Economics of Adaptation to Climate Change:

Acknowledgements

This report has been prepared by a core team led by Sergio Margulis (TTL) and Urvashi Narain and
comprising Paul Chinowsky, Laurent Cretegny, Gordon Hughes, Paul Kirshen, Anne Kuriakose, Glenn
Marie Lange, Gerald Nelson, James Neumann, Robert Nicholls, Kiran Pandey, Jason Price, Adam
Schlosser, Robert Schneider, Roger Sedjo, Kenneth Strzepek, Rashid Sumaila, Philip Ward, and David
Wheeler. Major contributions were made by Jeroen Aerts, Carina Bachofen, Brian Blankespoor, Ana
Bucher, Steve Commins, David Corderi, Susmita Dasgupta, Timothy Essam, William Farmer, Eihab
Fathelrahman, Prodipto Ghosh, Dave Johnson, James Juana, Tom Kemeny, Benoit Laplante, Robin
Mearns, Siobhan Murray, Hawanty Page, Mark Rosegrant, Klas Sanders, Arathi Sundaravadanan,
Timothy Thomas, Jasna Vukoje, and Tingju Zhu. Sally Brown and Susan Hanson made important
contributions to the coastal sector report, Miroslac Batka, Jawoo Koo, David Lee, Marilia Magalhaes,
Siwa Msangi, Amanda Palazzo, Claudia Ringler, Richard Robertson, and Timothy Sulser to the
agriculture sector report, William Cheung to the fishery sector report, and Pieter Pauw and Luke M.
Brander to the water sector report.

Since the beginning, the EACC team has had intense interaction with the Environment Department’s
management, particularly Warren Evans and Michelle de Nevers, who should, in fact, be considered part
of the EACC team. The team is also grateful to Sam Fankhauser and Ravi Kanbur for serving on the
advisory committee and to Julia Bucknall, Shanta Devarajan, Marianne Fay, Gherson Feder, Armin
Fidler, Kirk Hamilton, Tamer Samah Rabie, Peter Rogers, Jim Shortle, Joel Smith, Michael Toman, and
Gary Yohe for acting as peer reviewers. Numerous comments and suggestions were also received from a
very large number of colleagues and the team is most thankful to all of them. From the World Bank they
include Vahid Alavian, Aziz Bouzaher, Jan Bojo, Henrike Brecht, Kenneth Chomitz, Vivian Foster,
Alexander Lotsch, Kseniya Lvovsky, Dominique van Der Mensbrughe, John Nash, Ian Noble, Giovanni
Ruta, Apurva Sanghi, Robert Townsend, Walter Vergara, and Winston Yu. From outside the Bank they
include Marten van al Aast, Roy Brouwer, Maureen Cropper, Anton Hilbert, Christine Pirenne, Tamsin
Vernon, and Peter Wooders. None of these colleagues and reviewers are, in any way, responsible for the
contents and eventual errors of this report, which remain sole responsibility of the EACC Team.

This study is being conducted in partnership between the World Bank (leading its technical aspects), the
governments of the United Kingdom, Netherlands, and Switzerland (funding the study), and the
participating case study countries. The findings, interpretations, and conclusions expressed in this paper
do not necessarily reflect the views of the Executive Directors of the World Bank or the governments they
represent. The World Bank 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 concerning the legal status of any territory or the
endorsement or acceptance of such boundaries. 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 International Bank for Reconstruction and Development / World Bank encourages dissemination of
its work and will normally grant permission to reproduce portions of the work promptly.

Abbreviations

AR4

4th Assessment Report

CIAT

International Center for Tropical Agriculture

CLIRUN

The Climate and Runoff Model

CMI

Climate moisture index

CSIRO

Commonwealth Scientific and Industrial Research Organization

CRED

Centre for Research on the Epidemiology of Disasters

DALY

Disability-adjusted life year

DCCP2

Disease Control Priorities in Developing Countries Project

DIVA

Dynamic and Interactive Vulnerability Assessment

EACC

Economics of Adaptation to Climate Changes

EAP

East Asia and Pacific (World Bank region)

ECA

Europe and Central Asia (World Bank region)

EIA

Environmental impact analysis

ENSO

El Niño Southern Oscillation

FPUs

Food production units

FUND

Climate Framework for Uncertainty, Negotiation, and Distribution

GCM

Global climate model

GDP

Gross domestic product

GIS

Geographic information system

GHF

Global Humanitarian Forum

GPW

Gridded population of the world

HDI

UNDP’s Human Development Index

IFPRI

International Food Policy Research Institute

IMPACT

International Model for Policy Analysis of Agricultural Commodities and Trade

IPCC

Intergovernmental Panel on Climate Change

LAC

Latin America and Caribbean Region

MNA

Middle East and North Africa (World Bank region)

NAPA

National Adaptation Program of Action

NCAR

National Centre for Atmospheric Research

NGO

Nongovernmental organization

NPP

Net primary productivity

NREGA

National Rural Employment Guarantee Act

ODA

Official development assistance

OECD

Organisation for Economic Co-operation and Development

O&M

Operation and maintenance

PESP

Primary Education Stipend Program

Ppm

Parts per million

PPP

Purchasing power parity

PSD

Participatory scenario development

PSNP

Productive Safety Nets Program

RICE99

Regional Dynamic Integrated Model of Climate and the Economy

SAS

South Asia (World Bank region)

SSA

Sub-Saharan Africa (World Bank region)

SRES

Special Report on Emissions Scenarios of the IPCC

UIUC

University of Illinois at Urbana–Champaign

E xecutive Summary

Even with global emissions of greenhouse gases drastically reduced in the coming years, the global
annual average temperature is expected to be 2

C above pre-industrial levels by 2050. A 2

C warmer

world will experience more intense rainfall and more frequent and more intense droughts, floods, heat
waves, and other extreme weather events. Households, communities, and planners need to put in place
measures and initiatives that “reduce the vulnerability of natural and human systems against actual and
expected climate change effects” (IPCC 2007). Without such adaptation, development progress will be
threatened—perhaps even reversed.

While countries need to adapt to manage the unavoidable, they need to take decisive mitigation measures
to avoid the unmanageable. Unless the world begins immediately to reduce greenhouse gas emissions
significantly, global annual average temperature will increase by about 2.5

–7

C above pre-industrial

levels by the end of the century. Temperature increases higher than 2

C—say on the order of 4

C—are

predicted to significantly increase the likelihood of irreversible and potentially catastrophic impacts such
as the extinction of half of species worldwide, inundation of 30 percent of coastal wetlands, and
substantial increases in malnutrition and diarrheal and cardio-respiratory diseases. Even with substantive
public interventions, societies and ecosystems will not be able to adapt to these impacts.

Under the December 2007 Bali Action Plan, adopted at the United Nations Climate Change Conference,
developed countries have agreed to “adequate, predictable, and sustainable financial resources and the
provision of new and additional resources, including official and concessional funding for developing
country parties” (UNFCCC 2008) to help them adapt to climate change.

Yet, existing studies on adaptation costs provide only a wide range of estimates, from $4 billion to $109
billion a year, and have many gaps. Similarly, National Adaptation Programs of Action (prepared by
Least Developed Countries under the United Nations Framework Convention on Climate Change,
UNFCCC) identify and cost only urgent and immediate adaptation needs, and countries do not typically
incorporate adaptation measures into long-term development plans.

Putting a price tag on adaptation

To shed light on adaptation costs—and with the global climate change negotiations resuming in
December 2009 in Copenhagen—the Economics of Adaptation to Climate Change (EACC) study was
initiated by the World Bank in early 2008, funded by the governments of the Netherlands, Switzerland,
and the United Kingdom. Its objectives are to develop an estimate of adaptation costs for developing
countries and to help decision makers in developing countries understand and assess the risks posed by
climate change and design better strategies to adapt to climate change.

The initial study report, which focuses on the first objective, finds that the cost between 2010 and 2050 of
adapting to an approximately 2

C warmer world by 2050 is in the range of $75 billion to $100 billion a

year. This sum is of the same order of magnitude as the foreign aid that developed countries now give
developing countries each year, but it is still a very low percentage of the wealth of countries as measured
by their GDP. A second report, based on seven country case studies (Bangladesh, Plurinational State of

Bolivia, Ethiopia, Ghana, Mozambique, Samoa, and Vietnam) and expected by March 2010, will focus on
the second objective.

Using a consistent methodology

The intuitive approach to costing adaptation involves comparing a future world without climate change
with a future world with climate change. The difference between these two worlds entails a series of
actions to adapt to the new world conditions. And the costs of these additional actions are the costs of
adapting to climate change. With that in mind, the study took the following four steps:

•

Picking a baseline. For the timeframe, the world in 2050 was chosen, not beyond (forecasting climate
change and its economic impacts becomes even more uncertain beyond this period). Development
baselines were crafted for each sector, essentially establishing a growth path in the absence of climate
change that determines sector-level performance indicators (such as stock of infrastructure assets,
level of nutrition, and water supply availability). The baselines used a consistent set of GDP and
population forecasts for 2010–50.

•

Choosing climate projections. Two climate scenarios were chosen to capture as large as possible a
range of model predictions. Although model predictions do not diverge much in projected
temperatures increases by 2050, precipitation changes vary substantially across models. For this
reason, model extremes were captured by using the two model scenarios that yielded extremes of dry
and wet climate projections. Catastrophic events were not captured, however.

•

Predicting impacts. An analysis was done to predict what the world would look like under the new
climate conditions. This meant translating the impacts of changes in climate on the various economic
activities (agriculture, fisheries), on people’s behavior (consumption, health), on environmental
conditions (water availability, oceans, forests), and on physical capital (infrastructure).

•

Identifying adaptation alternatives and costing. Adaptation costs were estimated by major economic
sector—infrastructure, coastal zones, water supply and flood management, agriculture, fisheries,
human health, and forestry and ecosystem services. Cost implications of changes in the frequency of
extreme weather events were also considered. Cross-sectoral analysis of costs was not feasible.

Putting the methodology to work

The next step was adjusting and tailoring each step to the data and information available, a
distinctive feature of the EACC study. The study used extensive global and national data sets,
including World Bank projects and global economic indicators. In the process, several questions
arose.

What exactly is “adaptation”? Is development adaptation? In reality, developing countries face not only
a deficit in adapting to current climate variation, let alone future climate change, but also deficits in

providing education, housing, health, and other services. Thus, many countries face a more general
“development deficit,” of which the part related to climate events is termed the “adaptation deficit.”

There are two ways to estimate the costs of adaptation: with the adaptation deficit or without it. This
study chose to make the adaptation deficit a part of the development baseline, so that adaptation costs
cover only the additional costs to cope with future climate change. Thus, the costs of measures that would
have been undertaken even without climate change are not included in adaptation costs, but the costs of
doing more, doing different things (policy and investment choices), and doing things differently are.

Which adaptation measures? Adaptation measures can be classified by the initiating economic sector—
public or private. This study includes planned adaptation (adaptation that results from a deliberate public
policy decision) but not autonomous or spontaneous adaptation (adaptation by households and
communities acting on their own without public interventions but within an existing public policy
framework). Since the objective is to help governments plan for risks, it is important to have an idea of
what problems private markets will solve on their own, how public policies and investments can
complement markets, and what measures are needed to protect public assets and vulnerable people—that
is, planned adaptation.

In all sectors, “hard” options involving engineering solutions were favored over “soft” options based on
policy changes and social capital mobilization—except in the study of extreme weather events where the
emphasis is on investment in human resources, particularly those of women. Although hard adaptation
options are feasible in nearly all settings, while soft options depend on social and institutional capital and
thus may not be available in many settings, this focus on hard options was largely to ease computation of
adaptation costs and not to suggest that these are always preferable.

How much adaptation is appropriate? Countries have several options. They can try to fully adapt, so that
society is at least as well off as it was before climate change. They can choose to do nothing—to suffer
(or enjoy the benefits from) the full impact of climate change. Or they can decide to adapt to the level
where the benefits from adaptation equal their costs, at the margin. The study assumes that countries will
adapt up to the level at which they enjoy the same level of welfare in the (future) world as they would
have without climate change. This is not necessarily the most economically rational decision, but it is a
practical rule that greatly simplifies the exercise.

How should benefits be costed? What happens if climate changes lead to lower investment or
expenditure requirements for some sectors in some countries—for example, changes in demand for
electricity or water lead to lower requirements for electricity generating capacity, water storage, and water
treatment? In such cases, the “costs” of adaptation are negative. For calculating global costs, this becomes
a summation problem. Rather than making an explicit decision on whether to offset potential benefits of
climate change against costs of adaptation, whether across sectors or countries, the study presents costs
using three aggregation methods—gross (no netting of costs), net (benefits are netted across sectors and
countries), and X-sums (positive and negative items are netted within countries but not across countries).
The study opted to use X-sums in reporting most adaptation costs in the interest of space, although similar
trends hold for the other aggregation methods.

The global price tag

Overall, the study estimates that the cost between 2010 and 2050 of adapting to an approximately 2

warmer world by 2050 is in the range of $75 billion to $100 billion a year (table 1). This sum is the
same order of magnitude as the foreign aid that developed countries now give developing countries each
year, but it is still a very low percentage of the wealth of countries (measured by their GDP).

Table 1. Total annual costs of adaptation for all sectors, by region, 2010–50 ($ billions at 2005
prices, no discounting)

Cost
aggregation
type

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

National Centre for Atmospheric Research (NCAR), wettest scenario

Gross sum

28.7

10.5

22.5

4.1

17.1

18.9

101.8

X-sum

25.0

9.4

21.5

3.0

12.6

18.1

89.6

Net sum

25.0

9.3

21.5

3.0

12.6

18.1

89.5

Commonwealth Scientific and Industrial Research Organization (CSIRO), driest scenario

Gross sum

21.8

6.5

18.8

3.7

19.4

18.1

88.3

X-sum

19.6

5.6

16.9

3.0

15.6

16.9

77.6

Net sum

19.5

5.2

16.8

2.9

15.5

16.9

76.8

Note: The gross aggregation method sets negative costs in any sector in a country to zero before costs are aggregated
for the country and for all developing countries. The X-sums net positive and negative items within countries but not
across countries and include costs for a country in the aggregate as long as the net cost across sectors is positive for
the country. The net aggregate measure nets negative costs within and across countries.

Source: Economics of Adaptation to Climate Change study team.

Total adaptation costs calculated by the gross sum method average $10 billion a year more than
by the other two methods (the insignificant difference between the X-sum and net sum figures is
largely a coincidence). The difference is driven by countries that appear to benefit from climate
change in the water supply and flood protection sector, especially in East Asia and Pacific and
South Asia.

The drier scenario (Commonwealth Scientific and Industrial Research Organization, CSIRO) requires
lower total adaptation costs than does the wetter scenario (National Centre for Atmospheric Research,
NCAR), largely because of the sharply lower costs for infrastructure, which outweigh the higher costs for
water and flood management. In both scenarios, infrastructure, coastal zones, and water supply and flood
protection account for the bulk of the costs. Infrastructure adaptation costs are highest for the wetter
scenario, and coastal zones costs are highest for the drier scenario.

