70271 S A M OA CO U N T RY ST U DY i Economics of Adaptation to Climate Change SAMOA 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 S A M OA CO U N T RY ST U DY i Economics of Adaptation to Climate Change SAMOA Ministry of Foreign Affairs Government of the Netherlands 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 © 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 © Shutterstock, except Pages 4, 10, 18, and 55 © iStockphoto S A M OA CO U N T RY ST U DY iii Contents Acronyms vii Acknowledgments ix 1 Overview 1 2 Vulnerability To Climate Change 5 Background 5 Projections of Climate Change—Global Scenarios 8 Sea Level Rise 9 3 Impacts Of Climate Change And Natural Hazards 11 Historical Experience 11 Modeling the Frequency of Extreme Events 12 Economic Damage Caused by Natural Disasters 13 Estimating the Costs of Adapting to Climate Change 14 The Potential Impact of Climate Change on the Agricultural Sector 16 Model of Climate-Economy Interactions 17 4 Adaptation Plans And Costs 19 Strategy for the Development of Samoa (SDS) (2008–12) 19 National Adaptation Programme of Action (NAPA) (2005) 20 Coastal Infrastructure Management (CIM) Strategy and Plans 22 Evidence of Prioritization and Sequencing—Donor Influence 23 Adaptation and Design Standards for Infrastructure 24 Coastal Protection 27 NAPA Adaptation Options 28 Agriculture 32 Adaptation Costs and Benefits 32 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 5 Institutional and Social Analysis 39 Incorporation of Existing Data on Potential Risks and Hazards 39 Consistency Across and Within Sector Plans 40 Local Perspectives 40 Implementation: How Is It Working in Practice? 41 6 Lessons and Future Work 43 7 References 44 Annex 1. Storm Return Periods and Economic Damages 47 Annex 2. El Niño Events and Agricultural Production 49 S A M OA CO U N T RY ST U DY v Tables 1. Projected Changes in Climate Variables by GCM and Region 8 2. Economic Damage Caused by Recent Natural Disasters in Samoa 14 3. Losses Due to Climate Change Without Adaptation 17 4. Adaptation Costs by Infrastructure Category for 10 Year Standard 26 5. Adaptation Costs as Percent of Baseline Expenditures 27 6. Adaptation Costs by Infrastructure Category for 50 Year Standard 28 7. Costs of Coastal Protection and Residual Damage from DIVA Model 29 8. Priority Adaptation Activities Based on the NAPA 30 9. Potential Adaptation Options for NAPA And CIM Plans 31 10. Dates for Implementation of NAPA Adaptation Measures by Region 32 11. The Impact of Climate Change with and without Adaptation 34 12. Total Cost of Adaptation for 10 Year Design Standards by Scenario and Decade 34 13. Total Cost of Adaptation for 50 Year Design Standards by Scenario and Decade 34 Figures 1. Main Islands in Samoa 5 2. Thirty-Year Climate Averages for Rainfall and Wind Distribution 6 3. Monthly Temperatures in Apia 1993-2008 7 4. Regions of Samoa Used in the Climate-Economy Model 7 5. Illustration of Cyclone Return Periods by Peak Wind Speed 13 6. Links Between the NAPA and the SDS with Reporting Relationships 21 7. Approval Process for Adaptation Projects 22 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 S A M OA CO U N T RY ST U DY vii Acronyms ADB Asian Development Bank CIM Coastal infrastructure management CSIRO Commonwealth Scientific and Industrial Research Organisation DIVA Dynamic and interactive vulnerability assessment ENSO El Niño southern oscillation GCM General circulation model IFPRI International Food Policy Research Institute MNRE Ministry of Natural Resources, Environment, and Meteorology MWCSD Ministry of Women, Community and Social Development NAPA National Adaptation Program of Action NCAR National Center for Atmospheric Research NCCCT National Climate Change Country Team NOAA National Oceanic and Atmospheric Administration SDS Strategy for the Development of Samoa SLR Sea level rise SPSLCMP South Pacific Sea Level and Climate Monitoring Project SRES Special Report on Emissions Scenarios Note: Unless otherwise noted, all dollars are U.S. dollars. 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 S A M OA CO U N T RY ST U DY ix Acknowledgments The EACC study was conducted by a partnership coordinated production, and Hugo Mansilla pro- consisting of the World Bank (leading its techni- vided editorial and production support. cal aspects); the governments of The Netherlands, The United Kingdom, and Switzerland (funding The study team would like to thank the part- the study); and the participating case study coun- nership, which initiated, funded and actively tries. The EACC study includes a global study and engaged with the study team through its mul- seven country studies including the Samoa study. tiyear journey. The team is also grateful for the active engagement throughout the study of the The Samoa Country Study was coordinated by government of Samoa, in particular the Min- Laurent Cretegny, Sergio Margulis, and Kiran istry of Natural Resource, Environment, and Pandey. This report was written by Gordon Hughes Meteorology, Ministry of Finance, Ministry of with contributions and other material provided by Agriculture and Fisheries, Ministry of Women, Laurent Cretegny, Simon Bannock, Michele Daly, Community and Social Development, Ministry Erwin Kalvelagen, Peter King, Thakoor Persaud, of Works, Infrastructure and Transport, Samoa Isikuki Punivalu, Graeme Roberts,and Fernanda Water Authority, Samoa Bureau of Statistics, Zermoglio (consultants). The peer reviewers were Samoa Tourism Authority, and the Land Trans- Jan Bojo, Kirk Hamilton, Ian Noble, and Michael port Authority. We would also like to thank Rob- Jacobsen. Robert Livernash provided editorial ser- ert Livernash for editorial services and Hugo vices, Jim Cantrell contributed editorial input and Mansilla for production support. x 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 S A M OA CO U N T RY ST U DY 1 Overview Over the last two decades Samoa has suffered with the greater wind stresses and more intense major damage from two cyclones in 1990–91, precipitation associated with worse storms. minor damage from a third cyclone in 2004, and an earthquake tsunami in 2009. Changes The impact of climate change on agriculture is in the scale and impact of these types of nat- uncertain. Historical data show that variations in ural disasters are likely to be important con- ENSO indices are correlated with variations in sequences of climate change for the country precipitation in the following months. There is because the increases in sea level and in aver- also a weaker correlation between El Nino events age sea surface temperatures will increase the and declines in production of taro (the staple crop intensity and damage from major storms. Other of Samoa) and rises in agricultural imports. If cli- potential impacts are linked to changes in the mate change increases the magnitude and/or fre- weather patterns associated with El Niño South- quency of strong El Niño events, this would have ern Oscillation (ENSO) events. The primary an effect on agriculture. Still, the potential scale concern focuses on the impact on agriculture, of the impact seems likely to be small relative to especially in periods of lower precipitation fol- existing risks due to weather variability, disease, lowing strong El Niño episodes. and external market conditions. This study examines the consequences of an A macroeconomic model of the interactions increase in average temperatures of up to 1°C between climate and the economy suggests that by 2050 and up to 2.75°C by 2100 for the fre- the present value of the damage to the Samoan quency and intensity of major cyclones that hit economy through 2050 due to climate change— the islands. Estimates of the economic damage and without additional adaptation—may be caused by storms in the past have been used to $104–$212 million; this is equivalent to 0.6–1.3 calibrate a damage function that yields an esti- percent of the present value of GDP over the mated increase in the expected value of economic same period. The model assumes that sound damage as the peak wind speeds for storms with development policies are adopted to minimize return periods of 10, 50, or 100 years rise over the impact of existing weather risks and other time. In this framework the key element of adap- natural hazards along with those from climate tation is to ensure that buildings and other assets change. The loss of income falls on consump- are designed to standards that enable them to cope tion rather than investment, so the reduction in 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 present value of consumption is estimated at For agriculture the key element of an adaptation 0.9–1.9 percent. strategy is to increase expenditures on research, development, and advisory services to mitigate the The starting point for examining adaptation higher risks that are likely to be associated with options is the observation that the Samoan econ- climate change. Again, this is building upon poli- omy is resilient to storms with a return period of cies that would form part of a sound development 10 years but suffers significant damage from storms strategy. Samoan agriculture was hit hard by the with a greater return period, i.e., storms of greater taro blight in the mid-1990s, which devastated taro severity. This 10-year threshold is low by interna- production and eliminated taro export revenues tional standards. The analysis suggests that raising for nearly 15 years. Vulnerability to disease, pests, the threshold to 50 years would generate benefits and storm damage means that diversification of that greatly outweigh the costs involved even with- both crop varieties and crops is an important ele- out any consideration of climate change, reducing ment in any policy to limit the impact of these risks the expected annual value of storm losses from on agricultural households. Since climate change is 5.5 percent to 0.7 percent of GDP. Implementing likely to reinforce these risks, the appropriate level measures to ensure that buildings, infrastructure, of expenditure will be higher to reflect the greater and similar assets are able to withstand storms with value placed upon risk reduction. a return period of up to 50 years is a clear “no regrets� strategy. It is justified even without climate The Government of Samoa has undertaken change, but it will also greatly reduce the economic extensive consultations to identify community pri- impact of climate change under the climate sce- orities for adaptation to climate change under the narios that were examined. These actions can National Adaptation Program of Action (NAPA). quicken the attainment of development of goals These include protection of community water sup- including the MDGs. plies, support of agriculture and forestry sectors, implementation of coastal infrastructure man- The corollary of this strategy is to ensure that agement plans and integrated catchment man- design standards are adjusted to take account of agement. This study applies a cost-benefit test to prospective changes in the distribution of storms assess the appropriate timing of adaptation proj- as a consequence of climate change. This leads ects identified in the National Adaptation Program to the adoption of forward-looking design stan- of Action (NAPA). Some of these projects—e.g., dards rather than ones based on current climate upgrading water systems—are good development conditions as an adaptation strategy. In practi- projects under current climate conditions. How- cal terms this means that buildings and other ever, large investments in relocating coastal infra- assets constructed in the near future should be structure should only be implemented if and when designed to withstand wind speeds up to 160 the reduction in the expected value of storm dam- kph —the projected 50-year storm in the decade age exceeds the annualized costs. If a 50-year storm 2050–59 under the Commonwealth Scientific design standard were implemented, the analysis and Industrial Research Organisation (CSIRO) suggests that this type of adaptation may not be scenario—rather than 148 kph (the 50–year justified before 2050 based on climate change con- storm under current climate conditions). Strictly, siderations because the general gain from reducing the design standards will vary for different cli- storm damage is not sufficient to warrant addi- mate scenarios, and the additional costs of com- tional expenditures. However, these actions may plying with the worst scenario may be viewed be justified as a measure to reduce the risks associ- as a form of insurance against uncertainty over ated with non-climate hazards (for instance risk of future climate outcomes. tsunamis associated with earthquakes). S A M OA CO U N T RY ST U DY 3 The overall cost of adaptation is much higher change, especially with respect to activities which under the CSIRO scenario than under the have a large overlap between development and National Center for Atmospheric Research adaptation benefits. On the other hand, imple- (NCAR) scenario because the former projects a menting an overall approach to adaptation based much greater increase in the severity of flood- on an assessment of the risks associated with ing and storms. Under the CSIRO scenario, the storms and other natural hazards is hampered cost of adaptation would rise from $3.3 million by limitations in the collection and interpreta- per year in 2010–19 to $10.9 million per year tion of consistent and relevant information. It in 2040–49. The main cost arises from looking will be important to focus on strengthening the forward to the end of the century in setting the government’s capacity to develop early warn- design standards for buildings and infrastructure ing systems, to effectively utilize these warning constructed in the 2030s and 2040s. systems to prepare for and prevent losses, to develop, update and implement design stan- The Government of Samoa, with external assis- dards, and other measures required to mitigate tance, is implementing a program to act on the damage caused by projected changes in the community priorities for adaptation to climate frequency and severity of storms. 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 S A M OA CO U N T RY ST U DY 5 Vulnerability to Climate Change Background is projected to increase slowly to about 210,000 by 2050. Average life expectancy at birth was 72.8 Samoa is a small island country located in the years in 2006, which is higher than the median South-West Pacific between latitudes 13°–15°S for Pacific Island countries. About 76 percent of and longitudes 168°–173°W. It has four main the population live on the island of Upolu. The inhabited islands and six small uninhabited islands capital and only significant urban center is Apia in with a total land area of 2,935 sq. km. The two North Upolu with a population of about 40,000. main islands are Savai’i (the larger of the two but much less densely populated) and Upolu (Figure The islands are volcanic in origin, so the topog- 1). The population is approximately 183,000 and raphy includes mountains up to 1,850 m as well Figure 1 MAIN ISLANDS IN SAMOA 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 Figure 2 THIRTY-YEAR CLIMATE AvERAGES FOR RAINFALL AND WIND DISTRIBUTION Source: MNRE, Meteorology Division - Figure 2 in Government of Samoa (2005) as low-lying coastal areas. About 70 percent of occur in the period from December to February. the country’s population live in the coastal zone. The islands are also affected by dry spells that The coastal zone is vulnerable to sea level rise, coincide with the El Niño Southern Oscillation and all areas are subject to damage caused by the (ENSO). Figure 2 shows the geographical distri- high winds, storm surges, and torrential rainfall bution of annual precipitation and wind speeds associated with severe tropical cyclones. It is esti- based on historic averages from 1961 to 1990, mated that two cyclones—Ofa in 1990 and Val in while Figure 3 shows temperatures in Apia over 1991—caused damage to agriculture, infrastruc- the last two decades. ture, and other assets valued at 2.5 to 3 times the country’s GDP in 1990. A review of historical climate trends for Apia suggests that the daily maximum temperature The climate is tropical with a wet season from increased by about 0.7°C over the 20th century November to April and a dry season from May while the daily minimum temperature increased to October. Temperatures vary little over the by 0.2°C. Average annual precipitation decreased year with a typical daily range of 24° to 32°C. by about 49 mm over the century. There is some Average annual rainfall is high at 3,000 mm; evidence that the severity and/or the frequency about two-thirds of annual rainfall falls in the of tropical cyclones have increased in the SW wet season. Severe tropical cyclones tend to Pacific; see Chapter 1 of World Bank (2006). S A M OA CO U N T RY ST U DY 7 Figure 3 MONTHLY TEMPERATURES IN APIA, 1993-2008 33° 32° 31° MAXIMUM 30° 29° 28° 27° MEAN 26° 25° 24° MINIMUM 23° 22° 21° 20° 19° 18° 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Source: SPSLCMP data; Figure 18 in SPSLCMP (2008). Figure 4 REGIONS OF SAMOA USED IN THE CLIMATE-ECONOMY MODEL CLIMATIC HAZARDS: WINDS, LIMITED OR ERRATIC RAINFALL KEY ECONOMIC ACTIVITY: TOURISM SAVAII NORTH SAVAII SOUTH KEY ECONOMIC ACTIVITY: TOURISM KEY ECONOMIC ACTIVITY: MAJOR ECONOMIC ZONE ECONOMIC GROWTH: HIGH UPOLU NORTH UPOLU SOUTH Source: World Bank. KEY ECONOMIC ACTIVITY: TOURISM ECONOMIC GROWTH: MODERATE This study has divided the country into four climate-economy model. Figure 4 illustrates the regions: Savai’i North (SN), Savai’i South (SS), regional division and highlights the major eco- Upolu North (UN), and Upolu South (US). This nomic characteristics of each region. Upolu allows for regional differences in climate sce- North contains Apia and has the highest popu- narios, economic activity, and incomes in the lation, including all of the urban population. 