the constraints on data the output generated by this life-cycle perspective, required for the CGE analysis (annual series of which is used as input to the CGE models. damages from 2010 to 2050), this was the only feasible option given the availability of GCM DAmAge imPACtS output. Road and highway flood losses were calculated A custom damage function was used to gener- based on monthly runoff estimates generated ate loss estimates based on the return period of from CliRun simulations. While road and high- the precipitation intensities. The damage func- way damages actually occur under flooding con- tion given below in Figure 4 is a general damage ditions over a range of time scales (hourly, daily, function for flooding impacts on roads, based on weekly), only monthly GCM data were available assumed reasonable design standards and engi- for the time frame required for the EACC study. neering judgment. Climate change will have an By using the monthly data to estimate damages, effect on the curve in that the frequency and the underlying implication is that the distribution intensity of floods may change; for example, of damages resulting from sub-monthly precipi- what originally was a 70-year flood may occur tation follows the monthly rainfall distribution. more frequently, such as a 50-year flood. This will While this assumption is not likely to be strictly translate to damage becoming more severe on a accurate, neither is it likely to introduce strongly more frequent basis. Figure 11 FLOOD DAMAGE RELATIONSHIP 30 25 20 PERCENT DAMAGE 15 10 5 0 0 20 40 60 80 100 120 140 RETURN PERIOD (YEARS) 20 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E life-CyCle ConCernS repaving operations. These measures—which include increased drainage, increased road thick- The first component of the flooding methodol- ness, or a change in asphalt mix—are applied at ogy translates information on floods into actual the end of a 20-year design life span, which then kilometers of road that will be damaged. Flood climate-proofs the roads and provides resistance modeling generates information to calculate the to additional flooding levels induced by climate percentage of roads anticipated to be damaged change. For example, if 1,000 kilometers have the in each climate region, based on the intensity potential to be damaged, but 200 kilometers have and recurrence of the floods as well as the cor- been upgraded previously, then 800 kilometers responding damage curve. These percentages are are considered the actual pool from which the provided in a time series from 2010 through 2050. damage calculation will be applied. For example the damage calculator may return that in a specific climate zone, in 2025, that .12 For unpaved roads, the potential number of dam- percent of the roads will be damaged. This pro- aged roads is reduced by the number of roads vides the basic input for the system to determine that have been damaged and then had an adap- the specific number of roads that will be damaged tation approach applied over the previous five in a specific year. years of its anticipated design life span prior to re-grading. Adaptations for this group of roads The determination of how many kilometers will include increasing the thickness of the gravel and be damaged is dependent on two factors; whether applying a sealer to reduce the amount of erosion the roads are paved or unpaved, and how many due to flooding. In the non-adaptation scenario, it roads have had adaptations applied to them. In is assumed that no adaptation policies have been terms of the former, the road inventory is used to applied and all roads in the region remain as determine the number of primary, secondary, and potential roads that are susceptible to flooding. tertiary roads that exist in each region. Addition- ally, the inventory is used to determine the num- ber of roads that are paved and unpaved within IMPEND Model Description those classifications. From the combination of the damage projection for the specific zone and Hydropower simulation was done using a hydro- the inventory for that zone, the total number of power planning model developed for Ethiopia, potential damaged roads is determined. This is the IMPEND model (Investment Model for Planning illustrated by the following formula. Ethiopian Nile Development) (Block and Strzepek 2009). IMPEND was developed to plan reservoirs PDR = (∑PR * DE) + (∑UR * DE) and power generation facilities on the Upper Blue PDR = (∑PR * DE) + (∑UR * DE) Nile River in Ethiopia. It is a water accounting and optimization program written in the General Where: PDR = Potential Damaged Roads Algebraic Modeling System Software (GAMS PR = Paved Roads 2005) and requires measurements or estimates DE = Damage Estimate of monthly stream flow, net evaporation at each UR = Unpaved Roads reservoir, and discount rate, as well as reservoir attributes such as the surface area of each reser- The overall total is modified by one of two fac- voir, design head, and peak energy output. Out- tors. For paved roads, the potential number is put includes a time series of energy generation reduced if the roads have had adaptation mea- and associated project costs. sures applied to them during the most recent E T H I O P I A CO U N T RY ST U DY 21 The IMPEND simulation required estimates of to investigate intersectoral conflicts between monthly flow and net evaporation from the hydro- water demands. In particular, the study focuses logic model CliRun. The CliRun model was used on projected interactions between climate-driven to estimate flow into the hydropower generation changes in water availability and growing demand facilities for the four future climate realizations as for water in the hydropower and irrigation sectors. described above. These flow estimates were used The analysis projects perturbations, or “shocks,� to in IMPEND to estimate the potential power gen- hydropower production and irrigated crop yields eration available under these hydrologic condi- resulting from these conflicts from 2001 to 2050 tions. All other assumptions and conditions were across Ethiopia. These questions are addressed identical with the baseline; operating assump- using the water evaluation And planning (WEAP) tions, surface areas of the reservoirs, etc. were all tool (Sieber and Purkey 2007), which is a soft- held constant. Only influent flow changed. ware tool for integrated water resources planning. WEAP provides a mathematical representation Given the time scale utilized (monthly time-steps), of the river basins encompassing the configura- the modeling does capture seasonal peaking but tion of the main rivers and their tributaries, the not daily or hourly peaking, an important aspect hydrology of the basin in space and time, existing of hydropower design and operation. The analy- as well as potential major schemes, and their vari- sis therefore does not purport to inform project- ous demands for water. More information on the level decisions, but rather to estimate (a) climate WEAP tool is provided in Annexes 5 and 6. change impacts on annual power production up to 2050, as an input to an annual economy- Computations are performed on a monthly time wide economic model; and (b) the cost to restore scale for 50 years for a base-case scenario (i.e., no hydropower generation to levels attainable in the climate change) and the four climate change sce- no-climate-change scenario by constructing addi- narios. Each climate change scenario is charac- tional hydropower capacity. terized by unique inflows and growing demand, hydropower, and reservoir storage. Unmanaged inflows are modeled using CLIRUN-II (Strzepek wEAP Model Description et al. 2008), a hydrologic model used in climate change hydrologic assessments that simulates A water planning model is used to evaluate the runoff from a lumped watershed. The output potential interactions between growing munici- runoff projections from CLIRUN-II are used as pal and industrial (M&I) water use, irrigation, the available runoff in WEAP. Municipal and and hydropower demands under climate change. industrial (M&I) and irrigation demands both The model evaluates these intersectoral effects withdraw water from available runoff, and are between 2001 and 2050, and generates time projected based on World Bank and in-country series of impacts to irrigated agricultural yields sources. Hydropower production is calculated for and hydropower generation under each of the existing and planned dams based on an expected climate scenarios. These time series are used as investment and construction schedule. The with- perturbations to irrigation yields and hydropower drawals and hydropower production were vali- generation estimates in the CGE under each of dated with historical values. the climate scenarios. Twenty-one basins in Ethiopia are expected to This analysis consists of modeling surface water have hydropower capacity by 2050. Therefore availability(runoff), reservoir storage, hydro- these basins were used as the basis for hydrologic power, and major demands in Ethiopia in order simulation in CLIRUN-II and WEAP, as shown 22 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 12 ETHIOPIA RIVER BASINS MODELED wITH wEAP in the shaded areas in Figure 12. These 21 basins have access to water, depending on their distance may be aggregated to form six larger basins that from the supply. If hydropower is generated in a have outlets exiting Ethiopia; in the figure, basins basin, the reservoir and turbine characteristics of like color form these six larger basins. are calibrated to ensure that the power produced is validated with historical values. Basins that Surface water inflows from CLIRUN-II were included storage in 2000 (the base year) include used as inflows to an aggregated river in each the Akoba, Awash Wenz 2, Blue Nile 4, and Blue basin modeled in WEAP. At the six-basin level, Nile 5; by 2010, this list also included the Omo the outflows from each of the rivers either served and Tekeze Wenz. Several additional reservoirs as additional inflow to a downstream river, or are added to the WEAP model between 2010 and flowed out of Ethiopia; that is, water supplies and 2050, as discussed below. Major demand nodes demands are linked between the basins forming in the Ethiopia WEAP were for municipal, indus- the six larger basins. In some basins, an aggregate trial, and irrigation withdrawals from the aggre- reservoir was included at the head of the modeled gated rivers. river. The analysis assumes that all demands in a basin would have access to storage regardless of In general, the representation of the basin struc- potential constraints on the transmission of water. ture is similar to the basin schematic shown in Fig- This approach may overestimate water availability ure 13, which shows the linkage between supply in some cases because all demands may not always and demand in the Blue Nile 4, the Blue Nile 5, E T H I O P I A CO U N T RY ST U DY 23 Figure 13 ExAMPLE OF BASIN SCHEMATIC wITH SUPPLY AND DEMAND IN wEAP Figure 14 OVERVIEw OF ETHIOPIA wEAP MODEL 24 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E and the Jema Shet basins. In the Blue Nile 5, for and factor markets across the economy. Market example, the inflow data into the basin was sup- interactions—with changing prices, supplies, plied by CLIRUN-II and stored in the aggregated trade, and allocation of factors across production reservoir (resBN5). Irrigation and M&I withdraw activities—determine how the economy adjusts water for nodes miBN5 and irrBN5. While the over time under different scenarios. The CGE reservoir and aggregate demands are physically model provides a “simulation laboratory� for located within each basin, they are not located to analyzing the direct and indirect impacts of dif- represent any specific reservoir or demand site. ferent climate change and adaptation investment Also of note are the two run-of-river hydropower scenarios. facilities on the Blue Nile 5 (rorBN51 and 2), which produce hydropower but have no storage, Within the existing structure and subject to mac- and outflows from Jema Shet flow into the Blue roeconomic constraints, producers in the model Nile 5. The 21 basins have a similar schematic maximize profits under constant returns to scale representation, and vary slightly due to individual production technologies, with the choice between basin topology. An overview of the entire study factors governed by a constant elasticity of substi- area is shown in Figure 14. tution (CES) function. Factors are then combined with fixed-share intermediates using a Leontief specification. Under profit maximization, factors CGE Model Description are employed such that marginal revenue equals marginal cost based on endogenous relative prices. The economic impact of climate change is simu- Substitution possibilities exist between production lated using a dynamic computable general equi- for domestic and foreign markets. This decision of librium model (CGE), which is described in Box 3. producers is governed by a constant elasticity of The model simulates the operation of commodity transformation (CET) function that distinguishes E T H I O P I A CO U N T RY ST U DY 25 box 3 COMPUTABLE GENERAL EqUILIBRIUM MODELS (CGES) Computable general equilibrium (CGE) models are often applied to issues of trade strategy, income distribution, and structural change in developing countries. These models have features making them suitable for such analyses. First, they simulate the functioning of a market economy, including markets for factors (land, labor, capital) and commodities, and so provide a useful perspective on how changes in economic conditions are mediated through prices and markets. Secondly, the struc- tural nature of these models permits consideration of new phenomena, such as climate change. Thirdly, these models assure that all economywide constraints are respected. This is a critical disci- pline that should be imposed on long-run projections, such as those necessary for climate change. For instance, suppose climate change worsens growing conditions, forcing Ethiopia to import food. These imports require foreign exchange earnings. CGE models track the balance of payments and require that a sufficient quantity of foreign exchange is available to finance imports. Finally, CGE models contain detailed sector breakdowns and provide a “simulation laboratory� for quantitatively examining how various impact channels influence the performance and structure of the economy. In CGE models, economic decision making is the outcome of decentralized optimization by produc- ers and consumers within a coherent economywide framework. A variety of substitution mecha- nisms occur in response to variations in relative prices, including substitution between labor types; land and capital; between imports and domestic goods; and between exports and domestic sales. Dynamic CGE models The features described above apply to a single-period “static� CGE model. However, because climate change will unfold over decades, the model must be capable of forward-looking growth trajectories. Therefore, the model must be “dynamized� by building in a set of accumulation and updating rules; for example, investment adding to capital stock, labor force growth by skill cate- gory, and productivity growth. In addition, expectation formations must be specified. Expectations are a distinguishing feature of macroeconomic models. In our CGE model, a simple set of adaptive expectations rules are chosen so that investment is allocated according to current relative prices under the expectation that climate realization in the upcoming year will be an average of recent experience. A series of dynamic equations “update� various parameters and variables from one year to the next. For the most part, the relationships are straightforward. Growth in the total supply of each labor category and land is specified exogenously. Sector capital stocks are adjusted each year based on investment, net of depreciation. Factor returns adjust so that factor supply equals demand. The model adopts a “putty-clay� formulation, whereby new investment can be directed to any sector in response to differential rates of return, but installed equipment remains immobile; for example, a factory cannot be converted into a railroad. Sector- and factor-specific productivity growth is specified exogenously. Using these simple relationships to update key variables, we can generate a series of growth trajectories, based on different climate scenarios. 26 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E between exported and domestic goods, and by a linear expenditure system (LES) of demand. The doing so, captures any time- or quality-related government receives revenues from activity taxes, differences between the two products. Profit sales taxes, direct taxes, and import tariffs, and maximization drives producers to sell in markets then makes transfers to households, enterprises, where they can achieve the highest returns. These and the rest of the world. The government also returns are based on domestic and export prices; purchases commodities in the form of government the latter is determined by the world price times consumption expenditures, and the remaining the exchange rate adjusted for any taxes and sub- income of the government is saved (with budget sidies. Under the small-country assumption, Ethi- deficits representing negative savings). All savings opia faces a perfectly elastic world demand curve from households, enterprises, government, and the for its exports at fixed world prices. The final ratio rest of the world (foreign savings) are collected in a of exports to domestic goods is determined by the savings pool from which investment is financed. endogenous interaction of the relative prices for these two commodity types. The model includes three macroeconomic accounts: government balance, current account, Substitution possibilities also exist between and a savings-investment account. In order to imported and domestic goods under a CES Arm- bring about balance in the macro accounts, it is ington specification. This takes place both in necessary to specify a set of “macroclosure� rules, intermediate and final usage. These elasticities which provide a mechanism through which bal- vary across sectors, with lower elasticities reflect- ance is achieved. A �balanced� macro closure ing greater differences between domestic and is assumed such that investment, government imported goods. Again, under the small-country demand, and aggregate consumption are fixed assumption, Ethiopia faces an infinitely elastic shares of total aggregate demand. Savings rates world supply at fixed world prices. The final ratio are assumed to adjust to finance investment. of imports to domestic goods is determined by Implicit in this macro closure is that government the cost-minimizing decision making of domestic policy is assumed to be designed to share the bur- agents (firms/ consumers) based on the relative den of any fall in aggregate demand equally across prices of imports and domestic goods (both of the demand aggregates: consumption, investment, which include relevant taxes).6 and government. For the current account, a flex- ible exchange rate adjusts in order to maintain The model distinguishes among various institu- a fixed level of foreign savings (i.e., the external tions, including enterprises, the government, and balance is held fixed in foreign currency terms). five rural and five urban representative household Finally, in the government account, the fiscal groups in each region. Households and enterprises deficit is endogenous, with government demand receive income in payment for the producers’ use a fixed share of aggregate demand and all tax of their factors of production. Both institutions rates held constant, so that government income pay direct taxes (based on fixed tax rates) and save depends on the level of economic activity—the (based on marginal propensities to save). Enterprises tax base. Labor is assumed to be mobile