On a regional basis, for both climate scenarios, the East Asia and Pacific Region bears the highest
adaptation cost, and the Middle East and North Africa the lowest. Latin America and the Caribbean and
Sub-Saharan Africa follow East Asia and Pacific in both scenarios (figures 1 and 2). On a sector
breakdown, the highest costs for East Asia and the Pacific are in infrastructure and coastal zones; for Sub-
Saharan Africa, water supply and flood protection and agriculture; for Latin America and the Caribbean,
water supply and flood protection and coastal zones; and for South Asia, infrastructure and agriculture.

Figure 1. East Asia and Pacific has the highest cost of adpatation in the wetter scenario, followed by
Latin America and the Caribbean

Total annual cost of adaptation and share of costs for National Centre for Atmospheric Research
(NCAR) scenario, by region ($ billions at 2005 prices, no discounting)

Note: EAP is East Asia and Pacific, ECA is Europe and Central Asia, LAC is Latin America and Caribbean, MNA
is Middle East and North Africa, SAS is South Asia, and SSA is Sub-Saharan Africa.
Source: Economics of Adaptation to Climate Change study team.

28%

10%

24%

14%

20%

$25.0

$9.4

$21.5

$3.0

$12.6

$18.1

NCAR

EAP

ECA

LAC

MNA

SAS

SSA

Figure 2. East Asia and Pacific has the highest cost of adpatation in the drier scenario, followed by
Latin America and the Caribbean and Sub-Saharan Africa

Total annual cost of adaptation and share of costs for Commonwealth Scientific and Industrial
Research Organization (CSIRO) scenario, by region ($ billions at 2005 prices, no discounting)

Not surprisingly, both climate scenarios show costs increasing over time, although falling as a
percentage of GDP—suggesting that countries become less vulnerable to climate change as their
economies grow (figures 3 and 4). There are considerable regional variations, however. Adaptation costs
as a percentage of GDP are considerably higher in Sub-Saharan Africa than in any other region, in large
part because of the lower GDPs in this region.

25%

22%

20%

22%

$19.6

$5.6

$16.9

$3.0

$15.6

$16.9

EAP

ECA

LAC

MNA

SAS

SSA

Figure 4. ...but fall as a share of GDP

Total annual costs of adaptation for National Centre for Atmospheric Research (NCAR) scenario as
share of GDP, by decade and region (percent, at 2005 prices, no discounting)

Turning to the EACC analyses of sectors and extreme events, the findings offer some insights for
policymakers who must make tough choices in the face of great uncertainty.

Infrastructure. This sector has accounted for the largest share of adaptation costs in past studies and takes
up a major share in the EACC study—in fact, the biggest share for the NCAR (wettest) scenario because
the adaptation costs for infrastructure are especially sensitive to levels of annual and maximum monthly
precipitation. Urban infrastructure—urban drainage, public buildings and similar assets—accounts for
about 54 percent of the infrastructure adaptation costs, followed by roads (mainly paved) at 23 percent.
East Asia and the Pacific and South Asia face the highest costs, reflecting their relative populations. Sub-
Saharan Africa experiences the greatest increase over time with its adaptation costs rising from $1.1
billion a year for 2010–19 to $6 billion a year for 2040–49.

Coastal zones. Coastal zones are home to an ever growing concentration of people and economic activity,
yet they are also subject to a number of climate risks, including sea-level rise and possible increased
intensity of tropical storms and cyclones. These factors make adaptation to climate change critical. The
EACC study shows that coastal adaptation costs are significant and vary with the magnitude of sea-level
rise, making it essential for policymakers to plan while accounting for the uncertainty. One of the most
striking results is that Latin America and the Caribbean and East Asia and the Pacific account for about
two-thirds of the total adaptation costs (see figures 1 and 2).

0.00%

0.10%

0.20%

0.30%

0.40%

0.50%

0.60%

0.70%

0.80%

EAP

ECA

LAC

MNA

SAS

SSA

ts as

e
r
c
e
n

t of

World Bank Region

2010-19

2020-29

2030-39

2040-49

Water supply. Climate change has already affected the hydrological cycle, a process that is expected to
intensify over the course of the 21

century. In some parts of the world, water availability has increased

and will continue to increase, but in other parts, it has decreased and will continue to do so. Moreover, the
frequency and magnitude of floods are expected to rise, because of projected increases in the intensity of
rainfall. Accounting for the climate impacts, the study shows that water supply and flood management
ranks as one of the top three adaptation costs in both the wetter and drier scenarios, with Sub-Saharan
Africa footing by far the highest costs. Latin America and the Caribbean also sustain high costs under
both models, and South Asia sustains high costs under CSIRO.

Agriculture. Climate change affects agriculture by altering yields and changing areas where crops can be
grown. The EACC study shows that changes in temperature and precipitation from both climate scenarios
will significantly hurt crop yields and production—with irrigated and rainfed wheat and irrigated rice the
hardest hit. South Asia shoulders the biggest declines in production but developing countries fare worse
for almost all crops compared to developed countries. Moreover, the changes in trade flow patterns are
dramatic. Under the NCAR, developed country exports increase by 28 percent while under the CSIRO
they increase by 75 percent compared with 2000 levels. South Asia becomes a much larger importer of
food under both scenarios, and East Asia and Pacific becomes a net food exporter under the NCAR. In
addition, the decline in calorie availability brought about by climate change raises the number of
malnourished children.

Human health. The key human health impacts of climate change include increases in the incidence of
vector-borne disease (malaria), water-borne diseases (diarrhea), heat- and cold-related deaths, injuries and
deaths from flooding, and the prevalence of malnutrition. The EACC study, which focuses on malaria and
diarrhea, finds adaptation costs falling in absolute terms over time to less than half the 2010 estimates of
adaptation costs by 2050. Why do costs decline in the face of higher risks? The answer lies in the benefits
expected from economic growth and development. While the declines are consistent across regions, the
rate of decline in South Asia and East Asia and Pacific is more rapid than in Sub-Saharan Africa. As a
result, by 2050 more than 80 percent of the health sector adaption costs will be shouldered by Sub-
Saharan Africa.

Extreme weather events. In the absence of reliable data on emergency management costs, the EACC
study tries to shed light on the role of socioeconomic development in increasing climate resilience. It
asks: As climate change increases potential vulnerability to extreme weather events, how many additional
young women would have to be educated to neutralize this increased vulnerability? And how much would
it cost? The findings show that by 2050, neutralizing the impact of extreme weather events requires
educating an additional 18 million to 23 million young women at a cost of $12 billion to $15 billion a
year. For the period 2000–50 as a whole, the tab reaches about $300 billion in new outlays. This means
that in the developing world, neutralizing the impact of worsening weather over the coming decades will
require educating a large new cohort of young women at a cost that will steadily escalate to several billion
dollars a year. However, it will be enormously worthwhile on other margins to invest in education for
millions of young women who might otherwise be denied its many benefits.

Putting the findings in context

How does this study compare with earlier studies? The EACC estimates are in the upper end of
estimates provided by the UNFCCC (2007), the study closest in approach to the EACC (table 2),
although not as high as suggested by a recent critique of the UNFCCC study by Parry and others
(2009).

Why are the EACC estimates so much higher than those of the UNFCCC? To begin with, even
though a comparison of the studies is limited by a number of methodological differences (in
particular, the use of a consistent set of climate models to link impacts to adaptation costs and an
explicit separation of costs of development from those of adaptation in the EACC study), the
major difference between them is the sixfold increase in the cost of coastal zone management
and defense under the EACC study. This difference reflects several improvements to the earlier
UNFCCC estimates under the EACC study: better unit cost estimates, including maintenance
costs, and the inclusion of costs of port upgrading and risks from both sea-level rise and storm
surges.

Table 2. Comparison of adaptation cost estimates by the United Nations Framework Convention on
Climate Change and the Economics of Adaptation to Climate Change

Sector

United Nations

Framework

Convention on

Climate Change

(2007)

Economics of Adaptation to Climate

Change study

National Centre

for Atmospheric

Research (NCAR),

wettest scenario

Commonwealth

Scientific and

Industrial Research

Climate

(

CSIRO),

driest scenario

Infrastructure

2-41

29.5

13.5

Coastal zones

30.1

29.6

Water supply and flood
protection

13.7

19.2

Agriculture, forestry, fisheries

7.6

7.3

Human health

1.6

Extreme weather events

—

6.7

6.5

Total

28-67

89.6

77.7

Source: UNFCCC (2007) and Economics of Adaptation to Climate Change study team.

Another reason for the higher estimates is the higher costs of adaptation for water supply and
flood protection under the EACC study, particularly for the drier climate scenario, CSIRO. This
difference is explained in part by the inclusion of riverine flood protection costs under the EACC
study. Also pushing up the EACC study estimate is the study’s comprehensive sector coverage,
especially inclusion of the cost of adaptation to extreme weather events.

The infrastructure costs of adaptation in the EACC study fall in the middle of the UNFCCC
range because of two contrary forces. Pushing up the EACC estimate is the more detailed
coverage of infrastructure. Previous studies estimated adaptation costs as the costs of climate-
proofing new investment flows and did not differentiate risks or costs by type of infrastructure.
The EACC study extended this work to estimate costs by types of infrastructure services—
energy, transport, water and sanitation, communications, and urban and social infrastructure.
Pushing down the EACC study estimate are measurements of adaptation against a consistently
projected development baseline and use of a smaller multiplier on baseline investments than in
the previous literature, based on a detailed analysis of climate proofing, including adjustments to
design standards and maintenance costs.

The one sector where the EACC study estimates are actually lower than the UNFCCC study is
human health. The reason for this divergence is in part because of the inclusion of the
development baseline, which reduces the number of additional cases of malaria, and thereby
adaptation costs, by some 50 percent by 2030 under the EACC study.

The bottom line is that calculating the global cost of adaptation remains a complex problem, requiring
projections of economic growth, structural change, climate change, human behavior, and government
investments 40 years in the future. The EACC study has tried to establish a new benchmark for research
of this nature, as it adopted a consistent approach across countries and sectors and over time. But in the
process, it had to make important assumptions and simplifications, to some degree biasing the estimates.

•

Adaptation costs are calculated as though decisionmakers knew with certainty what the future climate
will be, when in reality the current climate knowledge does not permit even probabilistic statements
about country-level climate outcomes. In a world where decisionmakers hedge against a range of
outcomes, the costs of adaptation could be potentially higher.

•

Of the many global climate projections available for the baseline, only the set reporting maximum
and minimum temperatures—and within that set, only the two yielding the wettest and the driest
outcomes—were used. In addition, only one growth path was applied. A limited sensitivity analysis
finds that a small number of countries face enormous variability in the costs of adapting to climate
change given the uncertainty about the extent and nature of climate change. Moreover, the costs of
managing these risks could be substantially higher.

•

Climate science tells us that the impacts will increase over time and that major effects such as melting
of ice sheets will occur further into the future. Even so, the study opted for projecting what is known
today with greater certainty rather than making even less reliable longer-term estimates. Thus the
investment horizon of this study is 2050 only. A longer time horizon would increase total costs of
adaptation.

•

The study looks only at additional public sector (budgetary) costs imposed by climate change, not the
costs incurred by individuals and private agents. Similarly, the study generally opted for hard
adaptation measures that require an engineering response rather than an institutional or behavioral
response. Soft adaptation measures often can be more effective and can avoid the need for more
expensive physical investment. But as a first-cut global study, it was not possible to know whether
effective institutions and community-level collective action, which are preconditions for the
implementation of soft actions, exist in a given setting. While incorporating private adaptation would
increase cost estimates, including soft measures could potentially decrease them.

•

Other limitations include not being able to incorporate innovation and technical change; leaving out
local-level impacts, particularly the incidence on more vulnerable groups and the distributional
consequences of adaptation; not examining migration; and only partially accounting for adaptation
costs related to ecosystem services because of gaps in scientific understanding of the impact of
climate change on ecosystems. Relaxing the first of these limitations could lead to significant
reductions in adaptation costs, while a more comprehensive assessment of ecosystem services would
lead to an increase.

Lessons and recommendations

Four lessons stand out from the study.

First, adaptation to a 2

C warmer world will be costly. The study puts the cost of adapting between 2010

and 2050 to an approximately 2

C warmer world by 2050 at $75 billion to $100 billion a year. The

estimate is in the upper range of existing estimates, which vary from $4 billion to $109 billion. Although
the estimate involves considerable uncertainty (especially on the science side), it gives policymakers—for
the first time—a carefully calculated number to work with. The value added of the study lies in the
consistent methodology used to estimate the cost of adaptation—in particular, the way the study
operationalizes the concept of adaptation.

Second, the world cannot afford to neglect mitigation. Adapting to an even warmer world than the 2

assumed for the study—on the order of 4

C above pre-industrial levels by the end of the century—would

be much more costly. Adaptation minimizes the impacts of climate change, but it does not tackle the
causes. If we are to avoid living in a world that must cope with the extinction of half of its species, the
inundation of 30 percent of coastal wetlands, and a large increase in malnutrition and diarrheal and
cardio-respiratory diseases, countries must take steps immediately to sharply reduce greenhouse gas
emissions.

Third, development is imperative, but it must take a new form. Development is the most powerful form
of adaptation. It makes economies less reliant on climate-sensitive sectors, such as agriculture. It boosts
the capacity of households to adapt by increasing levels of incomes, health, and education. It enhances the
ability of governments to assist by improving the institutional infrastructure. And it dramatically reduces
the number of people killed by floods and affected by floods and droughts. But adaptation requires that
we go about development differently: breeding crops that are drought and flood tolerant, climate-proofing

infrastructure, reducing overcapacity in the fisheries industry, and accounting for the uncertainty in future
climate projections in development planning.

Countries may have to shift patterns of development or manage resources in ways that take account of the
potential impacts of climate change. Often, the reluctance to change reflects the political and economic
costs of changing policies and (quasi-) property rights that have underpinned decades or even centuries of
development. Countries experiencing rapid economic growth have an opportunity to reduce the costs
associated with the legacy of past development by ensuring that future development takes account of
prospective changes in climate conditions. The clearest, and probably most rewarding, opportunities to
reduce adaptation costs lie in the water sector, with coastal and flood protection. But other sectors also
stand to benefit.

Fourth, uncertainties are large, so robust and flexible policies and more research are needed. The
imprecision of models projecting the future climate is the major source of uncertainty and risk for
decision makers. Thus, it is crucial to undertake research, collect data, and disseminate information so
that if climate change turns out to have worse impacts than anticipated in 20 or 30 years, countries can
respond more quickly and effectively. In the meantime, countries should pursue low-cost policies and
investments on the basis of the best or median forecast of climate change at the country level. At the same
time, countries should avoid making investments that will be highly vulnerable to adverse climate change
outcomes. For durable climate-sensitive investments, strategies should maximize the flexibility to
incorporate new climate knowledge as it emerges. Hedging against varying climate outcomes, for
example by preparing for both drier and wetter conditions for agriculture, would raise the cost of adapting
well beyond what has been estimated here.