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 Table 1 PROjECTED CHANGES IN CLIMATE vARIABLES BY GCM AND REGION Baseline values for NoCC Deviations in 2050 relative to NoCC Precipita- Precipita- Mean Precipita- Precipita- Mean Total pre- tion tion tempera- Total pre- tion tion tempera- Climate cipitation Dec–Feb Nov–Apr ture cipitation Dec–Feb Nov–Apr ture model Region (mm) (mm) (mm) (°C) (mm) (mm) (mm) (°C) NCAR Savai’i North 2,958 1,062 1,921 26.87 -17 -39 5 0.99 NCAR Savai’i South 3,002 1,107 1,971 26.86 -19 -41 3 0.99 NCAR Upolu North 3,048 1,154 2,024 26.83 -21 -42 0 0.99 NCAR Upolu South 2,929 1,090 1,942 26.67 106 -8 118 0.97 CSIRO Savai’i North 2,958 1,062 1,921 26.87 277 43 197 0.81 CSIRO Savai’i South 3,002 1,107 1,971 26.86 343 65 215 0.83 CSIRO Upolu North 3,048 1,154 2,024 26.83 344 68 218 0.83 CSIRO Upolu South 2,929 1,090 1,942 26.67 335 66 213 0.83 Source: World Bank analysis; see World Bank (2009); NoCC= no climate change The situation is rather more complicated for Projections of Climate precipitation. Notwithstanding its name, the Change – Global Scenarios Global Wet scenario projects changes in total annual precipitation by region in the range -21 to +106 mm by 2050 with a value of -21 mm for Samoa is covered by four of the 0.5° grid cells for Upolu North. In contrast, the Global Dry sce- which projections of climate variables have been nario projects changes in the range +277 to +344 downscaled from the results of the general circula- mm by 2050 with a value of +344 mm for Apia. tion models (GCMs) used in this study. These can There are also changes in the seasonal distribu- be mapped to the four regions shown in Figure 4 tion of rainfall. For example, total precipitation with the largest population in the grid cell cen- for December–February—the prime period for tered on 13.75°S, 171.75°W, which corresponds cyclones—falls by about 40 mm by 2050 in the to Upolu North and covers Apia. The Global Wet Global Wet scenario, whereas precipitation dur- (NCAR) and Dry (CSIRO) scenarios differ little ing the rainy season from November to April is with respect to the increase in the annual average stable or increases slightly in Upolu South. Thus, temperature (Table 1). The Global Wet scenario the dry season becomes drier while the transi- projects an increase of 0.97–0.99°C by 2050 for tional months of the wet season become wetter the four regions, while the Global Dry scenario in this scenario. The shift is not as marked for the for 2050 projects an increase of 0.81–0.83°C by Global Dry scenario but still roughly one-half of 2050 for the four regions. Since the differences the increase in annual precipitation occurs during between regions are much smaller than the stan- the transitional months of the wet season. dard errors of the projections, it is reasonable to assume a uniform increase of about 1°C for Observation and modeling suggest that higher the Global Wet scenario and about 0.8°C for the global temperatures, including sea surface tem- Global Dry scenario. Changes in average daily peratures, will mean that the peak wind speeds maximum and daily minimum temperatures (not and probably precipitation and flooding associ- shown) are almost identical to the changes in aver- ated with severe cyclones will increase. This is age daily mean temperatures. equivalent to a shift in the tail of the distribution S A M OA CO U N T RY ST U DY 9 of extreme weather events (see Annex 1). These scenario in the coastal protection component of considerations point to a more frequent return the global study. The main scenario included in period of cyclones with the severity of Cyclones the global estimates assumes a rise of 87 cm from Ofa and Val, which had a devastating impact on 1990 to 2100. Other forecasts for sea level rise Samoa in 1990–91 (see Section 3). However, since in the South Pacific are considerably more pes- these cyclones rank second and third in the list of simistic. Evidence about changes in the height of the most damaging cyclones in the South Pacific storm surges is even more difficult to obtain, but region as a whole in the last 50 years, it is very dif- recent trends indicate that the difference between ficult to quantify the potential change in exposure average sea level and the maximum value of to extreme storm damage for different scenarios. hourly sea level measured in each year has been increasing at about 3 cm per decade (Young 2007). There is a separate but less well–documented concern that the severity—and perhaps the fre- The main earthquake zone in the South Pacific quency—of ENSO cyclical variations in weather lies about 200 km south of Samoa, with more conditions will increase. This could mean that than 12 earthquakes of magnitude 7 or greater the severity of ENSO dry periods may increase, since 1900. This means that the risks to coastal especially under the Global Wet scenario for infrastructure associated with climate change which average precipitation during the dry sea- must be assessed in a framework of vulnerability son is expected to fall. This is discussed further to other natural shocks. in Section 3. Data collected by SPSLCMP show 18 separate tsunami events over 15 years, with the largest Sea Level Rise trough-to-peak height of 570 cm after an earth- quake near the Kuril Islands in November 2006. The South Pacific Sea Level and Climate Moni- The tsunami that followed the magnitude 8.1 toring Project (SPSLCMP) has collected sea level earthquake, which occurred about 190 km south and climate data for Samoa since 1993. The most of Apia in September 2009, is reported as hav- recent country report published in December 2008 ing a trough-to-peak height of 140 cm in Apia. gives an average increase in sea level of 4.9 mm per The waves that struck parts of South Upolu were year at Apia.1 If sustained over the 21st century much greater; for example, the tsunami generated this would imply an increase of 54 cm from 1990 a trough-to-peak height of 314 cm in Pago Pago to 2100, which is a little greater than the assump- (American Samoa), and there were reports of tion of 40 cm used for the low sea level rise (SLR) waves of up to 450 cm in parts of Upolu. 1 Figures for another sea level gauge yield an average increase of 2.1 mm per year over a longer period, but this estimate is regarded as less reliable because of less precision and poorer datum control. 10 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 S A M OA CO U N T RY ST U DY 11 Impacts of Climate Change and Natural Hazards Historical Experience the Apia Observatory suggested a return period of about 25 years with a storm surge of 1.6 m and Over the last two decades Samoa has suffered onshore sustained wind velocity of 130 kph. serious economic shocks caused by four major natural disasters. The worst of these were two severe cyclones in successive seasons—1989–90 CyClone Val (DeCember 1991) and 1990–91—which caused massive damage to infrastructure and other assets. Cyclone Val was classified as a Category 5 hur- ricane with wind speeds up to 240 kph. It was CyClone ofa (february 1990) judged to be the most destructive storm to hit the Samoan islands in the 20th century because of Cyclone Ofa was classified as a Category 4 hur- its intensity, irregular path, and slow speed (Fair- ricane with wind speeds up to 215 kph. It struck bairn 1997). Damage was caused by a combina- Samoa as the storm was strengthening and con- tion of high winds, heavy rains, and wave action. tinued to affect Samoan waters during the worst The duration of the cyclone and the shifts in wind phase of the storm. Extensive damage occurred direction exacerbated the destructive effects of along the northern coasts of both Upolu and high winds, so that crops and plantations, natural Savai’i. A review of the damage on Upolu caused vegetation, buildings, and other structures all suf- by the storm reported seven deaths; major dam- fered extensive damage. The main effects of high age to buildings and infrastructure caused by waves were felt on the south coasts of Upolu and waves that inundated coastal zones and high Savai’i, an unfortunate complement to the impact winds inland; disruption to shipping and commu- of Cyclone Ofa. The storm was reported as hav- nications, partly caused by damage to infrastruc- ing caused 12 deaths in Samoa. It defoliated a ture and partly as a consequence of the beaching high proportion of trees on the main islands, and of the country’s main passenger/cargo ferry; and caused the complete loss of nearly 50 percent of loss of reclaimed land, plus damage to agricultural coconut trees. Analysis of wind speeds and the assets such as plantations caused by flooding and associated storm surge suggested that the storm wind (Rearic 1990). A comparison of the severity had a return period of about 100 years, with an of the storm with a theoretical 100–year storm by onshore sustained wind velocity of 165 kph. 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 CyClone Heta (January 2004) systems can minimize loss of life, though they cannot prevent widespread structural damage Cyclone Heta passed near to or struck Samoa, with high economic costs. American Samoa, Tonga, and Niue over a period of about a week from January 2 to January 8, 2004. Samoa was relatively fortunate, since the Modeling the Frequency center of the storm passed about 110 km west of of Extreme Events the country at a point when it was still classed as a Category 2 hurricane. Even so, sustained wind speeds of 110 kph were reported in Apia with Statisticians and others interested in analyzing gusts up to 145 kph. Damage on Samoa included natural hazards such as tropical storms, floods, or the loss of or disruption to power, water, and tele- earthquakes have developed standard methods of communications on Savai’i plus limited flooding modeling the distribution of extreme events over in Apia. On the other hand, Alofi—Niue’s capi- a period of time. Suppose that the severity of tal city—was devastated by the direct impact of tropical storms is measured by its peak wind speed Cyclone Heta, which at that stage was classified measured over a period of 10 minutes and that St* as a Category 5 hurricane. denotes the peak wind speed for the worst storm in year t. This is closely correlated to the amount tSunami (SePtember 2009) of wind damage caused by the storm and is likely to provide a reasonable proxy for storm damage In the early morning of September 29, 2009, an caused by rain and flooding since higher wind earthquake of magnitude 8.1 in the Kermadec- speeds tend to be associated with more intense Tonga Subduction Zone (part of the Pacific Ring precipitation over periods of 6 to 12 hours. of Fire) generated a tsunami that struck the south- ern coastal areas of Samoa about 10 minutes The probability that a storm with a wind speed after the earthquake. It is reported that the waves greater than S will occur in year t at a particu- caused damage up to 14 m above sea level on the lar location—prob (St* ≥ S)—is usually assumed coast of South Upolu and traveled up to 0.7 km to follow the cumulative distribution function for inland. Twenty coastal villages in the south and some variant of the generalized extreme value southeast of Upolu were mostly destroyed, with (GEV) distribution (Annex 1). The return period widespread damage to transport, power, and tele- of a cyclone with peak wind speed S is the recipro- communications infrastructure. About 5,300 peo- cal of this probability. Carter (1990) examines the ple were affected, with 3,000 losing their homes. return period for cyclones in the South Pacific, the Samoa 5-degree square, and Apia.2 The reported death toll caused by the tsunami in Samoa was 155—more than 5 times the total The “historic� curve in Figure 5 shows the esti- number of deaths caused by cyclones from 1990 mated return periods for peak wind speeds of to 2010. One lesson is that loss of life and eco- cyclones during the 20th century, while the nomic damage caused by natural disasters are not closely correlated. Geological events such as 2 Unfortunately, the expressions given in the paper for peak wind speed, wave height, etc., as a function of the return period— earthquakes and tsunamis, which are concen- equations (4) to (8)—are clearly wrong, though the graphs appear trated in their impact and occur with little or no to be correct. For example, using Figure 4 the text states that the wind speed with a 40–year return interval is 77 knots, whereas warning, cause much heavier loss of life relative equation (4) yields a value of 438 knots (p. 15). Thus, the return to their economic costs than weather events for periods quoted in the text have been used to derive correct values of the Gumbel distribution parameters used for the calculation which effective disaster planning and warning of historic return periods in Figure 5. S A M OA CO U N T RY ST U DY 13 Figure 5 ILLUSTRATION OF CYCLONE RETURN PERIODS BY PEAk WIND SPEED RETURN PERIOD (YEARS) 200 180 160 140 120 100 80 60 40 20 0 100 110 120 130 140 150 160 170 180 190 200 210 220 WIND SPEED (KPH) HISTORIC CLIMATE FUTURE CLIMATE - LOW FUTURE CLIMATE - HIGH “future� curve illustrates how the return periods infrastructure, and other assets shortly after the might fall if it is assumed that both the location event, and (b) the present value (using a real dis- and scale parameters of the storm distribution count rate of 5 percent) of the shortfall in GDP increase by either 10 percent (Future climate– relative to the trend rate of economic growth low) or 25 percent (Future climate–high) due to prior to the shortfall in economic growth.3 climate change. These increases correspond to the range for the year 2100 discussed below. The Such figures must be treated with some caution. changes have the effect of reducing the return It is often unclear how the value of damage to period of a storm with a peak wind speed of 165 capital assets is obtained. The usual procedure kph (89 knots) from 100 years to approximately is to estimate the cost of repairing or replacing 55 years for the low scenario and 26 years for the damaged assets, but this is rarely carried out on high scenario. a like-for-like basis. Existing assets may be old and will have been partly or fully depreciated. They are usually less efficient or they may be less Economic Damage Caused suitable for current requirements. Indeed, own- by Natural Disasters ers may decide that it is not economic to replace them. Hence, the economic value of assets that Table 2 summarizes estimates of the economic 3 For example, over the years 1992–94 GDP fell 7–8 percent below damage—expressed as one-off capital losses— the cyclically adjusted trend in GDP, apparently as a consequence of the combined effects of Cyclones Ofa and Val. The shortfall caused by the natural disasters that have affected was about 12 percent for agricultural GDP in 1994 since agricul- Samoa over the past two decades. The estimates tural assets, particularly coconut palms, were severely affected by Cyclone Val. However, by 1996 both total and agricultural GDP are based upon (a) reports of damage to buildings, had recovered to match the pre-disaster trends. 