across pay their remaining incomes to households in the sectors and fully employed. Under the full employ- form of dividends. Households, unlike enterprises, ment closure, for example, expanding biofuels use their incomes to consume commodities under production implies reduced use of labor elsewhere in the economy. The assumption of full employ- 6 For both the CES and the CET functions, a relatively flexible ment is consistent with widespread evidence that, value of 3.0 was applied for the substitution parameter across all while relatively few people have formal sector jobs, sectors. Qualitative results are robust to the choice of CES and CET parameter values. the large majority of working-age people engage E T H I O P I A CO U N T RY ST U DY 27 in activities that contribute to GDP. The model numeraire is the consumer price index (CPI). Social Analysis Approach The CGE model of the Ethiopian economy To complement the economic studies undertaken employed for the simulation analysis is calibrated within the EACC study in Ethiopia, a “social to a social accounting matrix (SAM) for the year component� was developed that used a bottom-up 2005/06 (EDRI-IDS, 2009). The SAM provides perspective to vulnerability assessment and iden- a detailed representation of the structure of pro- tification of adaptation investment options. The duction, demand, international trade, and income social component views vulnerability as encom- distribution. It contains a regional disaggregation passing both physical and socioeconomic ele- of agricultural activities, household income, and ments. It adopts IPCC definitions of vulnerability household consumption. The five regions distin- as comprising physical exposure, socioeconomic guished in this database are differentiated by their sensitivity, and adaptive capacity components agroecological characteristics, as summarized in (including levels of skills, institutional “thickness,� Table 9. The Ethiopia CGE model contains 22 and degree of market integration). commodity groups and 46 activities, including 35 regionally differentiated agricultural sectors. The vulnerability assessment included a litera- Fifteen primary factors of production are iden- ture review; identification of select “hotspots,� tified: four types of labor, agricultural land, and representing both physically exposed and socio- livestock capital in each of the five agroecological economically vulnerable areas from across the zones, and nonagricultural capital employed in country; and fieldwork in these areas, includ- industry and the service sector. ing focus group discussions and a survey of 294 households. The identification of an adaptation options element comprised a series of four par- Table 9 FIVE AGROECOLOGICAL ZONES ticipatory scenario development (PSD) work- shops at local/regional and national levels in SAM Region Temperature and Moisture Regime both highland and lowland areas to determine local stakeholders’ development visions for the Zone 1 Humid lowlands, moisture reliable area, their assessment of livelihood and other Zone 2 Moisture-sufficient highlands, cereals- based impacts of climate change in the area, and pre- Zone 3 Moisture-sufficient highlands, enset- ferred adaptation options for investment. based* Zone 4 Drought-prone (highlands) Zone 5 Pastoralist (arid lowland plains) Key Assumptions and Note: *Enset is a root crop. Limitations On the household side, the SAM-based model One of the strengths of the EACC study is its identifies 14 distinct household groups compris- use of mathematical tools, which impose intel- ing “poor� and “non-poor� rural households lectual discipline. Examples of this discipline residing in each of the five regional zones, as well are the use of a well-defined baseline and the as poor and non-poor households distinguished requirement under CGE models that the national by big and small urban settlements. income accounting identities balance at the end of each year. This mathematical approach is indi- cated when the objective of the exercise includes 28 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E quantitative and monetary impact evaluation of productivity, and hence GDP per capita. The costs and benefits. Apart from providing estimates base run also yields the structure of production of both, the models indicate the relative impor- and degree of urbanization, which, along with tance of some factors vis-à-vis others, and also income, largely drive the structure of demand the effects of changing certain variables onto oth- and required investment in infrastructure. ers. This is important for concretely supporting the decision-making process. By making a choice Technological uncertainty. Most parts of the for quantitative rigor, however, all the well-known study do not allow for the unknowable effects of limitations of using econometric and other math- innovation and technical change on adaptation ematical models apply. costs. In effect, these costs are based on what is known today rather than what might be possible unCertAinty in 20–40 years. Sustained growth in per capita GDP for the world economy rests on technical Uncertainty complicates the analysis of adapta- change, which is likely to reduce the real costs of tion to climate change in three different ways. adaptation over time. First, for most countries there is no consensus whether future climate will be wetter or drier, or inStitutionS the degree to which future storms will be stronger. The second major uncertainty concerns economic The EACC study deliberately never intended to growth. As the global results made clear, the more include institutional, political, and cultural factors rapid economic growth, the more assets are at risk, in the analyses of adaptation costs. As examples, but also the better prepared is a country to absorb many adaptive measures are best implemented and defend against climate-induced changes in through effective collective action at the com- productivity and adverse climate events. Finally, munity level. Soft adaptation measures—such as technological change over the next 50 years will early warning systems, community preparedness affect adaptation in currently unknowable ways. programs, watershed management, urban and These issues are discussed below. rural zoning, and water pricing—generally rely on effective institutions supported by collective Climate uncertainty. As indicated earlier, this action. Because of the desire to produce quanti- study uses four climate models to bracket uncer- tative results, and the availability of models to do tainty on future climate outcomes. However there so, both the global and country studies tended to is a wider number of climate models available, focus on physical adaptation measures (sometimes and thus many other climate outcomes are pos- referred to as “hard adaptation�), which basically sible. Of the 26 climate projections available for require an engineering response. Country-level the A2 SRES, an assessment of adaptation costs studies have explored numerous soft options, but was possible only with 2 for the global track, and again, hard numbers tend to crowd out important 2–4 for the country studies. concepts: the difficulty of assigning costs and ben- efits to good institutions and good public policy Growth uncertainty. A key contribution of this has led country teams to focus most intently on study is to separate the costs of adaptation from hard options. Furthermore, in some cases discus- those of development by defining a development sions of future investment might be less politically baseline. The study specifies as a comparator contentious than discussions of possible institu- “base run� a future development path for Ethio- tional reforms—especially in sensitive areas such pia that is consistent with forecasts of growth in as water and utility pricing, property rights, and population, labor force, aggregate investment, zoning and land-use policies. E T H I O P I A CO U N T RY ST U DY 29 Any type of outside assistance to bring about the stated development strategy being pursued institutional reform in developing countries needs by Ethiopia. As climate change impacts become to build on changes that have internal sources and more clearly defined in the future, further explora- support. There are no magic recipes for dealing tion of strategic options is a highly recommended with the institutional aspects of adaptation. We follow-up activity. observe that certain countries manage to cope with extreme events better than others, and we The second issue is the integration of the social can describe the mechanics of their actions, but analyses with the economic models. The origi- the incentives and mechanisms to make it hap- nal intent of the EACC was to translate the very pen has to come from internal concerns and rich information coming out of field work in eco- conviction. This is not a limitation of the study nomic terms, so that different adaptation actions itself, simply recognition of what we do or do not indicated by the local populations could somehow know. appear in the models as explicit adaptation alter- natives. The two difficulties were that (1) a greater moDeling level of effort would be required to obtain the necessary economic information as part of the Most of the results of this study are based on field work, and (2) the level of aggregation of the biophysical, engineering, and economic models. models—whether national or global—would not These models use mathematical techniques to accommodate the integration of these very spe- represent physical and economic processes. The cific local measures. Nonetheless, it was possible, more the phenomena being simulated are gener- as registered in the social analyses of the countries, ated by deterministic physical processes, the bet- to observe the consistency of the adaptation mea- ter performance of the models. As phenomena sures proposed at the two different levels. Specific become increasingly influenced by uncertainty or economic analyses of adaptation measures at the by human behavior and institutional change, the local level would be an important exercise to be ability to simulate weakens. pursued in future work. Efficiency in adaptation can be explored either One final important qualification must be made through extremely complex inter-temporal and to the overall limitations of the EACC: while cross-sectoral optimization (for which mod- uncertainty is pervasive when dealing with cli- els capable of policy analysis do not exist), or mate change, the basic lesson is no different from through comparing the results of a wide range of any other area of economic policy—do not act alternative investment programs, including those on fixed assumptions about the future, build flex- that implement projects at differing points in the ibility into both policies and hardware. No study, future. This study explores a limited number of however careful and detailed, can remove this adaptation investment strategies consistent with uncertainty. 30 TH REE E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 31 Impacts Agriculture The country is divided into five agroecological zones (Figure 15). Around 45 percent of the coun- SeCtor BACkgrounD try consists of a high plateau comprising zones 2 to 4 with mountain ranges divided by the East In 2006/07, agricultural production generated African Rift Valley. Almost 90 percent of the pop- around 46 percent of Ethiopia’s gross domestic ulation resides in these highland regions (1,500m product and employed 80 percent of the working above sea level). Within the highlands, zones population. According to Deressa (2006), Ethio- 2 and 3 generally have sufficient moisture for, pia has about 16.4 million hectares of arable respectively, the cultivation of cereals and enset land (14.6 percent of its total land area), of which (a root crop), whereas zone 4 is prone to droughts. about 8 million hectares are currently used for The arid lowlands in the east of the country— crop production. zone 5—are mostly populated by pastoralists. Figure 15 AGROECOLOGICAL ZONES IN The agricultural sector is dominated by mixed ETHIOPIA rainfed small-scale farming based on traditional technologies. Small-scale subsistence farming (about 8 million peasant households) accounts for 95 percent of the total area under crops and more than 90 percent of the total agricultural output. Most food crops (94 percent) and coffee (98 per- cent) are produced by small-scale farmers, while the remainder is generated by private and state commercial farms. Production technologies are predominantly characterized by the use of ox- drawn wooden ploughs with steel pikes and other traditional farm implements, minimal applica- tion of fertilizer and pesticides due to high input 32 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E prices in the presence of credit constraints7 and at home as household consumption. As recently weak extension services, and low use of improved as 2001/02, the Central Statistical Agency’s seeds (Deressa 2006). (CSA) agriculture census estimated that farmers consumed at home about 63 percent of their total By the end of the 2005/06 growing season, only output, and that less than 30 percent was in fact 13 percent of the potentially irrigable land area marketed. was irrigated. A typical farming household in the semi-arid areas owns just a small portion of imPACtS on CroPS land (generally less than one hectare) for crop and livestock (cattle, goats, sheep, poultry, donkeys) With the use of CliCrop, the changes in CO2 production. In addition to these constraints, the concentration, precipitation, and temperature country continues to experience persistent drought from the four GCMs are used to estimate the episodes because of its prominent location in the changes in production (yield) each year for the Sahel region, characterized by erratic rainfall and major crops. The yield effects reflect the reduc- unpredictable climatic variability. Factors contrib- tions in yield due to either the lack of available uting to the low productivity of the agricultural water, or due to the overabundance of water sector—beside droughts and floods—include that causes waterlogging. Regional and temporal declining farm sizes due to population growth; trends over the four scenarios are depicted below land degradation due to inappropriate use of land in Figure 16. such as cultivation of steep slopes, overcultivation and overgrazing; tenure insecurity; weak agri- CO2 fertilization is included in the analysis but cultural research and extension services; lack of does not make a significant difference. Current agricultural marketing; an inadequate transport research is suggesting much smaller CO2 fertil- network; low use of fertilizers, improved seeds, ization than initially thought. Additionally, new and pesticides; poor nutrition of livestock; low lev- research shows that under higher CO2 levels, els of veterinary care; and livestock diseases. ozone will also be present and that has a negative impact on crop yields. Zones 2 and 4 comprise the major agricultural production zone. These regions face both water Climate impacts are significant, but variable over shortages and waterlogging throughout the grow- regions and crop type. The impact of these trends ing season because they are primarily rainfed. tends to grow stronger in time. The Dry2 scenario Portions of these regions are able to accommo- is the most damaging scenario due to the frequent date multiple crops due to climate conditions. occurrence of droughts. The impacts of climate Parts of these regions and many others are suit- on yields are first-order effects that trigger direct able for irrigation-based multi-cropping. and indirect economic impacts—such as reduc- tions in income, employment, savings and invest- The eastern arid regions of zone 5 are dominated ments. These impacts are captured in the CGE by a grassland-based livestock production. Very analysis as described in chapter 4. limited rainfall in this area leads to a very vulner- able livestock sector. A large proportion of the farm output is consumed 7 Using a nationally representative data set, Croppenstedt, Derneke and Meschi (2003) identify credit constraints and low value-cost ratios due to high procurement and distribution costs as major hurdles to the adoption and intensity of fertilizer use in Ethiopia. E T H I O P I A CO U N T RY ST U DY 33 Figure 16 RANGE OF PERCENT YIELD DEVIATIONS FROM NO-CC BASE (2006–2050) FOR SELECTED CROPS BARLEY WHEAT 10 6 8 6 4 4 2 2 0 0 -2 DRY1 -2 DRY1 WET1 -4 WET1 -6 -4 -8 WET2 -6 -10 DRY2 -12 DRY2 -8 WET2 min ave max min ave max MAIZE SORGHUM 20 6 15 4 10 5 2 0 0 DRY1 DRY1 WET2 WET2 -5 WET1 -2 WET1 -10 -15 -4 -20 -6 DRY2 -25 -8 DRY2 min ave max min ave max Note: The effects of climate change on the different crops are weighted averages across regions, using the regions’ shares in crop total production as weights. Baseline yields include a technology growth component reflecting historical trends. 34 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 17 BIOPHYSICAL COMPONENT: RATIO OF FUTURE LIVESTOCK ExPECTED INCOMES TO ExPECTED INCOMES UNDER MEAN BASELINE CONDITIONS, ETHIOPIA BASE SCENARIO, 2001–50 AEZ1 AEZ2 AEZ3 1.2 AEZ4 AEZ5 RATIO OF FUTURE TO MEAN BASELINE NET REVENUES 1 0.8 0.6 0.4 0.2 0 2000s 2010s 2020s 2030s 2040s DECADE imPACtS on liveStoCk below mean baseline conditions as temperatures fluctuate from year to year. Based on the steps outlined in the methodology section above, the analysis developed two sets of Figure 18 presents the vectors of ratios for the time series that collectively make up the livestock Ethiopia dry (on right) and wet (on left) scenarios, impact vectors. Each of these sets of time series and Figure 19 presents results under the global and the combined final livestock productivity vec- dry (on right) and wet (on left) scenarios. In all sce- tors are described below. narios, the projected impacts on livestock incomes are severe by 2050. Under the Ethiopia wet and Biophysical results. The first is the biophysi- the two global scenarios, income in each AEZ falls cal component, which is a unitless time series of to 70 to 80 percent of baseline levels. Under the future livestock net revenues relative to mean more pronounced temperature effects of the dry baseline conditions; that is, mean baseline cli- scenario, livestock incomes fall to roughly 60 per- mate conditions would produce a value of one. cent of mean baseline levels, although in AEZ 1 Figure 17 presents the biophysical ratios for they reach a low of roughly 55 percent. Generally, AEZs under the base scenario from 2001 to these effects do not vary significantly across AEZs. 2050. Note that there are fluctuations above and E T H I O P I A CO U N T RY ST U DY 35 Figure 18 BIOPHYSICAL COMPONENT: RATIO OF FUTURE LIVESTOCK NET REVENUES TO NET REVENUES UNDER MEAN BASELINE CONDITIONS, ETHIOPIA DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 RATIO OF FUTURE TO MEAN BASELINE NET REVENUES AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Figure 19 BIOPHYSICAL COMPONENT: RATIO OF FUTURE LIVESTOCK NET REVENUES TO NET REVENUES UNDER MEAN BASELINE CONDITIONS, GLOBAL DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 RATIO OF FUTURE TO MEAN BASELINE NET REVENUES AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Feed results. Next is the feed component, which scenario from 2001 to 2050 are presented in Fig- is a unitless time series of projected millet yields ure 13. Notice that feed ratios vary much more relative to mean baseline yields generated by Cli- widely across AEZs than the biophysical ratios. Crop. The feed ratios for AEZs under the base 36 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 20 FEED COMPONENT: RATIO OF MILLET YIELDS TO MEAN BASELINE YIELDS, ETHIOPIA BASE SCENARIO, 2001−50 AEZ1 AEZ2 AEZ3 1.2 AEZ4 AEZ5 1 0.8 0.6 FEED EFFECT 0.4 0.2 0 2000s 2010s 2020s 2030s 2040s DECADE The vectors of ratios for the Ethiopia dry (on left) scenario than the dry scenario, particularly in and wet (on right) scenarios are presented in Fig- AEZ 5, which has yield ratios up to 20 percent ure 21, and Figure 22 presents results under the higher than mean baseline levels. Results are global dry (on left) and wet (on right) scenarios. mixed in the global scenarios, with AEZ 5 show- The projected impacts on yields differ consider- ing a positive yield response in the dry relative to ably between the four scenarios and across the wet scenario, but others (such as AEZ 1) showing five AEZs. Under the Ethiopia scenarios, yields the opposite relationship. tend to respond more positively under the wet E T H I O P I A CO U N T RY ST U DY 37 Figure 21 FEED COMPONENT: RATIO OF MILLET YIELDS TO MEAN BASELINE YIELDS, ETHIOPIA DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 FEED EFFECT 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Figure 22 FEED COMPONENT: RATIO OF MILLET YIELDS TO MEAN BASELINE YIELDS, GLOBAL DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 FEED EFFECT 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Combined results. Each of the 25 pairs of livestock productivity under each of the scenarios biophysical and feed vectors for the AEZs and relative to productivity under mean baseline cli- scenarios are averaged to produce the vectors matic conditions. The combined vectors for the presented in Figures 23 through 25 for 2001 to baseline scenario are presented in Figure 23. 