Section 1. B ackground and M otivation

All countries, developing and developed, need to adapt to climate change. Even if global emissions of
greenhouse gases are drastically reduced and concentrations are stabilized at 450 parts per million (ppm)
of equivalent carbon dioxide (CO

e), the annual global mean average temperature is expected to be 2

above pre-industrial levels by the middle of the century.

With a 2

C rise will come a higher incidence of

intense rainfall events and a greater frequency and intensity of droughts, floods, heat waves, and other
extreme weather events. Households, communities, and planners will need to take measures that “reduce
the vulnerability of natural and human systems against actual and expected climate change effects” (IPCC
2007, p. 3). Development will require such adaptation, and development progress may even be reversed
as the increased incidence of extreme weather events and rising sea levels results in higher mortality and
loss of assets, drawing resources from development; as greater incidence of infectious and diarrheal
diseases reverses development gains in health standards; and as temperature and precipitation changes
reduce agricultural productivity and the payoffs from agricultural investments.

While countries need to adapt to manage the unavoidable, decisive mitigation is required to avoid
the unmanageable. Unless the world begins immediately to substantially reduce greenhouse gas
emissions, annual global mean average temperature will rise by some 2.5–7

C over pre-industrial levels

by the end of the century. Temperature increases of more than 2

C will substantially increase the

likelihood of irreversible and potentially catastrophic impacts such as the extinction of half of all species,
inundation of 30 percent of coastal wetlands, and massive increases in malnutrition and diarrheal and
cardio-respiratory diseases (World Bank 2010). Even with government interventions, societies and
ecosystems will not be able to adapt to impacts of this magnitude. Mitigation, to avoid a further rise in
greenhouse gas emissions, is the only way to deal with climate change that is not already inevitable.

Adaptation will be costly, but there is little information about just how costly. Under the Bali Action
Plan adopted at the 2007 United Nations Climate Change Conference, developed countries agreed to
allocate “adequate, predictable, and sustainable financial resources and [to provide] new and additional
resources, including official and concessional funding for developing country parties” (UNFCCC 2008)
to help them adapt to climate change. The plan views international cooperation as essential for building
capacity to integrate adaptation measures into sectoral and national development plans. Yet studies on the
costs of adaptation (discussed in more detail later in the report) offer a wide range of estimates, from $4
billion to $109 billion a year. A recent critique of existing estimates suggests that these may be substantial
underestimates (Parry and others 2009). Similarly, National Adaptation Programmes of Action, developed
by the Least Developed Countries under Article 4.9 of the United Nations Framework Convention on
Climate Change (UNFCCC), identify and cost only urgent and immediate adaptation measures and do not
incorporate the measures into long-term development plans.

With current greenhouse gas concentrations at about 400 parts per million, annual average global

temperature is already 0.8

C above pre-industrial levels.

Mitigation is not discussed in this report, which focuses on adaptation.

Section 2. Study Objectives and Structure

The EACC study has two broad objectives: to develop a global estimate of adaptation costs for informing
the international community’s efforts to help the developing countries most vulnerable to climate change
meet adaptation costs, and to help decisionmakers in developing countries assess the risks posed by
climate change and design strategies for adapting to climate change. That requires costing, prioritizing,
sequencing, and integrating robust adaptation strategies into development plans and budgets. And it
requires strategies to deal with high uncertainty, potentially high future damages, and competing needs for
investments for social and economic development.

Supporting developing country efforts to design adaptation strategies requires incorporating country-
specific characteristics and sociocultural and economic conditions into analyses. Providing macro-level
information to developed and developing countries to support international negotiations and to identify
the overall costs of adaptation to climate change requires analysis at a more aggregate level. Reconciling
the two needs involves a tradeoff between the specifics of individual countries and a global picture.

The methodology developed for this study met both objectives by linking the country-level analysis with
the analysis for estimating the global costs of adaptation. Initially, the intention was to use country case
studies to develop unit least costs of adaptation and then to apply them to similar adaptation conditions in
other developing countries. As the country level analysis got under way, however, it became clear that
generalizing from the seven country cases (the seven partnering countries) would not work. A two-track
approach—a global track to meet the first study objective and a case study track to meet the second—
would yield a more robust estimate.

For the global track, country-level data sets with global coverage are used to estimate adaptation costs for
all developing countries by sector—infrastructure, coastal zones, water supply and flood protection,
agriculture, fisheries and ecosystem services, human health, and forestry. The cost implications of
changes in the frequency of extreme weather events are also considered. For most sectors, a consistent set
of future climate and precipitation projections are used to establish the nature of climate change, and a
consistent set of GDP and population projections are used to establish a baseline of how development
would look in the absence of climate change. This information is used to establish economic and social
impacts and the costs of adaptation (left side of figure 1).

For the country track, the impacts of climate change and adaptation costs are being established only for
the major economic sectors in each case study country (see right side of figure 1). To complement the
global analysis, vulnerability assessments and participatory scenario development workshops are being
used to highlight the impact of climate change on vulnerable groups and to identify appropriate adaptation
strategies (see box 1). Macroeconomic analyses are being used to integrate the sectoral analyses and to
identify cross-sector effects, such as relative price changes. Finally, in two country case studies (Bolivia
and Samoa), an investment model is being developed to prioritize and sequence adaptation measures (see
box 2).

Figure 1. Economics of Adaptation to Climate Change study structure: global and country tracks

Global track

Country track

Source: Economics of Adaptation to Climate Change study team.

The two tracks are intended to inform each other, to improve the overall quality of the analysis. This
report presents the methodology and the results for the global track. The report for the case study track
will be released early in 2010, by which time lessons from the country studies will be used to validate and
improve the estimate of total adaptation costs, resulting in a final report of the global track in early 2010.

Though the current report has undergone intensive review, with an internal World Bank review of the
concept note, methodology note, and draft report and reviews of draft sector chapters by an external and
an internal expert, the current report is nonetheless considered a consultation draft. Revisions to account
for comments received during the consultation process with a wide range of stakeholders will also be
incorporated in the final report.

Box 1. Understanding what adaptation means for the most vulnerable social groups

The negative impacts of climate change will be experienced most intensely by the poorest people in
developing countries. Just as development alone will not be enough to equip all countries or regions to
adapt to climate change, neither do all individuals or households within a country or region enjoy the
same levels of adaptive capacity (Mearns and Norton forthcoming). Drivers of physical, economic, and
social vulnerability (socioeconomic status, dependence on natural resource based livelihood sources, and
physical location, compounded by factors that shape social exclusion such as gender, ethnicity, and
migrant status) act as multipliers of climate risk for poor households. Social variables further interact with
institutional arrangements that are crucial in promoting adaptive capacity, including those that increase
access to information, voice, and civic representation in setting priorities in climate policy and action
(World Bank 2010).

Work is under way in six developing countries (Bangladesh, Plurinational State of Bolivia, Ethiopia,
Ghana, Mozambique, and Vietnam) under the EACC study to understand what adaptation means for
social groups that are most vulnerable to the effects of climate change and what external support they
need to help them take adaptation measures. This social component of the study combines vulnerability

assessments in selected geographic hotspots with facilitated workshops applying participatory scenario
development approaches. In the workshops, participants representing the interests of vulnerable groups
identify preferred adaptation options and sequences of interventions based on local and national climate
and economic projections. This approach complements the sectoral analyses of the costs of climate
change adaptation in those countries. The findings on what forms of adaptation support various groups
consider to be most effective—including “soft” adaptation options such as land use planning, greater
public access to information, institutional capacity building, and integrated watershed management—have
implications for the costs of adaptation. While this work is ongoing, some preliminary results from the
country investigations in Bangladesh, Bolivia, Ethiopia, Ghana, and Mozambique are presented
throughout this report to illustrate the range of adaptation options that are being suggested.

Box 2. Climate-resilient investment planning

A three-step methodology has been developed to help planners integrate climate risk and resilience into
development policies and planning. The first is to identify and validate climate-resilient investment
alternatives using a multicriteria decision analysis. This involves qualitative and quantitative impact
assessments for each sector, consultation at the national level (government, policymakers, technical
experts), and participatory workshops with community representatives and local authorities at the county
level. The second step is to conduct a cost-benefit analysis for identified climate-resilient investment
alternatives at a specific geographic unit. The final step is implementation of an investment planning
model that allows the government to prioritize and sequence robust adaptation strategies into
development plans and budgets.

Section 3. Operational Definition of Adaptation C osts

One of the biggest challenges of the study has been to operationalize the definition of adaptation costs.
The concept is intuitively understood as the costs incurred by societies to adapt to changes in climate. The
Intergovernmental Panel on Climate Change (IPCC) defines adaptation costs as the costs of planning,
preparing for, facilitating, and implementing adaptation measures, including transaction costs. But this
definition is hard to operationalize. For one thing, “development as usual” needs to be conceptually
separated from adaptation. That requires deciding whether the costs of development initiatives that
enhance climate resilience ought to be counted as part of adaptation costs. It also requires deciding how to
incorporate in those costs the adaptation deficit, defined as countries’ inability to deal with current and
future climate variability. It requires defining how to deal with uncertainty about climate projections and
impacts. And it requires specifying how potential benefits from climate change in some sectors and
countries offset, if at all, adaptation costs in another sector or country.

L inks between adaptation and development

The climate change literature examines several links between adaptation and development. Many studies
argue that economic development is the best hope for adaptation to climate change: development enables
an economy to diversify and become less reliant on sectors such as agriculture that are most likely to be
vulnerable to the effects of climate change. Development also makes more resources available for abating
risk. And often the same measures promote development and adaptation. For example, progress in
eradicating malaria helps countries develop and also helps societies adapt to the rising incidence of
malaria that may accompany climate change.

Adaptation to climate change is also viewed as essential for development: unless agricultural societies
adapt to changes in temperature and precipitation (through changes in cropping patterns, for example),
development will be delayed. Finally, adaptation requires a new type of climate-smart development that
makes countries more resilient to the effects of climate change. Urban development without attention to
drainage, for example, will exacerbate the flooding caused by heavy rains.

These links suggest that adaptation measures range from discrete adaptation (interventions for which
“adaptation to climate change is the primary objective”; WRI 2007) to climate-smart development
(interventions to achieve development objectives that also enhance climate resilience) to development not
as usual (interventions that can exacerbate the impacts of climate change and that therefore should not be
undertaken). Since the Bali Action Plan calls for “new and additional” resources to meet adaptation costs,
this report defines adaptation costs as additional to the costs of development. Consequently, the costs of
measures that would have been undertaken even in the absence of climate change are not included in
adaptation costs, while the costs of doing more, doing different things, and doing things differently are
included.

Defining the adaptation deficit

Adaptation deficit has two meanings in the literature on climate change and development. One captures
the notion that countries are underprepared for current climate conditions, much less for future climate
change. Presumably, these shortfalls occur because people are underinformed about climate uncertainty
and therefore do not rationally allocate resources to adapt to current climate events. The shortfall is not
the result of low levels of development but of less than optimal allocations of limited resources resulting

in, say, insufficient urban drainage infrastructure. The cost of closing this shortfall and bringing countries
up to an “acceptable” standard for dealing with current climate conditions given their level of
development is one definition of the adaptation deficit (figure 2). The second, perhaps more common, use
of the term captures the notion that poor countries have less capacity to adapt to change, whether induced
by climate change or other factors, because of their lower stage of development. A country’s adaptive
capacity is thus expected to increase with development. This meaning is perhaps better captured by the
term development deficit.

Figure 2. A simplified interpretation of adaptation deficit

Source: Economics of Adaptation to Climate Change study team.

The adaptation deficit is important in this study for establishing the development baseline from which to
measure the independent, additional effects of climate change. For example, should the costs of climate-
proofing infrastructure be measured relative to current provisions or to the levels of infrastructure
countries would have had if they had no adaptation deficit? Because the adaptation deficit deals with
current climate variability, the cost of closing the deficit is part of the baseline and not of the adaptation
costs. Unfortunately, except in the most abstract modeling exercises, the costs of closing the adaptation
deficit cannot be made operational (see box 3). This study therefore does not estimate the costs of closing
the adaptation deficit and does not measure adaptation costs relative to a baseline under which the
adaptation deficit has been closed.

It is not obvious whether analyses that take a different approach and measure costs of adaptation relative
to a baseline in which the adaptation deficit has been closed would estimate higher or lower adaptation
costs. In infrastructure, for example, closing the adaptation deficit implies that a larger stock of
infrastructure assets need be to climate-proofed, so closing the deficit in this sector could increase
adaptation costs. In contrast, closing the adaptation deficit in agriculture might imply a lower percentage
of rain-fed agriculture and therefore a lower impact of climate-change-induced droughts. Adaptation costs
are likely to be reduced in the agricultural sector as a result. Analyses that include the costs of closing the
adaptation deficit in the costs of adaptation are likely to estimate higher adaptation costs than those in this
study.

Additional capacity needed to

handle future climate change

ADAPTATION COSTS

Capacity to address current

climate variation

ADAPTATION DEFICIT

Appropriate capacity
to deal with current
climate variation

Appropriate capacity to deal

with future climate change

Box 3. Difficulties in operationalizing the adaptation deficit

Determining an acceptable level of adaptation to current climate variability is challenging. Some
observers consider the cost of closing the adaptation deficit as the cost of making all developing
countries—whatever their level of development—as prepared for current climate events as developed
countries are. Others argue that the amount countries spend should depend on conditions in the country.
For example, a poor country may devote fewer resources (than a rich country) on preventing loss of
lives from storm surges and more resources on fighting malaria if more lives can be saved for the same
amount of resources.

Because these hard choices are necessary in a resource-constrained world, differences in the amount of
resources devoted to adapting to current climate variability cannot be used as a proxy for the adaptation
deficit. Establishing the existence of an adaptation deficit requires first establishing that the benefit-cost
ratio of expenditures in climate-sensitive areas exceed those of expenditures in all other sectors. Then
estimating the size of the adaptation deficit requires estimating the degree of government
underspending in climate-sensitive areas relative to all other areas of the economy. Deficits for all
developing countries would then need to be estimate to estimate the “global” adaptation deficit—
clearly not feasible.

E stablishing the development baseline

Establishing the magnitude of the adaptation deficit is not relevant for this study. Establishing the
development baseline is. This is done sector by sector and assumes that countries grow along a
“reasonable” development path. In agriculture, it is done by imposing exogenous, reasonable growth
conditions on current development achievements, such as exogenous productivity growth, area expansion,
and investments in irrigation. In other sectors, such as infrastructure, the baseline is established by
considering historical levels of infrastructure provision, such as paved road density and length of sewer
pipes, in countries at different levels of development. Table 1 shows the definition of the development
baseline adopted for each sector.