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 2 ECONOMIC DAMAGE CAUSED BY RECENT NATURAL DISASTERS IN SAMOA Return Asset damage Loss of GDP Total GDP Loss as period US$ million, US$ million, US$ million, percent Event Years 2005 prices 2005 prices 2005 prices of GDP Notes Cyclone Ofa 25 166 15 161 113 Buildings & infrastructure Cyclone Val 100 388 36 163 260 30% agricultural assets Cyclone Heta 10 1 4 236 2 Limited damage Tsunami 2009 50 54 50 277 38 Buildings, infrastructure & tourism Source: World Bank estimates are damaged is likely to be significantly less than Estimating the Costs of their replacement or even repair costs. Adapting to Climate Change Similarly, the estimates of the loss of GDP caused by Cyclones Ofa and Val shown in the table are The technical basis for calculating the change in somewhat overstated, because they include the the expected value of storm damage as a conse- impact of the taro blight, which affected crop pro- quence of a shift in the probability distribution of duction from 1994 onward and delayed the full cyclones is set out in Annex 1. This focuses atten- recovery from the disasters by up to two years. tion on two components of the costs of adapting Further, the higher rate of economic growth asso- to climate change: ciated with recovery from the damage caused by the storms persisted after the recovery and a. The changes in design standards required to the long-term rate of economic growth clearly ensure that new or upgraded assets can resist increased in the late 1990s. The storms were cer- the higher wind speeds to which they are likely tainly not a benefit in disguise, but their long-term to be exposed as a consequence of the shift in impact has been reduced by pushing the Samoan the distribution of storms hitting Samoa. This economy toward sectors that appear to support a approach is built into the estimates of the costs higher long-term rate of economic growth. These of adaptation for infrastructure and related assets considerations mean that the estimates shown in prepared for Samoa under the global analysis of Table 2 set an upper bound on the economic losses the cost of adaptation. These global estimates associated with storms of different return periods. include an allowance for the increased costs of maintenance and/or accelerated replacement Policies to minimize the economic damage caused of existing assets because they were built to by cyclones will require the adoption of building lower design standards and are more likely to be standards capable of withstanding much greater damaged by storms with higher wind speeds. wind stresses or relocating infrastructure so that it is less susceptible to damage caused by flood- b. Expenditures and investments for other mea- ing and/or wave action. These measures will also sures to adapt to climate change that were reduce the impact of geological events, though identified in the NAPA. These are discussed in the attraction of coastal areas for tourism will more detail in Section 4 below. inevitably mean that people and assets are vulner- able to events triggered by earthquakes. There is, however, an important constraint on any detailed examination of the impact of S A M OA CO U N T RY ST U DY 15 climate change or the costs of adaptation. It is scenario for Samoa—it has been assumed that clear that a key feature of climate change for the key parameters of the distribution increase Samoa is the possibility that the distribution of by 4 percent to 2050 and by 10 percent to 2100. extreme weather events—primarily cyclones but These figures fall within the range identified by also ENSO droughts—will shift to more frequent Knutson, et al. (2010) and they correspond to or severe cyclones hitting the islands. Unfortu- the increases in average temperatures for this nately, the empirical evidence on how the distri- scenario combined with the middle of the range bution might shift over time for different climate of the sensitivity of cyclone intensity to ocean scenarios is very limited (IWTC 2006). temperatures.4 As a high estimate—linked to the relatively wet CSIRO scenario for Samoa— The damage caused by tropical cyclones is due it has been assumed that the key parameters of to a combination of high winds and precipitation the distribution increase by 8 percent to 2050 and leading to flooding. Both Emanuel (2005) and by 25 percent to 2100. These figures allow for a Knutson, et al. (2010) emphasize the link between strong shift toward more intense cyclones and a sea surface temperatures and cyclone intensity, greater intensity of precipitation associated with with higher ocean temperatures being associ- such cyclones. The assumptions for the CSIRO ated with both higher sustained wind speeds and scenario are intended to provide a worst-case higher rainfall. Generally, changes in sea surface scenario for the change in the distribution of temperatures follow changes in average land tem- cyclones given current knowledge. peratures but with a considerable lag. For this analysis there is little alternative other than to rely The analysis has also taken account of a signifi- upon changes in average land temperatures as the cant change in the distribution of precipitation basis for distributing changes in storm intensity over the year, especially for the CSIRO scenario. up to 2050 and 2100. Total precipitation during the rainy season aver- aged over the grid cells is 1,965 mm for the In the most recent assessment of probable shifts NoCC scenario and increases to 2,175 mm for in the intensity and frequency of cyclones as a the CSIRO scenario. This is likely to increase the result of climate change, Knutson, et al (2010) overall probability of severe flooding during the suggest that the overall intensity of cyclones rainy season and will require higher design stan- may increase by 2–11 percent in the period to dards for buildings, roads, storm water drainage, 2100, with the main driver being the increase in and similar infrastructure to cope with an increase ocean temperatures. This change may involve a in the intensity and volume of rainfall. Hence, the decrease in the frequency of cyclones but with a calculations of the cost of adaptation include an large margin of uncertainty (6–34 percent) off- allowance for the costs of providing greater resil- set by an increase in the frequency of the most ience to flooding linked to precipitation in the intense cyclones. In addition, increases on the rainy season. This is separate from and additional order of 20 percent are likely in the intensity of to the costs associated with the increase in precip- precipitation within 100 km of the storm for the itation intensity associated with severe cyclones. most severe cyclones. To capture the uncertainty about changes in both the peak wind speeds and precipitation associated 4 The statement prepared by a WMO expert group (IWTC 2006) states that both theory and observation support a conclusion that with any shift in the distribution of cyclones by the intensity of tropical cyclones will increase by 3–5 percent per intensity, a range of values has been used. As a 1°C rise in sea surface temperature. Emanuel (2005) uses a figure of 5 percent in considering changes in his power dissipation low estimate—linked to the relatively dry NCAR index for tropical cyclones. 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 The Potential Impact of Annex 2 provides details of a statistical analysis of the links between ENSO events, precipitation, Climate Change on the and agricultural production. This confirms that Agricultural Sector strong El Niño periods lead to a reduction of about 18 percent in precipitation in the months that fol- The agricultural sector in Samoa is a crucial source low, with little difference between the impact in of employment—providing the main occupation the wet and dry seasons. If climate change were for 32 percent of the workforce in the 2006 Cen- to lead to a 50 percent increase in the variation sus—but is a much smaller contributor to GDP. between normal and extreme ENSO conditions, Agriculture and fishing accounted for 10 percent then strong El Niño periods might be associated of GDP at constant prices in 2008, down from 20 with a reduction of up to 27 percent in precipita- percent in 1988, before hurricanes Ofa and Val tion. However, there is one important qualifica- and the taro blight. Value-added in agriculture tion. The analysis also indicates that the influence and fishing at constant prices reached a peak in of ENSO events on precipitation may have been 1996 and fell by 30 percent from 1996 to 2004. declining since 1950. If this apparent trend is real Since 2004 agricultural output has grown but less and continues in future, the impact of climate rapidly than the rest of the economy, so that value- change on the variability of ENSO-related pre- added in agriculture on its own has fallen from 8.1 cipitation would be small. percent to 6.2 percent of GDP over the five-year period 2004–09. This share is likely to fall below 5 In any case, the relationship between ENSO- percent in the decades up to 2050. related changes in precipitation and agricultural production has not been large in the past. The Research into the potential impacts of climate second part of the statistical analysis looked at the change on crop yields and viability in Samoa and extent to which ENSO events have been associ- other South Pacific islands is very limited. The ated with fluctuations in crop production, crop general assessment seems to be that an increase of yields, and trade in agricultural products. Again, 1°C in average temperatures and minor changes the results confirm that there is a link primarily in total precipitation up to 2050 will not have a driven by the impact of ENSO events on taro large effect on yields of the main crops such as production. A typical strong El Niño event has taro, bananas, and coconuts; see, for example, the been associated with a decline of 5–8 percent in analysis for Vita Levu, Fiji in World Bank (2000). taro production and an increase of 3–5 percent The changes are within the range of variation in the value of agricultural imports. The effects observed across countries where the same or simi- might rise to a decline of 8–12 percent in taro lar varieties are grown.5 Nonetheless, uncertainty production and a rise of 5–7 percent in agricul- about the effects of climate change would war- tural imports if the climate change were to lead rant expenditures on agricultural research and to a 50 percent increase in the variation between development as an insurance policy. normal and extreme ENSO conditions. A separate concern has been expressed about the To put these potential changes in context, the trend prospect of more severe dry periods associated with rate of growth in agricultural imports has been 6.2 ENSO events. Again, the evidence on this issue percent per year over the last two decades, while is limited and open to different interpretations. variations in agricultural production caused by the cyclones in 1990–91 or the taro blight were an 5 This is documented in the relevant data sheets from FAO’s order of magnitude greater than the potential vari- Ecocrop database, which can be accessed at: http://ecocrop.fao. org/ecocrop/srv/en/home. ations associated with the effect of climate change S A M OA CO U N T RY ST U DY 17 Table 3 LOSSES DUE TO CLIMATE CHANGE WITHOUT ADAPTATION No adaptation NCAR (1) CSIRO (2) Present value @ 5 percent, $ million 103.9 212.4 Annualized equivalent, $ million per year 5.9 12.1 Loss/benefit as percent of baseline GDP 0.6% 1.3% Loss/benefit as percent of baseline consumption 0.9% 1.9% Source: World Bank estimates. on the intensity of ENSO events. This means that (c) Growth rates for population and GDP without a sound strategy for reducing the effects of existing climate change correspond to the baseline data weather and agricultural risks on farmers and the used for the Global Track estimates. These economy should be the starting point for adapt- assumptions determine the growth in capital ing to climate change. Crop diversification, reli- stock and total factor productivity. ance upon varieties that can cope better with dry periods or resist wind damage, the development of (d) Gross investment is calculated assuming a stan- alternative sources of employment and income, dard depreciation rate applied to the current and similar measures will limit the damage caused capital stock plus net capital accumulation. by existing risks and facilitate adaptation to climate change. It will then be easier and less expensive (e) Total consumption is equal to total output to implement additional measures to reduce vul- minus climate damage, adaptation costs, and nerability to risk if concerns about changes in the gross investment. The present value in 2010 of severity of ENSO events due to climate change are total consumption from 2010 to 2050 is used sustained by evidence from experience over the to assess the impact of climate change and the coming decades. effects of adaptation. The expected value of the economic damage Model of Climate-Economy caused by cyclones is calculated in the manner Interactions described in Annex 1 using the shifts in the prob- ability distribution of storms by 2050 for the NCAR and CSIRO scenarios based on changes A simple climate-economy model has been used to in average temperature. To maintain investment examine the impact of climate change on economic and economic growth, the damage caused by activity and the effects of spending on adaptation storms has the effect of reducing consumption, measures. The key features of the model are: so that the cost of climate change for each sce- nario is measured as the reduction in the pres- (a) There are four regions, North and South ent value of consumption in USD at 2005 prices Upolu/Savai’i, with populations and value- over the period 2010–50. Without any adapta- added derived from the 2006 census. tion, Table 3 shows that the loss of consumption is $104 million for the NCAR scenario or $212 (b) Total output in each region is calculated using million for the CSIRO scenario. The annual- a Cobb-Douglas production function in capital ized equivalent is $5.9 million per year for the and labor, with the capital share calibrated to NCAR scenario and $12.1 million per year for match input shares and investment in 2006. the CSIRO scenario. 18 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 S A M OA CO U N T RY ST U DY 19 Adaptation Plans and Costs Samoa has a rapidly developing and increasingly Development of Samoa (SDS) (2008–12), which sophisticated framework of strategies, plans, and seeks to “ensure sustainable economic and social regulations that have developed over time to meet progress.� The SDS provides a framework on international conventions, international monitor- national priorities under key development sectors. ing and reporting requirements, and governance The government has identified seven goals under best practice goals. Many of the older national- this framework. Climate change is mentioned in level strategies have been developed at different Goal 7: Environmental Sustainability and Disas- times as a result of different drivers, but an effort ter Risk Reduction. has been made to integrate planning in different sectors with a broad development strategy. An insti- In the SDS under Goal 7, climate change adapta- tutional reform in 2003, which merged 26 depart- tion is identified as a cross-cutting issue alongside ments into 13 ministries, provided the opportunity environmental sustainability. Adaptation to climate to review and update some of the legislation. change is linked with deforestation and cyclone fre- quency (natural disasters). Mitigation activities are Adaptation to climate change as a concept and also identified and linked to renewable energy pri- priority is reflected through all government and orities. The primary focus is on the link between planning levels, most noticeably in high-level plans climate change and disaster management, which and strategies. Translating these general goals into reflects the critical concern that climate change will regular corporate and departmental plans is more lead to increased frequency and intensity of storms. difficult, though the programs and projects under Coastal communities in Samoa are already at high way are consistent with development priorities and risk from cyclones, and their vulnerability to coastal could be considered as adaptation activities. inundation and erosion has already been assessed. Priority activities to address Goal 7 in the related Strategy for the Development areas of climate change and disaster management of Samoa (SDS) (2008–12) are to “implement the Disaster Management Act 2007 through various programs and projects aimed at both climate mitigation (greenhouse gas reduc- The overarching document in Samoa that guides tions) and disaster readiness.� Increasing resilience all other development is the Strategy for the to the adverse impacts of climate change will be 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 addressed through continued work on coastal Significantly, NAPA led to: management and adaptation programs for vulner- able villages and other coastal locations, as well ■■ The creation of the Samoa National Climate as activities such as promotion of energy-efficient Change Country Team (NCCCT) and NAPA building design. Task Team, both cross-sectoral teams compris- ing representatives from mainly government The environment is to feature prominently as a departments but also NGOs and other interests cross-cutting consideration in all planning activities, including the formulation of sector plans. There ■■ The preparation of a Climate Synthesis will be a focus on improved environmental man- Report, which assessed the current vulnerabil- agement, compliance, and monitoring in 2008– ity and potential increase in climate hazards 2012, with the Ministry of Natural Resources, and associated risks of critical sectors Environment and Meteorology (MNRE) the key implementing agency (including for climate change ■■ Discussion of adaptation needs and priorities adaptation). The Planning and Urban Manage- ment Act 2004 aims to “implement a framework ■■ Consideration of whether and how to sepa- for planning the use, development, management rate climate change adaptation activities from and protection of land in Samoa in the present and other development activities. long-term interests of all Samoans and for related purposes.