2050. These hybrid vectors represent a ratio of 38 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 23 COMBINED COMPONENTS: MEAN OF BIOPHYSICAL AND FEED VECTORS, ETHIOPIA BASE SCENARIO, 2001−50 AEZ1 AEZ2 AEZ3 1.2 AEZ4 AEZ5 1 0.8 COMBINED BIOPHYSICAL AND FEED EFFECTS 0.6 0.4 0.2 0 2000s 2010s 2020s 2030s 2040s DECADE Figure 24 presents the vectors of combined ratios shows the ratio falling to a low value of approxi- for the Ethiopia dry (on left) and wet (on right) mately 0.70 in AEZ 1, or a 30 percent decline in scenarios, and Figure 25 presents the results productivity. Under each of the scenarios, there is under the global dry (on left) and wet (on right) a downward trend in productivity over the 2001 scenarios. Livestock productivity is affected most to 2050 period. severely under the Ethiopia dry scenario, which E T H I O P I A CO U N T RY ST U DY 39 Figure 24 COMBINED COMPONENTS: MEAN OF BIOPHYSICAL AND FEED VECTORS, ETHIOPIA DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 COMBINED BIOPHYSICAL AND FEED EFFECTS AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Figure 25 COMBINED COMPONENTS: MEAN OF BIOPHYSICAL AND FEED VECTORS, GLOBAL DRY (ON LEFT) AND wET (ON RIGHT) SCENARIOS, 2001−50 AEZ1 AEZ1 COMBINED BIOPHYSICAL AND FEED EFFECTS AEZ2 AEZ2 1.2 AEZ3 1.2 AEZ3 AEZ4 AEZ4 AEZ5 AEZ5 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 2000s 2010s 2020s 2030s 2040s 2000s 2010s 2020s 2030s 2040s DECADE DECADE Drought Expenditures of $1.2 billion in the Ethiopia dry scenario in the 2030s. This compares to an average annual The average annual projected expenditure on recurrent drought expenditure between 1997/98 droughts by decade under the baseline and four and 2005/06 of roughly $696 million and a climate scenarios is presented in Figure26. These maximum annual expenditure of $1.8 billion (see vary from a low of $7.3 million annually under Annex 9). the Ethiopia wet scenario in the 2040s to a high 40 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 26 MEAN ANNUAL PROjECTED ETHIOPIAN GOVERNMENT ExPENDITURES IN THE 2000S THROUGH 2040S (MILLIONS OF 2010 $) 1400 BASE GLOBAL WET GLOBAL DRY 1200 ETHIOPIA WET ETHIOPIA DRY 1000 EXPENDITURES MILLIONS OF 2010 US$ 800 600 400 200 0 2000s 2010s 2020s 2030s 2040s DECADE Road Transport in extreme events and the costs associated with repairing the roads from those extreme events. The road transport sector is impacted by climate change in two areas; standard maintenance and flood-induced maintenance. The former repre- roAD trAnSPort BACkgrounD sents costs that are incurred due to precipitation and/or temperature changes that occur during Ethiopia’s strategy for the road sector stated that the life span of the road. These changes represent the total road length in the country was 56,113 km differences in the average climate conditions that as of April 2006. Unpaved roads represent about exist for the road and thus change the conditions 85 percent of the total road length (47,612), while under which the road is intended to perform on paved roads represent the remaining 15 percent an everyday basis. The latter represents changes (Table 10). E T H I O P I A CO U N T RY ST U DY 41 impacts are realized on existing road inventory Table 10 BASE CLASSIFIED AND URBAN that is not designed for increased temperature ROAD NETwORKS, 2006 (KM) and precipitation. These maintenance costs drop off over time as new inventory is assumed to be Class Unpaved Paved Total adapted to the future climate change impacts with enhanced design standards. Primary 1,155 3,490 4,625 Secondary 24,869 2,443 27,282 Similarly, the increased maintenance cost for Tertiary 21,588 1,047 22,635 unpaved roads is estimated between $2 mil- Subtotal 54,613 Classified lion and $14 million per year depending on the climate model used. In contrast to the paved urban 1,500 1,500 roads that see reductions in maintenance costs grand total 47,612 8,480 56,113 due to enhanced design standards, the unpaved Source: Ethiopia Ministry of Transport roads continue to see maintenance costs that are dependent on the climate scenario due to limited Table 11 provides estimates of unit maintenance options for making unpaved roads resistant to cli- costs for the existing road network. mate change effects. Overall, the total increase in maintenance costs roAD trAnSPort imPACtS due to climate change is therefore estimated to be between $15 million and $31 million per year The stressor equations introduced earlier pro- depending on the climate model used. These vide the basis for determining the impact of cli- numbers are on the lower bound of what may be mate change on the maintenance of paved and expected due to the key assumptions that (1) regu- unpaved roads. Based on the road inventory in lar maintenance is implemented on the roads so Ethiopia, it is estimated that maintenance on that compounding damage effects do not occur; paved roads that is directly attributable to climate (2) the effects do not include flooding damage, change ranges from $5 million to $13 million per which is a major factor for road damage as seen year depending on the climate model used for the in the next section; and (3) adaptation to paved projection. As illustrated in Table 12 and Figures roads is completed when roads are repaved. If 27 and 28, maintenance costs on paved roads are any of these assumptions fails to materialize, then the highest in the first decades as climate change the costs could be significantly higher. Table 11 UNIT MAINTENANCE COST RATES (US$) Type of Maintenance Transitability Routine Periodic Rehabilitation road Class un-paved un-paved Paved un-paved Paved un-paved Paved Primary N/A 1,500 1,100 35,000 55,000 80,000 300,000 Secondary N/A 1,200 880 28,000 44,000 50,000 240,000 Tertiary 300 750 660 10,000 44,000 25,000 200,000 Note: Values in US$ per km, per year for transitability and routine. Source: Data obtained from World Bank projects in-country and in region. 42 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 27 AVERAGE DECADE COSTS FOR EACH OF THE FOUR GCMS FOR MAINTAINING GRAVEL AND EARTH ROADS DUE TO CLIMATE CHANGE INCREASE IN PRECIPITATION DECADE COST INCREASE FOR MAINTAINING GRAVEL AND EARTH ROADS $160,000,000 $140,000,000 $120,000,000 $100,000,000 WET 2 $80,000,000 DRY 2 WET 1 $60,000,000 DRY 1 $40,000,000 $20,000,000 $- 2020 2030 2040 2050 Figure 28 AVERAGE DECADE COSTS FOR EACH OF THE FOUR GCMS FOR MAINTAINING PAVED ROADS DUE TO CLIMATE CHANGE INCREASE IN PRECIPITATION AND TEMPERATURE DECADE COST INCREASE FOR MAINTAINING PAVED ROADS—NO ADAPTATION $300,000,000 $250,000,000 $200,000,000 WET 2 $150,000,000 DRY 2 WET 1 DRY 1 $100,000,000 $50,000,000 $- 2020 2030 2040 2050 E T H I O P I A CO U N T RY ST U DY 43 the maintenance costs required to repair the roads Table 12 CUMULATIVE CLIMATE CHANGE to functioning order. For this focus, a combina- IMPACT ON PAVED AND UNPAVED ROADS BASED ON INCREASED MAINTENANCE tion of the COWI report on making transport COSTS FOR THE FOUR CLIMATE GCMS climate resilient and previous work conducted (US$ MILLION) by the research team for the U.S. Federal High- Wet 2 Dry 2 Wet 1 Dry 1 way Administration—together with actual num- bers obtained from local transport ministries—is Cumulative $849.2 $538.4 $371.8 $531.7 cost increase used as a basis for maintenance costs. Specifically, for maintaining paved roads these efforts yield a cost for maintenance based on Cumulative $408.8 $308.7 $221.7 $273.8 original construction costs. cost increase for maintaining gravel and earth Primary roads have a 15 percent maintenance roads cost, while 29 percent and 35 percent are used for Total cumulative $1,258.0 $847.1 $593.5 $805.5 secondary and tertiary roads respectively to obtain maintenance costs from a cost per kilometer for maintenance. The primary climate change roads number is lower due to the roads being built (paved and unpaved roads) to a higher standard at original construction. For unpaved roads, we use these same sources to esti- mate maintenance costs. However, unpaved roads differ in that they are based on the severity of the flooDing CoStS flood rather than the road type, since the impact focuses on the road surface rather than the road Flooding costs for the road sector focus on the base or the type of pavement, as is the case for need to maintain roads after a flooding event. The paved roads. Therefore, the gravel and dirt surfaces methodology—introduced later in the section on on these roads will erode based on the amount of economywide impacts—is utilized to illustrate the floodwater rather than the type of road that is con- costs that flooding will impose on the road sector. structed. Given this relationship to water rather The focus of the flooding analysis is to determine than construction materials, the impact percentage Table 13 TIME-SERIES INPUTS TO CGE MODEL FOR TRANSPORT COSTS Vector Definition 1. Base Maintenance costs derived from historic flood This is the base yearly maintenance costs that would be data anticipated from historic floods. 2. The total percentage of road capital loss for both This is the projected capital road loss percentage from paved & unpaved roads, using a weighted average over the historic storm levels. This is the # of damaged road the country based on road density kilometers / number of total kilometers. This number is weighted over the climate zones to reflect the relative amount of roads contained in that zone. 3. Total maintenance costs for existing paved and This is the projected cost for maintenance of the roads unpaved roads WITH NO ADAPTATION due to climate-change-induced flooding if no adaptation were put in place. 4. Total road capital % loss due to climate change— This is the projected capital road loss percentage from paved & unpaved—weighted each region based on total the projected climate-change-induced storms over and kms above the historic levels. This is the # of damaged road kilometers / number of total kilometers. This number is weighted over the climate zones to reflect relative amount of roads contained in that zone. 44 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 29 BASE MAINTENANCE COSTS TO REPAIR FLOOD DAMAGE BASED ON PRO- jECTED FLOODS FROM HISTORIC CLIMATE PATTERNS $18,000,000 $16,000,000 $14,000,000 $12,000,000 WET 2 DRY 2 $10,000,000 WET 1 DRY 1 $8,000,000 $6,000,000 $4,000,000 $2,000,000 $- 2020 2030 2040 2050 Figure 30 TOTAL ROAD CAPITAL % LOSS BASED ON HISTORIC PROjECTED FLOOD EVENTS 4.00% 3.50% 3.00% HISTORIC MAINTENANCE FROM FLOODING WET 2 2.50% DRY 2 WET 1 2.00% DRY 1 1.50% 1.00% 0.50% 0.00% 2020 2030 2040 2050 E T H I O P I A CO U N T RY ST U DY 45 Figure 31 ANNUAL AVERAGE MAINTENANCE COST FOR ExISTING PAVED AND UNPAVED ROADS PER DECADE wITH NO ADAPTATION $400,000,000 $350,000,000 $300,000,000 WET 2 $250,000,000 DRY 2 WET 1 $200,000,000 DRY 1 $150,000,000 $100,000,000 $50,000,000 $0 2020 2030 2040 2050 for unpaved roads does not vary according to in the time series. The application of this process whether they are primary, secondary, or tertiary to the time series of damages results in a time roads. Therefore, an 11 percent maintenance ratio series of maintenance costs. The flooding anal- is used for maintenance costs for all categories of ysis results in the generation of the cost vectors unpaved roads. This ratio indicates that mainte- reported in Table 13 , which are used as input for nance cost for flood repairs is equal to 11 percent the CGE modeling. of the construction cost each time a flood event occurs. It is assumed that the repairs can be com- For each vector, the data conveyed from the cost pleted in a single year. module are divided among the three road types— primary, secondary, and tertiary. The vectors com- Using these maintenance ratios for the non-adap- bine the climate zones together since the CGE tation scenario, the potential roads are then mul- models do not divide the economywide analysis tiplied by the cost per kilometer to obtain a total into separate zones. An example of each vector is maintenance rate required for that particular year supplied as follows. 46 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Table 14 SUMMARY OF IMPACTS COSTS ON ROADS (US$ MILLION, ANNUAL AVERAGE) Hydropower Wet 2 Dry 2 Wet 1 Dry 1 hyDroPower BACkgrounD Impacts 31.45 21.12 14.84 20.14 on regular maintenance Associated with water resource use is the genera- Flood 340.86 296.92 265.7 257.73 tion of energy by constructing renewable hydro- impacts electricity generating dams. Current generation Total 372.31 318.04 280.54 277.87 impacts capacity is approximately 2,000 MW, of which 95 costs percent is generated from hydroelectric genera- tors. Presently, five additional hydroelectric dams and one wind farm are under construction. When finalized, they are expected to increase power Table 15 PLANNED POwER GENERATION PROjECTS INCLUDED IN THE ANALYSIS Year of Completion & 2010 value Plant Online ($Million) Fixed (started) Wind - Ashegoba 2011 259 Gilgel Gibe III 2013 1730 Tendaho Geothermal (near Djibouti) 2013 305 Planned Gilgel Gibe IV 2014 2,930 Halele Werabesa 2014 725 Chemoga-Yeda 2014 318 Geba I & II 2016 593 Genale III 2018 362 Genale IV 2018 456 Tekeze II 2020 694 Karadobi 2023 2,411 Border 2026 1,741 Mendaia 2030 2,990 Baro 2034 914 Aleltu 2038 1,444 Didessa 2038 523 Dabus 2042 1,805 Birbir 2042 1,199 Tams 2046 1,805 E T H I O P I A CO U N T RY ST U DY 47 Figure 32 HYDROPOwER GENERATION UNDER TwO CHANGE SCENARIOS, 2008–50 70 60 ANNUAL ENERGY [TWATT HRS/YR] 50 40 30 BASE 20 GFDL (DRY) NCAR (WET) 10 0 2010 2015 2020 2025 2030 2035 2040 2045 2050 YEAR Source: IMPEND – CliRun Simulations supply to a total of 3,270 MW and provide access As the figure shows, climate change does not to 50 percent of the total number of households change the variability of hydropower generation in the country. Moreover, the hydroelectric gen- but impacts the mean annual energy generation. erating dams are also expected to regulate water The wet scenario produces more hydropower than flows, control floods, and channel water for irri- the base and the dry scenario produces less than gation purposes. The government’s plan for the the base. However, the deviations from the base expansion of hydropower generation capacity is only start after 23 years because initially there is so summarized in Table 15. little installed hydro capacity that changes in flows have no impact. This is an important result as it hyDroPower imPACtS shows that Ethiopia has a chance to gain or lose from climate, but these impacts have a threshold The IMPEND model analysis provided an esti- and then increase nonlinearly with the size of the mate of the potential change in hydropower gen- hydropower development. This suggests a careful eration capability for the plants, under the above risk-based approach to hydropower investments investment schedule in Table 15 using a monthly after 2030, which accounts for close to half the modeling of streamflow and energy generation. value of the projects listed in Table 15. The results of this comparison between the base, Ethiopian dry, and Ethiopian wet scenarios are shown in Figure 32. The economywide implica- reSultS of the weAP AnAlySiS tions of these alternative hydroenergy paths are explored in chapter 4. The water evaluation and planning (WEAP) anal- ysis first establishes the baseline (i.e., no climate 48 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E change) projections through 2050 for key water production and irrigated crop yields relative to demands, including municipal and industrial maximum potential generation and yields. In (M&I), hydropower, and irrigation (described addition, the analysis evaluates the effects of these above). Next, it analyzes baseline competition rising demands on flows from Ethiopia into other among water end-users given baseline water avail- countries. Finally, the analysis incorporates the ability; that is, competition for water that would runoff under the climate change scenarios to see take place even in the absence of climate change. if competition among end-users intensifies and/ Specifically, the impacts of the competition or changes, as well as how flows from Ethiopia are measured in terms of reduced hydropower are affected. Figure 33 TOTAL HYDROPOwER PRODUCTION IN THE 21 ETHIOPIA RIVER BASINS, ASSUMING GROwING M&I DEMANDS AND IRRIGATION TO 3.7 MILLION HA, 2001–50 100 NoCC Ethiopia Dry 90 Ethiopia Wet Global Dry Global Wet 80 HYDROPOWER PRODUCTION (MILLION MWH) 70 60 50 40 30 20 10 0 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 YEAR E T H I O P I A CO U N T RY ST U DY 49 Figure 34 MEAN DECADAL CHANGES IN HYDROPOwER PRODUCTION GIVEN INCREASING M&I AND IRRIGATION DEMANDS, RELATIVE TO A NO-DEMAND SCENARIO 0 PERCENT CHANGE IN HYDROPOWER GENERATION -0.5 -1 -1.5 -2 BASE ETHIOPIA DRY ETHIOPIA WET GLOBAL DRY GLOBAL WET -2.5 2000s 2010s 2020s 2030s 2040s DECADE The hydropower and irrigation results are used and no transboundary flow requirements.8 The to produce adjustments in the hydropower and results and assumptions of the analysis of compe- crop impact results generated by IMPEND and tition under both baseline and climate change are the crop models under both baseline and climate described for hydropower, irrigation, and unman- change. Both of the other models consider cli- aged flows below. mate change effects but no intersectoral effects. Intersectoral effects on hydropower produc- The analytical scenario assumes increasing M&I tion. The IMPEND model was used to evaluate demands, increasing irrigation to 3.7 million ha the potential impacts of climate change on energy by 2050, full expansion of hydropower to levels outlined by the Ministry of Water Resources’ 8 This analysis was unable to evaluate the water resources implica- Water Sector Development Plan (MWR 2002),), tions of the existing Nile Basin treaty, so does not impose trans- boundary flow requirements. Instead, the analysis evaluates the implications of rising Ethiopian demand for water on transbound- ary flow. 50 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E production in Ethiopia. For comparison purposes, The maximum effect is under the Ethiopia dry sce- the total hydropower output in megawatt-hours nario, where an average decadal effect for the 2040s (MWH) shown in Figure 33 is produced under reaches -2.5 percent. The findings of the analysis the baseline and four climate scenarios. Baseline in terms of total volume of power produced (Mil- hydropower production over the 50-year period is lion of MWH) are presented in Table 16. 