Table 1. Definition of development baseline, by sector

Sector

Development Baseline

Infrastructure

Average sector performance by income groups

Coastal zones

Efficient protection of coastline

Water supply and flood
protection

Average municipal and industrial water demand by income groups;
efficient protection against monthly flood with given return period

Agriculture

Exogenous productivity growth, area expansion, investment in
irrigation

Fisheries

Maintenance of 2010 fish stocks

Human health

Health standard by income groups

Forestry and ecosystem
services

Not established

Extreme weather events

GDP-induced changes in mortality and numbers affected

a. For reasons discussed in section 5, development baselines were not established for this sector.
Source: Economics of Adaptation to Climate Change study team.

H ow much to adapt

The next issue is how much to adapt. One possibility is to adapt completely, so that society is at least as
well off as it was before climate change. At the other extreme, countries could choose to do nothing,
experiencing the full impact of climate change. Or countries could invest in adaptation using the same
criteria as for other development projects, investing until the marginal benefits of the adaptation measure
exceed the costs, which could lead to either to an improvement or a deterioration in social welfare relative
to a baseline without climate change.

How much to adapt is consequently an economic problem—how to allocate resources to adapt to climate
change while also meeting other needs. And herein lies the challenge. Poor urban workers who live in a
fragile slum dwelling might find it difficult to decide whether to spend money to strengthen their hut to
make it less vulnerable to more intense rainfall, or to buy school books or first-aid equipment for their
family—or how to allocate between the two. Poor rural peasants might find it difficult to choose between
meeting these basic education and health needs and some simple form of irrigation to compensate for
increased temperatures and their impact on agricultural productivity. These examples suggest that
desirable and feasible levels of adaptation depend on both available income and other resources.

Corresponding to a chosen level of adaptation is an operational definition of adaptation costs. If the policy
objective is to adapt fully, then the cost of adaptation can be defined as the minimum cost of adaptation
initiatives needed to restore welfare to levels prevailing before climate change. Restoring welfare may be
prohibitively costly, however, and policymakers may choose an efficient level of adaptation instead.
Adaptation costs would then be defined as the cost of restoring pre-climate change welfare standards to
levels at which marginal benefits exceed marginal costs. Because welfare would not be fully restored,
there would be residual damage from climate change after allowing for adaptation.

In this study, largely due to limitations of existing models, adaptation costs are generally defined as the
costs of development initiatives needed to restore welfare to levels prevailing before climate change and
not as optimal levels of adaptation plus residual damage (to the extent that residual damages are
compensated, original welfare is restored). The one exception is coastal zones, where adaptation costs are
defined as the cost of measures to establish the optimal level of protection plus residual damage. This
study assumption is expected to bias the estimates upwards.

Since costs are estimated by sector, sectoral proxies for welfare were identified (table 2). In agriculture,
for example, welfare is defined by the number of malnourished children and per capita calorie
consumption.

Table 2. Welfare proxies for defining sectoral adaptation costs

Sector

Welfare proxy

Infrastructure

Level of services

Coastal zones

Optimal level of protection plus residual damage

Water supply and flood management

Level of industrial and municipal water availability ;
availability of flood protection

Agriculture

Number of malnourished children and per capita calorie
consumption

Fisheries

Level of revenue

Human health

Health standard defined by burden of disease

Forestry and ecosystem services

Stock of forests; level of services

Extreme weather events

Number of deaths and people affected

Source: Economics of Adaptation to Climate Change study team.

Adapt to what? Uncertainty about climate outcomes

Operationalizing adaptation costs requires dealing with the considerable uncertainty about future climate
projections. Studies indicate that annual global mean average temperatures will increase (with a 2

increase by 2050 now considered inevitable), rainfall will become more intense in most places and
possibly less frequent, sea levels will rise, other extreme climate events will become more frequent and
more intense, and regional climate systems such as the El Niño Southern Oscillation phenomenon and the
Asian monsoon will be altered.

While there is considerable consensus among climate scientists on these general outlines of climate
change, there is much less agreement on how climate change will affect a given location. Maps 1 and 2
give a glimpse of this uncertainty for two global climate models—that of the Commonwealth Scientific
and Industrial Research Organization (CSIRO) and that of the National Centre for Atmospheric Research
(NCAR)—for the A2 scenario (“storyline”) of the IPCC Special Report on Emissions Scenarios (SRES).
These maps illustrate qualitatively the range of potential climate outcomes with current modeling
capabilities and thus are indicative of the uncertainty in climate change impacts. For example, the NCAR
model has substantially higher average maximum temperatures than does the CSIRO model and a larger
average increase in precipitation on land. The CSIRO model has substantial precipitation declines in the
western Amazon, while NCAR shows declines in the eastern Amazon. CSIRO has substantial
precipitation declines in Sub-Saharan Africa, while NCAR has increases there.

Map 1. Projected change in average maximum temperature based on two climate models, 2000–50

Commonwealth Scientific and Industrial

Research Organization (CSIRO), driest

scenario

National Centre for Atmospheric Research

(NCAR), wettest scenario

Note: Projections are based on the A2 scenario of the IPCC Special Report on Emissions Scenarios (SRES).
The Economics of Adaptation to Climate Change study team acknowledges the Program for Climate Model
Diagnosis and Intercomparison and the World Climate Research Programme's (WCRP) Working Group on
Coupled Modelling for their roles in making available the WCRP’s Coupled Model Intercomparison Project
phase 3 (CMIP3) multimodel dataset. Support of this dataset is provided by the Office of Science, U.S.
Department of Energy.

Source: Maps are based on data developed at the MIT Joint Program for the Science and Policy of Global
Change using CMIP3 data (the WCRP’s CMIP3) multimodel dataset. Maps were produced by the
International Food Policy Research Institute.

Map 2. Projected change in average annual precipitation based on two climate models,
2000–50

Commonwealth Scientific and Industrial

Research Organization (CSIRO), driest

scenario

National Centre for Atmospheric Research

(NCAR), wettest scenario

Note: Projections are based on the A2 scenario of the IPCC Special Report on Emissions Scenarios (SRES). The
Economics of Adaptation to Climate Change study team acknowledges the Program for Climate Model Diagnosis
and Intercomparison and the World Climate Research Programme's (WCRP) Working Group on Coupled Modelling
for their roles in making available the WCRP’s Coupled Model Intercomparison Project phase 3 (CMIP3)
multimodel dataset. Support of this dataset is provided by the Office of Science, U.S. Department of Energy.

Source: Maps are based on data developed at the MIT Joint Program for the Science and Policy of Global Change
using CMIP3 data (the WCRP’s CMIP3) multimodel dataset. Maps were produced by the International Food Policy
Research Institute.

Large-scale discontinuities create even greater uncertainty. Most uncertain are risks related to systemic
changes, such as the melting of the Greenland and West Antarctic ice sheets, the collapse of the Atlantic
thermohaline circulation, and the die-back of the Amazon, all hard to predict and subject to sudden
threshold changes that can trigger potentially irreversible processes. The precise timing and level of these
triggers cannot be projected with confidence, but the science is clear that these risks are substantial.

Such inherent uncertainties in climate projections suggest that a range of adaptation costs should be
estimated for a range of climate scenarios. They also suggest that policymakers will have to hedge when
making decisions with long-term consequences, weighing the current costs of investments against their
benefits over a wide range of potential climate outcomes (see box 4).

The EACC has calculated the range

of adaptation costs over wet (CSIRO) and dry (NCAR) scenarios to bracket adaptation costs between the
two extreme scenarios. In the real world, where decisionmakers must hedge against a range of outcomes,
actual expenditures are potentially much higher than these estimates.

Box 4. Taking climate uncertainty into account: how should national policymakers interpret
global numbers?

Total adaptation costs for a specific climate projection are an estimate of the costs the world would
incur if policymakers knew with certainty that that particular climate projection would materialize. But
national policymakers do not have such certainty. At present, climate scientists agree that no climate
model projection can be considered more likely than another. The current disparities in precipitation
projections mean, for example, that ministers of agriculture have to consider the risks of both the
wettest and the driest scenarios and thus whether to invest in irrigation to cope with droughts or in
drainage to minimize flood damage, while urban planners in flood-prone areas have to decide whether
to build dikes (and how high) without knowing whether the future will be wetter or drier.

The EACC has calculated the range of adaptation costs over wet (CSIRO) and dry (NCAR) scenarios
to bracket adaptation costs between the two extreme scenarios. This provides a range of estimates for a
world in which decisionmakers have perfect foresight. In the real world, where decisionmakers must
hedge against a range of outcomes, actual expenditures are potentially much higher than these
estimates. With such high costs involved, improving the certainty of the climate model projections is
urgent, as are strategies that permit decisionmakers to remain flexible until better climate information is
available.

Summing potential costs and benefits

This study estimates adaptation costs relative to a baseline of what would have happened in the absence
of climate change. One possible outcome is that changes in climate lead to lower investment or
expenditure requirements for some sectors in some countries—for example, changes in demand for
electricity or water that reduce requirements for electricity generating capacity, water storage, and water
treatment. In these cases, the “costs” of adaptation are negative. This is straightforward, but it gives rise to

another question: how should positive and negative costs be summed across sectors or countries? It is
easy to envisage that higher expenditures on coastal protection could be offset by lower expenditures on
electricity generation in the same country, but it is unlikely that higher expenditures on electricity
generation in country A can be offset by lower expenditures in the same sector in country B.

How then to

define aggregates that add up consistently across sectors and countries?

Box 5 illustrates three options for summing positive and negative costs when there are restrictions on
offsetting negative and positive items: gross, net, and X-sums. Under the gross aggregation method,
negative costs in any sector in a country are set to zero before costs are aggregated for the country and for
all developing countries. Under X-sums positive and negative items are netted within countries but not
across countries, and costs for a country are included in the aggregate as long as the net cost across
sectors is positive for the country. In the net aggregate measure, negative costs are netted within and
across countries. The net calculation is carried out by decade. Of 146 developing countries, 10 have
negative net adaptation costs in at least one decade across all sectors with the CSIRO scenario and 5 with
the NCAR scenario. Most of these countries are landlocked, buffering them from the substantial costs for
coastal protection that constitute a large part of the adaptation costs for coastal countries.

All three options are used in the study to estimate adaptation costs, though costs are mainly reported as X-
sums in the interest of space.

A simple example helps to illustrate the situation. Suppose that Brazil has a positive cost in both

agriculture and water, meaning that both sectors will be negatively affected by climate change (relative to
the no-climate-change scenario), and suppose that India has a negative cost in agriculture and a positive
cost in water, meaning that agriculture benefits but the water sector suffers from climate change. It may
be reasonable to assume that in India the gains in agriculture can compensate to some extent for the losses
in the water sector. But it is unlikely that Brazil will be compensated by India because Brazil incurs a cost
and India a benefit in the agriculture sector.

Box 5. Calculating aggregate costs—gross, net, and X-sums

In summing positive and negative adaptation costs across countries, whether for a single sector or all
sectors, three types of aggregate can be constructed (as illustrated by the hypothetical figures in the
table).

Summing positive and negative adaptation costs

Sector and
type of
aggregate

Country

Sector aggregate

Sector

gross

Sector net

Sector X-

sum

Sector 1

—

Sector 2

–4

–2

—

Sector 3

–4

—

Country gross

—

Country net

–4

—

Country X-sum

—

— is not applicable.

Gross sum. The gross sum represents the aggregate costs incurred by countries with positive costs for
a particular sector, ignoring all country and sector combinations resulting in negative costs. One
difficulty with gross sums is that the results vary depending on how sectors are defined. This can be
illustrated by recalculating the gross sums after combining sectors 1 and 2, giving an overall sectoral
gross sum of 18 rather than 22, even though nothing else has changed (not shown in table).

Net sum. The net sum treats positive and negative values symmetrically. It represents the pooled costs
incurred by each country or each sector without restrictions on pooling across country borders.

X-sum. X-sums take account of restrictions on pooling across countries, so all entries for a given
country are set to zero if the net sum for the country is negative (see country C in the table).

For the hypothetical data in the table, the overall gross sum is 22, and the overall net sums is 12. The
difference between the two values is the absolute value of negative entries for sectors 2 and 3 in
countries B and C. The overall X-sum, which must fall between the overall gross and net sums, is 16.
The difference between the overall X-sum and the overall net sum is 4, equal to the loss of pooling
because of the net negative cost for country C.

Section 4. M ethodology and V alue Added

Although the methodology used to estimate the impacts of climate change and the costs of adaptation is
specific to each sector, the sectoral methodologies share several elements. Adaptation costs in most
sectors were calculated for 2010–50 from a common trajectory of population and GDP growth used to
establish the development baseline and a common set of global climate models used to simulate climate
effects. For all sectors, adaptation costs include the costs of planned, public policy adaptation measures
and exclude the costs of private adaptation. For agriculture, for example, the methodology allows for the
effects of autonomous adjustments in the private sector, such as changes in production, consumption, and
trade flows in response to world price changes, but does not include the costs of those adjustments in
adaptation costs. These common methodological elements, along with wide and in-depth sectoral
coverage and a consistent definition of adaptation costs, allow the study to substantially improve on
earlier estimates (box 6).

Box 6. Previous estimates of global adaptation costs

World Bank (2006). The first estimate of costs of adaptation to climate change for developing countries
was produced by the World Bank in 2006. Its report defined adaptation costs as the cost of climate-
proofing three categories of investment flows: official development assistance and concessional finance,
foreign direct investment, and gross domestic investment. The study defined the proportion of total
investments in each category that was likely to be climate sensitive and then estimated the percentage
increases in costs to climate-proof these investments. Adaptation cost estimates ranged from $9 billion to
$41 billion a year.

Stern (2007) and UNDP (2007). Using the same methodology as World Bank (2006) but different values
for the proportion of climate-sensitive investments and the increases in costs for climate-proofing
investments, the Stern Report (Stern 2007) estimated costs of adaptation of $4– $37 billion a year by
2050, somewhat lower than the World Bank estimate, while Human Development Report 2007/2008
(UNDP 2007) estimated costs of $5–67 billion a year by 2015, somewhat higher than the World Bank
estimate. In addition to the cost of climate-proofing investments, Human Development Report 2007/2008
also estimated that by $40 billion a year would be needed by 2015 to strengthen social protection
programs and scale up aid in other key areas and $2 billion a year to strengthen disaster response systems,
boosting overall adaptation costs to $47–109 billion a year by 2015.

Oxfam International (2007). In contrast to this these top-down approaches, Oxfam International (2007)
used a bottom-up approach, estimating adaptation costs by assessing National Action Plans for
Adaptation and the costs of adaptation projects initiated by nongovernment organizations. Assuming
average warming of 2

C, the report estimated global adaptation costs of at least $50 billion a year: $7.5

billion a year to support adaptation efforts initiated by nongovernmental organizations,

$8–33 billion a

year to meet the costs of the most urgent adaptation measures being proposed under the National Action
Plans for Adaptation, and $5–15 billion a year to address unknown and unexpected impacts. Though
richer in the range of potential adaptation measures, this methodology uses a small and likely
unrepresentative sample of projects and countries to generalize to all developing countries.