� Under the act, any development activ- The relationship of the NAPA to the SDS and ity requires a development consent with a support- its potential for influence on the development of ing environmental impact assessment (EIA) unless sector plans is shown in Figure 6. a sustainable management plan or regulations pro- vide otherwise. The capacity of the Planning and From interviews with participants and others, a Urban Management Agency to enforce the regula- number of issues were identified: tions and undertake or facilitate a greater level of community consultation will need to be strength- ■■ The original Climate Risk Profile was based on ened for this to occur. fairly limited climate data available at the time. This has been identified as an issue and improv- ing climate monitoring features as a priority National Adaptation program area. An updated Climate Risk Profile Program of Action (NAPA) has now been prepared (Young 2007). (2005) ■■ The ranking of activities was based on a con- sensus approach rather than a more objective NAPA and its development were discussed in multicriteria analysis or similar method outlined the desk-top study by Beca (2010). NAPA is the in the annotated guidelines for the preparation key document for the identification of the most of NAPA. In part, this was a consequence of urgent and immediate adaptation needs from the the lack of good information on both the eco- adverse impacts of climate change. Samoa was nomics and effectiveness of adaptation options. one of the first countries to receive funding from A consensus-based approach fitted in well with the Global Environment Facility (GEF) to develop the cultural norms of working together. Lim- its NAPA, which took two years of comprehensive ited understanding among the community of information and data collection, as well as exten- climate change and of the effect of different sive countrywide consultation. adaptation activities influenced the ranking of S A M OA CO U N T RY ST U DY 21 Figure 6 LINkS BETWEEN THE NAPA AND THE SDS WITH REPORTING RELATIONSHIPS STRATEGY FOR DEVELOPMENT OF SAMOA (SDS) 2008-2012 NATIONAL ADAPTATION SECTION PLAN FOR EACH PROGRAMME OF ACTION OF 15 SECTORS, EACH (NAPA) 2005 PLAN BASED ON THE SDS. CORPORATE PLAN FOR MINISTRY EACH MINISTRY STRATEGIC ANNUAL MAY HAVE LINKS TO MORE MANAGEMENT PLAN THAN ONE SECTION PLAN SERVICE MINISTRY CHARTERS BUDGET CAPABILITY NATIONAL PLANS BUDGET Source: Beca International Consultants Ltd (2010) alternatives. For example, construction of sea- sooner or later. Projects are designed so that walls features as an urgent community need, they clip onto existing programs largely funded even though other strategies emphasize the role through bilateral agreements and sector pool of more cost-effective measures. funds. Figure 7 illustrates the approval process. Assistance in setting priorities—stages 1 to 3— ■■ Economic assessments were limited to analyses would improve the allocation of funds. of cost-effectiveness, feasibility, and long-term sustainability by activity rather than overall. ■■ The significance of the energy sector has been downplayed, possibly as a result of a lack of ■■ There was an extensive series of countrywide information but also possibly due to responsi- workshops, but some were not particularly bility for it being in another ministry (Minis- well-attended. try of Finance). While energy featured on the original list of sectors, it was dropped from the ■■ The ready availability of donor funds for final ranked list. The comment was made that adaptation projects has been a disincentive to energy priorities were being met through a dif- develop clear priorities among potential proj- ferent strategy and as part of climate change ects and programs. Everything gets funded mitigation programs. 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 ■■ The NAPA has a high level of awareness among and awareness-raising, use and management agencies and there is commitment to, and sup- of resources, and for undertaking intervention port for, its implementation by MNRE. actions. The CIM Strategy set out the need for coastal infrastructure management plans (CIM Plans) and defined goals, objectives, policies, and implementation methods across a broad range of Coastal Infrastructure coastal considerations (Daly et al. 2010). Management Strategy The CIM plans were supported by the compila- and Plans tion of data on the state of coastal resources based upon (a) a survey of Samoa’s 403 km coastline; The Samoa Coastal Infrastructure Management (b) mapping the extent and condition of natural (CIM) Strategy (2001) provided a series of national environments (such as landforms, mangroves, and and local priorities for coastal management. The lagoons); (c) identifying natural resources (such as strategy developed objectives, policies, and imple- aggregate and offshore sand resources); and (d) mentation methods for hazard and environmental mapping coastal hazards (coastal inundation and information-gathering and monitoring, education erosion). The analysis of coastal hazards focused Figure 7 APPROvAL PROCESS FOR ADAPTATION PROjECTS NATIONAL ADAPTATION PROGRAMME OF ACTION (NAPA) 2005 A S S I S TA N C E W I T H P R I O R I T I Z AT I O N R E Q U I R E D PRIORITY PROFILES SELECTED CABINET AID 1 ADAPTATION 2 DEVELOPMENT 3 CO-ORDINATION 1. Securing community ACTIVITIES COMMITTEE COMMITTEE APPROVED FOR EACH (FOR APPROVAL (DETERMINE water resource. & SDS REVIEW) FUNDING) PRIORITY IF 2. Reforestation, PROFILE rehabilitation and community forest �re protection. 3. Climate health program. INTERNAL CRITERIA 4. Climate early warning (MET BY GOVERNMENT system. EXTERNAL RESOURCES- EXISTING CRITERIA PROGRAMS) 5. Agriculture & food security/sustainability. 6. Zoning and strategic management planning. 7. CIM plan PARALLEL POD FINANCING FINANCING implementation. (BILATERAL SECTOR (INDIVIDUAL PODS) 8. Marine and terrestrial DONORS) concession programs. Source: Beca International Consultants Ltd (2010) S A M OA CO U N T RY ST U DY 23 on reducing tsunami risks as well as climate- ■■ Coastline monitoring at a number of points related hazards. was established as a baseline to measure coast- line morphology changes. There was no evi- The project resulted in a total of 41 CIM plans dence that this monitoring was continuing. highlighting vulnerable areas and identifying pri- ority actions to reduce community vulnerability. Updating and implementing the CIM plans Adaptation activities took account of community should be a priority. This would include updating consultations and fa’a Samoa: a traditional model of the plans, prioritizing the actions, and undertak- community decision making by consensus under ing actions of high priority. the leadership of the matai (chief). Community priorities were based on strong anecdotal (history and traditional knowledge) and physical evidence Evidence of Prioritization (such as hazard zone mapping). “Hard�, i.e. physi- and Sequencing: cal, measures—such as coastal protection and road Donor Influence relocation—were subjected to an economic analy- sis to determine their level of viability. NAPA provided a ranked list of sectors and pre- Implementation of the CIM plans—progressively ferred adaptation actions for each. There was no working through the list of prioritized mitigation attempt to prioritize the actual adaptation actions and adaptation options per village—has been themselves. Activities were grouped into pro- identified in the NAPA as a priority. The next stage grams, some of them cross-cutting, and all have will be to develop a prioritization tool to rank proj- been funded (or partially funded) through avail- ects within villages and between districts. able donor funds. The interviews highlighted the following points: Donors are engaging more in partnership with government, whereby donor support is guided ■■ Implementation of the CIM strategy and and designed by the government. There is con- plans is viewed, particularly by MNRE, as a siderable effort on building capacity to allow for key element in adapting to climate change. the proper management of provided resources, although capacity remains an issue. There are ■■ The CIM plans have been used to guide recov- also numerous country strategies (some specific ery in the regions of Samoa affected by the to climate change) and donors have organized 2009 tsunami. themselves into supporting specific sectors based ■■ The CIM plans are being referred to during on their country strategies; for example, the Euro- the consenting process for new developments pean Union (EU) for water sector projects. in order to ensure they are not located in pre- viously identified hazard zones. Currently there is little incentive to prioritize or sequence beyond the level currently afforded by ■■ Awareness of the CIM plans and the value of the NAPA process, as most projects eventually the outcomes of the consultation is declining get funded through available funding mecha- in other ministries and agencies due to staff nisms. However, both the government and the changes, time, and other project priorities. donors spoken to recognize the need for better There is a likelihood of duplication of effort in prioritization to make the best use of the avail- other (smaller) projects if awareness of the CIM able funds. plans and value of the consultation is lost. 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 Adaptation and Design ■■ Does a return period of 10 years offer the right balance between actual costs of construction Standards for Infrastructure and the expected damage caused by storms now and in the future? Or, would it be bet- As discussed in Section 3, the key component of ter for Samoa to adopt a higher value for the adaptation to climate change is the revision of return period—say 20 or 50 years—when set- design standards and associated planning require- ting minimum design standards? ments to ensure that buildings and infrastructure are capable of coping with the peak wind speeds ■■ What will be the additional costs of construct- and precipitation associated with storms that have ing public assets—buildings and infrastruc- a return period of X years under future climate ture—in compliance with the efficient design conditions. The reference to X rather than some standard—that is, a return period of 10, 20 or specific number makes the point that choosing 50 years—under the future climate scenarios? an appropriate value of X involves an economic tradeoff that needs detailed investigation. Build- The time and resources required to answer these ings that will withstand storms with a higher value questions in detail are beyond the scope of this of X—that is, higher wind speeds—cost more, case study. Nonetheless, it is possible to give an but the expected annual value of damage will be initial assessment on the evidence that is available. lower. Rich countries tend to use return periods Since the issue is critical for Samoa’s response to of 50 or 100 years in setting design standards for climate change, the initial analysis should be con- most infrastructure and buildings, but the addi- firmed and/or refined by more detailed investiga- tional cost may not be warranted in Samoa. tion in future. A reasonable interpretation of the damage caused Managing climate risks. A simple test can be by cyclones that have hit Samoa in the past is that used to consider whether the current design stan- the average design standard for the country’s dards based on a 10-year period (peak wind speed capital stock was rather low—a return period of 108 kph) is appropriate. This is based on com- of about 5 years—up to 1990, so that Cyclones paring the reduction in the expected annual losses Ofa and Val caused massive damage. Following from storm damage if a higher design standard those cyclones, the replacement assets are capable were adopted with the annualized value of the of withstanding storms with a return period of additional investment and operating costs required 10 years—such as Cyclone Heta—with mini- to construct and maintain assets to the higher stan- mal damage.6 Estimating the cost of adaptation dard. If new design standards were based on pro- depends on answering two questions: tecting buildings and other assets from storms with return periods up to 50 years (peak wind speed of 148 kph) the model indicates that the expected 6 It is important to be careful in interpreting what this means. annual value of storm losses would fall from about Many buildings in Samoa have been built to design standards that enable them to withstand storms with a historic return 5.5 percent of GDP to about 0.7 percent of GDP, period of 50 or more years. However, this is not true of the country’s entire capital stock. In practice, the design standard for giving an expected annual benefit of 4.8 percent some buildings and infrastructure is much lower, so that a storm of GDP or about $30 million per year in 2008. like Cyclone Ofa with a historic return period of 25 years would cause substantial damage if it were to occur now. What matters is In the long run, the cost of adopting such design the minimum design standard across either all assets or the public standards, calculated using the methods described assets that are covered by this study. Hence, the focus is on those assets that are most vulnerable to storm damage rather than below, will be 2–3 percent of the annualized cost those that have already been built to a standard that will enable of the capital stock, which is less than $5 million them to cope with storms with a historic return period of 50 or 100 years. per year in 2008. S A M OA CO U N T RY ST U DY 25 It is clear that moving to a design standard of 50 (a) upgrading design standards over a period of years will generate benefits that outweigh the costs 40 years, and (b) adapting to climate change over by a large margin. On the other hand, applying the same period. The first of these components the same method to a move from design stan- involves an increase in the peak wind speed for dards based on a 50-year return period to ones which buildings are designed from 108 kph to 148 based on a 100-year return period yields results kph, while the second component extends this to that are much less clear. The additional benefits 170 kph in 2050. would only be $3–4 million per year, while the additional costs would be $2–3 million per year. The costs of adaptation. The infrastructure On this basis, the analysis of the costs of adapta- analysis carried out for the global study—see tion considers two scenarios: chapter 5 in World Bank (2009)—has been modified to take account of the specific circum- Scenario A: Adaptation takes place on the basis stances of Samoa, particularly in the use of (a) of maintaining design standards at their exist- average temperature as a proxy for the shift in ing level—for storms with a return period of 10 the probability distributions of storms under the years—up to 2050. However, there is one fur- two climate scenarios, and (b) total precipitation ther aspect of adaptation to reflect the impact during the rainy season as a proxy for the dam- of climate change. Instead of being designed to age caused by flooding. The key concept is that withstand current storms with a return period building standards are updated in discrete steps of 10 years, which might be appropriate without to take account of expected changes in weather climate change, it is assumed that they are con- stresses over the life of a building or other asset. structed to withstand equivalent storms over the For each update in building standards, addi- life of the assets; for simplicity, a period of 50 tional costs of construction and maintenance are years ahead is used. This forward-looking basis incurred and these make up the costs of adapt- for setting design standards confers significantly ing to climate change. For Samoa it has been greater protection today than at the end of the assumed that building standards are updated for life of the assets, which is appropriate since the each 10 kph increment in peak wind speeds with destruction of an asset is much more costly when a 50-year return period. The extra costs cover it is new than when it is near the end of its eco- both improvements in resistance to wind damage nomic life. Under this scenario, the peak wind and protection against flood damage. For assets speed for which buildings and other assets are in place in 2010, it has been assumed that the designed increases gradually from 115 kph now annual cost of maintenance over the remainder to 125 kph in 2050. of their life is increased by 50 percent of the base cost of maintenance for each 10 kph increment Scenario B: Adaptation is combined with the in the current 50-year wind speed. This assump- gradual implementation of forward-looking tion is designed to reflect additional spending on design standards based on storms with a return measures to strengthen buildings to make them period of 50 years. Part of the gross cost of more resilient to potential storms in the imme- adaptation that is calculated by the model arises diate future, and is based on recommendations because of the progressive shift from a 10- year for Caribbean countries with histories of severe basis for design standards to a 50-year basis hurricane damage. between now and 2050. This component is esti- mated by applying the same method of analysis to This modified approach has been used for all the development baseline with no climate change. buildings including housing. The standard Hence, it is possible to assess the separate costs of assumptions—reflecting changes in maximum 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 temperatures, annual and maximum monthly the costs associated with a shift in the probability precipitation, and so on—have been used for distribution of storms. The range is generated by other assets such as roads, bridges, electricity, combining the range of parameter values for the and water systems. Allowance has also been increase in storm intensity per 1°C increase in made for improvements in urban storm drainage sea surface temperature and the range of values to enable this to limit the flooding and associated for the increase in average temperature. For the damage caused by more intense precipitation NCAR climate scenario, it has been assumed that during storms. the key parameters of the probability distribution increase by 4 percent up to 2050 and by 10 per- As shown in Table 1, the NCAR and CSIRO cent up to 2100. For the CSIRO climate scenario, climate scenarios generate rather similar projec- the equivalent assumptions are increases of 8 per- tions for the increase in average temperature, so cent to 2050 and 25 percent to 2100. Inevitably, that the costs of adaptation under the two sce- these are only approximations, but the ranges for narios will not differ by much. To get a better 2050 and 2100 provide a reasonable indication of sense of the range of possible costs of adapta- the degree of uncertainty when assessing the costs tion, a range of values has been used in analyzing of adaptation to climate change. Table 4 ADAPTATION COSTS BY INFRASTRUCTURE