1.89 billion MWH; as expected, under the Ethio- pia wet scenario outputs are highest (13.6 percent Table 16 TOTAL HYDROPOwER PRODUC- TION UNDER DIFFERENT DEMAND AND CLI- higher than baseline), and under the Ethiopia dry MATE SCENARIOS (MILLION MwH, 2010–40) scenario outputs are lowest (8.7 percent lower than baseline). The two global scenarios hover Baseline (no much more closely to the baseline (dry is 2.2 per- Demand scenario Wet 2 Dry 2 climate change) cent higher and wet is 9.3 percent higher than baseline). These closely reflect the runoff projec- No competing demand to be met 1,884 2,136 1,722 tions presented in Annex 7. Demand for M/I to be met 1,881 2,133 1,718 Figure 34 presents the mean decadal reductions Demand for M/I and for irrigation to in hydropower production given increased M&I be met 1,858 2,113 1,696 and irrigation demands through 2050 (described below).9 These changes are evaluated under the baseline and four climate change scenarios Changes in irrigation water availability and (i.e., five bars per decade in Figure 34.) relative crop yields. Intersectoral water conflicts will to hydropower production under a hypotheti- also impact the availability of water for irriga- cal scenario where Ethiopia has no extractive tion, which will decrease crop yields in irrigated water demands (i.e., no M&I or irrigation). Note areas. Based on communications with the Minis- that the baseline scenario includes growth in the try of Water Resources, irrigated agriculture will M&I and irrigation sectors but no changes in expand considerably by 2050, from 1.6 percent to climate, whereas the hypothetical “no demand� 35 percent (3.7 million ha) of the approximately scenario (which mirrors the scenario used in 10 million ha of agricultural land in Ethiopia the IMPEND analysis) includes no changes in (total agricultural area is assumed to remain rela- demand or climate. tively constant). WEAP generates unmet demand estimates based on a balancing of demands in Changes in hydropower production occur because the system. These unmet irrigation demands (all the higher-priority downstream irrigation and M&I demands are met because it has top priority) M&I demands affect the reservoir water release are produced for each basin and summed across schedules and thus cause suboptimal production basins to generate percent reductions in irrigation conditions. For example, if irrigation demand water availability between 2001 and 2050 under causes water to be released in May rather than the baseline and each of the climate scenarios. stored to increase hydraulic head, overall power Table 17 illustrates these unmet irrigation water production can decline. demands, which rise to an annual average of up to 965 million m3 under the Ethiopia dry scenario 9 To evaluate the worst-case outcome for hydropower, this scenario in the 2040s. assumes that municipal and industrial demands have the first pri- ority, followed by irrigation, then hydropower. E T H I O P I A CO U N T RY ST U DY 51 Figure 35 MEAN DECADAL CHANGES IN CROP YIELDS GIVEN INCREASING M&I AND IRRIGATION DEMANDS 0 -0.5 -1 PERCENT CHANGE IN CROP YIELDS -1.5 -2 -2.5 -3 -3.5 -4 BASE ETHIOPIA DRY -4.5 ETHIOPIA WET GLOBAL DRY GLOBAL WET -5 2000s 2010s 2020s 2030s 2040s DECADE Table 17 UNMET DEMAND FOR IRRIGA- TION (MILLION M3, ANNUAL AVERAGE BY percent reduction in water availability generates a DECADE) 5 percent yield reduction). Based on this assump- tion, the resulting decadal reductions in yield are Scenario 2010 2020 2030 2040 presented in Figure 35. Note that impacts on Baseline 3.98 23.0 135 443 yield under the Ethiopia dry scenario are most Wet 2 0.00 9.63 61.9 160 severe at an average loss of approximately 4.4 Dry 2 8.33 122 496 965 percent in the 2040s (i.e., unmet irrigation water demands of 8.8 percent). The maximum annual loss in crop yield between 2001 and 2050 is also The unmet water demands can also be converted under the dry scenario at approximately 9 per- to changes in crop yields. Because farmers can cent in 2049 (i.e., unmet irrigation demands of adapt to restrictions on water and because the 18 percent). Under the scenario where 4.1 mil- relationship between water and yield is nonlin- lion hectares are irrigated by 2050 instead of 3.7 ear, this analysis assumes that each percentage million, the maximum annual yield reduction is reduction in irrigation water availability results in approximately 10 percent, also in 2049. a one-half percent reduction in yields (i.e., a 10 52 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E social protection measures organized at village Vulnerability: A level through their socially constructed “entitle- Participatory Analysis ments� to these measures. Analysis shows that private-measure benefits accrue largely to wealth- ier households facing crises (given they are likely To complement the numerical simulation of bio- to offer future reciprocal assistance). Specifically, physical sector performance, a participatory anal- at the household level, low levels of entitlements ysis was also performed. Vulnerability was found to resources means that households are more to stem from a number of factors. These included vulnerable to climate and other shocks. Cluster elements of physical location/ hazard-proneness; analysis of household survey responses found economic geography/ regional development lev- that informal village social assistance institutions els; socioeconomic status; and social differentia- tended to benefit large, landowning farmers (52 tion, including ethnicity and gender. percent of these large farmers accessed the social assistance institutions, compared to an average Physical Geography. Physical location and access rate for all income groups of 34 percent). hazard-proneness greatly affect household vulner- This finding points to the likelihood of inequity in ability, as in the drought-prone lowlands, which collective social protection mechanisms through are chronically exposed to low rainfall. Just as asset community institutions such as idir and kire (infor- depletion occurs in a chronic form at the house- mal insurance organizations), jiggie and debbo hold level, at the area level too, repeated hazard (labor sharing organizations), and iqub (informal events can reduce a region’s adaptive capacity. rotating savings associations). Economic Geography. Vulnerability can also This finding underlines the importance of ensuring arise from the existing livelihood systems and public social protection transfers are also available policy regimes governing these systems. For to households, with transparent targeting processes, example, single-sector pastoral livelihoods, which as now being undertaken by the Productive Safety are dependent on natural resources and there- Nets Program (PSNP) in Ethiopia. Such programs fore vulnerable to climate change, are also being help smooth household consumption, and through threatened by state policies regarding nonfarm NRM-targeted public works, help build area resil- external investment in the area that reduces land ience, but may also thereby help households take access for pastoralists. on more of the necessary risk needed to diversify successfully through long-term adaptation mea- Socioeconomic status. Poverty status (including sures in which more vulnerable households are low physical, financial, and human capital asset unable to invest (for example, education). levels) lead to extreme vulnerability of households. Common factors here include high dependency Social differentiation, including gender. ratios in the household, reliance on one liveli- Impacts of physical hazards have differential hood, and low education levels. Cluster analysis effects on diverse groups. Social vulnerability of 294 households revealed that small poor farm- factors identified in the Ethiopia study included ers were in the worst position in the field sites, ethnic status, migration status, presence of social with young agropastoralists also vulnerable. conflict, and female-headed household status. Vulnerable groups identified through commu- Crucially, the perception of households’ wealth nity discussions included asset-poor households, status by others (i.e., their “fallback position�) the expanding group of rural landless; the urban affects the degree to which they can access private poor living in flood-prone areas of cities, and the E T H I O P I A CO U N T RY ST U DY 53 elderly and the sick due to their limited adaptive household consumption, and charcoal and timber capacity. Women and children left behind as male sales; that is, the less common/ dominant and less adults migrate for employment during drought- remunerative strategies. related production failures were identified as vul- nerable during and after extreme events. Other Area asset status, including social capital vulnerable groups identified included communi- and infrastructure. Social capital is an impor- ties living on already-degraded lands, and pas- tant input to adaptive capacity of both house- toral communities who face severe conflicts over holds and areas. access to land. At the area level, social capital matters. Ethnic Gender.: There is also a relationship between differences underlying diverse production systems household wealth/ livelihood composition/ asset can also exacerbate conflict over natural resources. ownership/ education levels, and female-headed Afar pastoralists’ mobility has been restricted in household status. In sum, a quarter of households Fentalle district in the Oromia region of east- from the “small, poor farm� category were female- ern Ethiopia, leading to recurrent conflicts with headed, in contrast to other, better-off household neighboring agricultural communities (including categories where female-headed households com- in-migrant highlanders) over resource access and prised just 5-7% of the total. Migration appeared land rights. The violence in the area over these to be of the “pull� factor, undertaken by those issues was presented as currently posing as much households cultivating more land. of a threat to pastoralist livelihoods as physical exposure factors of recurrent droughts, erratic Gendered norms and division of labor dispropor- rainfall, floods, and the growth of invasive bush tionately harm women in resource-constrained species that destroy the rangelands. environments such as those experiencing drought. In Birko-Debele kebele in South Wollo, Amhara Without good governance in the natural resource region in the midlands of Ethiopia, a male farmer management sector (including at the local level), stated that women suffer most from food short- conflicts over natural resources will increase in cli- ages at the household level, saying : mate-stressed environments. Water management at the local level poses a particular governance “When I realize that there is not enough food in the house, challenge. Fentalle district in the pastoralist area I go out to the nearby town or to my friends [to eat]. The receives rainfall from the highland region, but the woman cannot go out because the children will be waiting water does not move across the area due to the on her to get some food. In such cases, it is common that she dikes in Gola that have been built to protect the cooks the little she has in the house and gives to her children local sugar estate. This local water governance and puts some aside for the husband, and goes hungry her- issue further inflames tensions in an area where self. As a result, the women get sick easily.� the pastoralists have already lost land to the com- mercial sugar estates. There is a strong gender division of labor regarding particular adaptation strategies and who decides Infrastructure assets also matter to reduce sensi- about whether to undertake them. According to tivity and improve adaptive capacity. In highland the survey, men decide on agriculture and live- Ethiopia, for example, livelihood diversification was stock sales, use of savings, and all pastoralist prac- said by workshop respondents to be constrained by tices. Men often tend to also decide seed selection, a number of missing linkages such as poor road planting dates, and tillage practices, while women and telecommunication infrastructure, poor mar- control decisions regarding handicraft production, ket information, and weak credit systems. 54 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E trends, policies, and priorities in the absence of cli- Economywide Impacts mate change, but incorporating a historical pattern of climate shocks. The baseline is not a forecast, BACkgrounD but instead provides a counterfactual—a reason- able trajectory for growth and structural change of The Ethiopia CGE model contains 22 commod- the economy in the absence of climate change that ity groups and 46 activities, including 35 region- can be used as a basis for comparison with various ally differentiated agricultural sectors. Agriculture climate change scenarios. and food processing (AgFood) account for 42 percent of gross production value and generate Within the CGE model, the decisions of consum- around 50 percent of Ethiopia’s GDP at factor ers, producers, and investors change in response cost in 2005/06. AgFood imports account for only to changes in economic conditions driven by dif- 8.4 percent of Ethiopia’s total import bill, and the ferent sets of climate outcomes, as do market out- share of AgFood imports in domestic AgFood comes. The model allows a degree of endogenous demand is also fairly low (5.3 percent). The only adaptation within periods, with changes in labor agricultural commodity with a large share of allocation across sectors and crops in response to imports in domestic demand is wheat. Teff, maize, shocks. In agriculture, land cannot be reallocated barley, sorghum, and enset are all virtually non- across crops within a period in response to climate traded goods. On the other hand, agriculture shocks—cropping decisions are assumed to be makes a significant contribution to Ethiopia’s made in the beginning of the period, before the total export revenue. Agricultural exports, which realization of climate shocks are imposed. Between consist primarily of coffee and oilseeds, account periods, land and capital can shift in response to for nearly 80 percent of agricultural exports. changes in the economic environment. Agriculture represents 63 percent of total house- hold consumption, including non-marketed home In the baseline, underlying rates of productiv- production for own home consumption, with a far ity growth, world prices, foreign aid inflows, tax higher share for rural poor households. Region- rates, and government investment policies are ally, zone 2 produces nearly 50 percent of Ethio- imposed exogenously. In the climate change sce- pia’s total agricultural output and has the largest narios, climate shocks affect various parameters production share in all agricultural commodities and exogenous variables, as described below. By except enset, while zone 1’s contribution is mar- comparing results from the baseline path with ginal. Ninety-six percent of zone 5’s agricultural those from the CC scenarios, the CGE model output value is livestock production, and livestock provides an estimate of the economywide impact accounts for 31 percent of Ethiopia’s total agri- of climate change. cultural gross production value. Because comparisons are made with specific BASeline changes imposed and everything else held con- stant, the interesting results—the differences in The CGE model provides a simulation laboratory outcomes between an experiment and the base- that allows us to estimate the economic impacts of line—are generally insensitive to changes in the climate change by doing controlled experiments, assumptions underlying the baseline. Results are comparing the results of various scenarios. In order generally more sensitive to the trajectory of base- to use the model to estimate costs imposed on Ethi- line variables that are also policy variables. In the opia by global warming, we start by specifying a adaptation section, potential strategic options for “baseline� path to 2050 that reflects development adapting to climate change are presented. For E T H I O P I A CO U N T RY ST U DY 55 Table 18 SCENARIOS Name GCM CMI (%) Description Base Historical Climate Historical climate shocks Wet2 Ncar_ccsm3_0-sresa1b 23 Very wet CC shocks for Ethiopia Wet1 Ncar_ccsm3_0-sresa2 10 Global wet CC shocks Dry1 Csiro_mk3_0-sresa2 -5 Global dry CC shocks Dry2 Gfdl_cm2_1-sresa1b -15 Very dry CC shocks for Ethiopia Wet2A Ncar_ccsm3_0-sresa1b 23 Very wet CC shocks, with adaptation Wet1A Ncar_ccsm3_0-sresa2 10 Wet CC shocks, with adaptation Dry1A Csiro_mk3_0-sresa2 -5 Dry CC shocks, with adaptation Dry2A Gfdl_cm2_1-sresa1b -15 Very dry CC shocks, with adaptation Wet2AC Ncar_ccsm3_0-sresa1b 23 Very wet CC shocks, with adaptation and costs Wet1AC Ncar_ccsm3_0-sresa2 10 Wet CC shocks, with adaptation and costs Dry1AC Csiro_mk3_0-sresa2 -5 Dry CC shocks, with adaptation and costs Dry2AC Gfdl_cm2_1-sresa1b -15 Very dry CC shocks, with adaptation and costs Notes: GCM=Global Climate Model. GCM scenarios are from Strzepek et al. (2008). CMI=Crop moisture index change. “Adaptation� is the set of adaptation investment projects described in the text. example, increasing irrigation is an important period. To recapitulate from earlier sections, there adaptation investment options. If the baseline are four climate change (CC) scenarios based plan were to expand irrigation to the limits of on output from various global climate models land or water availability, then there would be (GCMs): Ethiopia very wet (Wet2), globally wet little policy space for expanded irrigation in the (Wet1), globally dry (Dry1), and Ethiopia very dry adaptation scenarios. (Dry2). The wet1 and dry1 scenarios are based on separate GCMs that yield global wet and global In defining the baseline, policy documents as well dry scenarios, and use the same GCM models as sectoral planning documents were very helpful across the case studies. The country very wet and in establishing the expected path of sector devel- country very dry scenarios are country-specific, opment, and these plans were largely incorporated using GCMs that yielded very dry and very wet in the baseline. In some cases, such as road invest- scenarios for each country. ment, the time horizon of the plan did not extend to 2050, so we had to extrapolate the existing In the scenarios with a suffix “A,� a program of plan. In the case of dams and hydropower, exist- adaptation investments are included, but with no ing plans extend to the end of the period (2050). explicit costs. These scenarios provide an upper bound of potential gains from adaption. The The scenarios considered in this report are scenarios with a suffix “AC� include costs of the described in Table 18. There is a “Base� sce- various adaptation investment programs. In this nario, which provides the baseline with histori- section, we report results for the shock scenarios. cal climate shocks, using monthly historical data Adaptation scenarios will be considered later. from the past fifty years, extended over the future 56 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E In the CGE model, the climate scenarios involve of more or fewer dams for electricity output shocks to (1) agricultural productivity by crop and and flow further downstream. Finally, crop region, (2) hydroelectric power production, and models determine the maximum potential area (3) flood damage by region to roads, crop yields, that could be irrigated given available water and livestock. flow. The CGE model incorporates directly the fluctuations in hydropower production due to eConomiC imPACtS variation in river flow, which are exogenous in the model. Climate change is expected to influence the growth and development of Ethiopia through a 3. Road Infrastructure maintenance and series of mechanisms. Four principal mechanisms upkeep. Changes in temperature and precipi- likely to alter growth and development are con- tation can influence maintenance requirements sidered. These mechanisms are: for infrastructure, particularly roads. Rainfall 1. Productivity changes in dry-land agricul- or temperature realizations outside of the