UNFCCC (2007). Whereas previous efforts considered only the costs of planned adaptation, the United
Nations Framework Convention on Climate Change study considered the costs of both planned and

private adaptation measures. Also, whereas previous studies had considered costs across all sectors, this
report estimated the costs of adaptation by major sectors (agriculture, forestry, and fisheries; water
supply; human health; coastal zones; and infrastructure), yielding total costs of $26–67 billion a year by
2030.

A recent critique of the UNFCCC estimates (Parry and others 2009) suggests that these estimates may be
too low because some sectors were excluded (ecosystems, energy, manufacturing, retailing, and tourism),
included sectors were not fully accounted for, climate-proofing of infrastructure stocks ignored the need
for additional stocks (financed through full funding of development) for handling current climate
variability, and residual damages (impacts remaining after adaptation) were not accounted for.

Project Catalyst (2009). The final estimate was produced in 2009 by the Climate Works Foundation’s
Project Catalyst initiative. This study estimated that annual average adaptation funding requirements for
developing countries lie between $15 billion and $30 billion for the period 2010–20 and between $30
billion and $90 billion by 2030. Softer measures, such as capacity building, planning, and research, are
the focus of adaptation policy in the first decade, followed by more expensive structural investments in
the second decade. Unlike previous estimates, the study accounts for potential co-benefits of adaptation
actions and reduces the cost estimate to reflect these benefits.

C hoosing the timeframe

The choice of timeframe for the analysis of the costs of adapting to climate change will likely affect the
overall cost estimates, with a longer timeframe producing higher costs than would a shorter one. The
timeframe up to 2050 was selected largely because forecasting climate change and its impacts on an
economy becomes even more uncertain beyond this period, and the complexity of the analysis favors
getting more precise (or less imprecise) estimates in the near term rather than less precise estimates over a
more extended timeline.

Related to the issue of timeframe is the choice of the discount rate, which is related to the timing of
investments. The timing of all investments in the sector models is determined by the outcomes of specific
climate projections. Given the expected climate outcome within the useful life of an investment, each new
investment must be designed to restore welfare (as defined in table 2) to levels that would have existed
without climate change. Because of the complexity of modeling sectors at a global level, none of the
sectoral models is capable of choosing the optimal timing of investments. This implies that the time-paths
of investments is insensitive to changes in the discount rate and therefore all results are presented for a
zero discount rate though costs have been expressed in 2005 constant prices. Obviously, discounting the
time stream of investment costs would lower the net present value of total investment or adaptation costs,
but it would not influence the choice of investments or the underlying investment costs. The inability to
model policy tradeoffs across time is a clear limitation imposed by the global nature of this study. The
selection of the discount rate and intertemporal choices will be explored in depth in some of the country
case studies.

Using baseline G DP and population projections to account for continuing

development

Most studies of adaptation to climate change hold developing countries at their current level of
development when estimating adaptation costs even over the medium term. Yet most developing
countries will become economically more advanced over the medium term, which will alter the economic

impact of climate change and affect the type and extent of adaptation needed. As already explained, the
EACC study accounts for the impact of development on estimates of adaptation costs by establishing
development baselines by sector (see table 1). These baselines establish a fictional growth path in the
absence of climate change that determines sectoral performance indicators, such as stock of infrastructure
assets, level of nutrition, and water supply availability. Climate change impacts and costs of adaptation
are examined in relation to this baseline.

Baselines are established across sectors using a consistent set of future population and GDP projections.
The population trajectory is aligned with the United Nations Population Division middle-fertility
projections for 2006. To ensure consistency with emissions projections, the GDP trajectory is based on
the average of the GDP growth projections of the three major integrated assessment models of global
emissions growth—Climate Framework for Uncertainty, Negotiation, and Distribution (FUND; Anthoff
and Tol 2008); PAGE2002 (Hope 2006); and Regional Dynamic Integrated Model of Climate and the
Economy (RICE99; Nordhaus 2001), and growth projections used by the International Energy Agency
and the Energy Information Administration of the US Department of Energy to forecast energy demand.
All these sources provide growth estimates at a regionally disaggregated level.

The global average annual real GDP per capita growth rate constructed in this way is 2.1 percent, similar
to global growth rates assumed in the United Nations Framework Convention on Climate Change
(UNFCCC) A2 emissions scenario from the IPCC 4

Assessment Report (AR4), once considered an

extreme scenario but no longer (IPCC 2007). The regionally downscaled GDP projections under different
IPCC scenarios (available from the Center for International Earth Science Information Network,
Columbia University) were not used because they are based on older data.

C hoosing climate scenarios and global climate models

Twenty-six global climate models provide climate projections based on the IPCC A2 Special Report on
Emission Scenarios (SRES) (see box 7). In this study, the National Center for Atmospheric Research
(NCAR) CCSM3 and Commonwealth Scientific and Industrial Research Organization (CISRO) Mk3.0
models were used to model climate change for the analysis of most sectors because they capture a full
spread of model predictions to represent inherent uncertainty and they report specific climate variables
(minimum and maximum temperature changes) needed for sector analyses. Though the model predictions
do not diverge much for projected temperature increases by 2050 (both projecting increases of
approximately 2

C above pre-industrial levels), they vary substantially for precipitation changes. Among

the models reporting minimum and maximum temperature changes, the NCAR was the wettest and the
CSIRO the driest scenario (globally, not necessarily the wettest and driest in every location) based on the
climate moisture index. Climate projections for these two models were created at a 0.5 by 0.5 spatial
degree scale and a monthly time scale by applying model predictions through 2050 to a historical climate
baseline obtained from the

University of East Anglia

Climate Research Unit’s Global Climate Database

time series 2.1.

Analysis was limited to two specific scenarios rather than the mean multiple of the global climate models
because the mean masks extreme values. A model average of near zero could be the result of models
predicting near-zero change, but just as well the result of two opposing changes that differ in sign. Using
a group of global climate models (multimodel ensembles), as opposed to one model, can somewhat
correct for biases and errors. The question with an ensemble approach is how to capture the full range of
results from model runs.

Box 7. Special Report on Emissions Scenarios of the Intergovernmental Panel on Climate Change

Adaptation requires understanding the potential impacts of climate change on human, economic, and
ecological systems. Yet attempts to estimate such impacts have to take on a cascade of uncertainty.
Uncertainty starts with the selection of an appropriate underlying emission scenario that is determined by
economic and population growth and by energy use choices. Will the world grow rapidly or slowly? Will
developing country populations soon adopt the consumption habits of high-income countries? And what
kind of energy future are we to look forward to? To account for these questions, the Intergovernmental
Panel on Climate Change (IPCC) has developed six socioeconomic scenarios that characterize possible
trajectories of emissions.

A scenario is a coherent, internally consistent, plausible description of a possible future state of the world.
It is not a forecast; rather, each scenario is one alternative image of how the future can unfold, given a
specific set of assumptions described in a set of four narrative storylines for the climate scenarios: A1
(focus on economic growth and globalization), A2 (regional focus), B1 (environmental focused), and B2
(regional focus). According to the IPCC, all families of scenarios from each storyline are equally valid,
with no assigned probabilities of occurrence.

The choice of climate and related nonclimate scenarios is important because it can determine the outcome
of a climate impact assessment. According to the IPCC, however, all scenarios have more or less the same
projected temperature increase up to 2050 (a timeframe arguably more relevant for adaptation), even
though there are large uncertainties regarding carbon dioxide emissions within each scenario. Therefore,
the selection of scenarios for this study depends largely on the availability of global climate model data as
well as some range of most “likely” future scenarios for the location of interest.

Selecting adaptation measures

Adaptation measures can be classified by the types of economic agent initiating the measure—public or
private. The literature distinguishes between autonomous or spontaneous adaptation (adaptation by
households and communities acting on their own without public interventions but within an existing
public policy framework) and planned adaptation (adaptation that results from a deliberate public policy
decision). This study focuses on planned adaptation. This focus is not to imply that autonomous
adaptation is costless. But since the objective is to help governments plan for risks, it is important to have
an idea of what problems private markets will solve on their own, how public policies and investments
can complement markets, and what measures are needed to protect public assets and vulnerable people.
For that, assessment of planned adaptation is needed.

In all sectors except extreme weather events, “hard” options involving engineering solutions are favored
over “soft” options based on policy changes and social capital mobilization (table 3). For adaptation to
extreme weather events, the emphasis is on investment in human resources, particularly those of women.
The decision to focus on hard options for the global cost assessment was motivated largely by fact that
these are easier to cost. Though hard adaptation options are feasible in nearly all settings, while soft
options depend on social and institutional capital, the focus on hard options is not to suggest that they are
always preferable. As discussed in box 8, adaptation measures being identified in the companion case
studies through participatory scenario workshops span both hard and soft measures. Since hard options
are typically more expensive than soft ones, this study assumption is likely to give the estimates an
upward bias.

Box 8. Adaptation measures identified in participatory workshops

Participants in the participatory scenario development workshop identified several cross-cutting climate
change impacts in the infrastructure, natural resource management and agriculture, health and education,
land tenure, governance and service delivery, and migration support sectors. Participatory scenario
development methods were particularly good at eliciting information on intersectoral linkages among
climate impacts and investments and the need for complementary investments. For example, female
farmers and others in a local workshop in Kalu, Ethiopia, noted the multiple effects of climate variability
on livelihood outcomes in the midland region. They noted that drought and water scarcity led to livestock
disease, human health impacts, and reduced household farm productivity and income, resulting in the
withdrawal of children from school, distress migration, and more detahs. Calls for adaptation support
included investments in watershed management, drought-resistant crop varieties, nonfarm diversification,
and capacity building. Local workshop participants in Xai-Xai, Mozambique, highlighted the different
income groups within broad sectoral categories (such as commercial producers and nontimber forest
collectors within agroforestry) and noted their varied preferences for adaptation investments (see table).
In addition, participants in both workshops identified not only vulnerable populations but also dynamic
processes of migration, urbanization, and market development that were leaving some households more
vulnerable than others.

Livelihood groups identified in southern Mozambique participatory scenario development
workshop

Sector

Income tiers

Key climate impacts

Select adaptation options

sought

Fishing

•

Commercial fishers

•

Artisanal fishers

•

Sea level rise,
abandonment of fishing

•

Increased salinity in
estuaries, reduced
fluvial fisheries

•

Introduction of new fish
species

•

Coastal zone pollution
reduction measures

Agroforestry

•

Harvesters (including
commercial harvesters)

•

Charcoal producers and
fuelwood collectors

•

Construction pole
gatherers

•

Nontimber forestry
product and food
gatherers

•

Cyclones, loss of
coastal vegetation,
ecosystem change

•

Floods destroying forest
access routes

•

Drought, increased
physical vulnerability
and species change

•

Reforestation and dune
protection

•

Improved road
construction planning

•

Community
involvement and
education

Trade and
commerce

•

Informal and formal
sector trading

•

Cyclones destroying
infrastructure and

•

Climate-proof
infrastructure;
improved early warning

•

Differential access to
market (seasonal
traders, retail traders,
wholesale)

displacing people

•

Sea level rise, coastal
erosion and reduced
land for development

systems

•

Improved erosion
control through public
works

Agriculture
and ranching

•

Large, medium, and
subsistence farmers
(both rainfed highland
farmers and lowland/
floodplain farmers with
irrigation)

•

Floods and droughts,
loss of production,
increased livestock
disease and death

•

Cyclones, loss of lives,
crops, infrastructure

•

Salinity intrusion

•

Barns for animals

•

Improved early warning
systems

•

Better siting of farms

•

Dam, floodgate
construction

Source: Xai-Xai, Mozambique, participatory scenario development workshop report.

In all sectors except extreme weather events, “hard” options involving engineering solutions are favored
over “soft” options based on policy changes and social capital mobilization (table 3). For adaptation to
extreme weather events, the emphasis is on investment in human resources, particularly those of women.
The decision to focus on hard options for the global cost assessment was largely motivated by fact that
these are easier to cost. Though hard adaptation options are feasible in nearly all settings, while soft
options depend on social and institutional capital, the focus on hard options is not to suggest that they are
always preferable. As discussed in box 8 adaptation measures being identified in the companion case
studies through participatory scenario workshops span both hard and soft measures. Since hard options
are typically more expensive than soft ones, this study assumption is likely to give the estimates an
upward bias.

Table 3. Types of adaptation measures considered, by sector

Sector

Adaptation measure

Infrastructure

Design standards, climate-proofing maintenance

Coastal zones

River and sea dikes, beach nourishment, port upgrades

Water supply and flood
protection

Reservoir storage, recycling, rainwater harvesting, desalination; flood
protection dikes and polders

Agriculture

Agricultural research, rural roads, irrigation infrastructure expansion and
efficiency improvements

Fisheries

Fisheries buybacks, individual transferable quotas, fish farming, livelihood
diversification measures, marine protected areas

Human health

Prevention and treatment of disease

Extreme weather events

Investment in human resources

Source: Economics of Adaptation to Climate Change study team.

Understanding the limitations of this study

Calculating the cost of adaptation for developing countries requires simplifying a complex problem
involving multiple countries, institutions, decisionmakers, and projections of government investments
into a world 40 years in the future. This requires constructing projections of economic growth, structural
change, climate change, and human behavior over a long time horizon and for numerous sectors. Subject
to these constraints, the study has adopted a consistent approach across countries and sectors and over
time, establishing a new benchmark for research of this kind.

To do so, however, several important assumptions and simplifications had to be made. The features and
limitations of the analysis for each sector are discussed in the sector analyses in section 5. This section
looks at five important limitations of the overall study methodology that arise from the need to simplify
the problem sufficiently to derive adaptation costs for all developing countries: characterization of
government decisionmaking environment, limited range of climate and growth outcomes, limited scope in
time and economic breadth, simplified characterization of human behavior, and top-down versus bottom
up analysis.

Stylized characterization of government decision-making environment

The characterization of government decisionmaking is the most problematic element of the study. As
have all other attempts to estimate the total costs of adaptation, this study calculates adaptation costs as if
decisionmakers knew with certainty what the future climate will be. In truth, current climate knowledge
does not permit even probabilistic statements about country-level climate outcomes and therefore
provides virtually no help in informing country-level decisionmakers’ investment decisions.

In fact, with current climate knowledge, country-level decisionmakers face a different problem—how to
maximize the flexibility of investment programs to take advantage of new climate knowledge as it
becomes available. While this decision problem can be explored at the country level, it is intractable in a
global study. Without the assumption of perfect foresight, it would be impossible to calculate adaptation
cost for developing countries in all but the most highly stylized and aggregated models. If such an
analysis were possible, though, costs of adaptation to climate change would likely be higher than those in
this study.

For most

durable investment decisions, decisionmakers know with certainty only that climate in the future will
differ from climate today. The adaptation costs calculated in this study and in all other global studies are
based on the fiction that decisionmakers know what future climate will be and act to prevent its damages.

L imited range of climate and growth outcomes

Even with this strongly stylized characterization of the decision problem, overall model complexity
permits systematic exploration of only a small range of potential outcomes. The two major drivers of
adaptation costs are climate outcome and economic growth. Of the 26 climate projections available for
the A2 SRES,

a complete assessment of adaptation costs was possible only with 2. Exploration of

Although some researchers have, as a practical expedient, constructed triangular probability densities to represent

the range of global climate model outcomes, most climate scientist would object to this use of their data.

alternative growth paths was even more restricted, with only one future applied across all sectors.