CATEGORY FOR 10-YEAR STANDARD (DECADE AvERAGES WITHOUT DISCOUNTING, $ MILLION PER YEAR AT 2005 PRICES) 2010-19 2020-29 2030-39 2040-49 A. NCAR CLIMATE SCENARIO Education and health 0.01 0.02 0.03 0.08 Electricity and telecoms 0.01 0.02 0.03 0.06 Housing 0.00 0.01 0.00 0.25 Municipal 0.05 0.08 0.14 0.36 Other transport 0.01 0.01 0.01 0.05 Roads 0.03 0.19 0.07 0.03 Water and sewers 0.00 0.00 0.00 0.02 Total 0.11 0.33 0.28 0.84 Total excl. housing 0.11 0.32 0.28 0.59 B. CSIRO CLIMATE SCENARIO Education and health 0.09 0.16 0.25 0.39 Electricity and telecoms 0.05 0.08 0.12 0.16 Housing 0.45 0.93 1.64 2.79 Municipal 0.76 1.25 1.95 2.69 Other transport 0.06 0.11 0.19 0.31 Roads 0.52 0.68 0.82 0.93 Water and sewers 0.02 0.03 0.05 0.08 Total 1.94 3.23 5.02 7.33 Total excl. housing 1.49 2.31 3.38 4.54 Source: World Bank estimates. S A M OA CO U N T RY ST U DY 27 Table 5 ADAPTATION COSTS AS PERCENT OF BASELINE ExPENDITURES CSIRO CLIMATE SCENARIO AND 10-YEAR STANDARD 2010-19 2020-29 2030-39 2040-49 Education and health 1.3 1.9 2.5 3.2 Electricity and telecoms 0.6 0.8 1.0 1.2 Housing 2.6 3.7 4.9 6.3 Municipal 2.4 3.1 3.7 4.6 Other transport 1.6 2.3 3.0 3.7 Roads 3.4 4.3 5.2 6.2 Water and sewers 0.7 0.9 1.3 1.7 Total 2.2 3.0 3.7 4.7 Total excl. housing 2.1 2.8 3.3 4.0 Source: World Bank estimates. The costs of adaptation for each category of infra- period from 2010 to 2050, the increase is 3.2 per- structure reported in the tables cover both (a) the cent for infrastructure excluding housing and 3.6 incremental costs of implementing new design percent including housing. As might be expected, standards to ensure that new (and replacement) the burden of adaptation rises over time as the infrastructure assets can cope with greater weather probability distribution of severe storms shifts. stresses associated with higher wind speeds, pre- cipitation, and temperatures; and (b) the higher Table 6 shows the net costs of adaptation to cli- costs of maintenance for existing infrastructure mate change for the two climate scenarios on the assets as a result of the same weather stresses. assumption that the 50-year design standard has been or is being implemented. The total costs Table 4 shows the costs of adaptation for Samoa are about 12 percent higher than the total costs for the two climate scenarios under the assumption of adaptation for the 10-year standard, partly that the current 10-year basis for design standards because the base investment costs are higher due is retained. The costs of adaptation are much to the stricter design standards and partly because higher for the CSIRO scenario, partly because the higher return period means a larger increase this involves protection against higher peak wind in peak wind speeds due to climate change. None- speeds and partly because the level and pattern theless, the overall cost of adaptation is small of precipitation is an important driver of some relative to the benefits of reducing the expected of the costs of responding to climate change. The values of losses caused by storms. main costs are incurred for housing and munici- pal infrastructure, which covers public buildings and storm water drainage. Coastal Protection Table 5 shows the relative magnitude of these The global analysis for coastal protection is based adaptation costs under the CSIRO scenario when upon use of the dynamic and interactive vulner- expressed as a percentage of the cost of providing ability assessment (DIVA) model to estimate the the relevant services in the baseline scenario. The costs of sea walls, beach nourishment, estuarial increases in costs are about 6 percent for housing flood protection, and port upgrades (World Bank and roads in the period 2040–49. Over the whole 2009). An important feature of the DIVA analysis 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 Table 6 ADAPTATION COSTS BY INFRASTRUCTURE CATEGORY FOR 50-YEAR STANDARD (DECADE AvERAGES WITHOUT DISCOUNTING, $ MILLION PER YEAR AT 2005 PRICES) 2010-19 2020-29 2030-39 2040-49 A. NCAR CLIMATE SCENARIO Education and health 0.01 0.05 0.06 0.08 Electricity and telecoms 0.01 0.05 0.06 0.06 Housing 0.00 0.15 0.19 0.25 Municipal 0.05 0.23 0.31 0.36 Other transport 0.01 0.03 0.04 0.05 Roads 0.03 0.19 0.07 0.03 Water and sewers 0.00 0.01 0.02 0.02 Total 0.11 0.70 0.75 0.84 Total excl. housing 0.11 0.55 0.55 0.59 B. CSIRO CLIMATE SCENARIO Education and health 0.11 0.18 0.30 0.44 Electricity and telecoms 0.08 0.11 0.17 0.20 Housing 0.55 1.07 1.93 3.16 Municipal 0.86 1.39 2.22 2.98 Other transport 0.08 0.13 0.22 0.36 Roads 0.52 0.68 0.82 0.93 Water and sewers 0.03 0.04 0.07 0.10 Total 2.22 3.60 5.72 8.16 Total excl. housing 1.67 2.53 3.79 5.00 Source: World Bank estimates. is that expenditures on coastal protection are sub- to $40,000 per year at 2005 prices under the High ject to a cost-benefit test that examines whether SLR scenario—are required to upgrade the port at the value of the assets protected from the effects Apia, but otherwise no expenditures can be justi- of sea level rise and storm surges is sufficient to fied. The consequence is that some people will be justify the costs involved. In the case of Samoa, affected by flooding in the absence of coastal pro- the cost-benefit test is only satisfied for expendi- tection. The average number affected will increase tures on port upgrades. This finding seems to be to about 170 per year in the decade 2040–49 under consistent with the analysis undertaken for the the High SLR scenario. Other, less expensive, mea- CIM Strategy, which indicated that other mea- sures or compensation for the resulting damage will sures—such as moving infrastructure and other be required to offset this impact of climate change. assets—are more cost-effective than building sea walls in providing protection against the combi- nation of sea level rise and storm surges. NAPA adaptation options The results of the DIVA analysis for Samoa are The study reviewed NAPA and identified the shown in Table 7. Relatively small expenditures—up adaptation activities in order of the priority given S A M OA CO U N T RY ST U DY 29 Table 7 COSTS OF COASTAL PROTECTION AND RESIDUAL DAMAGE FROM DIvA MODEL (DECADE AvERAGES WITHOUT DISCOUNTING, $ MILLION PER YEAR AT 2005 PRICES) Total costs and residual damage 2010-19 2020-29 2030-39 2040-49 LOW SLR SCENARIO Port upgrades 0.017 0.017 0.017 0.017 Sea walls 0.000 0.000 0.000 0.000 Beach nourishment 0.000 0.000 0.000 0.000 Land loss (sq km per year) 0.000 0.000 0.000 0.000 People flooded (000s per year) 0.000 0.010 0.080 0.160 MEDIUM SLR SCENARIO Port upgrades 0.032 0.032 0.032 0.032 Sea walls 0.000 0.000 0.000 0.000 Beach nourishment 0.000 0.000 0.000 0.000 Land loss (sq km per year) 0.000 0.000 0.000 0.000 People flooded (000s per year) 0.010 0.010 0.090 0.170 HIGH SLR SCENARIO Port upgrades 0.042 0.042 0.042 0.042 Sea walls 0.000 0.000 0.000 0.000 Beach nourishment 0.000 0.000 0.000 0.000 Land loss (sq km per year) 0.000 0.000 0.000 0.000 People flooded (000s per year) 0.010 0.010 0.100 0.170 Source: World Bank estimates. both to the relevant sectors and the project goals highlights the difficulty of drawing clear distinc- (Table 8). This exercise provided the basis for tions between development aid and assistance identifying specific projects and programs that with adaptation to climate change. would be consistent with the NAPA and the CIM plans. The details of these projects together with Nonetheless, in order to illustrate how decisions estimated costs are given in Table 9. can be made about which adaptation options should implemented and when, all of the options Many of the activities and projects identified in in Table 9 have been included in the analysis. Tables 8 and 9 are really development policies This involves a series of steps. or projects in the sense that they would almost certainly be justified without any concern about (a) The adaptation costs in Table 9 are con- climate change. This is particularly the case for verted to annualized values. No conversion is improvements in water supply, health programs, required for recurrent costs, but capital costs agriculture, and even tourism. Samoa’s vulner- are converted using annualization rates based ability to natural disasters under current condi- upon a 5 percent discount rate and life spans tions means that many activities would fall into from 10 years—for studies and similar mea- the category of implementing effective policies sures—to 40 years for coastal infrastructure. for disaster planning and management. This The annualization rates include an allowance 30 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 8 PRIORITY ADAPTATION ACTIvITIES BASED ON NAPA Rank Sector Goal Activities 1 Water Securing community A. Develop water purification programs for communities water resources B. Develop watershed management program for other communities C. Alternative water storage systems D. Restoration of coastal springs 2 Forestry Reforestation, A. Sustainable forest management rehabilitation, and B. Forest fire prevention program forest fire prevention 3 Health Climate health A. Establish climate health cooperation program cooperation program 4 All Climate early A. Develop climate early warning system and emergency warning system measures 5 Agriculture Agriculture and A. Investment in annual crops and home vegetable farming food security B. Alternative farming systems 6 Urban Zoning & strategic A. Strengthen zoning, disaster planning, and urban planning settlements planning to increase resilience to cyclone damage 7 Coastal Implement Coastal Infra- A. Implement coastal zone management infrastructure structure Management B. Coastal infrastructure protection, incl. sea walls Plans for highly vulnerable C. Relocation of roads districts D. Relocation of communities and associated infrastructure 8 Biodiversity & Conservation programs for A. Establish conservation areas and marine reserves environments highly vulnerable marine and terrestrial areas 9 Tourism Sustainable tourism A. Develop sustainable tourism policy and ventures Source: Beca International Consultants Ltd based on Government of Samoa (2005). for maintenance costs at 1 percent of capital documentation and analysis of the effects of expenditure for fixed assets. the various adaptation options that is not avail- able in this case study. In addition, it is neces- (b) In the case of studies and policy measures, it is sary to make explicit assumptions about how assumed that other measures with an equiva- far the measures should be treated as devel- lent cost will be implemented to sustain the opment or adaptation policies. For example, benefits of the initial policy changes. Further, improving water supplies and building new it is assumed that physical assets—coastal schools would generate significant benefits infrastructure and improvements to water sup- even if there were no climate change. Strictly, ply systems—are replaced at the end of their these non-climate benefits should be deducted working life. Thus, implementation of any from the gross project cost to obtain an esti- adaptation measure represents a permanent mate of the net cost of adaptation, which commitment to annual expenditure equal to has to be compared with the climate benefits the annualized value of the project cost. that are generated. None of this is easy and it is certainly not practical with the informa- (c) Detailed cost-benefit analysis of each mea- tion that is available for Samoa. Instead, it sure requires an estimate of the proportion has been assumed that the total annualized of damage associated with climate change cost of the adaptation options for each region that would be mitigated by implementing (dominated by the costs of relocating coastal the measure. This implies a level of detailed infrastructure) can be used as a proxy for the S A M OA CO U N T RY ST U DY 31 Table 9 POTENTIAL ADAPTATION OPTIONS FOR NAPA AND CIM PLANS Cost ($ 000 at Region Key sector Adaptation measure 2005 prices) Type Water Improve water treatment at source for existing 120 Capital boreholes Health Education, monitoring, and control of pest plants and 40 Annual animal species that would otherwise adversely impact upon health and biodiversity North Savai’i Coastal Build new school away from coastal flood hazard zone 1,000 Capital infrastructure Tourism Develop and support inland ecotourism venture 50 Capital Agriculture Identify and establish village/home-based vegetable 120 Capital farming areas Agriculture Strategy and farming adviser 40 Annual Water Improve water storage capacity of villages for use in 150 Capital drought periods Health Education, monitoring, and control of pest plants & 40 Annual animal species that would otherwise adversely impact upon health and biodiversity Coastal Place remaining overhead electrical lines underground 1,000 Capital infrastructure South Savai’i Tourism Development and implementation of a Sustainable 100 Capital Tourism Charter for Savai’i Tourism Adviser on sustainable tourism for businesses and 40 Annual villages Agriculture Increase efficiency of existing plantation areas 80 Capital Agriculture Inspection management and advisory program 40 Annual Water Repair leaks in reticulated water supply network 100 Annual Health Collection of better health, meteorological, environ- 65 Capital mental, and socioeconomic data for health planning, incl. development of a health vulnerability indicator Coastal Provide addition access road inland to reduce reliance 23,000 Capital infrastructure upon coast route North Upolu Tourism Development and implementation of a Sustainable 100 Capital Tourism Charter for Upolu Tourism Adviser on sustainable tourism for businesses and 55 Annual villages Urban Develop structure plan for Apia to encourage appropri- 400 Capital development ate urban design and land use Water Develop an integrated watershed management pro- 115 Capital gram with villages in the catchment Health Develop early intervention health services to deal with 150 Capital water and vector-borne diseases Upgrade clinic and fund district nurse 40 Annual Coastal Relocate village inland out of coastal hazard zone, 32,000 Capital South Upolu infrastructure incl. expansion of power grid, sealing inland plantation roads, water supply reticulation, and telecommunications Tourism Develop and support inland ecotourism venture 50 Capital Agriculture Identify and implement sustainable management of 80 Capital fish and shellfish resources Agriculture Strategy and adviser position 40 Annual Source: Beca International Consultants Ltd (2010). 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 Table 10 DATES FOR IMPLEMENTATION OF NAPA ADAPTATION MEASURES BY REGION Upolu North Upolu South Savai’i North Savai’i North A. DESIGN STANDARDS FOR 10-YEAR RETURN PERIOD NCAR 2045–49 > 2050 2035–39 2035–39 CSIRO 2035–39 > 2050 2025–29 2025–29 B. DESIGN STANDARDS FOR 50-YEAR RETURN PERIOD NCAR > 2050 > 2050 > 2050 > 2050 CSIRO > 2050 > 2050 > 2050 > 2050 Source: World Bank estimates. costs of fully adapting to climate change in A second conclusion is that the projects proposed the region. for Upolu South—in particular the relocation of coastal infrastructure—are hard to justify over (d) The assumptions outlined in (c) mean that the next 40 years under any of the scenarios. the cost-benefit decision rule for adaptation But, again there is another consideration. Upolu involves a set of simple comparisons of the South was the region most severely affected by annualized costs of the adaptation options for the 2009 tsunami and is most vulnerable to the each region against the damage due to climate impact of future earthquakes and the resulting change in the region. Provided that the direc- tsunamis. Relocating coastal infrastructure may tion of change of estimated damage does not be justified as a measure to reduce the risks associ- reverse, once implementation of adaptation ated with non-climate hazards. The lesson is that options becomes economic there will be no it is important to focus on climate change within reason to alter the decision in future. the context of other natural hazards and broader development goals. Table 10 shows the period during which the NAPA adaptation measures should be imple- mented in order to obtain the most efficient bal- Agriculture ance between the costs of climate change and the costs of adaptation under the two climate sce- Based upon the potential impact of climate narios on alternative assumptions about the set change on agriculture discussed earlier, two types of design standards that are adopted. The results of public expenditure would contribute to the indicate that the NAPA projects are not urgent capacity of the agricultural sector in Samoa to when considered purely in terms of adaptation adapt to climate change. The estimates given here to climate change. Some of the projects have sig- are very broad and are intended only as an indi- nificant development benefits that would justify cation of orders of magnitude. the required expenditures without any climate change, but the lesson is to focus on the design Agricultural research and development. An and implementation of the projects that are bene- increase in spending on agricultural research and ficial under any climate scenario. This conclusion development would provide farmers with a wider is reinforced by the significant differences in opti- range of crop varieties and cultivation options mal timing under the alternative combinations of to respond to the risks associated with climate design standards and climate scenarios. change, including a greater degree of weather variability within and between seasons and S A M OA CO U N T RY ST U DY 33 changes in patterns of pests and/or diseases. The a