band ture. The influence of climate variables on of design tolerances are likely to require more agricultural productivity is obtained from the frequent or more expensive maintenance costs. crop models (CliCrop) using the crop moisture In the CGE model, these greater maintenance index (CMI) generated by the GCM on a daily requirements result in either less rapid expan- basis. The CGE model determines how much sion in the road network for a given level of land, labor, capital, and intermediate inputs are spending on roads, or an actual shrinkage in allocated to a crop, as well as an estimated level the network if the resources necessary to main- of production under the assumption of normal tain the network are unavailable. climatic conditions. CliCrop determines devia- tions from this level as a consequence of realized 4. Extreme events. Extreme and damaging climate. The resource allocations determined in events such as floods and extended droughts the CGE and the deviations obtained from Cli- may become more frequent under climate Crop jointly determine the level of production. change, and are an output of the GCMs. The Livestock production is estimated separately simulations presented below take flood damage from CliCrop, as described in chapter 4. to crops, livestock, and road infrastructure into account. “Extreme events� refers to extreme 2. Water availability. There are three principal monthly streamflow events, not daily-scale sources of demand for water: municipal needs, flooding. The extreme events are obtained by hydroelectric power, and irrigation. The river rainfall runoff modeling using the monthly cli- basin models described earlier track water mate data from CGMs. Floods affect crop yields availability under alternative climates. Avail- directly and also indirectly through damage to able water is allocated according to a hierarchy the transport sector that increases agricultural of use. First, municipal demand is satisfied. Sec- costs. Flood damage to roads affects the capi- ond, flow is used to generate hydropower from tal stock in the transport sector (differently for available dams. Third, flow is used to irrigate paved and unpaved roads), raising transport cropland. The river basin models pass their costs. In estimating flood impacts, we have results to hydroelectric power planning models, drawn on data on the road system in Ethiopia which estimate power output given available that distinguish primary, secondary, and ter- flow, and are affected by CC shocks and con- tiary roads and the transportation model out- struction of dams over time. In addition, these lined earlier that estimates the impact of floods models assess the implications of construction on these different types of roads. E T H I O P I A CO U N T RY ST U DY 57 Figure 36 wELFARE LOSS FROM CC SCENARIOS 0.00 -1.00 WET 2 WET 1 DRY 1 DRY 2 -2.00 RATIO TO NPV OF BASE RUN GDP -3.00 -4.00 -5.00 -6.00 -7.00 -8.00 -9.00 -10.00 Note: Difference in net present value (NPV) of total absorption (defined as GDP plus imports, minus exports) from the base run as a percent of NPV of base GDP. All figures are calculated over the 2010–50 time horizon reduced production potential. In the same vein, Other potential impacts are recognized but not increased infrastructure maintenance costs imply explicitly considered. For example, climate change less infrastructure investment, which results in less may alter the incidence of malaria in parts of infrastructure capital both now and in the future. Ethiopia, with potential implications for the pat- Extreme events can destroy infrastructure capital, tern of economic activity and rates of economic which can only be replaced by additional invest- growth. Health-related implications are not con- ment over time. Generally, even small differences sidered at this stage. in rates of accumulation can lead to large differ- ences in economic outcomes over long time peri- Economic development is very much about the ods. The CGE model employed is well-positioned accumulation of factors of production such as to capture these effects. The key findings of the physical capital and human capital, and growth CGE analysis are summarized below. in technology. These factors, combined with the necessary institutional frameworks to make them Climate change has significant negative productive, determine the material wellbeing of impacts on welfare. The Dry2 and Wet2 sce- a country. The dynamic CGE model captures narios are the most damaging in net present value these accumulation processes, which largely (NPV) terms (Figure 36), with the Dry2 having drive growth. To the extent that climate change the largest negative impact. reduces agricultural or hydropower output in a given year, it also reduces income and hence The Wet2 scenario is especially damaging in the savings. This reduction in savings translates into final decade due to extreme floods, with GDP reduced investment, which translates into future loss of nearly 8 percent compared to the base. 58 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 37 DEVIATION OF GDP FROM BASE SCENARIO 2015 2025 2035 2045 0.00 -2.00 WET 2 PERCENT DEVIATION FROM BASE -4.00 WET 1 DRY 1 -6.00 DRY 2 -8.00 -10.00 -12.00 Note: 10-year average centered on year. Figure 38 STANDARD DEVIATION OF AGRICULTURAL YEAR-TO-YEAR GROwTH RATES 7.00 6.00 5.00 PERCENTAGE POINTS 4.00 3.00 2.00 1.00 0.00 BASE DRY 2 WET 2 Note: Standard deviation (SD) of annual growth rates of agricultural GDP for the different scenarios over the entire period. E T H I O P I A CO U N T RY ST U DY 59 The damage from the wet scenarios is concen- The wet scenarios tend to be better for crop yields trated in the latter years, while the damage from than dry scenarios, but floods are damaging, espe- the dry scenarios is larger and is spread more cially in the final decade (Figure 39). evenly over the period. Under the Dry2 sce- nario, by 2050 GDP is projected to be some 10 The impacts of CC shocks differ greatly across percent smaller than in the no-climate change regions for different scenarios. In zone 5, agricul- baseline (Figure 37). ture is almost exclusively based on livestock, and is very sensitive to water availability and tempera- Climate change increases variability in agriculture ture. (Figures 40 and 41). income. The variance in yields seems to increase with time, and the shocks become more negative, The impacts of the shocks on electricity genera- which is consistent with the view that the CC tion are significant. However, when we account shocks will become more intense and damaging for the construction of new dams, the supply of over time, as demonstrated in Figure 38. electricity grows faster than domestic demand and there are significant exports within a few Figure 39 AGRICULTURAL GDP, DEVIATION FROM BASE 4.00 2.00 0.00 WET 2 PERCENT DEVIATION (%) WET 1 DRY 1 -2.00 DRY 2 -4.00 -6.00 -8.00 2015 2025 2035 2045 60 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 40 REGIONAL GDP, DEVIATION FROM BASE, wET2 40.00 30.00 rrR1 PERCENT DEVIATION FROM BASE 20.00 rrR2 rrR3 10.00 rrR4 0.00 rrR5 rrURB -10.00 -20.00 2015 2025 2035 2045 Note: Regions R1 to R5 and Urban. Figure 41 REGIONAL GDP, DEVIATION FROM BASE, DRY2 10.00 5.00 0.00 rrR1 PERCENT DEVIATION FROM BASE -5.00 rrR2 -10.00 rrR3 -15.00 rrR4 -20.00 rrR5 -25.00 rrURB -30.00 -35.00 2015 2025 2035 2045 Note: Regions R1 to R5 and Urban. REGIONAL GDP, DRY2 SCENARIO E T H I O P I A CO U N T RY ST U DY 61 Figure 42 ExPORT SHARE OF ELECTRICITY PRODUCTION 100.00 90.00 80.00 70.00 PERCENT SHARE (%) 60.00 BASE TREND 50.00 40.00 WET 2 30.00 DRY 2 20.00 10.00 0.00 2048 2006 2009 2012 2015 2018 2021 2024 2027 2030 2033 2036 2039 2042 2045 Table 19 STATISTICS ON YEAR-TO-YEAR GROwTH RATES (%) OF HOUSEHOLD CONSUMPTION Scenario Household type Mean SD Min Max base Poor 5.29 1.82 2.01 8.97 base Non-Poor 5.57 1.78 0.61 9.25 Wet2 Poor 5.18 3.37 -4.50 12.77 Wet2 Non-Poor 5.44 3.23 -4.19 12.61 Dry2 Poor 5.09 2.50 -0.65 11.65 Dry2 Non-Poor 5.34 2.43 0.65 11.03 Note: Statistics on year-to-year growth rates over the entire period. Mean=simple mean of year-to-year growth rates; SD=standard deviation; Min=minimum; Max=maximum years. The CC shock scenarios lead to large year growth rates of household consumption for variations in exports, but in no scenario is there poor and non-poor households for the Dry2 and a significant shortage or price rise in the domes- Wet2 scenarios. In general, the poor suffer slightly tic market (Figure 42). less in terms of means, but have to adjust to more variability in income and hence aggregate con- Climate change impacts tend to hurt the poor sumption than non-poor households. more. Table 19 provides statistics on the year-to- 62 FO UR E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 63 Adaptation Options This chapter identifies a set of adaptation mea- Agriculture sures in agriculture, roads, and hydropower. The following adaptation options are considered: Taking into account the government’s recent development activities in the sector (Box 4), as n Increase irrigated area well as the significant (and yet largely untapped) n Increase research and development for potential for irrigation growth (Table 20), this agriculture report proposes a “portfolio strategy� approach to n Modify plans for expansion of hydroelectric adaptation in agriculture. Such an approach com- power; build more or fewer dams bines (a) programs in research and development n Build climate-resistant road infrastructure; (R&D) and farm management practices, aimed for example, increase the capacity of roads primarily at boosting yields in rainfed areas; and and bridges to withstand greater heat and (b) investments in irrigation and drainage infra- precipitation. structure. The proposed approach is consistent with the EACC “global track� analysis (Nelson et By and large, these options are identified by tak- al. 2009), which analyzed R&D and irrigation/ ing as given certain sector development objec- drainage as a direct adaptation strategy; and the tives—for example, the road network expansion expansion of rural roads as an indirect strategy. plan, or the target production of electricity from In the case of Ethiopia, adaptation in the road hydropower—and defining ways to achieve those sector is discussed in the next section. objectives even under varying climate conditions. Generally, however, adaptation might also involve changing sector development plans, or promoting a different allocation of resources across sectors. An illustrative investigation of this different line of reasoning is summarized below. 64 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E box 4 GOVERNMENT ACTIVITIES IN THE in precipitation patterns (in varying directions and AGRICULTURE SECTOR magnitudes, depending on the climate change scenarios). Recent government activities to accelerate the development of the agriculture sector have focused on promoting better integra- Under all scenarios, temperature increases, with tion into market networks of both small negative impacts on crops yields. Investment in and large farmers. This is being pursued R&D is thus intended to maintain the technol- through construction of farm-to-market ogy-induced productivity growth in the agricul- roads, increases in drainage to address tural sector at the base, no-climate-change rate, waterlogging and salinization, and increases by developing new crop varieties optimized for in irrigation area, including by means of mul- the changed climate. In each scenario an initial tipurpose dams. In addition, the government period of extensive R&D activities over the first has been supporting reforms to improve the 10 years was assumed to allow time to learn what availability of fertilizer and improved seeds, direction climate change was taking for Ethio- and specialized extension services for dif- pia. This would allow infrastructure designs to ferentiated agricultural zones and types of minimize the risk of “regrets� associated with the commercial agriculture. selection of the “wrong� adaptation response. The government is actively promoting irri- Table 20 ExISTING IRRIGATION SCHEMES gation, which has large expansion poten- (2005/06) AND IRRIGATION POTENTIAL tial, but remains largely untapped. In terms of small-scale irrigation, schemes are being Area (000 Irrigation activity hectares) supported—such as such as river diversion, Small-scale irrigation implemented micro-dam construction, and groundwater (including traditional & modern) abstraction for supplementary and double 2004/05 306.70 cropping—through provision of techni- 2005/06 242.84 cal and material support for expansion and Total—end of 2005/06 549.54 improved water use efficiency. Efforts are Medium & large-scale irrigation 61.06 also being made to strengthen water har- implemented vesting, encourage the adoption of technol- Private implemented 5.41 ogies for supplementary irrigation, and to Total irrigation potential 4,250.00 promote low-cost manual, mechanical, and Source: Ahmed, Arndt, Robinson and Willenbockel (2009). electrical water lifting mechanisms, as well as mini-drip irrigation methods and family drip- In all four adaption scenarios the baseline irriga- kits. In addition, medium- and large-scale tion development plan of 3.7 million hectares by irrigation projects are also being promoted 2050 is increased gradually to 4.1 million hectares (Ahmed, Arndt, Robinson and Willenbockel by 2050. The level of irrigation infrastructure is 2009). matched to the magnitude of climate-change- induced irrigation deficit. Note that it is possible to have increases in irrigation deficit even in wet The two pillars of the adaptation approach ana- scenarios. As warming increases, crops demand lyzed here (R&D and irrigation/ drainage) are an amount of water greater than the increases in meant to capture key aspects of a strategy capa- precipitation during the growing season. Changes ble of tackling the essential features of the climate in precipitation intensity and seasonality call for of the future—that is, an increase in temperature, increased installation of drainage systems, espe- which is common to all scenarios—and changes cially in wet scenarios. E T H I O P I A CO U N T RY ST U DY 65 The order of magnitude of investments in agri- While the adaptation strategy analyzed here cultural adaptation was determined taking addresses key aspects of sector vulnerability, it into account the opportunity cost of diverting is admittedly defined in relatively coarse terms, resources from other sectors. Using CGE model- given the aggregate level of analysis of this ing, an average annual cost of about $70 million report. Future work will be needed to spell out was assessed by expert judgment to be in a rea- in further detail individual components (and sonable range so as to avoid an excessive drain on costs) of a more comprehensive adaptation strat- economywide growth.10 egy for agriculture, such as for example, specific technologies for livestock, soil, or water manage- Irrigation infrastructure is installed on 400,000 ment; changes in planting dates, crop varieties, hectares for all scenarios. For the Wet2 sce- and cultivars; enhancement of large irrigation nario, the design was for only stream diversion schemes; development of small irrigation projects for supplemental irrigation For the Wet1, again and associated reservoirs in water-short areas; on- only stream diversions were considered, but for farm water harvesting projects; and installation of a greater amount of supplemental irrigation. For agricultural tile drainage in waterlogged areas. Dry1, the designs included water harvesting and small-scale storage reservoirs. For Dry2, small, medium, and large-scale irrigation systems were Road Transport part of the adaptation design, with correspond- ingly different levels of cost per hectare. The adaptation approach recommended for Ethiopia in the road sector is to adopt a “design Drainage infrastructure was installed on 1.4 mil- strategy� approach that emphasizes enhanced lion, 1 million, 0.9 million, and 0.4 million hect- design standards for roads and bridges such that ares for the Wet2, Wet1, Dry1, and Dry2 scenarios these new designs consider the risk of increased respectively. The distribution of the costs between climate change stressors. These design strategies the adaptation components is listed in Table 21. encourage building infrastructure with enhanced materials and technologies that are able to with- Table 21 SUMMARY OF ADAPTATION COSTS stand the increased climate stressors. IN AGRICULTURE (ANNUAL AVERAGE 2010–50, US$ MILLION) Cost Elements of the Wet 2 Wet 1 Dry 1 Dry 2 Adaptation Scenario Irrigation 16.00 30.00 32.00 50.00 Costs Drainage 36.78 23.79 21.17 7.50 Costs R&D, Farm and Water- 16.84 17.14 16.93 10.34 shed Mgt Total 69.63 70.93 70.10 67.84 10 In an economy-wide CGE analysis, after each climate shock the model reallocates inputs to activities and products so as to maxi- mize profits. In certain cases, even after adaptation, marginal returns to inputs may be higher in sectors other than agriculture, so the baseline, no-climate-change level of production may not be attained in the scenarios with climate change. 