L imited scope in economic breadth and time

Sensitivity analysis was performed in various sectors, however (as described later). For climate outcomes,
sensitivity analysis suggests that one or two global climate models predict adaptation costs in several
South Asian countries that are orders of magnitude greater than those of the other climate models. For
growth, sensitivity analysis indicates that the results are much less sensitive than for climate outcomes, as
would be expected. While more growth increases the assets at risk, it raises incomes and reduces
vulnerability.

To make calculations tractable, the study had to limit both the breadth of economic analysis and the
length of the time horizon. For the economic analysis, this means that the study has estimated only the
additional public sector (budgetary) costs imposed by climate change, not overall economic damages.
These additional costs for the provision of public goods must not be confused with overall economic
damages and cannot be usefully compared with mitigation costs. The investment horizon of this study is
to 2050 only. Climate science tells us that adaptation costs and damages will increase over time, and that
major effects such as melting of major ice sheets are more likely to occur well beyond this horizon.

Simplified characterization of human behavior

H ard adaptation versus soft adaptation. Hardest of all to project is human behavior, especially
developments in institutions and the political economy. Many adaptive measures are best implemented
through effective collective action at the community level. However, the circumstances that elicit
effective collective action are complex (Ostrom 1990). “Soft” adaptation measures, such as early warning
systems, community preparedness programs, watershed management, urban and rural zoning, and water
pricing, generally rely on effective institutions supported by collective action. Because it is easier to cost
hard measure and because it is impossible to know, in a global study, whether such preconditions exist in
a given setting, this study has generally opted for hard adaptation measures that require an engineering
response.

M igration behavior. Decisions to migrate are also strongly mediated by community processes and social
capital. Because social processes that create poverty and marginality are more important determinants of
likely migration outcomes than are environmental changes themselves, in theory it should be possible to
reduce the likelihood of migration arising from climate change. However, in the absence of vastly
improved political and economic structures that can reduce poverty, environmental change will continue
to be an important proximate factor in migration decisions (box 9). The estimates in this study are based
on demographic projections by the United Nations Population Division that do not take climate change
into account. Population movements across countries may impose heavy infrastructure costs in areas

Not a recommendation, this is rather a simplifying assumption to make the study tractable. To

the extent that local institutions exist that can employ more effective and less expensive soft adaptation
measures, this assumption imparts an upward bias to the global cost.

However, the growth path used in this study represents a consensus growth path among climate modelers and is

chosen to be consistent with the emissions level underlying the A2 SRES.

An exception is the inclusion in the agriculture sector assessment of a number of soft measures such as water

harvesting in the adaptation measure “irrigation reform”.

receiving substantial numbers of migrants. This is, however, more likely to become a serious issue in the
second half of the century.

Box 9. Migration and climate change—Ghana’s experience

Climate change impacts are expected to induce large new migration flows. The number of environmental
migrants (people moving in response to environmental degradation, extreme events, or related economic
conditions; see Warner and others 2009)

is projected to rise in coming decades, with the vast majority

seeking residence in large cities.

Migration was a recurring theme in the EACC participatory scenario

development workshops, as well as in field-level investigations. This box highlights some key findings
from Ghana.

Drought in the northern savannah region of Ghana has long triggered migration to the country’s coastal
cities. Rural-urban migration creates vulnerabilities at a number of levels. New migrants live in informal
housing and often in peri-urban areas without services. They also typically lack social ties and access to
information in their new locations. Recent migrants to Accra reside in unplanned developments in highly
risky sites including flood-prone and malarial marshlands. Migrants are disproportionately young men,
leaving women, children, and the elderly to tend to agricultural lands and putting household farm
production and food security at risk because of lack of household labor.

Rural-rural migration also leads to problems, especially in land access and ownership for production.
Resource rights are tenuous for recent migrants—at least 80 percent of land in Ghana is administered
through customary law institutions including local chiefdoms that can be exclusionary. Focus group
participants in Dzatakpo, Ghana, stated that the local chief has given only land use rights to immigrants,
rather than full land rights. Immigrants in Buoyem, Ghana, were reluctant to plant long-gestation (and
higher value) crops because of insecure access to land. Sharecropping and use-right rules in the Western
Region of Ghana also impede sustainable land management. Because failure to clear a piece of forested
land for cultivation within two years of acquisition results in forfeiture, the rule leads to destruction of
forest resources. Despite rising numbers of migrants to the Western Region, this customary practice has
not changed, highlighting the slow pace of adaptation of some local institutions to changing
circumstances.

Key policy responses to environmental migration include social protection support to migrants, such as
easing place-based residence requirements for accessing social services; investing in sending regions so
as to reduce the flow of migrants, as Ghana is doing with its northern development strategy; and
considering rights-based resettlement for populations directly displaced by climate impacts, such as sea-
level rise.

T he efficiency of adaptation. Economic models normally assume fully rational behavior—producers
maximize profits, consumers maximize welfare, governments provide public goods using cost-benefit
criteria to choose the most efficient projects, and projects are implemented optimally through time to
maximizes the net present value of the government’s future investment stream. None of the sector models
used in this study is capable of intertemporal optimization. Calculations in each sector ensure that service
levels are maintained despite climate change, but no effort was made to identify whether the resources
invested in one sector to counter the effects of climate change would have yielded a higher benefit-cost
ratio in another sector (except in the sea-level rise component) or whether cash transfers would maintain
welfare at less cost. As a result, the adaptation costs calculated in this study are almost certainly

inefficient, even within the framework of the study. This simplification imparts an upward bias to the
adaptation costs.

I nnovation and technical change. Most parts of the study do not allow for the unknowable effects of
innovation and technical change on adaptation costs. In effect, these costs are based on what is known
today rather than what might be possible in 20–40 years. Sustained growth in per capita GDP for the
world economy rests on technical change, which is likely to reduce the real costs of adaptation over time.
This treatment of technological change also contributes to an upward bias in the calculated costs. The
exception is agriculture. Growth in total factor productivity in agriculture, based on historical trends and
expert opinion, is built into the model, and explicit investment in research is included in the costs.

T op-down or bottom-up analysis

In the final report of this study, this global approach will be supplemented by country case studies. But
this report on the global track relies on a mixed top-down sectoral approach to country analysis because
of the difficulty of generalizing from country studies when there is no clear basis for scaling up country
results. It is “mixed” because, for countries that are too large and too heterogeneous to be treated as a
single analytical unit, the basic analytical units include river basins and food production units. It would
have been preferable to estimate the costs of adaptation for infrastructure at the subnational rather than
national level in all countries with a population of, perhaps, 50–100 million or more. However, data
availability and economic consistency are difficult to ensure at the subnational level.

Section 5. K ey R esults

This section presents the key results of the EACC global track study of the costs of adaptation to
climate change for developing countries. Results by sector are followed by a discussion of
consolidated global costs and the results of sensitivity analysis.

Sector analyses

The sector analyses cover infrastructure, coastal zones, water supply and flood management,
agriculture, fisheries, human health, forestry and ecosystem services, and extreme weather events.

I nfrastructure

Adaptation costs for infrastructure assets have been one of the largest components of total
adaptation costs in past estimates—the largest in the UNFCCC (2007) study (the closest in
approach to this study). Previous studies have estimated adaptation costs for infrastructure as the
costs of climate-proofing new investment flows (see box 6). The percentage of new investment
flows likely to be climate sensitive is multiplied by the percentage increase in construction costs
(table 4). However, none of these studies provides a strong analytic basis for its choice of
parameter values for climate proofing. And none accounts for the costs of climate-proofing
existing stocks of capital.

Table 4. Estimates of adaptation costs for infrastructure from previous studies (billions)

Study

New

investment

flows ($

billions)

Percent of new

investment
sensitive to

climate

Additional costs to

reduce risk from

climate change

(percent)

Costs

($ billions)

World Bank (2006)

1,760

2–40

10–20

9–41

Stern (2007)

1,760

2–20

5–20

4–37

UNDP (2007)

3,112

2–33

5–20

5–67

UNFCCC (2007)

5,417

1–3

5–20

2–41

a. In 2000.
b. In 2005.
c. In 2030, backed out as mean of upper and lower bounds.
Source: Economics of Adaptation to Climate Change study team analysis of listed sources.

In this study, analysis of the infrastructure sector begins by projecting stocks of major types of
infrastructure over 2010–50 that would have existed under the development baseline without
climate change. Infrastructure services include transport (mainly roads, rail, and ports),
electricity, water and sanitation, communications, urban and social infrastructure such as urban
drainage, health and education facilities (rural and urban), and general public buildings.
Adaptation cost is computed as the additional cost of constructing and operating and maintaining
these baseline levels of infrastructure services under the new climate conditions projected by the
NCAR and CSIRO global climate models. This cost is referred to as the delta-P cost of adaptation
because it focuses on price and cost changes for fixed quantities of infrastructure (see box 10 for
details).

Considerable work went into developing infrastructure-specific dose-response relationships
between climate variables (dose) and the unit costs of construction (response) and between

climate variables (dose) and operation and maintenance (response), which were used to estimate
adaptation costs (table 5 presents details for one type of infrastructure, paved roads). For most
types of infrastructure, dose-response functions for construction costs captured adjustments in
building standards to enable assets to withstand predicted changes in climate conditions.
Standards, assumed to be forward looking, were adjusted to withstand changes for 50 years from
the date of construction, reflecting the typical life of infrastructure assets. Maintenance costs were
distinguished for existing assets in 2010 and for new assets constructed after 2010. Existing assets
require more maintenance and perhaps modification of short-lived components to cope with
climate stresses not taken into account when they were built, such as resurfacing roads or
replacing heating and cooling equipment. New assets, built to standards that take climate change
into account, require only normal maintenance. Finally, allowances were made for the impact of
climate change on the efficiency of power generation and water and sewage treatment—
particularly in response to higher maximum temperatures.

Table 5. Examples of dose-response relationships for paved roads, 2010–50

Type of cost

Precipitation

Temperature

Construction costs

Change in costs of constructing 1
kilometer (km) of paved road per 10
centimeter (cm) change in annual
precipitation projected during
lifespan relative to baseline climate;
dose-response represents change in
costs for every 10 cm increment

Change in cost of constructing 1 km of
paved road per stepwise increase in
maximum of monthly maximum
temperature values projected during
lifespan relative to baseline climate; the
first increase occurs after a 1

C change in

maximum temperature. Every other step
occurs at 3

C beyond that

Maintenance costs

Existing assets

Change in annual maintenance costs
for 1 km of paved road per 10 cm
change in annual rainfall projected
during lifespan relative to baseline
climate

Change in annual maintenance costs for 1
km per 3

C change in maximum of

monthly maximum temperature projected
during lifespan

New assets

Paved roads constructed after 2010 would have no maintenance impact if designed
for changes in climate expected during their lifetime

Source: Economics of Adaptation to Climate Change study team.

Box 10. Infrastructure sector methodology

The starting point for estimating the costs of adaptation are baseline projections of infrastructure
demand in physical units by country at five-year intervals with no climate change. These
projections are derived from econometric equations estimated using historic panel data, including
GDP per person at purchasing power parity exchange rates, population structure, urbanization,
country characteristics, and climate variables as independent variables. Two econometric
specifications were used: panel regressions representing average levels of infrastructure, and
stochastic frontier regressions representing the “efficient” levels of infrastructure given the values
of the independent variables.

In the period from t to t + 1, say from 2010 to 2015, the country will have to invest to meet the
new level of infrastructure in t + 1and to replace infrastructure existing at date t that reaches the
end of its useful life during the period. Thus, the total value of investment in infrastructure of type
i in country j and period t is

(1)

where C

ijt

is the unit cost of investment, (Q

ijt+1

– Q

ijt

) is the quantity of new investment in

infrastructure, and R

ijt

is the quantity of existing infrastructure that has to be replaced. The change

in the total cost of infrastructure investment may be expressed as the total differential of equation
(1) with respect to the climate variables that affect either unit costs or efficient levels of provision
for infrastructure of type i:

(2)

An equivalent equation may be derived for operation and maintenance costs. The first part of the
right side of equation (2) is referred to as the delta-P component of the cost of adaptation, and the
second part as the delta-Q component. These components cover several ways in which climate
change might cause changes in the costs or quantities of infrastructure services.

The delta-P component combines the baseline projections of infrastructure assuming no climate
change with estimates of the percentage changes in the unit costs of constructing, operating, and
maintaining infrastructure as a consequence of climate change. The changes in unit costs are
derived

from dose-response relationships estimated from the engineering-economic literature on

the costs of adjusting asset design and operational standards to hold infrastructure performance
constant under different climate conditions. The factors that drive the costs include average and
maximum monthly temperatures, total annual and maximum monthly precipitation, and
maximum wind speed. The dose-response relationships for operating and maintenance costs for
existing assets differ from those for newly constructed assets, which are designed to cope with the
projected climate over the life of the assets.

The delta-Q component of equation (2) captures the impact of climate change on demand for
infrastructure services, taking account of the higher unit costs of constructing and operating
infrastructure. This has two dimensions. Climate change may change the level or composition of
demand for energy, transport and water at given levels of income, so the net impact on capital and
operating costs has to be calculated. Climate change will also mean that countries have to invest
in additional assets to maintain standards of protection for noninfrastructure activities or services.
For water management and flood management and for coastal protection, this dimension of the
delta-Q component is addressed in specific sector studies, while for infrastructure the analysis
includes the first dimension plus other adjustments that are not captured elsewhere, such as
changes in health infrastructure.

The econometric analysis involves estimating a reduced form equation describing demand for
infrastructure:

(3)

[

]

ijt

C Q

−

[

]

(

)[

]

ijt

∆

= ∆

−

+ ∆

∆

− ∆

+ ∆

{

, }

ijt

h P Y

V t

where P

is the population of country j in period t; Y

is average income per capita for country j in

period t; X

is a vector of country characteristics for country j in period t (including an index of

construction costs); and V

is a vector of climate variables for country j in period t.

Since there are no strong priors on the appropriate functional forms, a standard flexible functional
form is used to represent the demand equation h

{ } in terms of the explanatory variables using a

restricted version of the translog specification for variables other than population. Because
practice, it is often difficult to estimate the full translog specification using the more complex
econometric models, the analysis started with the log-linear specification and then tested whether
the coefficients on the quadratic and cross-product terms were significant.

To deal with the claim that climate variables—especially average temperature—may act as a
proxy for institutional and other factors that shaped past patterns of economic development, the
values of demographic variables in 1950 are used in the models as instruments for institutional
development, following the approach of Acemoglu, Johnson, and Robinson (2001). Other country
fixed effects include country size and the proportions of land area that are desert, arid, semi-arid,
steep, or very steep and the proportion of land with no significant soil constraints for agriculture
using standard Food and Agriculture Organization land classifications. The use of differently
weighted climate variables (population-weighted and inverse population-weighted mean
temperature, total precipitation, temperature range, and precipitation range) captures the
differences between climate conditions in more and less densely populated areas.