damage function similar to that used for eco- EACC global study of agriculture carried out by nomic losses with a power of 1.5 for the increase the International Food Policy Research Institute in damage for increases in wind speed above the (IFPRI) estimated that spending on agricultural threshold; (c) an average capital-output ratio research and development would need to increase (excluding land) of 3 for agriculture; and (d) losses by 25–30 percent relative to the baseline of no of 20 percent of agricultural assets for a storm climate change to hold levels of malnutrition con- with a peak wind speed of 165 kph (based on the stant under the NCAR and CSIRO scenarios. losses caused by Cyclone Val). On this basis, the expected cost of asset insurance associated with As a general indication of the cost of adapta- storm damage will increase from 0.95 percent of tion, it has been assumed that public spending agricultural value-added without climate change on agricultural research, development, and advi- to 1.15 percent for the NCAR scenario in 2050 sory services, which amounted to $1.6 million in and 1.37 percent for the CSIRO scenario. The 2007, should be increased by 30 percent to fund increase in insurance costs is relatively small, additional activities to mitigate the effects of cli- reaching an average of $0.28 million per year for mate change. This is applied to a baseline level 2040–49 for the CSIRO scenario. of expenditure that is assumed to increase at 2.5 percent per year in real terms. This estimate of the cost of adaptation rises from an average of Adaptation Costs and $0.58 million per year in the decade 2010–19 to Benefits $1.22 million per year for 2040–49. Agricultural asset insurance. Storms damage Table 11 is an extended version of Table 3 with or destroy agricultural assets including livestock, the addition of the costs of climate change if trees, machinery, and some irrigation facilities. adaptation measures are implemented as dis- These are not protected by the adoption of design cussed in this section under the two alternative standards designed to enable buildings and other assumptions about design standards. In this case infrastructure to withstand more severe storms. the cost of climate change with adaptation for Hence, the shift in the probability distribution of each region is the minimum of (a) the climate storms will lead to higher expected losses of agri- damage estimated using the appropriate damage cultural assets when severe storms occur. While function, and (b) the annualized cost of adapta- farmers may be expected to bear the costs of such tion measures. losses up to some threshold, it is likely that the gov- ernment will offer implicit or explicit insurance for The most important observation is that the adop- the losses associated with the worst storms as a way tion of 50-year design standards reduces the of supporting agriculture. Thus, the cost of adap- impact of climate change on consumption by tation will include the increase in the expected 80-90 percent under both of the climate scenarios. value of losses due to the shift in the probability This is a classic example of a “no regrets� strat- distribution of storms due to climate change. egy: It can be justified in economic terms even without climate change, but its net benefits are The scale of this item is estimated by calibrating even larger if climate change occurs. Additional a loss function on the following assumptions: (a) adaptation measures further reduce the impact of the threshold for farmers up to which farmers climate change so that the net cost is equivalent bear losses is 90 kph (a storm return period of to $0.3 million per year on an annualized basis five years under historic climate conditions); (b) after allowing for the cost of implementing the 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 Table 11 THE IMPACT OF CLIMATE CHANGE WITH AND WITHOUT ADAPTATION No adaptation With adaptation Benefit of adaptation NCAR CSIRO NCAR CSIRO NCAR CSIRO (1) (2) (3) (4) (5) (6) A. DESIGN STANDARDS FOR 10-YEAR RETURN PERIOD Present value @ 5 percent, $ million 103.9 212.4 34.8 24.5 69.1 187.9 Annualized equivalent, $ million per year 5.9 12.1 2.0 1.4 3.9 10.7 Loss/benefit as percent of baseline GDP 0.6 1.3 0.2 0.2 0.4 1.2 Loss/benefit as percent of baseline 0.9 1.9 0.3 0.2 0.6 1.7 consumption B. DESIGN STANDARDS FOR 50-YEAR RETURN PERIOD Present value @ 5 percent, $ million 19.9 37.0 4.5 5.4 15.4 31.6 Annualized equivalent, $ million per year 1.1 2.1 0.3 0.3 0.9 1.8 Loss/benefit as percent of baseline GDP 0.1 0.2 0.0 0.0 0.1 0.2 Loss/benefit as percent of baseline 0.2 0.3 0.0 0.0 0.1 0.3 consumption Source: World Bank estimates. Table 12 TOTAL COST OF ADAPTATION FOR 10-YEAR DESIGN STANDARDS BY SCENARIO AND DECADE (DECADE AvERAGES WITHOUT DISCOUNTING, $ MILLION PER YEAR AT 2005 PRICES) 2010–19 2020–29 2030–39 2040–s49 NCAR SCENARIO Coastal protection 0.032 0.032 0.032 0.032 Infrastructure excl housing 0.108 0.320 0.277 0.589 Housing 0.000 0.007 0.001 0.252 Health 0.044 0.000 0.000 0.000 Agriculture 0.606 0.826 1.110 1.475 Fisheries 0.379 0.640 0.901 1.163 NAPA projects 0.000 0.000 0.150 1.179 Total 1.169 1.825 2.472 4.689 CSIRO SCENARIO Coastal protection 0.032 0.032 0.032 0.032 Infrastructure excl housing 1.491 2.307 3.381 4.544 Housing 0.449 0.927 1.635 2.789 Health 0.023 0.000 0.000 0.000 Agriculture 0.608 0.834 1.126 1.500 Fisheries 0.379 0.640 0.901 1.163 NAPA projects 0.000 0.169 1.370 2.329 Total 2.982 4.909 8.445 12.357 Source: World Bank estimates. S A M OA CO U N T RY ST U DY 35 Table 13 TOTAL COST OF ADAPTATION FOR 50-YEAR DESIGN STANDARDS BY SCENARIO AND DECADE (DECADE AvERAGES WITHOUT DISCOUNTING, US$ MILLION PER YEAR AT 2005 PRICES) 2010–19 2020–29 2030–39 2040–49 NCAR scenario Coastal protection 0.032 0.032 0.032 0.032 Infrastructure excl housing 0.108 0.552 0.552 0.589 Housing 0.000 0.152 0.193 0.252 Health 0.044 0.000 0.000 0.000 Agriculture 0.606 0.826 1.110 1.475 Fisheries 0.379 0.640 0.901 1.163 NAPA projects 0.000 0.000 0.000 0.000 Total 1.169 2.202 2.788 3.510 CSIRO scenario Coastal protection 0.032 0.032 0.032 0.032 Infrastructure excl housing 1.667 2.533 3.794 5.002 Housing 0.548 1.069 1.929 3.158 Health 0.023 0.000 0.000 0.000 Agriculture 0.608 0.834 1.126 1.500 Fisheries 0.379 0.640 0.901 1.163 NAPA projects 0.000 0.000 0.000 0.000 Total 3.257 5.108 7.782 10.855 Source: World Bank estimates. measures. In present value terms the combination included in the climate-economy model. In the of adopting 50-year design standards and associ- case of health, the expenditures are required to ated adaptation yields a net benefit of $88 million offset the projected impact of changes in temper- under the NCAR scenario with a relatively small ature and precipitation on the incidence of diar- increase in cyclone intensity and one of $181 mil- rhoeal diseases, after controlling for income and lion under the CSIRO scenario with a relatively economic development. In the case of fisheries, high increase in cyclone intensity. the expenditures are calculated as the amount of compensation required to offset the reduction in Finally, Tables 12 and 13 show the total cost of the value-added generated by fishing in Samoa’s adaptation for Samoa by decade and climate sce- exclusive economic zone because of climate nario for the 10- and 50-year design standards. change (World Bank 2009). In addition to the items that have already been discussed in this section, these costs include The decline in fish catches is the largest element expenditures on health and fisheries based on of the cost of adaptation for the NCAR scenario. the calculations prepared for the EACC global The total cost of adaptation is quite small in this analysis. These are not linked to changes in the scenario. In contrast, the total cost of adaptation probability of cyclones, so that they have not been in the CSIRO scenario is much higher than for 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 the NCAR scenario and increases by 3.5–4.5 adopted for the global analysis. As a general times from 2010–19 to 2040–49. This growth is principle, the EACC study focuses on adaptation driven partly by the increase in the cost of pro- within the public sector and for public assets. The tecting buildings from storm damage and partly boundary between public and private assets var- by the implementation of the NAPA projects— ies from country to country, but in most countries in particular the relocation of coastal infrastruc- housing is predominantly owned and managed ture—in the second half of the period under by individuals rather than public organizations, review. Underpinning these increases is the pro- and housing has been treated as lying outside the jected risk in the peak wind speed associated with scope of public assets in estimating the global cost 1-in-50-year storms after 2050 in this climate of adaptation. scenario because this determines the design stan- dards required to ensure that buildings and other Notwithstanding this general principle, the real- structures can cope with predicted wind stresses ity is that damage to a significant proportion during their economic lives. of a country’s housing stock caused by a major storm—or, similarly, major earthquakes—invari- Tables 12 and 13 include estimates of the addi- ably leads to public intervention and support tional costs of upgrading design standards for to repair or replace damaged housing. This is housing. This is a departure from the coverage true in countries as diverse as the United States, S A M OA CO U N T RY ST U DY 37 Honduras, or Haiti after major natural disasters. of more stringent design standards. Providing tax The manner in which governments bear much of incentives or subsidies to offset the incremental the cost of dealing with the damage caused by cost of ensuring that houses are more resilient natural disasters varies from emergency grants to to future storms would be a worthwhile invest- subsidized insurance arrangements, but the prac- ment for the Samoan government in terms of the tice is nearly universal. If the public sector is likely long-run consequences for public revenues and to pay for much of the cost of repairing storm expenditures. Because of the particular nature of damage, then it would be sensible for the public climate risks in Samoa, housing adaptation has sector to bear part or most of the cost of ensuring been included in the overall cost of adaptation in that such damage is minimized by the adoption this case study. 38 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 S A M OA CO U N T RY ST U DY 39 Institutional and Social Analysis Incorporation of Existing projected climate changes to determine what Data on Potential Risks areas will be more suitable than others in the future for different crop types. and Hazards From the interviews and information reviewed, a Typically, explicit data and information on poten- number of observations can be made: tial risks and hazards are limited to a narrative of existing data, an analysis of trends (where sufficient ■■ Samoa has developed its own Climate Synthesis data exists), and a qualitative assessment of vulner- Report, which summarizes trends and analyzes ability and risk. Examples of this would be NAPA data from Samoa’s climate monitoring sta- and the National Disaster and Emergency Man- tions. There are other stations being established agement Plan (2005). The use of data to guide (through Electric Power Corporation) that in policy changes is not yet well-developed, partly time will provide additional data. EPC and because of the lack of baseline monitoring data MNRE (Meteorological Division) are working over a sufficient period, but also due to a lack of together, and information from EPC’s network awareness about how to interpret information in a will be available to MNRE. Accessibility of this way that is meaningful for policy and planning. data to external (non-profit) agencies is cur- rently limited and consequently difficult. A number of projects are attempting to better integrate information into policy: ■■ There is still not enough information to allow for local trend analysis or tools to support deci- ■■ The CIM plans integrated environmental sion making. information, hazard, and risk information into the plans and based mitigation options directly ■■ Environmental data continues to be expanded on the vulnerability and risk profile of the vil- and collected—examples include a forest lages concerned. inventory, water availability, and coastal haz- ard maps. Socioeconomic information—such Monitoring the effect of changing climate on as land valuation and the Human Develop- crop suitability. Approximately 60 thematic maps ment Index—is more limited. NAPA identifies are being produced based on different crops and climate monitoring (including climate early 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 warning systems) as a priority. One senior etc. needed for development consents very MNRE official commented that more training difficult. A lot of information has been lost, was needed for staff in cost-benefit analysis. archiving is difficult, and security is an issue. Much of this information is needed to sup- ■■ There is scope for more work integrating climate port disputes at the land titles court. science with climate change adaptation policy. Consistency Across and There is considerable scope for more data and ■■ information collection in the private sector. For Within Sector Plans example, the Samoa Tourism Authority could expand the information it uses (currently lim- A comprehensive assessment of all available plans ited to arrivals/ departure information com- could not be completed in the time available. How- piled by the Bureau of Statistics) by building its ever, the impression from the interviews and from relationship with the Samoa Hotel Association the few plans accessed is that there is relatively and improving relations with local hotel oper- good vertical and horizontal integration within ators. Information needs include occupancy and across government departments. A number rates and accommodation stock (number of of sector plans have been developed as part of the beds and rooms). With the right incentives SDS, the development of which has involved vari- and commercial confidentiality protected, this ous agencies (both inside and outside government) information could be collected by the hotels with interests and responsibilities in that particular themselves for little effort and aggregated by sector. Sector plans inform corporate plans, which the STA or the Hotels Association. in turn inform capability plans—all of which are linked to budget development. Strategies such as ■■ Information management is an issue in Samoa the NAPA provide a coordinating and integrating and needs considerable investment. Informa- role across the sector plans. tion is not always stored in robust databases, integrated, or shared across ministries, and is While input into the development of a sector plan not always easy to retrieve. Data ownership doesn’t necessarily translate to horizontal inte- and access costs (user pays) make accessing this gration, the establishment of a number of cross- information difficult. A number of examples agency committees and task groups—such as of discrepancies and lost data emerged during the Disaster Advisory Committee (DAC) and the the visit, for example: Samoa National Climate Change Country Team (NCCCT)—provide forums for ongoing discussion ■■ There were discrepancies in immigration and promote integration. Some of these commit- data (10,000 cards not accounted for affect- tees, such as DAC, are a statutory requirement. ing arrivals statistics). ■■ The car registration database (providing number and type of vehicles on the road) was Local Perspectives corrupted and all data was lost. Currently this is on a manual system and needs inte- Samoa’s cultural context is an important factor grating, for example, with police databases. when selecting adaptation measures. The traditional ■■ Site-specific (land-based) information is model of community decision making is by consen- stored on a yearly basis (rather than site sus under the leadership of the matai (chief). The basis), making correlating historical and cur- authority of a village matai and customary land own- rent land use activity, ownership, hazards, ership rights are respected, so negotiations between S A M OA CO U N T RY ST U DY 41 the government and village matai can often take a varieties and more resistant crops, or rezoning long time. There is a commitment to supporting agriculture and training farmers on sustain- village-based consultations, which include women able land management and young adults. Raising awareness of climate ■■ Limiting casualties from cyclones by imple- change and other development concerns through menting early warning systems village-based consultation is an effective and sus- tainable way of supporting the traditional decision- ■■ Promoting alternative sources of energy making model. Nevertheless, women and migrants ■■ Climate-proofing ecotourism enterprises in the poorer communities remain among the most vulnerable groups. Stakeholders at workshops held ■■ Introducing improved fishing methods to during the preparation of the NAPA identified the respond to the impact of climate change on following areas as critical to a strategy for adapting marine ecosystems. to climate change: the protection of community water supplies, early warning systems, support for Much of this represents a continuation of the pri- agriculture and forestry sectors, implementation of orities and proposals put forward in the CIM plans coastal infrastructure management plans, and inte- and NAPA. Project costs are typically small. As dis- grated catchment management. cussed in Section 4, there is a large overlap between activities that would be undertaken to promote A consultation program across the entire coun- social and/or economic development and those try funded by UNDP and led by the Ministry of that contribute to adaptation to climate change. Women, Community and Social Development (MWCSD) has started. It is envisaged that this will lead to the implementation of small-scale proj- Implementation—How is it ects designed to reduce vulnerability to climate Working in Practice? change. This Community Centered Sustainable Development Program (CCSDP) will comple- ment the earlier CIM plan consultation by focus- The overall impression in Samoa is that climate ing on village activity such as tourism markets, change adaptation and development issues are food security, women in business, and disaster risk being addressed and implemented at the “business- reduction initiatives. There is an opportunity for as-usual� level, although the distinction between the outcomes of this consultation to feed into any climate change adaptation, mitigation, sustainable planned update of the CIM plans. Activities that development, and common sense is merged. may be supported under the CCSDP include: At the “business-as-usual� level, there is a mixed awareness of climate change adaptation as a ■■ Planting trees and mangroves to reduce coastal driver. The awareness was higher in government erosion agencies than in the private sector (although not ■■ Establishing conservation areas (marine or ter- many truly private sector agencies were inter- restrial) and/or coral gardens to increase resil- viewed as part of the project). ience to coral bleaching It is recommended that the private sector be ■■ Increasing the resilience of water resources encouraged to participate in and buy into cli- through the restoration of natural springs and mate change adaptation activities more through more efficient water uses the multisector groups and committees that have ■■ Securing food production by introducing new already been established to foster integration. 