66 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E box 5 DESIGN STRATEGY ADAPTATION FOR PAVED AND UNPAVED ROADS Paved roads. The design strategy approach focuses on the concept that new structures such as paved roads will be subject to code updates if it is anticipated that a significant climate change stressor will occur during their projected life span. Historic evidence provides a basis that a major update of design standards results in a 0.8 percent increase in construction costs (FEMA 1998). The readily available data suggest that such code updates would occur with every 10 centimeter (cm) increase in precipitation or 3°C maximum temperature increase for paved roads (Blacklidge Emul- sions 2009; Whitestone Research 2008). The general dose-response relationship for paved roads is expressed as follows: CP,BHP = 0.8% (BBHP) (Equation 4) Where CP,BHP = change in construction costs associated with a climate stressor BBHP = base construction costs for paved roads. We assume a construction cost of $500,000 per kilometer (km) for a new paved road in Ethiopia, which represents the average cost per km of constructing a 2-lane collector road in rural areas based on in-country data, and $117,700 per km for re-paving a road (World Bank 2009; Washington DOT 2009; Oregon DOT 2009). These numbers can be adjusted for specific instances where data are available, or can be adjusted to represent a composite or average value of roads within a spe- cific location. Using this approach, the total additional cost for adaptation is determined based on the number of stressor thresholds that are achieved during the projected 20-year design life span. For example, if it is estimated that precipitation will increase 11 cm over the next 20 years and temperature will increase 4°C, then one precipitation threshold and one temperature threshold has been exceeded. The adaptation cost for this threshold increase is 0.8 percent of the construction costs for precipitation and 0.8 percent of construction costs for temperature. Thus, a total increase of 1.6 percent of construction costs, $8,000 per km, is required to adapt to the projected change in climate. Unpaved roads. For unpaved roads, the adaptation approach costs are directly related to specific changes in climate or infrastructure design requirements. In general terms, this approach is sum- marized by Equation 5. CURBT = M X BURBT (Equation 5) Where CURB = change in construction costs for unpaved roads associated with a unit change in cli- mate stress or design requirements M = cost multiplier BURB = base construction costs for unpaved roads The stressor-response relationship represented by Equation 5 associates the change in construc- tion costs with a 1 percent change in maximum monthly precipitation. Research findings have demonstrated that 80 percent of unpaved road degradation can be attributed to precipitation (Ramos-Scharron and MacDonald 2007). The remaining 20 percent is attributed to factors such as the tonnage of traffic and traffic rates. Given this 80 percent attribution to precipitation, we assume that the base construction costs for unpaved roads increase by 80 percent of the total percentage increase in maximum monthly precipitation. For example, if the maximum monthly precipitation increases by 10 percent in a given location, then 80 percent of that increase is used (8 percent) as the increase in base construction costs. The readily available data suggest no relationship between temperature and the cost of building unpaved roads. E T H I O P I A CO U N T RY ST U DY 67 In this approach, when the road is re-paved at the due to the fact that each climate scenario indi- end of its 20-year life span, it is repaved accord- cates that over time, climate change impacts will ing to a design standard that compensates for the increase. The longer adaptation is delayed, the change in climate. Specifically, a road should be greater the expense that must be incurred doing designed so that it can withstand the stress of reactive maintenance. increased precipitation or temperatures that will occur over its planned 20-year life span. In this The climate impacts analysis on road infrastruc- manner, no increase in maintenance costs will be ture used engineering data from the United States incurred due to the road being overbuilt at the and Ethiopia. Unpaved roads in the United States beginning to meet any increased climate-based and Ethiopia face similar engineering threats from damages that may have been required mainte- climate, and impacts and costs have been vetted nance during its life span. against a companion study on climate change impacts on transportation in Ethiopia. The mag- The quantitative analysis of the potential total nitude of the undiscounted costs appear too high adaptation costs for Ethiopia utilizing the design compared to current road investment and main- strategy approach is illustrated in Table 20. In tenance, because by 2040 the number of kilo- addition to the enhanced design strategy for paved meters of road is projected to be many times the roads, the costs assume that unpaved roads are re- current network and extreme events from climate graded and re-sealed at five-year intervals, increas- change are projected to become much more fre- ing maintenance costs as required in-between the quent and severe. This results in a multiplicative five-year cycles. These costs are additional to regu- increase in impacts exhibiting, very large impacts lar maintenance or construction costs. at mid-century. Table 22 SUMMARY OF ADAPTATION COSTS Additionally, much of the planned road expan- FOR ROADS (ANNUAL AVERAGE 2010–50, US$ MILLION) sion is with unpaved roads. Unpaved roads are very susceptible to intense precipitation events Wet 2 Dry 2 Wet 1 Dry 1 and local flooding damage. Both of these condi- tions are projected by climate models to increase Cost increase for 2.1 1.9 1.69 1.93 new paved roads under all GHG emission scenarios and are being Cost increase 4.1 3.5 2.26 4.98 reflected in the large impacts and adaptation costs for maintaining at mid-21st century. paved roads Cost increase 10.2 7.7 5.54 6.85 for maintaining Adaptation to flooding. Similar to the process gravel and earth roads used for examining flooding under a mainte- Total 16.5 13.1 9.5 13.8 nance-only scenario, the adaptation scenario uti- lizes a focus of multiplying a cost-per-kilometer Total (Cumulative 658.4 524.4 379.8 550.4 total for entire factor times the actual pool of roads that may be period 2010–50) damaged. The difference in this process is that the rate per kilometer includes adaptation costs As illustrated, the costs of adaptation compared and strategies that when implemented will pro- to the potential impacts that result from no adap- tect the road from additional damage caused by tation make it an imperative to spend a mini- climate-change-induced flooding. The basis for mal amount up-front to avoid significant costs this approach for paved roads is derived from later due to no action. The cost-benefit from the report on making transport climate resilient this expense becomes greater as time progresses (COWI 2009), where investigations were made 68 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E to make roads climate resilient and climate proof. adaptation is a higher ratio than paved roads due As introduced earlier, the adaptation approach to the actions that are necessary to make unpaved assumes that the extra costs are used to improve roads resilient to flooding. Similar to paved roads, drainage, road characteristics, and pavements to the adaptation cost is applied to unpaved roads enhance flooding resilience. whenever the anticipated flood exceeds the design flood. The outcome of this research was the develop- ment of specific cost ratios for the adaptation pro- The conclusion of the flooding cost process is the cess. For paved roads, the percentage increase for generation of the adaptation cost vectors that adapting to climate change is applied equally to are used as input for the CGE modeling. For this all categories of roads. Specifically, an 11 percent output, the flooding cost component provides an increase in base construction cost is applied for adaptation vector as follows: adaptation charges.11 This cost increase is added to all roads that are being re-paved at the end of Vector Definition their design life cycle when it is anticipated that a flood will occur in the projected life span that This is the total cost of adapt- Total Adaptation exceeds a minimum flood threshold. For this ing both paved and unpaved Cost for new roads - roads in a given year to the Paved + Unpaved study, the projected flood must exceed a 20-year 1-in-50-year flood. flood event, since this is the standard level of design that a road would incorporate in current For each vector, the data conveyed from the cost design standards. If the projected floods exceed module is divided among the three road types, pri- this level, then the adaptation cost is included in mary, secondary, and tertiary. The vectors com- the rehabilitation costs. These roads are then con- bine the climate zones together, since the CGE sidered to be resilient to a climate induced 1-in- models do not divide the economywide analysis 50-year flood for their design life span, which is into separate zones. An example of each vector is traditionally the greatest level of flooding that supplied as follows. a road is designed to withstand under normal design conditions. Additional design and adapta- In summary, the greatest adaptation costs for tion criteria would be required to make the road the road sector arise from adaptation to flood- fully resilient to a revised 1-in-100-year flood. ing. Combined, the average annual adaptation costs for both maintenance and flooding are in Similarly, unpaved roads use the same approach, the $80–$90 million per year range except for the except since unpaved roads are on a 5-year design Wet1 scenario, where the total rises to $117 mil- cycle rather than a 20-year design cycle, the adap- lion per year. Although these costs appear large, tation process is incorporated on a more frequent the cost-benefit factor must be taken into account. basis. Other than this factor, the process is the Specifically, the costs of proactive adaptation same. For unpaved roads, the adaptation increase remain at only about 20 percent of the costs that factor is 35 percent, which was derived using the will be incurred if no adaptation is put in place same detailed adaptation analysis process as dis- and a reactive perspective is put in place. cussed previously for paved roads. The cost of this 11 The 11 percent cost is obtained when the individual adaptations of drainage, road base, and road surface treatments are examined, the total cost related to the original design is an 11 percent increase. This number reflects findings by the overall EACC research team and is previously reported in broader terms in the COWI report, Making Transport Climate Resilient (COWI 2009). E T H I O P I A CO U N T RY ST U DY 69 Figure 43 TOTAL ADAPTATION COST FOR NEw ROADS (AVERAGE ANNUAL COST PER DECADE) $250,000,000 $200,000,000 $150,000,000 WET 2 DRY 2 WET 1 $100,000,000 DRY 1 $50,000,000 $0 2020 2030 2040 2050 Table 23 SUMMARY OF ADAPTATION COSTS But the case for better standards is further IN THE ROAD SECTOR (ANNUAL AVERAGE OVER THE 2010–40 PERIOD, US$ MILLION) strengthened when one accounts for the climate shocks of the future, which are likely to increase Wet 2 Dry 2 Wet 1 Dry 1 (for the same cost of construction and mainte- nance) the benefits, in terms of extended life of Road 16.5 13.1 9.5 13.8 maintenance the network span, and of avoided higher mainte- Adaptation to 71.92 73.17 107.85 67.76 nance cost in the future. floods Total Cost 88.42 86.27 117.35 81.56 A benefit-cost analysis confirms these insights. The benefit/cost ratio of adopting higher design ADAPtAtion in A Benefit/CoSt standards is 17 percent to 75 percent higher than frAmework in the baseline under the Wet2 scenario, and 16 to 55 percent in the Dry2 scenario (Figure 44). The The interventions included in the adaptation additional benefit of higher standards become analysis—consisting of enhanced design stan- more pronounced in later decades of the time dards, so that roads require less maintenance horizon considered on account of the expected under ordinary weather conditions and are less higher frequency and/or severity of flood events. vulnerable to floods—are likely to make sense (in a benefit-cost sense) even under current climate. However, financial constraints, which limit access Hydropower to the required incremental capital at construc- tion stage, make their adoption difficult. Potential adaptation policy adjustments in the hydroelectric sector include altering the scale and 70 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 44 BENEFIT/COST RATIO OF UPGRADING ROAD STANDARDS WET 2 SCENARIO DRY 2 SCENARIO 5.00 4.00 3.50 4.00 3.00 2.50 3.00 2.00 2.00 1.50 1.00 1.00 0.50 0.00 0.00 2020 2030 2040 2050 2020 2030 2040 2050 NO CLIMATE CHANGE WITH CLIMATE CHANGE NO CLIMATE CHANGE WITH CLIMATE CHANGE timing of planned projects as well as constraining The goal of the adaptation analysis was to return downstream flow and irrigation flow. the annual energy generation to base (non-climate change) values. With reduced flows and reservoir Ethiopia is embarked on a very extensive hydro- levels resulting from a drying climate change sce- power development program. This program nario, energy generation would be reduced. Since includes major dams on the Blue Nile, Atbara, in the base development plan only a portion of all and Gibe rivers. Some of these dams are planned of Ethiopia’s potential hydropower is developed, as cascades of dams in series, and some as dams the adaptation strategy is to construct a series of on parallel rivers. Climate change may result in additional hydropower projects to generate the higher average river flow, so the proposed projects “energy lost� to climate change. The costs for could produce more energy than the base case. these new plants would be the adaptation costs. Additional economic benefits can be realized—or capital costs can therefore be saved—by building Figure 45 shows the additional cost for the dry fewer projects. scenario and reduced costs for the wet scenario from the base hydropower investment develop- Drier climate change scenarios result in lower ment plan. The additional costs are incurred by average river flows and thus reduced energy gen- bringing on additional power plants sooner than eration from the projects developed under base in the base case. The figure shows cumulative conditions. Additional dams and power stations discounted costs. The annual undiscounted cost can be used to develop greater energy generation average over the 2010 to 2050 period is estimated potential for the same river flow, as well as devel- at $100 million. oping new dam sites on parallel rivers. The cost of additional capital to keep up with the Ethio- pian base-energy generation plan is assumed to Summary of Sector-Level be the cost of adaptation in the energy sector. In Adaptation Costs some cases fossil fuel (CO2-emitting) power sta- tions could be built at a lower cost, but due to mitigation targets and high fuel cost this option By summarizing the results of the sector-wise was removed as an adaptation option. analyses reported above, it is possible to evaluate E T H I O P I A CO U N T RY ST U DY 71 Figure 45 CUMULATIVE DISCOUNTED COST OF HYDROPOwER FOR TwO CLIMATE CHANGE SCENARIOS 15 CUMULATIVE DISCOUNTED COST [$ BILLION] 10 5 BASE GFDL (DRY) NCAR (WET) 0 2010 2015 2020 2025 2030 2035 2040 2045 2050 YEAR Table 24 SUMMARY OF ADAPTATION the range of total costs (Table 24) that Ethiopia COSTS, ALL SECTORS ANALYZED (ANNUAL AVERAGE 2010–50, US$ MILLION) would need to incur in order to offset climate change impacts on the sectors analyzed. That is, the cost of actions that would need to be under- Wet 2 Dry 2 Wet 1 Dry 1 taken to achieve the sectors’ development goals, Agriculture 69.6 70.9 70.1 67.8 even in the harsher climate of the future. Road 88.4 86.3 117.4 81.6 transport Adaptation costs vary considerably, depending on Hydro- the climate scenarios considered. On an annual 100.4 25.0 power average basis, the range is between $158 mil- Total annual 158.0 257.6 187.4 174.4 average lion to $258 million per year. The highest cost is associated with the Dry2 scenario, which tends to Cumula- tive total generate damages (and therefore costs to reme- for entire 6,321.9 10,304.0 7,497.9 6,976.2 period diate them) in a consistent manner throughout (2010-2050) the period considered; whereas under the Wet2 scenario, damages (and adaptation costs) tend to cluster in the final decade. 72 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E analysis between hydropower and irrigation is Adaptation: Economywide discussed in chapter 3, and was used to develop Analysis this adaptation strategy in the CGE model. In the CGE model, the costs of the dams and hydro- power investments are assumed to be financed The sector-based analysis of adaptation reported by foreign borrowing. The annual costs are cal- above is just one part of the whole picture. To culated as the interest charges (at 5 percent) on get a better representation of efforts needed to the cumulated debt in each period, which is then adapt, it is necessary to consider the opportunity assumed to be a foreign exchange outflow in the cost of diverting scarce investment resources from current account balance. Note that the costs are ordinary development objectives to enhancing higher under the “dry� scenarios, reflecting the the economy’s resilience to climate change. To need for additional investments to manage lower exemplify, building stronger roads might entail stream flows. less resources to build hospital or schools. Irrigation. The model incorporates irrigation To include these indirect effects into the analysis, and water management investments. In the adap- sector-level adaptation costs are incorporated into tation scenarios, the assumption is that the share the CGE model described in chapter 2. In par- of irrigated land and land benefiting from bet- ticular, the economy-wide effects of undertaking ter water management increases over time more the following adaptation strategies are analyzed: than in the base (no CC) scenario. The incremen- tal cost of these investments has been estimated Roads. An expanded investment program in roads and incorporated in the model. There is also a is analyzed. Such a program includes increasing tradeoff between water used for irrigation and the share of paved and hardened roads, as well for power generation. In the Dry2 scenario with as “soft� measures such as changes in transpor- adaptation, we assume that policy favors irriga- tation operation and maintenance, development tion, with some loss of hydropower production of new design standards that consider projected and exports as a result. climate changes, transfer of relevant transporta- tion technology to stakeholders, and the enhance- welfAre AnAlySiS ment of transportation safety measures. The associated adaptation costs were included in the Figure 46 presents the effects of adaptation (with model by converting the annual flows to local cur- costs) on economic well-being—measured here as rency units (billions of birr) and treating them as a net present value (NPV)—of differences in total “claim� on gross fixed capital formation (GFCF). absorption as a ratio of the NPV of GDP for the In the model, the initial road construction costs different climate scenarios. The results indicate are incorporated in GFCF in the base run. These that adaptation investments significantly offset— adaption costs are incremental; they essentially but do not eliminate—the impact of CC shocks increase the road infrastructure investment plan and improve welfare in all scenarios. for Ethiopia, increasing the share of paved roads and of “hardened� unpaved roads. The ratio between the benefits of the adaptation strategy, and the project-level costs of implement- Dams. In this case, adaptation consists of altering ing it ranges between 6 and 10, suggesting that the scale and timing of planned dam construction while not fully effective at restoring baseline wel- and hydropower projects, as well as constraining fare, the strategy is quite efficient and capable of downstream flow and irrigation flow. The conflict delivering welfare gains at relatively low cost. E T H I O P I A CO U N T RY ST U DY 73 Figure 46 NET PRESENT VALUE (NPV) OF ABSORPTION DIFFERENCES 0.0 WET2 DRY2 WET1 DRY1 -1.0 -2.0 RATIO (%) TO NPV OF BASE GDP -3.0 -4.0 -5.0 -6.0 -7.0 -8.0 -9.0 -10.0 NO ADAPTATION ADAPTATION Note: NPV of Absorption, Difference from Base (% of NPV of GDP). Absorption is defined as GDP, plus imports minus exports Time profile of costs. Figure 47 provides information the indirect, economywide benefits implications on the impact of the various scenarios on average of enhanced design standards. A more climate- GDP for various years (10-year average centered resilient road network can avoid costly disruptions on the reported year). of communications links and supply chains that increased flood frequency might bring about. The Dry2 scenario has a much bigger impact in all periods, while the Wet2 scenario has a dra- Adaptation investment is also found to have matic impact in the last decade (with three 100- income smoothing benefits: it significantly reduces year floods in that period). variability of agriculture GD growth compared to the no adaptation scenario (Figure 48). There is some evidence of “regret� adaptation investment, particularly in the case of building ADAPtAtion CoStS costly dams that may not be needed in the Wet scenarios. On the other hand, the gains from We measure adaptation costs in two ways. First, improved transport infrastructure are significant. earlier in the chapter, we calculated the direct Improved transportation is an essential part of the costs of adaptation investment projects that were Ethiopian basic development strategy, and more considered to be additional to the investment pro- resilient roads make sense even in the absence of gram in the base run. These costs are summarized climate change. As discussed in above, the argu- in Table 24. Below we also provide a general equi- ment is strengthened by introducing climate. In librium measure of direct and indirect adaptation fact, the benefits are likely to be even higher than costs. In the general equilibrium approach, the those considered at the project level because of total cost of adaptation investments includes the 74 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 47 AVERAGE DEVIATION OF GDP FROM BASE RUN (%) 2015 2025 2035 2045 0.00 -2.00 PERCENT DEVIATION FROM BASE -4.00 -6.00 -8.00 -10.00 -12.00 WET 2 WET 2AC DRY 2 DRY 2AC opportunity cost of the resources diverted from would