Under the NCAR scenario, the total delta-P costs of adaptation average $29.5 billion a year over
of 2010–50 (table 6). The decade averages increases from nearly $16 billion a year in 2010–19 to
more than $44 billion a year for 2040–49. Adaptation costs are considerably lower under the
CSIRO scenario, averaging $13.5 billion a year for the period, though also increasing over time.
The NCAR scenario is significantly wetter than the CSIRO scenario in Asia and parts of Africa.
Because adaptation costs for infrastructure are particularly sensitive to levels of annual and
maximum monthly precipitation, the NCAR scenario has a larger impact on the costs of building
and maintaining roads, urban drainage, and buildings for countries in South Asia, Southeast Asia,
and Southern Africa.

Table 6. Annual delta-P costs of adaptation for infrastructure, by region and period, 2010–
50 ($ billions at 2005 prices, no discounting)

Period

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

National Centre for Atmospheric Research (NCAR), wettest scenario

2010–19

6.8

1.5

1.8

0.9

3.8

1.1

15.9

2020–29

9.5

1.9

2.8

1.2

6.6

2.3

24.3

2030–39

11.3

4.4

3.9

1.5

8.7

3.9

33.7

2040–49

14.8

5.3

5.4

1.8

10.7

6.1

44.1

Average

10.6

3.3

3.5

1.4

7.5

3.4

29.5

Commonwealth Scientific and Industrial Research Organization (CSIRO), driest scenario

2010–19

3.1

0.7

1.3

0.6

1.4

0.7

7.8

2020–29

3.3

1.1

1.6

0.5

1.5

1.0

9.0

2030–39

4.3

1.5

1.8

0.9

3.9

1.7

14.1

2040–49

5.6

2.1

1.4

9.1

2.6

22.9

Average

4.1

1.4

1.7

0.9

4.0

1.5

13.5

By far the largest delta-P costs of adaptation under the NCAR scenario are for constructing new
or replacing existing infrastructure (table 7). The share of maintenance costs rises gradually but is
still less than 10 percent in 2040–49. The pattern for the CSIRO scenario is similar (not shown).
The highest adaptation costs are in East Asia and Pacific and South Asia, reflecting their larger
populations. Sub-Saharan Africa experiences the largest increase over time, with its adaptation
cost rising from $1.1 billion a year for 2010–19 to $6 billion a year for 2040–49. This rapid rise is
associated with a low share of maintenance costs in total costs and is driven by the need for large
investments in infrastructure to support future economic growth. In contrast, countries in Europe
and Central Asia face maintenance costs that are larger than capital costs after 2030, reflecting the
pattern of climate change under the NCAR scenario for Russia and Central Asia. The same result
does not emerge for the CSIRO scenario, a reminder of how different climate scenarios can affect
the character and the magnitude of projected adaptation costs.

Table 7. Breakdown of annual delta-P costs of adaptation for infrastructure for the
National Centre for Atmospheric Research (NCAR) climate scenario, by region and cost
type, 2010–2050 ($ billions at 2005 prices, no discounting)

Period and
cost type

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

2010–19

Capital

6.7

1.2

1.8

0.9

3.8

1.1

15.5

Maintenance

0.1

0.2

0.3

Total

6.8

1.4

1.8

0.9

3.8

1.1

15.9

2020–29

Capital

9.3

1.7

2.7

1.2

6.5

2.3

23.7

Maintenance

0.2

0.1

0.7

Total

9.5

1.9

2.8

1.2

6.6

2.3

24.3

2030–39

Capital

11.1

1.9

3.8

1.3

8.6

3.9

30.6

Maintenance

0.3

2.5

0.2

0.1

3.4

Total

11.4

4.4

4.0

1.5

8.7

4.0

33.7

2040–49

Capital

14.1

2.3

5.0

1.5

10.4

5.9

39.2

Maintenance

0.7

3.0

0.4

0.2

0.1

4.6

Total

14.8

5.3

5.4

1.7

10.6

6.0

44.1

Urban infrastructure (urban drainage, public buildings, and similar assets) accounts for 54 percent
of the delta-P adaptation cost over 2010–50, followed by roads (mainly paved roads) at 23
percent. (Box 11 describes some of the private adaptation costs for urban housing that are not
covered by planned adaptation.) Networks and associated assets (power generation, electricity
transmission and distribution, fixed telephone lines, water and sewage treatment) account for less
than 9 percent of the estimated cost of adaptation even though they account for about 45 percent
of total infrastructure costs.

For comparison, table 8 also shows the total costs of providing each type of infrastructure (not
just climate proofing)—the baseline cost. The costs of adaptation are 4.6 percent of the total costs
of infrastructure provision over the period for urban infrastructure, 2.3–2.1 percent for roads and
other transport, and less than 1 percent for the other infrastructure categories. Overall, the
adaptation cost is 1.6 percent of total infrastructure costs. These shares contrast with those of
previous studies, which use ranges of 0.01 percent to 8 percent to estimate adaptation costs (see
table 4, where equivalent parameters are obtained by multiplying percentage of climate sensitive
new investments by percentage increases in costs) and fail to differentiate by type of asset. These
differences in parameter values explain in part why the EACC estimates of adaptation costs for
infrastructure ($15–30 billion a year) fall between the maximum and the minimum of past
estimates ($2–67 billion a year; see table 4).

Table 8. Breakdown of delta-P costs of adaptation for the National Centre for Atmospheric
Research (NCAR) climate scenario, by region and infrastructure category, 2010–50,
($ billions at 2005 prices, no discounting)

Infrastructure
category and
adaptation or
baseline cost type

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

Health and education

Adaptation

1.1

0.7

0.4

0.2

0.5

0.1

3.0

Baseline

123.6

87.8

36.3

54.3

16.2

386.2

Other transport

Adaptation

0.2

0.3

0.1

0.9

Baseline

9.6

16.2

5.6

1.8

4.7

3.7

41.6

Power and wires

Adaptation

0.6

0.2

0.1

0.3

0.1

1.9

Baseline

164.3

108.8

62.9

25.9

82.8

21.3

466

Roads

Adaptation

1.8

0.7

0.6

1.4

0.8

6.3

Baseline

60.1

47.9

43.1

23.4

57.2

36.5

268.2

Urban infrastructure

Adaptation

6.6

0.8

1.6

0.3

4.9

2.3

16.5

Baseline

105.1

83.3

40.3

12.2

85.7

355.6

Water and sewers

Adaptation

0.3

0.1

0.2

0.7

Baseline

140.7

26.5

67.8

23.4

382.4

All infrastructure

Adaptation

10.6

3.3

3.5

1.3

7.4

3.4

29.5

Baseline

603.5

405.1

282.8

126.2

352.5

130.1

1900.2

Note: Delta-P cost is the adaptation cost computed as the additional cost of constructing, operating, and
maintaining baseline levels of infrastructure services under the new climate conditions projected by the two
global climate models.
a. The baseline cost is defined as the sum of capital and maintenance expenditures over the lifetime of the
asset.

Source: Economics of Adaptation to Climate Change study team.

Box 11. Urban housing and climate change

Planned adaptation costs do not account for the high adaptation costs of urban housing, which are
largely individually provided. EACC estimates annual average household investments in urban
housing in response to climate change at $2.3 billion (in 2005 dollars) per year in 2010, rising to
$25.6 billion a year by 2050, under the CSIRO climate scenario. Under the NCAR scenario,
annual costs rise even more, from an average of $4.4 billion a year in 2010 to $45.5 billion by
2050. Under both scenarios, costs are highest in East Asia and Pacific (followed by Latin
America and the Caribbean under CSIRO and Europe and Central Asia under NCAR).

The costs of adaptation related to housing would be even higher if they also accounted for slums.
Most informal settlements in developing countries share characteristics that intensify the
vulnerability of their residents to climate change (Moser and Satterthwaite forthcoming). These
include poorly constructed buildings; inadequate infrastructure; lack of safe drinking water,
drainage, and sanitation services; and severe overcrowding with attendant public health impacts.
Municipal governments often neglect or even criminalize such settlements, exacerbating the
problem of underprovisioning of protective infrastructure and services. These factors combine
with high concentrations of poor people with few assets to make slums especially vulnerable to
flooding and other extreme events, which can lead to loss of lives and property and the spread of
diseases such as malaria.

As discussed in the participatory scenario workshops, Ghana presents considerable challenges in
adapting urban slums to climate change. Rural migrants to Accra and increasingly to Ghana’s
secondary towns cluster in slums prone to overcrowding and poor sanitation. Workshop
participants report that floods are more severe in these sprawling urban spaces of coastal Ghana
than in inland towns, in part because of weak urban planning. Urbanization, especially in slums,
increases the risk of climate-related disasters such as flooding and landslides, in part because
natural ecosystem-based storm breaks and rain catchment areas are increasingly converted to
public buildings and housing developments.

Thus far the analysis has assumed that climate change does not affect demand for infrastructure,
but only the cost of providing it under the no-climate change scenario. However, climate change
is likely to affect the demand for infrastructure services as well. For example, the optimal
investment in roads would vary depending on whether climate change alters the structure of the
economy and thus the location of economic activity, or more or higher dykes might be needed to
cope with sea level rise and storm surges (see the discussion of adaptation cost for coastal zones).
Called the delta-Q component of adaptation cost because it focuses on changes in the quantity of
infrastructure required in response to changes in demand, this component is difficult to estimate,
for reasons discussed in box 12, and in several cases it was also difficult to identify mechanisms
that could explain counterintuitive results. Therefore, although estimates of delta-Q are presented
in table 9, for illustrative purposes, they are not used to calculate adaptation costs for
infrastructure assets.

A comparison of infrastructure estimates with and without estimated changes in infrastructure
demand indicates that the demand equations do not imply any simple relationships between
climate and infrastructure demand (table 9). In most cases, the impact of climate change depends
on interactions among per capita GDP, urbanization, and the range between maximum and
minimum temperatures or precipitation. As a consequence, the predicted impact of climate
change on demand for infrastructure by country ranges from –5 percent to +5 percent of baseline
investment.

The overall impact is negative in most large countries, with the exception of the Europe and
Central Asia region, so the net quantity adjustment reduces the overall cost of adaptation by $19–
22 billion a year for the two scenarios. This is equivalent to a reduction of about 1 percent of the
baseline cost of infrastructure, demonstrating that small shifts in demand can have a very large
impact on the total cost of adaptation. With the NCAR scenario, the net cost of adaptation for
infrastructure declines from $29.5 billion a year to $7.3 billion a year when the delta-Q

adjustment is included. Since the total delta-P cost of adaptation is relatively small for the CSIRO
scenario, including delta-Q more than offsets the price effect of climate change, leaving a net
reduction in the cost of infrastructure due to climate change. On the other hand, using the gross
aggregate cost measure (not allowing for cross-country transfers and setting benefits to zero)
leads to higher estimates of adaptation costs because the larger negative delta-Q adjustments are
excluded (not shown).

Table 9. Alternative measures of the cost of adaptation per year for infrastructure, by
region ($ billions at 2005 prices, no discounting)

Cost component

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

National Centre for Atmospheric Research (NCAR), wettest scenario

Delta-P only

10.6

3.3

3.5

1.3

7.4

3.4

29.5

Delta-P + Delta-Q

(0.1)

8.9

(1.2)

(0.7)

0.2

7.3

Commonwealth Scientific and Industrial Research Organization (CSIRO), driest scenario

Delta-P only

4.1

1.4

1.7

0.9

4.0

1.5

13.6

Delta-P + Delta-Q

(3.0)

6.5

(3.0)

0.2

(2.7)

(3.0)

(5.0)

Note: Delta-P cost is the adaptation cost computed as the additional cost of constructing, operating, and
maintaining baseline levels of infrastructure services under the new climate conditions projected by NCAR
and CSIRO global climate models. Delta-Q cost accounts for changes in the quantity of infrastructure
required in response to changes in demand under the new climate conditions projected by the two climate
models.
Source: Economics of Adaptation to Climate Change study team.

his is an important area for further investigation. The economic viability of certain areas will

certainly be altered by climate change, which could lead to either more or less demand for
infrastructure. However, for the reasons outlined in box 12, the delta-Q values are not used in this
report’s estimates of the overall cost of climate change.

Box 12. Why the study reports only delta-P and not delta-Q adaptation costs

The econometric equations used for this study, based on historical data, reflect the location of
economic activity (and the consequent demand for infrastructure) in response to a given climate,
not the relocation of economic activity (and the consequent change in the demand for
infrastructure) as a result of a change in climate. Theses long-run relationships reflect an
equilibrium between the influences of climate and economic variables. The literature on path
dependency suggests that how a country responds to external shocks such as climate change may
depend critically on its current stock of assets, which is codetermined with the current location of
economic activity. The counterargument is that the effects of climate change on demand for
infrastructure are small relative to the impact of economic development over 40 years or more, so

the effects of path dependency are swamped by the structural changes implied by the
development baseline.

Another hurdle concerns econometric specification. The data used for the analysis are a
combination of time series and cross-country variables. Like other fixed country characteristics,
climate variables are constant over time, so their influence has to be estimated from cross-country
variation alone. Many studies rely on cross-country variation for key explanatory variables—for
example, studies of the effects of governance and trade policy on economic growth. The cross-
country variables help to explain a set of country fixed effects that are combined with the
influence of factors (GDP per capita, population, urbanization, and so on) that vary across time
and countries. The difficulty is that one or more climate variable might act as a proxy for country
characteristics that influence the demand for infrastructure but that are not included in the
analysis, so the coefficient on the climate variable will reflect both its direct influence on demand
and its correlation with the omitted factor. Omitted variables are a potential problem in all
econometric analysis, and it is impossible to demonstrate a negative—that the coefficients on the
climate variables are not affected by omitted variables. The most that can be done is to include
additional variables that may be better proxies for potential influences that cannot be included in
the equations and to use specifications—such as interactions with time-varying factors—that
reduce or eliminate correlation between omitted variables and climate variables.

The influence of climate variables on demand for infrastructure remains an open area of research.
There is ample evidence that some climate variables have an impact on specific types of
infrastructure, such as temperature on energy demand or precipitation on water use. There is
much less agreement on how these influences operate in the longer term and on whether the
relationships can be extended to all categories of infrastructure.

In view of these uncertainties, the final estimates of the costs of climate change exclude the delta-
Q adjustments.

C oastal zones

Coastal zones, home to an ever-growing concentration of people and economic activity, are
subject to several climate risks, including rising sea level and increased intensity of tropical
storms and cyclones, making adaptation to climate change critical, particularly in small islands
and deltaic countries (see box 13).

Box 13. Adaptation costs for deltaic countries and small islands states

Deltaic countries and small island states are particularly at risk for sea-level rise induced by climate
change. For deltaic communities, ongoing subsidence and land conversion may exacerbate the effects of
sea-level rise or extreme sea levels caused by intense storms. Small island states are vulnerable because of
their small size, limited resource base, and geographic isolation.