42 S Ix 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 S A M OA CO U N T RY ST U DY 43 Lessons and Future Work Samoa is a small island nation with most of its developing plans for managing coastal infra- population and infrastructure located along the structure are continuing with a focus on other coast, and like other SIDS, it is highly vulnerable development and adaptation issues. to extreme weather events. However, Samoa is also among the most climate-resilient Pacific island ■■ The key adaptation measure identified in countries, and there is much to learn from the way the study is the adoption of forward-looking it is approaching climate change and related devel- design standards that will enable buildings and opment issues. Over the last decade it has focused other assets to cope with storms with higher on increasing the capacity of its institutions, which peak wind speeds and associated precipitation are necessary for the implementation of soft under alternative climate scenarios. At present, approaches to adaptation, including land-use con- peak wind speeds above 110 kph may cause trols and coastal infrastructure management. significant damage. If the standard were raised to 135 kph by 2050—equivalent to a 1-in-10 The key lessons that may be drawn from this year storm in 2100 for the CSIRO scenario— country study are: the expected losses from climate change would be greatly reduced. ■■ Extreme weather variability in the coastal zone will involve significant costs for either invest- ■■ The analysis also suggests that the country ments in coastal protection or the relocation should consider, as an urgent matter of good of assets. In the longer term, the relocation of development policy even in the absence of cli- assets—or even whole villages—may be the mate change, the adoption of design standards best option, as it can shift economic activity that would enable buildings and infrastructure such as tourism, crops, and other businesses assets to cope with 1-in-50 year storms with- away from the coast. out significant damage. The benefits of this change would greatly outweigh the implemen- ■■ Uncertainty about climate outcomes and a lack tation costs. It would require the immediate of baseline data have led to a focus on the col- adoption of standards to ensure that buildings lection of information in Samoa. More effort and other assets can withstand storms with a is needed to support the collection and analysis peak wind speed up to 160 kph without dam- of this information and use of the information age and a gradual increase in this threshold to to inform decision making. 185 kph in 2050. This is a clear example of a “no regrets� strategy for adaptation because it ■■ Good development policies are a foundation for is justified without climate change, but it will climate change adaptation. The participatory also substantially reduce the future costs of consultations undertaken across the country in adapting to climate change. 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 References Beca International Consultants Ltd. 2010. “Samoa Climate Knutson, T. R., J. L. McBride, J. Chan, K. Emanuel, C. Hol- Change Adaptation Study—Desk Review.� Report for the land, C. Landsea, I. Held, J. P. Kossin, A.K. Srivastava, and World Bank, March 18, 2010. K. Sugi. 2010. “Tropical cyclones and climate change.� Nature Geoscience 3: 157–163, DOI: 10.1038/NGEO779. Carter, R. 1990. “Probability and recurrence of tropical cyclones in Western Samoa.� SOPAC Technical Report 106, Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. Suva, Fiji: Pacific Island Applied Geoscience Commission F. B. Mitchell, R. J. Stouffer, and K. E. Taylor. 2007. “The (SOPAC). WCRP CMIP3 multi-model dataset: A new era in climate change research.� Bulletin of the American Meteorological Society Daly, M., N. Poutasi, F. Nelson, and J. Kohlhase. 2010. “Reduc- 88: 1383–1394. ing the Climate Vulnerability of Coastal Communities in Samoa.� Journal of International Development 22: 265–281. Rearic, D.M. 1990. “Survey of Cyclone Ofa damage to the northern coast of Upolu, Western Samoa.� SOPAC Techni- Emanuel, K. 2005. “Increasing destructiveness of tropical cal Report 104. South Pacific Applied Geoscience Commis- cyclones over the past 30 years.� Nature 436 (August 4, 2005): sion, April 1990. 686–688. South Pacific Sea Level and Climate Monitoring Project. 2008. Evans, M., N. Hastings, and B. Peacock. 2000. Statistical Distribu- Pacific Country Report on Sea Level and Climate: Samoa. Adelaide: tions. 3rd Edition. New York: Wiley-Interscience. Bureau of Meteorology. Fairbairn, T.I.J. 1997. “The Economic Impact of Natural Disas- Trenberth, K.E., and T. J. Hoar. 1997. “El Niño and climate ters in the South Pacific.� United Nations. New York: South change.� Geophysical Research Letters 24 (23): 3057–3060. Pacific Disaster Reduction Program. World Bank. 2000. Cities, Sea and Storms: Managing Change in Pacific Government of Samoa. 2001. Coastal Infrastructure Management Island Economies. Volume IV: Adapting to Climate Change. Project: CIM Strategy. Apia: Beca International Consultants Washington, DC: The World Bank. Ltd. World Bank. 2006. Not if but when: Adapting to natural hazards in the Government of Samoa. 2005. National Adaptation Programme of Pacific Islands region. East Asia and Pacific Region Policy Note. Action. Apia: Ministry of Natural Resources, Environment Washington, DC: The World Bank. and Meteorology. World Bank. 2009. “The costs to developing countries of Hijmans, R.J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. adapting to climate change: New methods and estimates.� Jarvis. 2005. “Very high resolution interpolated climate sur- Economics of Adaptation to Climate Change Study, Consul- faces for global land areas.� International Journal of Climatology, tation Draft. Washington, DC: The World Bank. Vol. 25, pp. 1965–1978. Young, W.J. 2007. “Climate risk profile for Samoa.� Apia: Samoa IWTC (International Workshop on Tropical Cyclones). 2006. Meteorology Division. “Statement on Tropical Cyclones and Climate Change.� World Meteorological Organization International Workshop on Tropical Cyclones, San Jose, Costa Rica, November 2006. S A M OA CO U N T RY ST U DY 45 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 S A M OA CO U N T RY ST U DY 47 ANN Ex ONE Storm Return Periods and Economic Damages The Samoa country study has focused on the available. The two parameters are location α (the impact of and adaptation to changes in the distribu- mode of the distribution) and scale β > 0.7 The tion of tropical storms that strike the country as a probability of a storm with a peak wind speed of consequence of climate change. Since the modeling S* ≥ S is: relies upon concepts such as the “return period� of -(S - α ) prob(S* ≥ S) = 1- exp {- exp( β }. ) storms with different impacts, this annex is intended as a very brief introduction to the basic analysis of extreme events. It focuses on the probability distri- The return period of a storm with peak wind bution for cyclones characterized by their peak wind speed of S is the reciprocal of prob (S* ≥ S). speeds sustained over land for a period of at least Thus, the peak wind speed for a storm with a 10 minutes: referred to as peak wind speeds. Peak return period of N years is: 8 wind speeds are closely correlated with structural S(N) = α – β in[in(N) – in(N-1)]. and economic damage caused by high winds, storm surges, intense rainfall, and flooding. The economic damage caused by a storm with In any year (t), Samoa is struck by a series of more peak wind speed S is assumed to be a power func- or less severe storms (n=1, … Nt) with peak wind tion of the positive difference between S and the speed for each storm denoted by Stn. The maxi- wind speed that buildings are designed to resist mum peak wind speed in year t is St* = max (Stn) without damage, denoted by SD, i.e. for given t. The analysis focuses on the character- D = γ [max(S – SD, 0)] λ Y istics of the distribution of St* over many years. It is standard to use some variant of the general- 7 In the three-parameter variant of the GEV distribution, the ized extreme value (GEV) to describe the distri- inner exponential is replaced by a power function using a shape bution of extreme values of natural events such parameter γ. The two-parameter variant is a special case of the three-parameter specification with γ=0. as floods, storm surges, wind speeds, earthquakes, 8 Carter (1990) examines the return period for cyclones in the etc. (Evans, Hastings, and Peacock 2000). In this South Pacific, the Samoa 5-degree square, and Apia. Unfor- tunately, the expressions for wind speed, wave height, etc. as a case, the two parameter version of the GEV dis- function of the return period— equations (4) to (8)—are clearly tribution, also known as the Gumbel distribution, wrong, though the graphs appear to be correct. For example, using Figure 4, the text states that the wind speed with a 40-year will be used because of the limited data that is return interval is 77 knots, whereas equation (4) yields a value of 437.7 knots (p. 15). 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 where Y is total GDP and the parameters γ and λ However, if the design standard is adjusted to are chosen to reflect the actual economic damage maintain the 1–in–10 year storm assumption, caused by storms that have affected Samoa in the then SD would be increased to 125 kph and the past. Until Cyclones Ofa and Val, it seems that expected value of storm damage would be 5.4 design standards in Samoa meant that assets were percent of GDP. The specification of the damage typically able to cope with storms with a 5-year function means that the proportional adjustment return period—SD ≈ 90 kph—but improvements in SD required to hold expected damage constant is since the early 1990s mean that Cyclone Heta, a somewhat greater than the proportional change in storm with a 10-year return period, caused mini- the parameters of the probability distribution. In mal economic loss, implying that SD ≈ 108 kph. this case the design standard SD would have to be On this basis, the damage parameters are esti- 128 kph to restore the expected damage to about mated as λ = 1.5 and γ = 0.0041. 4.8 percent of GDP. The expected value of the economic damage Three points should be noted: caused by storms in any year is SM A These calculations only apply to damage that ∫ E(D) = p(S)D(S)dS can be prevented by implementation of appro- SD priate design standards. The potential damage where p(S) is the probability density function for to agricultural assets such as coconut planta- peak wind speed S, D(S) is the damage function tions will increase if peak wind speeds with a for S, and SM is the maximum wind speed used 10- or 50-year return period increase. This is for the calculation. A discrete approximation addressed separately. is used in place of the continuous integral with steps of 1 kph. The value of SM corresponds to B Existing assets have been built to older design the peak wind speed with a return period of 200 standards and will suffer more damage than new years since there is insufficient data to calibrate assets. In some cases their remaining economic either the probability distribution or the damage life may be relatively short, so that the increased function beyond this level. risk of storm damage may be relatively small. The calculations assume that a tradeoff is made between the options of accelerated deprecia- The impact of a shift in the probability distribu- tion (early replacement) of long-lived assets that tion of storms on the expected value of storm do not meet the new design standards or incur- damage can be largely offset by changing the ring higher costs of maintenance and repairs as design standards that are applied when building a consequence of more serious storm damage. new assets. As an illustration, Figure 1 in the main text shows the effect of increasing both the loca- C In some cases, the best strategy is to ensure that tion parameter α and the scale parameter β by buildings and infrastructure assets are located 15 percent. This reduces the return period for a out of harm’s way. This is particularly true storm with a peak wind speed of 165 kph from for vulnerability to storm surges and flooding 100 years to 40 years. If there were no change caused by intense rainfall. Thus, when thinking in design standards, then the expected value of about design standards it is important not just to annual storm damage would increase from 4.8 focus on resistance to wind damage but also to percent of GDP to 11.6 percent of GDP using ensure that planning and development policies the base design standard of a 1–in–10 year storm, take proper account of the impact of changes in which corresponds to SD ≈ 108 kph. the severity and frequency of storms in future. S A M OA CO U N T RY ST U DY 49 AN N Ex T WO El Niño Events and Agricultural Production There is evidence that ENSO events have been values ($ 000s), and import values ($ 000s). The associated with periods of drought, but the nature analysis of crop production focused primarily and magnitude of their impact on agricultural on Samoa’s three main crops—taro, bananas, production in Samoa has not been investigated. and coconuts (both nuts and oil)—with more This could be important if climate change were limited examination of yams, mangoes, papa- to alter the frequency or character of ENSO yas, pineapples, and avocadoes. The analysis events. To rectify this gap, a time series analysis of trade focused on the value of exports and of the links between indices of ENSO cycles, pre- imports of (a) all agricultural products, and (b) cipitation, and agricultural production has been food and animals. carried out. (b) Monthly precipitation (mm) by 0.5 degree The analysis is based upon the following data: grid cells covering eight countries and depen- dencies in the South Pacific—American (a) Agricultural production and trade statistics Samoa (ASM), Fiji (FJI), Niue (NIU), Tokelau for 1961–2008 for Samoa extracted from (TKL), Tonga (TON), Tuvalu (TUV), Wallis FAOSTAT, including area harvested (hect- and Futuna (WLF), and Samoa (WSM)—for ares), yields (tons per ha), total production 1901–2006 extracted from the CRU TS 3.0 (metric tons), export quantities (tons), export historical climate database. Grid cells were Table a2.1 AvERAGE vALUES OF ENSO INDICES BY ENSO EvENT CLASS Average value of index for each ENSO event class SOI TNI ONI MEI ENSO category ENSO code 1901–2009 1901–2009 1950–2009 1950–2009 Strong El Niño SE -2.30 -0.15 1.08 1.42 Moderate El Niño EN -1.15 -0.53 0.68 0.69 None N 0.02 -0.48 -0.15 -0.07 Moderate La Niña LN 1.14 0.07 -0.75 -1.05 Strong La Niña SL 2.11 0.82 -1.22 -1.33 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 linked to countries and estimates of the 2000 Table A2.2 provides some simple statistics by population in each grid cell. Using this infor- ENSO class on total precipitation averaged over mation, population-weighted average precipi- grid cells for Samoa for the wet and dry season tation average was calculated for each country and four dry season months from June to Sep- on a monthly basis from 1901 to 2005. Mod- tember. The ENSO classification is assigned by els specified in terms of the logs of monthly NOAA using data for June to November in each or seasonal precipitation generally performed year t from 1933 onwards. The corresponding wet best and are reported here. season is defined as November in year t to April in year t+1, while the