otherwise be available for use in Ethio- gross fixed capital formation (GFCF). It is mea- pia. The effect is to cut absorption, with the cut sured as the difference between total absorption shared between aggregate consumption, govern- in the various CC scenarios with costless adap- ment spending, and investment. Table 26 shows tation and with adaptation costs. This approach the impact on foreign savings of the servicing uses the CGE model to provide the counterfac- costs on foreign debt arising from hydropower tual experiments to measure total absorption, first investment. assuming costless adaptation, and then assuming adaptation is costed, based on the sector-wise Table 25 TOTAL DIRECT AND INDIRECT analysis in chapter 3. The general equilibrium ADAPTATION COSTS ($ BILLIONS) measure includes benefits that are foregone as a result of diverting toward adaptation investment, Sum NPV Average resources that would be otherwise employed to Wet2 100.30 29.31 2.457 support Ethiopian development programs. The results are given in Table 25, which shows total Wet1 32.19 8.90 0.79 direct and indirect adaptation costs in dollars. Dry1 38.71 10.74 0.94 Dry2 115.34 32.72 2.81 In the case of hydropower, as discussed above, Notes: Sum: Sum, 2010 to 2050; NPV: Net Present Value; Average: we assumed that the costs were funded by for- Annual average eign borrowing and measured the annual flow of costs as the cost of servicing the accumulated foreign debt, using scarce foreign exchange which E T H I O P I A CO U N T RY ST U DY 75 Figure 48 STANDARD DEVIATION OF YEAR-TO-YEAR AGRICULTURE GDP GROwTH RATES 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 NO WITH NO WITH ADAPTATION ADAPTATION ADAPTATION ADAPTATION BASELINE WET2 DRY2 Table 26 FOREIGN SAVINGS ($ BILLIONS) Table 26 and Figure 50 show the differences in Sum NPV Average foreign saving in the adaptation cost scenarios. The Dry2 scenario involves the largest increase in Base 103.65 $35.29 2.53 dams and hydropower investment, which results Wet2AC 106.23 $35.66 2.59 in the largest change in foreign savings as Ethio- Dry2AC 98.49 $33.62 2.40 pia services the increased foreign debt to finance Wet1AC 103.84 $35.20 2.53 the investment. From Table 23, average annual Dry1AC 101.68 $34.70 2.48 adaptation costs vary widely and are a large share relative to foreign savings annually. Notes: Sum, 2010 to 2050; NPV: Net Present Value; Average: Annual average Figure 50 and Table 26 show the effect of dam/ hydropower investment on the current account The results shown in Table 25 and Figure 49 balance during the period. Note that in the Dry2 indicate that adaption costs in the various sce- scenario, the additional investment costs impose narios differ a lot. Adaptation in the dry scenarios interest charges that significantly reduce foreign involves expensive increased investment in dams, saving (the current account balance). By contrast, irrigation, and hydropower, while adaptation in in the Wet2 scenario, the lower construction costs the wet scenarios involves relatively major invest- and hence lower foreign debt lead to increases in ments in improved roads, which is especially evi- foreign savings relative to the base run. dent in the later periods. Figures 51 and 52 show the impact of the Wet2 and Dry2 scenarios on annual GFCF. The dry 76 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 49 DIRECT AND INDIRECT ADAPTATION COSTS ($ BILLIONS) 6.000 5.000 4.000 $ BILLIONS 3.000 2.000 1.000 0.000 t1 t3 t5 t7 t9 t11 t13 t15 t17 t19 t21 t23 t25 t27 t29 t31 t33 t35 t37 t39 t41 t43 WET 2AC WET 1AC DRY 1AC DRY 2AC Notes: Direct and indirect costs of adaptation projects. Figure 50 FOREIGN SAVING, DIFFERENCES FROM BASE RUN VALUES ($ BILLIONS) 0.30 0.20 0.10 $ BILLIONS 0.00 t0 t2 t4 t6 t8 t10 t12 t14 t16 t18 t20 t22 t24 t26 t28 t30 t32 t34 t36 t38 t40 t42 t44 -0.10 -0.20 -0.30 -0.40 WET 2AC WET 1AC DRY 1AC DRY 2AC Note: Foreign saving is the current account balance. E T H I O P I A CO U N T RY ST U DY 77 scenarios are more damaging to investment over- manuals seek to approximate by partial equilib- all, while the wet scenarios are most damaging in rium analysis and shadow pricing. the latter years, with magnitudes close to those of the dry scenarios. Not surprisingly, costless adap- The values are not discounted because the time tation is the most beneficial. Costly adaptation sequence of the benefits depends on the sequence always yields higher GFCF in all years compared of climate change shocks, which represent one to the no-adjustment scenario, indicating the suc- draw from a stochastic process. While climate cess of costly adaptation to generate increased change effects become more pronounced in the savings and investment. latter part of the period, there is no certainty about year-to-year values. The “net gain� from Benefit-CoSt ASSeSSment adaptation is a general equilibrium measure, cap- turing all the direct and indirect benefits. It is the In addition to being able to reduce welfare losses appropriate measure for benefit-cost analysis in and GDP variability, the adaptation strategy project evaluation, which project evaluation man- analyzed here appears to be quite sensible in a uals seek to approximate by partial equilibrium benefit-cost perspective. Table 27 provides data analysis and shadow pricing. The costs include all on the net gains from adaptation and the direct adaptation projects undertaken simultaneously, so costs of the various adaptation projects assumed the benefit-cost ratio measures the ratio of gains to be undertaken in the adaptation scenarios. The to costs for an adaptation investment program, net gains are measured using the CGE model and rather than an evaluation of particular projects. are measured by the difference in total welfare (absorption) between the climate change shock The results are dramatic. The mix of infra- scenario and the same scenario with costly adapta- structure projects in the adaptation investment tion. This measure is the appropriate measure of programs (hydropower, irrigation, water man- direct and indirect benefits that project evaluation agement, and roads) yield very high benefit-cost 78 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 51 GFCF, RATIO TO BASE RUN, wET SCENARIOS 1.110 1.060 RATIO TO BASE 1.010 0.960 0.910 0.860 t0 t2 t4 t6 t8 t10 t12 t14 t16 t18 t20 t22 t24 t26 t28 t30 t32 t34 t36 t38 t40 t42 t44 WET 2 WET 2A WET 2AC Note: GFCF is Gross Fixed Capital Formation Figure 52 GFCF, RATIO TO BASE RUN, DRY SCENARIOS 1.050 1.000 RATIO TO BASE 0.950 0.900 0.850 0.800 t0 t2 t4 t6 t8 t10 t12 t14 t16 t18 t20 t22 t24 t26 t28 t30 t32 t34 t36 t38 t40 t42 t44 DRY 2 DRY 2A DRY 2AC E T H I O P I A CO U N T RY ST U DY 79 Table 27 NET BENEFITS AND ADAPTATION PROjECT COSTS, US$ BILLIONS Project Benefit-cost Scenarios Welfare Losses Net gain costs ratio With adaptation Without adaptation Wet2 -61.48 -131.80 70.32 6.32 11.1 Wet1 -17.67 -55.60 37.93 7.50 5.1 Dry1 -32.67 -88.41 55.74 6.98 8.0 Dry2 -124.06 -264.59 140.54 10.30 13.6 Notes: Cumulated losses and costs over the period 2010-2050, no discounting ratios. Adaptation reduces the losses from the cli- labor force upgrading in order to offset the nega- mate change shocks by over half, and the invest- tive impacts of CC on growth and to complement ment costs are much lower than this gain in all other adaptation investments. cases. In the worst scenario, Dry2, the benefit- cost ratio is highest (13.6), which indicates that Residual compensation costs. Residual com- these projects are very worthwhile. Discounting pensation costs are measured as the difference does not change these results. between total absorption in the base run (with historical climate) and various climate change scenarios, including adaptation investments. CloSing the welfAre gAP CAuSeD By The Dry2 scenario has the largest residual com- ClimAte ChAnge pensation costs, while the impact of the increased number of extreme floods in the last decade gen- Sector-wise adaptation enables the achievement erates large losses in the Wet2 scenario, which of sector-level objectives in a cost-effective man- are only partly reduced by adaptation investment ner. But it also entails the opportunity cost of projects (Figure 46). diverting resources toward investment in climate resilience, and therefore it does not necessar- Closing the “welfare gap� through residual com- ily enable Ethiopia to fully eliminate the welfare pensation would entail mobilizing a volume of impacts of climate change. What would it take to resources much larger (Wet2 and Dry2 scenarios) “make the economy whole� again? Two options than in the case including only adaptation costs. are explored in this section. Labor force upgrading. The second approach The first is to estimate the “residual compensa- considered to close the welfare gap is to imagine tion costs� as the transfers (in $ US) that would be a significant departure from the baseline develop- required to completely offset the loss of absorp- ment trajectory, rather than considering the lat- tion from CC shock, after implementing adap- ter as fixed and seeking ways to offset the impacts tation investments. The second is to explore an that climate change may have on it. In particular, alternative development strategy not considered a program of labor-force upgrading is analyzed, in the sector-wise analysis, namely a strategy of whereby it is assumed that unskilled rural labor 80 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Table 28 ADAPTATION COSTS AND RESIDUAL DAMAGE (ANNUAL AVERAGE, 2010-2050), US$ BILLIONS Scenario Adaptation costs Residual Damage Total Wet 2 2.45 1.52 3.97 Wet 1 0.79 0.43 1.22 Dry 1 0.94 0.81 1.75 Dry 2 2.81 3.03 5.84 Figure 53 RESIDUAL DAMAGE COSTS ($ BILLIONS) 12 10 8 BILLION US$ 6 4 2 0 -2 t1 t3 t5 t7 t9 t11 t13 t15 t17 t19 t21 t23 t25 t27 t29 t31 t33 t35 t37 t39 t41 t43 WET 2AC WET 1AC DRY 1AC DRY 2AC migrates to the urban area, lowering the rural significant potential benefits of accelerating the share of total labor by 0.1 percent a year. The diversification of the economy away from highly migrated labor is assumed to lead to skill upgrad- climate sensitive sectors, such as agriculture. ing, with all categories of urban labor increasing. imPACtS on Poor AnD non-Poor When tested under the Wet2 scenario, an adap- houSeholDS tation strategy including such a labor-upgrading program appears to be able to more than offset, The Ethiopia data separate poor and non-poor over the whole time horizon considered, the negative households, so it is possible to analyze the impact impacts of climate change on welfare (Figure 54). of climate change and adaptation scenarios on household welfare. The figures below provide While this scenario does not involve costing the information on these impacts. skill upgrading program, this finding points to the E T H I O P I A CO U N T RY ST U DY 81 Figure 54 DISCOUNTED DIFFERENCES IN ABSORPTION FROM BASELINE (2010–2050, wET2 SCENARIO) 8.00 6.00 RATIO TO NVP OF BASELINE GDP 4.00 2.00 NO ADAPTATION ADAPTATION 0.00 ADAPTATION AND SKILL UPGRADING -2.00 -4.00 Note: NPV of Absorption, Difference from Base (% of NPV of GDP). Absorption is defined as GDP, plus imports minus exports Figures 55 to 58 present results concerning the scenario. The effects are larger than in the Wet2 impact of the climate change scenarios on house- scenario, and spread over the entire period. hold welfare, measured by total household con- sumption. From Figure 55, the impact of climate Figure 58 shows the impact of the extreme wet change shocks on the welfare of poor and non- and dry scenarios, with and without adaptation, poor households is roughly the same. The Dry2 on the coefficient of variation of the year-to-year scenario is the most damaging in all periods, but growth rates of total household consumption. the Wet2 scenario is very damaging in the last The mean of the year-to-year growth rates for period, with serious floods affecting roads. both poor and non-poor households is around 5 percent, and the coefficients of variation (CV) Figure 56 shows the impact of the Wet2 scenario, range from a low of 0.3 to a high of 0.65. They with and without adaptation. Adaptation invest- represent the year-to-year changes in consump- ment is very effective, offsetting about half the tion to which households must adjust. A value impact of climate change in both the extreme dry of 0.3 in the base run indicates that households and wet scenarios. In the earlier periods, adapta- must manage annual swings in the change in tion greatly reduces the impact of the Wet2 sce- consumption of 30 percent. The CVs for poor nario—only in the final period, with extreme floods, households are slightly higher than those for non- does adaptation only partly offset the impact. The poor households—poor households must deal impacts of adaptation on household welfare are with more income variability than the non-poor. similar for both poor and non-poor households. The impact of the climate change scenarios on Figure 57 provides the same data for the Dry2 the CVs is significant—doubling in the extreme 82 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 55 HOUSEHOLD CONSUMPTION: DIFFERENCE FROM BASE, CC SHOCKS 2015 2025 2035 2045 0.00 -2.00 PERCENT CHANGE FROM BASE % -4.00 -6.00 -8.00 -10.00 -12.00 WET 2 HH POOR WET 2 HH NONP WET 1 HH POOR WET 1 HH NONP DRY 1 HH POOR DRY 1 HH NONP DRY 2 HH POOR DRY 2 HH NONP Figure 56 HOUSEHOLD CONSUMPTION, DIFFERENCE FROM BASE, wET2 CC SHOCKS AND ADAPTATION 2015 2025 2035 2045 1.00 0.00 PERCENT CHANGE FROM BASE �%� -1.00 -2.00 -3.00 -4.00 -5.00 -6.00 -7.00 -8.00 -9.00 WET 2 HH POOR WET 2 HH NONP WET 2 AC HH POOR WET 2 AC HH NONP E T H I O P I A CO U N T RY ST U DY 83 Figure 57 HOUSEHOLD CONSUMPTION, DIFFERENCE FROM BASE, DRY2 CC SHOCKS AND ADAPTATION 2015 2025 2035 2045 0 -2 PERCENT CHANGE FROM BASE �%� -4 -6 -8 -10 -12 DRY 2 HH POOR DRY 2 HH NONP DRY 2 AC HH POOR DRY 2 AC HH NONP Figure 58 COEFFICIENT OF VARIATION OF YEAR-TO-YEAR GROwTH RATES, HOUSEHOLD CONSUMPTION 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 POOR NON POOR POOR NON POOR POOR NON POOR BASE WET2 DRY2 Note: Coefficient of Variation (CV) is the standard deviation (SD) divided by the mean of the year-to-year growth rates. 84 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E wet scenario (Wet2) and increasing by half in the participation in cash for work programs. Specifi- extreme dry (Dry2) scenario. While not shown in cally, household coping strategies in Kalu district Figure 58, adaptation is effective in reducing the included migration to neighboring towns (with impact of the climate change shocks on the coef- migration distances becoming longer as the effects ficients of variation, moving them about half way of extreme events become more pronounced), to the base run values. and the sale of goats. In Fentalle district, coping strategies included the sale of livestock; camel migration during the dry season from October Adaptation: Priorities at the to January; reducing household expenditures to Community Level staple items only; firewood and charcoal sales in nearby towns; and participation in cash-for-work projects from January to June. None of the coping ADAPtAtion PrACtiCeS AnD CoPing strategies identified in the study could significantly StrAtegieS reduce the effect of climate impacts on communi- ties. Some also further contribute to environmen- Adaptation practices by households vary according tal degradation, as in the case of illegal firewood to hazard type, location and asset base holdings. and charcoal sales. The most common forms of adaptation practices currently used by households as identified by the As one farmer said in Choresa kebele, “I am aware survey were (a) crop selection of drought-toler- of the use of the forest for environment, for the ant crops (78 percent of households); (b) terrace soil and rainfall situation in the area, but I can’t let rehabilitation (72 percent); (c) soil erosion preven- my family die when there are some trees around tion programs (69 percent); and (d) homestead to cut and make money out of.� or forest restoration (62 percent) and change of planting dates (51 percent). Income diversifica- Long-term adaptation planning (beyond short tion strategies (including migration, non-timber and medium-term measures such as changing forest product sales, handicrafts, and timber sales) crop types and planting dates) include the need were not common. The households in these areas for economywide diversification to industry and (drought-prone highland, midland, and lowland services, as well as significant improvements in areas) already face significant deprivation cur- human capital levels (education and training) to rently, with 84 percent of surveyed households allow households to take advantage of risk-pre- reporting food shortages during part of the vention strategies at household and community year. This makes the selection of transformative levels. adaptation strategies more difficult for individual households. Private collective action was also not Finally, complex social responses to past extreme common. The outreach of agricultural extension events hold important lessons on the need to con- agencies in the field-site villages was high, though sider the indirect impacts of, for example, distress NGOs and cooperative activity much less preva- migration to cities. In Ethiopia, the influx of rural lent. This suggests the importance of consider- in-migrants during drought events led to slum ing the use of existing public service providers in overcrowding, decreases in urban wage rates, and adaptation planning and capacity-building. increases in food prices. The timing and scale of migrant flows to second- and first-tier cities also Focus group discussions identified additional changed as the crisis deepened, with the capi- elements of coping strategies. They included tal Addis Ababa ending up as the refuge of last some temporary migration, sale of assets, and resort for the entire country. Earlier public and E T H I O P I A CO U N T RY ST U DY 85 private actors (e.g., church institutions and family The PSD workshops conducted in case countries networks) played complementary social protec- revealed broad support for NAPA and related cli- tion roles in Ethiopia, providing formal state cash mate strategy priorities in-country, in such areas transfers and private housing respectively. State as agriculture and water resources management, support now predominates. land management, roads, and early warning sys- tems. However, they also revealed stakeholder ADAPtAtion PreferenCeS AriSing preferences for investments in governance, social from PSD workShoPS protection, training and education, and land ten- ure. Training and education were identified as a The PSD workshop process included participants need not only for livelihood diversification, but identifying their preferred long-term development also in the area of increased capacity building in vision for the area, as well as expected impacts of cli- community-based approaches to climate change mate change on that vision, and adaptation invest- adaptation and natural resource management. ments needed in order reach toward that vision. Specifically, the three local participatory scenario Notably, workshop participants’ future visions for development (PSD) workshops (in highland, the area were not limited to climate outcomes, midland, and lowland areas), and one national but sought reduction of social tension, improved workshop identified soil and forest rehabilitation, safety net programs, benefit-sharing with external irrigation and water harvesting, improved agri- investors in the area and Awash National Park, cultural techniques and drought-resistant variet- resettlement programs, improved participation/ ies, education, and land use rights for pastoralists use of consultation in development planning pro- as adaptation preferences. Regional development cesses, and improved livestock services. Similarly, and the need for structural shifts toward service the Kalu district workshop included a vision of and industry sectors to improve employment out- improvements in education, employment, and comes were also raised as issues. At the national infrastructure. In terms of impacts, Karrayu pas- level, similar options were identified, along with toralists in the Fentalle district PSD workshop a focus on early warning systems and flood con- identified how their vulnerability and worsening trol measures, agricultural technology, finance livelihood position was related not only to rain and market development, renewable energy, and shortages but also due to additional factors of urban planning. The adaptation options identi- forced land alienation, population growth, and fied at local and national levels generally aligned land conflict, which interacted with the naturally with the natural resource and agriculture focus degrading pasturelands to cause animal (and in the NAPA, which also identifies needed invest- eventually human) disease and death, as well as ments in crop insurance, wetlands protection, migration and withdrawal from school. carbon livelihoods, agroforestry, and anti-malaria initiatives. 86 F IV E E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 87 Recommendations The findings of this analysis suggest that impacts by the government for roads, dams, hydropower, of climate change will be quite significant, par- water management, and irrigation would signifi- ticularly as Ethiopia approaches the middle of cantly increase longer-term vulnerability to cli- the century. While the magnitude of the impacts mate change and thus make adaptation costlier. remains considerable—irrespective of whether However, there are a number of additional issues the climate of the future will be wetter or drier— that the government could consider to further several important adaptation decisions are sensi- enhance the contribution of GTP to Ethiopia’s tive to what climate is expected. climate resilience (and thus, ultimately, to the ability of the country to support sustained, longer Given the large uncertainties regarding future cli- term growth). In particular, two of the GTP pil- mate outcomes, the approach to enhance Ethio- lars deserve attention: pia’s climate resilience should be couched in terms of a gradual, adaptive, and learning paradigm. n The agricultural sector as the engine of Such an approach could be articulated for both growth. the shorter term—including the implementation n Expand the coverage and enhance the quality of the Growth and Transformation Plan (GTP) of infrastructure. recently issued by the government)—and for the longer term. AgriCulture The GTP purports to “Continue the on-going Shorter Term (up to 2015) effort of improving agriculture productivity in a sustainable manner so as to ensure its place of By and large, the Growth and Transformation the engine of growth.� The analysis of this report Plan supports a number of actions which, by indicates that under future climates many regions boosting growth, will contribute to the enhance- of Ethiopia will face decreases in agricultural ment of Ethiopia’s resilience to climatic shocks. production. This suggests that agricultural pro- Robust growth based on infrastructure investment duction as an engine of growth is vulnerable to is likely to be the first line of defense against cli- climate change and climate variability. While the mate change impacts. Relatively small deviations more pronounced effects on crops and livestock from the ambitious investment targets set forth are likely to materialize in later decades, efforts to 88 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E enhance the resilience to climate shocks of crop The GTP includes ambitious targets for upgrad- yields and livestock production should be stepped ing the road network, including 70,000 km of all- up as soon as possible, particularly on account of weather, woreda-level roads. Unpaved roads only the lead time needed to strengthen research sys- have a 5-year design span until resurfacing and tems and to transfer and adapt findings from the are very susceptible to flooding damage, which lab to the field. has very large indirect, economywide costs on supply chains and health and education services. Investments in improved agricultural productiv- The government may want to consider a more ity—such as watershed management, on-farm detailed economic analysis of the road expansion technology, access to extension service, transport, targets to determine if building fewer, but more fertilizers and improved seed varieties, and cli- climate resilient roads, is preferable to building a mate and weather forecasting—will enhance the larger number of roads, but which are likely to be resilience of agriculture, both to droughts and to more vulnerable to climate shocks. The case for waterlogging caused by floods. National and local the former option seems compelling under current actions will need to be supported by international climate, and will become even sounder under the efforts (for example, through the CGIAR system) climate of the future. In addition, should interna- to develop climate resilient agriculture technolo- tional climate finance resources become available gies, given the global public goods nature of these in the future for Ethiopia—from the Copenhagen innovations. Green Fund or other mechanisms—the govern- ment might consider utilizing these resources for roAD infrAStruCture enhancing the climate resilience of the road net- work expansion plans. The GTP aims to expand the coverage and enhance the quality of infrastructure: “Focus will energy be given to the development of roads, railways, energy, telecommunication, irrigation, drinking Current water resources and Ethiopian topog- water & sanitation and basic infrastructure devel- raphy indicate an overall potential of more than opments; (…) With regard to roads, rural roads 30,000 megawatts in economically viable hydro- will be constructed on all regions and all rural power generation capacity. The GTP approach is kebeles will be connected (through) standardized to focus on “the development of water and wind all weather roads with main highways.� energy options to fulfill the energy demand of the country,� with targets for hydropower of 6,000 to Modeling results show that existing infrastructure 8,000 MW in additional generation capacity. The design standards—the level of prevention against hydropower analyses of this report (conducted at extreme events, such as local and regional flood- the monthly scale, which is adequate for sector- ing—are inadequate to address current climate wide planning purposes, although not for plant- variability and will impact economic growth rates level design and operation), provides support, in the near and middle terms. Results from cli- from a climate change perspective, for the GTP mate change analyses show that this issue is likely targets. The projects likely to go online in the next to become worse in the middle to long term. The 5 years have very low risk of being impacted by government should consider enhancing infra- climate change. structure design standards as soon as possible. This would make sense—in benefit/ cost terms— While in the longer term hydropower develop- already under today’s climate, but even more so ment will become increasingly more climate under the harsher climate of the future. sensitive, projects in the current pipeline are E T H I O P I A CO U N T RY ST U DY 89 likely to be less vulnerable to shocks as the over- and infrastructure development programs. Early lap between their life span, and the time when planning for the more severe climate impacts of stronger climate change effects will materialize, is mid-century is desirable, so as to avoid locking the relatively limited. Some climate change scenarios country into a climate-vulnerable development actually project an increase in Ethiopian runoff, trajectory, particularly when it comes to economic resulting in larger volumes of hydropower gen- processes with a high degree of inertia, or invest- eration, and thus making the case for investment ment decisions concerning infrastructures with a in hydropower stronger. long life span. In the nearer term, the economics of hydropower Due to the uncertainty of future climate, a risk- investments will be influenced less by climate, and based investment planning approach should be more, on the demand side, by the evolution of adopted. Robust decision-making principles are domestic and external markets (regional African needed to minimize the “regrets� of climate- power grids). A sustained expansion of national sensitive decisions. As climate shocks become and foreign demand for power will be key to sup- more frequent and severe, the “opportunity costs port the expansion of Ethiopia’s hydropower of capital� invested in projects and programs sector, which in turn will be vital to support the viable only under a limited set of climate out- country’s accelerated economic growth. comes becomes too large. Some key areas to be considered to develop a climate risk management Thus, in the short run, expansion of hydropower approach to support long term development generation should be accelerated as a way to sup- include the following. port growth, and to facilitate the transition of the economy from being highly agriculture-dependent mACroeConomiC mAnAgement to having a broader productive base in industry and services. Given the vulnerability of the agri- Historically the Ethiopian economy has been vul- cultural sector to current climate shocks (let alone nerable to climate fluctuations (Figure 59). The those to be expected in the future), strengthening analysis of this report shows that climate variabil- of the electricity sector, and in particular the pro- ity will increase under all scenarios. Since agricul- motion of regional and Africa-wide power grids ture (the economy’s most climate sensitive sector) to receive Ethiopia’s excess power, should be a is likely to remain for some time one of Ethiopia’s priority in the investment strategy. Strengthened main engines of growth, climate-induced shocks hydropower development can both increase near- will continue to be a threat to macroeconomic sta- term economic growth and make the energy sys- bility because of the impacts on income, employ- tem more climate resilient. More reservoir storage ment, fiscal revenues, capital formation, and the distributed over the country would provide more drain on government expenditure and aid flows reliability and protection from regional droughts. to support disaster relief. Under climate change, renewed efforts will be Medium to Long Term necessary to buffer the economy from more frequent and/or severe climate shocks. These As Ethiopia looks into the next stages of develop- include strengthening social safety nets, access to ment (starting with preparation of the next growth relief funds, drought early warning systems, crop plan, which will follow the GTP 2011-2015), it insurance programs, grain banks, and strengthen- might want to evaluate more closely the impli- ing infrastructure design. cations of climate change for its overall policies 90 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Figure 59 ECONOMIC GROwTH AND CLIMATE 80 25 20 60 15 40 10 20 5 -0 % 0 -5 2000 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 -20 -10 RAINFALL VARIABILITY -15 -40 GDP GROWTH -20 -60 AG GDP GROWTH -25 -80 -30 YEAR Source: De Jong, The World Bank (2005) Promote DiverSifiCAtion ACroSS evAluAte the ClimAte reSilienCe of SeCtorS of inCome AnD emPloyment lArge infrAStruCture ProjeCtS In the longer term, however, accelerated diversi- As we move toward mid-century, the range of fication of income and employment sources away possible climate futures broadens to encompass from climate-sensitive sectors such as agriculture markedly different “wet� and “dry� scenarios. is likely to become increasingly important under This has implications for the optimal timing of a more erratic climate. It be should explored in dams and other investments in water infrastruc- closer detail, particularly because it holds prom- ture, which is likely to be quite sensitive to climate ise to be a cost-effective way to eliminate residual outcomes. Large projects of this type should be welfare damage caused by climate change. subject—on account of the large capital outlays involved—to careful climate-robustness tests. The government may want to look into ways to accelerate the absorption of the rural labor force To adequately inform the design of subsequent into non-agricultural activities, including through generations of water infrastructure projects, skills-upgrading programs and encouragement of investments in enhancing national hydrometeoro- growth poles around medium-size municipalities. logical services and data collection and analysis are E T H I O P I A CO U N T RY ST U DY 91 crucial to assist identifying which climate change Given the significant pay-off of addressing inter- path Ethiopia is actually on, and to provide inputs nal and transboundary conflicts on water use to the adaptive management process for resource before they arise, the government might want to management. Better data on hydrometeorologi- consider investments into river basin planning cal processes, and stronger capacity to analyze systems and institutional arrangements that can and model them, is key to making more informed facilitate information sharing, dialogue, and dis- decisions on issues such as the number of hydro- pute resolution. power plants, the design of individual plants, and the operation of the grid. ProACtively ADDreSS ConfliCtS in wAter uSeS Under “dry� future climate scenarios, competi- tion among users of water (e.g. municipal and industrial consumption, hydropower generation and irrigation) might become more acute, par- ticularly in certain river basins. The availability of water to downstream riparian countries might also be affected. 92 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E E T H I O P I A CO U N T RY ST U DY 93 Bibliographic References Adams, R.M., B.A. McCarl, K. Segerson, C. Rosenzweig, COWI. 2010. “Making Transport Climate Resilient, Country K.J. Bryant, B.L. Dixon, R.Connor, R.E. Evenson, and D. Report-Ethiopia.� Kongens Lyngby, Denmark: COWI A/S Ojima. 1999. “The economic effects of climate change on Croppenstedt, A., M. Derneke and M.M. Meschi (2003) US agriculture.� In R. Mendelsohn and J. E. Neumann, Technology Adoption in the Presence of Constraints: The eds. Impact Of Climate Change On The United States Economy. Case of Fertilizer Demand in Ethiopia. Review of Development Cambridge, UK: Cambridge University Press. Economics 7(1), 58-70. Ahmed, H., C. Arndt, S. Robinson, D. Willenbockel. 2009. Deressa, T.T. (2006) Measuring the Economic Impact of Climate “Ethiopia: A Country Study on the Economic Impacts of Change on Ethiopian Agriculture: Ricardian Approach. Climate Change.� World Bank Report No.46946-ET. The CEEPA Discussion Paper No.25. Centre for Environmental World Bank, Washington, DC. Economics and Policy in Africa: University of Pretoria. Bekele, Z. and K. Tilahun. 2006. “On-farm performance evalu- Diao, X. and A. Nin Pratt (2007) Growth Options and Poverty ation of improved traditional small-scale irrigation practices: Reduction in Ethiopia - An Economy-wide Model Analysis. A case study from Dire Dawa area, Ethiopia.� Irrigation and Food Policy 32, 205-28.Dimaranan, B.V. 2006. “Global Trade, Drainage Systems 20(1): 83–98. Assistance, and Production: The GTAP 6 Data Base.� West Blacklidge Emulsions, Inc., 2009. “SHRP Performance Graded Lafayette, IN: Center for Global Trade Analysis, Purdue Asphalt Binders.� Accessible at: (viewed May 31, 2009). EDRI – IDS (2009) Ethiopia: Input-Output Table and Social Block, P. 2010. “Analysis of the Economics of Adaptation to Cli- Accounting Matrix. Ethiopian Development Research mate Change for the Agriculture and Energy Sectors within Institute: Addis Ababa. Ethiopia.� Final Report for the World Bank EACC Study. FAO (Food and Agriculture Organization of the United Washington, DC: World Bank. January 27. Nations). 2005. Fertilizer Use in Ethiopia. FAO: Rome. Block, P., and K. Strzepek. 2009. “Economic Analysis of Large- FAO (Food and Agriculture Organization of the United scale Upstream River Basin Development on the Blue Nile Nations). 1989. “Irrigation Water Management: Irrigation in Ethiopia Considering Transient Conditions, Climate scheduling.� Series title: Irrigation water management, Variability, and Climate Change.� J. of Water Res. Planning and Training manuals-4. Accessible at: (accessed on Sept. 8, 2010). Breisinger, C., X. Diao, J. Thurlow, and R.M. Al-Hassan. 2008. FAO. 2000. “AQUASTAT: Information System on Water and “Agriculture for Development in Ethiopia: New Opportuni- Agriculture.� Accessible at: (accessed on July 18, 2010). Washington, DC: IFPRI. FAO. 2005. “Africa Dams Database.� Accessible at: (accessed on July 25, 2010). 2005. “The economic and food security implications of climate change in Mali.� Climatic Change 68 (3): 355–378. FAOSTAT. 2000. “Live Animals.� Accessible at: (accessed on August Cerlanek , W. D., C.M. Zeigler, and S.E.Torres. 2006. “Main- 15, 2010). tenance of Paved and Unpaved Roads in Alachua County.� Alachua, FL: Department of Public Works CIESIN (Center FDOT. 2009. “Generic Cost Per Mile Models.� Florida Depart- for International Earth Science Information Network ment of Transportation. Accessible at: (viewed Gridded Population of the World. Accessible at: (accessed August 4, 2010). 94 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E FEMA. 1998. Promoting the Adoption and Enforcement of Seismic Build- 14, 2010 from ftp://ftp.fao.org/docrep/fao/005/y4176E/ ing Codes. Federal Emergency Management Agency, FEMA y4176E00.pdf. 313. Washington, DC: FEMA.GAMS. 2005. “General Palmer, W.C. 1965. “Meteorological drought.� Research Paper Algebraic Modeling System.� Accessible at: Weather Bureau. Hinkel, Jochen, and Richard J.T. Klein. “DINAS-COAST: Postel, S.L., G.C. Daily, and P.R. Ehrlich. 1996. “Human Developing a Method and a Tool for Dynamic and Interac- appropriation of renewable fresh water.� Science 271:785. tive Vulnerability Assessment.� Accessible at: . Ramos-Scharron, C. E. and L.H. MacDonald. 2007. “Run- Hughes, G., P. Chinowsky, and K. Strzepek. 2009. “The costs of off and suspended sediment yields from an unpaved road adaptation to climate change for water infrastructure in Q5 segment, St. John, U.S. Virgin Islands.� Hydrological Processes OECD countries.� Global Environ. Change (submitted). 21(1): 35–50. ICPAC (2007) Climate Change and Human Development in Seo, S.N., and R. Mendelsohn. 2006. The Impact of Climate Change Africa: Assessing the Risks and Vulnerability of Climate on Livestock Management in Africa: A Structural Ricardian Analysis. Change in Kenya, Malawi and Ethiopia. Human Development Pretoria: Center for Environmental Economics and Policy in Report Office Occasional Paper No.2007/8. Africa (CEEPA), printed by the University of Pretoria. ILRI (International Livestock Research Institute). 2007. Sieber, J., and D. Purkey. 2007. User Guide for WEAP21. Somer- “Ethiopia Livestock.� GIS shapefile. Accessible at: ville, MA: Stockholm Environment Institute, US Center. (accessed on August 15, 2010). Accessible at: Climate Change: New Methods and Estimates. Washington, DC: (accessed July 1, 2009). World Bank. Otte, M.J. and P. Chilonda. 2002. Cattle and small ruminant Zeleke, G., W. Bewuket, and D. Alemu. 2010. “Learning from production systems in sub-Saharan Africa: A systematic Past Experiences with Extreme Climate Events.� Draft review. Livestock Information Sector Analysis and Policy Report, January 2010. Addis Ababa, Ethiopia. Branch, FAO Agriculture Department. Accessed on August E T H I O P I A CO U N T RY ST U DY 95 96 E C O N O M I C S O F A D A P TAT I O N T O C L I M AT E C H A N G E Ministry of Foreign Affairs Government of the Netherlands the world Bank group 1818 h Street, nw washington, D.C. 20433 uSA tel: 202 473 1000 fax: 202 477 6391 www.worldbank.org/eacc