corresponding dry season (c) Monthly values of four ENSO indices are is defined as May-October in year t+1. Hence, extracted from the NCAR and National the statistics in the table are based on data from Oceanic and Atmospheric Administration November 1933 to October 2005. (NOAA) databases. All of the indices are constructed as standardized deviations from There are differences between average levels of monthly normal values, though methods precipitation in the seasons following different of standardization vary. Two indices—the categories of ENSO events, and there is a gen- Southern Oscillation Index (SOI), based eral tendency for lower precipitation after an El upon atmospheric pressure at sea level for Niño event and higher precipitation after a La Tahiti and Darwin (using the Trenberth stan- Niña event. However, statistically these effects are dardization); and the Trans Niño Index (TNI) much stronger for the wet season immediately reflecting the gradient of sea surface tempera- after the event than for the following dry season. tures—are available for months from roughly The monthly pattern through the dry season 1870 onwards. The other two—NOAA’s shows differences that are statistically significant Oceanic Niño Index (ONI) based on sea sur- in June but not in later months. The implication is face temperatures, and Wolter’s Multivariate that relying upon a simple classification of ENSO ENSO Index (MEI) based on a set of different events is not an adequate method of capturing atmospheric and ocean indices—are available the impact of ENSO variations in sea tempera- for the period from 1950. Table A2.1 shows ture and other weather variables on precipita- the average values of the four indices for five tion. Instead a more detailed analysis based upon categories of the ENSO. It should be noted explicit analysis of the time profile of ENSO that the TNI is designed to be approximately influences is required. orthogonal to the primary ENSO indices, so it is used as an additional variable in the The first step was to examine whether vector model with the SOI index. autoregression (VAR) models revealed significant lagged cross-country effects in the relationships The ONI and MEI are reported as moving between precipitation and the ENSO indices. averages of data for three months (ONI) or The central idea is that there is a stochastic pro- two months (MEI). In the analysis of monthly cess that generating a vector of variables—in this precipitation, the monthly values of SOI and case the logs of monthly precipitation for all eight TNI were converted to 2-month moving aver- countries—whose evolution over time depend ages when assessing the relative performance upon their common history. VAR models can be of alterative ENSO indices in the models. rewritten as more complex time series models for The treatment of the ENSO indices in mod- each variable separately, but it is often possible eling seasonal precipitation is discussed in to obtain more parsimonious models and better detail below. forecasts by relying upon the VAR specification. S A M OA CO U N T RY ST U DY 51 Table a2.2 PRECIPITATION STATISTICS BY ENSO CLASS, 1933–2005 (MM) Wet season Dry season June July August September Strong El Niño Average 1,677 1,132 144 177 176 166 SD 390 408 64 97 119 68 Min 1,026 442 76 28 32 39 Max 2,192 1,683 288 322 464 250 Moderate El Niño Average 1,783 1,104 193 127 177 177 SD 309 240 68 54 85 96 Min 1,364 649 101 45 55 67 Max 2,403 1,402 306 219 337 375 None Average 1,973 1,092 140 143 160 156 SD 264 283 66 58 87 95 Min 1,382 438 40 46 17 3 Max 2,457 1,672 305 292 339 474 Moderate La Niña Average 2,201 1,110 168 121 142 194 SD 342 330 76 58 136 163 Min 1,790 626 107 70 36 64 Max 2,557 1,658 326 244 432 479 Strong La Niña Average 2,338 929 217 104 65 133 SD 556 238 75 54 53 121 Min 1,810 666 79 45 17 22 Max 3,446 1,386 300 202 175 390 All years Average 1,952 1,086 159 140 154 163 SD 380 299 72 67 98 101 Min 1,026 438 40 28 17 3 Max 3,446 1,683 326 322 464 479 Oneway ANOVA F-value 6.02 0.56 2.86 1.75 1.91 0.40 by ENSO class Prob 0.000 0.689 0.030 0.149 0.449 0.805 In this case, almost none of the lagged cross- 12 in(Pt) = ∑ αsMst + ( βdDdt +βwDwt)It + ut country effects were statistically significant. s=1 Further, the hypothesis of normal errors was decisively rejected in all cases, which may mean where Pt is precipitation in period t, Mst are dummy that the forecasting performance of the models variables for months s=1 … s=12 in period t, Ddt will be uncertain. and Dwt are dummy variables for dry and wet sea- sons in period t, It is the ENSO index in period t, and For these reasons, the remainder of the analysis ut is the error in period t which follows some form focused on ARMA time series models using data of ARMA process. This specification allows for the for Samoa on its own. The basic equation that is possibility that the influence of ENSO events may estimated is: be different during the dry and wet seasons. 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 Table a2.3 COEFFICIENTS ON ENSO INDICES IN EqUATIONS FOR MONTHLY PRECIPITATION Sample 1901–2006 Sample 1950–2006 (1) (2) (3) (4) (5) (6) (7) (8) (9) SOI 0.058*** 0.067*** 0.074*** (0.010) (0.010) (0.016) SOI * Dry 0.054*** 0.059*** Season (0.013) (0.013) SOI * 0.061*** 0.069*** Wet Season (0.013) (0.013) SOI * 0.037* 0.040** 1901– 1949 (0.019) (0.018) SOI * 0.075*** 0.090*** 1950– 2005 (0.012) (0.011) TNI -0.0710*** (0.013) TNI * Dry -0.095*** Season (0.016) TNI * -0.043** Wet Season (0.021) TNI * -0.046 1901– 1949 (0.036) TNI * -0.084*** 1950– 2005 (0.014) ONI -0.166*** (0.033) MEI -0.136*** (0.026) Log- 789.5 789.4 778.7 776.3 787.7 775.2 504.7 503.2 503.9 likelihood Test for equality of coeffs Chi- 0.19 4.67 2.95 6.14 square Prob 0.662 0.097 0.086 0.046 Notes: (a) Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.10. (b) In addition to the ENSO indices the model includes month dummies and an AR(3) error. S A M OA CO U N T RY ST U DY 53 The main results from estimating this model are variations and precipitation may have changed or shown in Table A2.3. The key conclusions that may be changing over time. However, an alterna- can be drawn from these results are: tive possibility is that the values of the SOI and TNI indices are less reliable for years before 1950 ■■ There is a clear and consistent relationship than for more recent years. To investigate this, between the ENSO indices (averaged over two sets of checks were carried out. First, the two or three months) and precipitation. Dur- timing of any structural break was examined by ing warm (El Niño) phases of the oscillation, considering breaks at decade intervals from 1930 precipitation is below average. to 1980. The statistical evidence for a break is strongest for 1950 and marginally less strong for ■■ The Oceanic Niño Index (ONI) yields the best 1940. Any break after 1950 is decisively rejected. fit for the data from 1950 onwards, but the dif- The second test was to examine whether the ferences between statistical performance using influence of ENSO variations on precipitation this index and the two other indices examined indicate sign of trends since 1950. The results of for the same period—SOI and MEI—are this test are shown in Table 2.4 which looks at marginal, so that other considerations may be decade trends in the coefficients on the ENSO more important. indices. The trend coefficients are significant at the 5 percent level for the SOI index and at ■■ The impact of ENSO variations on precipi- the 10 percent level for the ONI index but not tation is immediate in the sense that it is not for the other indices. However, the direction of necessary to include lagged values of the aver- the trend in each case is to weaken the effect of aged ENSO indices to explain variations in ENSO variations on precipitation. precipitation. Overall, the results suggest that the data for the ■■ There is no evidence that the impacts of ENSO SOI index before 1950 may not provide a reli- variations have different relative impacts on able basis for estimating the relationship between precipitation during the wet and dry seasons. ENSO variations and precipitation in Samoa. The best ENSO index for the period from 1950 ■■ The inclusion of the TNI index improves the onwards is the ONI index. Using equation (8), a overall statistical performance of the model. typical strong El Niño event will reduce precipi- A higher value for TNI—that is, a steeper tation in the following months by about 18 per- sea temperature gradient—is associated with cent with little difference between its effects in the lower precipitation. wet season and the dry season. If the downward trends in the coefficients on the SOI and ONI ■■ Some evidence suggests that there is some kind indices in equations (13) and (15) are accepted, of structural break in the relationships that then the impact of a typical strong El Niño event may have occurred in the middle of the last in the current decade (2010–19) may be close to century. The hypothesis that the coefficients zero (-2 percent for the ONI index to +2 percent on SOI and TNI before and after 1950 are for the SOI index). equal is rejected at the 10 percent level for SOI on its own and at the 5 percent level for SOI There is insufficient evidence to assess whether the and TNI together. apparent trend in the relationship between some ENSO indices and precipitation is a consequence The last of these conclusions suggests the possi- of climate change or other natural phenomena. bility that the basic relationship between ENSO It has been argued—see, for example, Trenberth 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 Table a2.4 DECADE TRENDS IN ENSO INFLUENCE ON MONTHLY PRECIPITATION Sample 1950–2005 (10) (11) (12) (13) SOI 0.133*** 0.142*** (0.031) (0.029) SOI * Decade -0.0235** -0.0214** (0.010) (0.010) TNI -0.0915* (0.049) TNI * Decade 0.0037 (0.014) ONI -0.264*** (0.064) ONI * Decade 0.0409* (0.022) MEI -0.184*** (0.052) MEI * Decade 0.0205 (0.018) Log-likelihood 502.3 494.3 501.4 503.3 Notes: (a) Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.10. (b) In addition to the ENSO indices the model includes month dummies and an AR(3) error. Table a2.5 COEFFICIENTS ON ENSO INDICES IN EqUATIONS FOR AGRICULTURAL TRADE Agricultural exports Agricultural imports (1) (2) (3) (4) (5) (6) (7) (8) SOI 0.011 0.011 -0.020** -0.021*** (0.023) (0.023) (0.008) (0.008) TNI -0.000 0.023 (0.045) (0.015) ONI -0.017 0.031** (0.041) (0.014) MEI -0.010 0.028** (0.035) (0.012) Log- -12.5 -12.5 -12.5 -12.6 30.2 31.6 29.7 29.9 likelihood Notes: (a) Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.10. (b) In addition to the ENSO indices the model includes dummy variables for years affected by cyclones and the taro blight (1994–1996) plus time and an AR(2) error. S A M OA CO U N T RY ST U DY 55 and Hoar (1997)—that there has been a change while the results for other crops such as yams, in the frequency and intensity of ENSO events mangoes, and avocadoes do not alter any of the since the mid-1970s, which may be linked to cli- conclusions. The analysis of agricultural trade mate change as a result of the increase in aver- focuses on the values of exports and imports of all age sea surface temperatures in the Pacific over agricultural products. Imports of food products the last 50 years. If ENSO events become more account for about 90 percent of imports of all intense, i.e., the peak values of the ENSO indices agricultural imports. The share of food products become larger, the effect of this on precipitation in total agricultural exports has been about 65 may be partly or wholly offset by the change in percent in recent years, but was as high as 90 per- the relationship that seems to be apparent in the cent immediately before the cyclones in the late data. Thus, it cannot be assumed that the effect 1980s. Samoa runs a large deficit on agricultural of climate change on ENSO events will necessar- trade, which has grown from about $15 million ily have a clear impact on variations in seasonal per year in the late 1990s to over $40 million per precipitation in Samoa. year in 2006–07. In view of this uncertainty, the second part of The basic model in this case is: the analysis examines the direct link between in(Vit) = αi + βi I t-1 + γi Dct) + ∂ibDbt + uit ENSO indices, crop production, and trade in agricultural products. The results presented here focus on production and yields for taro, coconuts, where Vit is the value of trade or crop production bananas, pineapples, and papayas, since these are for category i in period t, It-1 is the ENSO index the main Samoan crops by land area and value, in period t-1, Dct is a dummy variable taking the 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 value 1 in years (1990–92) affected by cyclones June-November of the previous year; hence the Ofa and Val, Dbt is a dummy variable taking the subscript t-1. value 1 in years (1994–1996) affected by the taro blight, and uit is the error in period t, which fol- Table 2.5 shows that ENSO events do not affect lows some form of ARMA process. Since produc- the level of agricultural exports to any significant tion and trade data is annual, the ENSO indices degree, but they do have a significant—though used in the equations are averages for the months modest—influence on the level of agricultural Table a2.6 COEFFICIENTS ON ENSO INDICES IN EqUATIONS FOR AGRICULTURAL PRODUCTION Crop production Crop Yields (1) (2) (3) (4) (5) TARO SOI 0.035*** 0.039*** 0.014 (0.011) (0.011) (0.008) TNI -0.039*** -0.020** (0.015) (0.008) ONI -0.050*** (0.019) MEI -0.047*** (0.016) COCONUTS SOI 0.014 0.015 0.060*** (0.012) (0.012) (0.018) TNI -0.008 -0.035** (0.016) (0.018) ONI -0.014 (0.019) MEI -0.015 (0.017) BANANAS SOI 0.174 0.200* 0.073 (0.116) (0.115) (0.150) TNI -0.220 0.064 (0.158) (0.150) ONI -0.262 (0.191) MEI -0.272* (0.164) Log-likelihood 239.8 244.6 238.4 238.8 262.4 Notes: (a) Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.10. (b) In addition to the ENSO indices the model includes dummy variables for years affected by cyclones (1990–92) and the taro blight (1994–1996) plus time and an AR(2) error. S A M OA CO U N T RY ST U DY 57 imports. Strong El Niño events are associated following months: lower for El Niño, higher for La with imports that are 3–5 percent higher than in Niña. In turn, this is reflected in lower or higher a normal year, with a corresponding reduction in production for taro and perhaps for bananas. strong La Niña events. The TNI index does not The effect on coconuts and other crops is small or appear to have a significant impact on trade. zero. Finally, the changes in crop production are accompanied by offsetting changes in the value of This equation was also estimated for the produc- agricultural imports. This may be due to a com- tion and yield of each crop on its own to examine bination of higher/lower volumes and higher/ the influence of ENSO events on crop produc- lower prices to the extent that ENSO events can tion and yields. None of the coefficients on the have a large impact on agricultural production in ENSO indices were significantly different from the Pacific basin. zero in this single equation specification. So, a VAR version of the model pooling data for If—and this is only a hypothesis at present— the five main crops—taro, coconuts, bananas, climate change leads to more intense or frequent pineapples, and papayas—was estimated as an ENSO events, the consequence for Samoa will be alternative specification so as to allow for the greater year-to-year variations in crop produc- covariance between different crops. Table 2.6 tion, agricultural incomes, and imports. This is summarizes the results of estimating the VAR largely a matter of risk management, since the models using different ENSO indices. It focuses results do not imply that average production, etc., on production and yields for taro, coconuts, and over a series of events will change. Further, the bananas because the coefficients on the ENSO magnitude of the impact of such changes must be indices for pineapples and papayas are not sig- put in context. If the intensity of a typical strong nificantly different from zero. El Niño event were to increase by 50 percent, the associated fall in taro production would rise from The main conclusion is that there is a clear rela- 5–8 percent to 8–12 percent. tionship between the ENSO indices and the production of taro. The coefficients mean that The increase in crop variability is a good reason to a strong El Niño event will cause a fall of 5–8 encourage farmers to diversify away from relying percent in taro production in the following crop upon taro as a primary crop, but it is not signifi- year. There is weak evidence for a large impact on cant relative to other natural hazards—particularly the production of bananas but the results may not cyclones and earthquakes—that can have a much be robust. Crop yield—but not production—for larger impact on Samoa’s economy. As an illustration coconuts is also influenced by ENSO events, so it of this point, changes in value-added in agriculture seems that farmers are able to offset lower yields and fisheries have been negatively correlated with by harvesting a greater area of trees. changes in total GDP since 1995. This reflects the overall decline in the share of agriculture in GDP, Overall, the analysis has shown that ENSO events together with the fact that the economy has become have a significant effect on precipitation in the much more subject to other economic risks. 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 S A M OA CO U N T RY ST U DY 59 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 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