Adaptation costs and residual damages for the medium sea-level rise scenario suggest that the costs of

adapting to climate change for deltaic countries are nearly $4.5 billion per year (see table), or about 15
percent of the total cost of adaptation estimated here. Adaptation costs for small island states are more
than $1 billion a year, or about 3 percent of the total cost of adaptation estimated here. In relative terms,
the adaptation costs are higher still, averaging 1 percent of GDP in small island states over 2010–50
compared with 0.03–0.1 percent for other developing countries. Residual damage costs as a percentage of
GDP are also higher in deltaic and small island states than in other countries, even after the large
adaptation investments considered here.

Average annual coastal adaptation costs and residual damage, 2010–50

Cost category

Deltaic

countries

Small

island

states

Brazil,

Russia,

India, and

China

Other

developing

countries

Adaptation cost

Amount ($ billions, 2005
prices, no discounting)

4.5

1.0

9.0

14.1

Share of GDP (percent)

0.1

1.0

0.03

0.1

Residual damage

Amount ($ billions, 2005
prices, no discounting)

0.62

0.01

0.52

0.36

Share of GDP (percent)

0.01

0.002

Note: Residual damages are impacts remaining after adaptation.
a. Includes Bangladesh, Burma, China, Egypt, French Guiana, Guyana, India, Iraq, Mozambique, Nigeria,
Pakistan, Romania, Suriname, Thailand, Ukraine, Venezuela, and Vietnam. While no country is entirely
deltaic, in these countries the coastal impacts and adaptation costs are strongly influenced by deltaic areas.
b. State or territory with a land area of less than 30,000 square kilometers (sq km) occupying an island or group
of islands that are separately less than 20,000 sq km. This definition excludes Cuba, Haiti, and the Dominican
Republic.
Source: Economics of Adaptation to Climate Change study team.

This study estimates costs for coastal adaptation over 2010–50 by building on the earlier estimate
of the UNFCCC (Nicholls 2007) in several ways. It considers the adaptation costs of more intense
storms as well as rising sea level, extends the UNFCCC estimates from 2030 to 2050, includes
maintenance as well as construction costs, and adds the costs of port upgrade. And, as discussed
in section 3, it defines costs as those needed to achieve an efficient level of adaptation. Selected
residual damages from climate change are also reported (impacts remaining after adaptation, such
as land loss costs and number of people flooded) and added to adaptation costs in estimating the
resources needed to restore welfare to pre-climate change levels. These improvements
significantly raise the cost of adaptation to climate change for coastal zones over the UNFCCC
estimate.

The analysis considers two main types of impact (coastal erosion, and sea and river flooding and
submergence) and three adaptation approaches (beach nourishment, particularly in areas with

high tourism revenue; sea and river

dike construction; and port upgrade). Impacts due to

salinization and wetland loss are not considered (see box 14 for details).

Box 14. Coastal zone methodology

Adaptation costs for coastal zones are derived mainly from the Dynamic and Interactive
Vulnerability Assessment (DIVA) model, based on 12,148 coastal segments that make up the
world’s coast (except for Antarctica) and a linked database and set of interacting algorithms
(MacFadden and others 2007; Nicholls and others 2007; Vafeidis and others 2008). The sea-level
rise scenarios are downscaled with an estimate of the vertical land movement in each segment.
The coastal erosion analysis considers only sandy coasts and takes account of the direct effect
(Bruun effect) and indirect effects of sea-level rise, as well as beach nourishment where it occurs.
The indirect effects occur at major estuaries and lagoons.

The flooding analysis determines the flood areas for different return periods and extreme water
levels, including the effects of dikes. Since empirical data on actual dike heights are not available
at a global level, “optimum” dikes heights were estimated for the base year of 1995 using a
demand for safety function.

Dike heights are then upgraded according to projected sea-level rise

to 2050. Increased flooding due to sea-level rise along the coastal-influenced reaches of major
global rivers (identified in the DIVA database) is also considered. Damages are evaluated in
terms of physical, social, and economic indicators such as land lost to erosion or submergence,
the number of people expected to be subject to annual flooding, the number of people forced to
migrate because of land loss, and the costs of this migration.

DIVA implements the adaptation options according to complementary adaptation strategies. For
beach nourishment, a cost-benefit adaptation strategy balances costs and benefits (damage
avoided) of adaptation, including the tourist value of beaches. For dike building, the demand
function for safety is applied over time, subject to population density. Dikes are built only when
population density exceeds 1 person per square kilometer, with an increasing proportion of the
recommended height being built as population density rises—for example, 98 percent of the dike
height is built at densities of 1,000 people per square kilometer. The unit costs of beach
nourishment, dikes, and port upgrades were derived from the global experience of Delft
Hydraulics (now Deltares).

For this analysis, DIVA was extended to include a sensitivity analysis of more intense tropical
storms. This influences adaptation costs only for dikes. The maintenance costs of sea and river
dikes and port upgrades globally are also computed outside DIVA. Port costs are based on a
strategy of continuously raising existing port areas

as sea levels rise.

__________

This concerns the incremental costs of upgrading river dikes in coastal lowlands where sea-level rise will

raise extreme water levels. Additional upgrade may be required if extreme river flows are increased, but
this is not investigated here.

1. The demand for safety function increases with per capita income and population density and decreases
with the costs of dike building, an approach that is posited as the solution to a cost-benefit analysis (Tol
2006).
2. All new port areas are assumed to include sea-level rise to 2050 in their design, so upgrade costs will be
effectively zero.

The analysis considers four scenarios of global sea-level rise: a no-rise (or reference) case of no
climate change and low, medium, and high sea-level rise scenarios based on IPCC AR4 (Meehl
and others 2007) and Rahmstorf (2007) (table 10). These useful and plausible scenarios reflect
the uncertainty in climate projections. They are not specifically linked to temperature rise,
however, because of uncertainties in the timing of deglaciation. An arbitrary 20 percent increase
in flood heights is assumed under the high sea-level rise scenario by 2100 to reflect
intensification of storms in areas currently subject to such storms.

Table 10. Sea-level rise under four scenarios, 2010–2100 (centimeters above 1990 levels)

Year

No sea-level

rise

Low sea-level

rise

Medium sea-

level rise

High sea-level

rise

2010

0.0

4.0

6.6

7.1

2020

0.0

6.5

10.7

12.3

2030

0.0

9.2

15.5

18.9

2040

0.0

12.2

21.4

27.1

2050

0.0

15.6

28.5

37.8

2060

0.0

19.4

37.0

50.9

2070

0.0

23.4

47.1

66.4

2080

0.0

28.1

58.8

84.4

2090

0.0

33.8

72.2

104.4

2100

0.0

40.2

87.2

126.3

Note: The low rise scenario is derived as the midpoint of the IPCC AR4 A2 range in 2090–99, a trajectory
consistent with a Model for the Assessment of Greenhouse-gas Induced Climate Change (MAGICC, a
coupled gas-cycle/climate model) IPCC Third Assessment Report A2 mid-melt 3

C sensitivity run. The

medium rise scenario is derived from the Rahmstorf (2007) A2 trajectory. The high rise scenario is derived
from the Rahmstorf (2007) maximum A2 trajectory.
Source: Neumann (2009).

Uniform population growth is imposed on the EACC projections of population and GDP growth,
so that coastal populations do not grow relative to other areas. However, a scenario of no
population growth is also considered, in which all future growth happens in areas that will not be
affected by sea-level rise.

Following best engineering practice for sea and river dikes, dike building anticipates sea-level
rise in terms of additional height needed 50 years into the future (thus dike heights in 2050 are
determined by expected extreme sea levels in 2100). For other adaptation measures, there is no
anticipation of future conditions, again reflecting best engineering practice. Adaptation methods
are applied in a standard way around all the world’s coasts using criteria that select optimum or
quasi-optimum adaptation strategies. Selected residual impacts that remain even with adaptation
are also reported (impacts remaining after adaptation, such as land loss costs, coastal flood costs,
and the number of people flooded), stressing that much larger investments would be required to
avoid all impacts of sea-level rise, if this is even possible or desirable.

Table 11. Annual costs of adaptation for coastal zone protection, by scenario and cost
component, 2010–50 ($ billions at 2005 prices, no discounting)

Coastal zone cost
component

Low sea-level

rise

Medium sea-

level rise

High sea-

level rise

High sea-level

rise with

cyclones

Beach nourishment

1.7

3.3

4.5

River dikes

0.2

0.4

0.6

Sea dikes

10.7

24.6

36.7

39.1

Port upgrades

0.2

0.4

0.5

Residual damages

0.7

1.5

Total

13.5

30.2

44.3

46.7

a. Includes impacts remaining after adaptation, such as land loss, coastal flooding, and number of people
flooded.
Source: Economics of Adaptation to Climate Change study team.

Coastal adaptation costs are substantial and vary with the magnitude of sea-level rise (table 11),
making it essential for policymakers to plan while accounting for the uncertainty. Flooding
dominates both the adaptation costs (of building dikes) and the costs of damages due to the
residual risk. Sea-level rise does not have a large effect on the size of residual damages—the
main effect is an increased investment in adaptation.

An analysis of how adaptation costs and residual damages are distributed for the medium sea-
level rise scenario indicates that Latin American and the Caribbean and East Asia and Pacific
together account for some two thirds of the total cost of adaptation (table 12). Deltaic countries
and small island states are particularly at risk (see box 13).

Increased tropical storm intensity does not raise annual costs substantially, and targeting future
population growth outside the coastal flood plain does not reduce costs substantially, as the
existing development already creates a substantial need for protection.

Clearly, a wider range of adaptation options than considered in DIVA are available in practice,
including retreating from coastal zones and accommodating higher water levels by raising
buildings above flood levels. These steps could reduce the need for protection measures and
could lead to lower adaptation costs than those estimated here (these softer measures are very
difficult to cost, which is why that was not done). Realizing these benefits will require long-term
strategic planning and more integration across coastal planning and management. Few if any
countries have this capacity today, and strengthening institutional capacity for integrated coastal
management would seem a prudent response to climate change (also yielding benefits in other
areas).

Table 12. Annual cost of adaptation for coastal zone protection and residual damages for
the medium seal-level rise scenario, by region, 2010–50 ($ billions at 2005 prices, no
discounting)

Type of adaptation
cost and period

East

Asia and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

Total adaptation cost

2010–19

7.6

2.4

8.5

1.0

1.6

3.2

24.3

2020–29

8.4

2.6

9.5

1.2

1.7

3.7

27.1

2030–39

9.2

2.8

10.6

1.3

1.9

4.2

30.0

2040–49

10.0

3.1

11.7

1.4

2.1

4.8

33.1

Residual damage

2010–19

0.3

0.0

0.1

0.0

0.4

2020–29

0.6

0.0

0.1

0.0

0.1

0.0

0.8

2030–39

1.1

0.0

0.2

0.0

0.3

0.0

1.6

2040–49

1.3

0.1

0.4

0.1

0.9

0.0

2.8

a. Includes beach nourishment, river and sea dikes, and port upgrades.
b. Includes impacts remaining after adaptation, such as land loss, coastal flooding, and number of people
flooded.
Source: Economics of Adaptation to Climate Change study team.

I ndustrial and municipal water supply and riverine flood protection

Climate change has already affected the hydrologic cycle, and the impacts are expected to
continue and intensify over the century. Where water availability has increased, the increase is
expected to continue, and where it has decreased it is expected to continue to decrease. Projected
increases in the intensity of rainfall are expected to boost the frequency and magnitude of floods.
Policymakers need to understand these changes and adapt to them.

The analysis of the costs of adaptation for water management includes industrial and municipal
water supply (box 15) and excludes water for agriculture and ecosystem services. Irrigation is
considered in the discussion of the agricultural sector, and water management for ecosystem
services is implicitly dealt with by limiting future withdrawals to no more than 80 percent of total
runoff, with no further withdrawals permitted in river basins where current water withdrawals are
already more than 80 percent.

Box 15. Water sector methodology

The effects of climate change on the water cycle were assessed by running the Climate and Runoff
model (CLIRUN-II) on a monthly time-step. The key parameters were monthly runoff and the
magnitude of the 10-year and 50-year maximum monthly runoff. The results were aggregated to

the 281 food production units of the International Model for Policy Analysis of Agricultural
Commodities and Trade (IMPACT) developed by the International Food Policy Research Institute.
The analysis considers industrial and municipal water supply and riverine flood protection.

Water supply. Costs of adaptation are defined as the cost of providing enough raw water to restore
future industrial and municipal water demand to the levels that would have existed without climate
change. Such demand is assumed to be met by increasing the capacity of surface reservoir storage,
except when that would raise withdrawals to more than 80 percent of river runoff and when the
cost of supplying water from reservoir yield is more than $0.30 a cubic meter. In these cases,
supply is assumed to be met through alternative measures, such as recycling, rainwater harvesting,
and desalination, at a cost of $0.30 a cubic meter.

Additional reservoir storage capacity to meet future water demand is calculated using storage-yield
curves showing the storage capacity needed to provide a given yield and reliability of water supply
over the year. The storage yield curves were developed using simulated time-series of monthly
runoff and evaporation from CLIRUN-II. The costs of reservoir construction were based on a
method relating topography to cost, and annual operation and maintenance costs were assumed to
be 2 percent of construction costs. Three scenarios were used to estimate the size distribution of
future reservoirs: small dams, with all future reservoirs having a storage capacity under 25 million
cubic meters; large dams, with all future reservoirs having a storage capacity greater than 12,335
million cubic meters; and best estimate, with future construction assumed to follow the same size
distribution as in the 20

century in the United States. The results in this section are shown for the

best estimate scenario.

Flood protection. Costs are defined as the cost of providing flood protection against the 50-year
monthly flood (maximum monthly runoff) in urban areas and the 10-year monthly flood in
agricultural areas. First, the baseline costs (without climate change) of providing flood protection
to all urban and agricultural areas were estimated. Then, the costs of adaptation were estimated by
assuming that the costs of providing flood protection rose by the same percentage as the
percentage change in the magnitude of the 50-year or 10-year monthly flood event. Flood
protection was assumed to be provided through a system of dikes and polders, at a cost of $50,000
per square kilometer in urban areas and $8,000 per square kilometer in agricultural areas (these
cost estimates were derived from World Bank case studies). Annual operation and maintenance
costs were assumed to be 0.5 percent of construction costs.

In a methodological improvement over previous studies (Kirshen 2007, subsequently modified by
UNFCCC 2007), the sectoral water balance is maintained, with any change in agricultural
withdrawals accounted for before computing the costs of adaptation for raw industrial and
municipal water supply. Other methodological improvements include use of a longer time
horizon (to 2050 rather than 2030); analyses of the baseline without climate change and of the
baseline changes under climate change, whereas the previous studies examined the combined
costs of adaptation to socioeconomic development and climate change and then assumed the costs
related to climate change to be 25 percent of the total; and use of hydrological models to estimate
change in generic reservoir capacity. In addition, this study estimates the global costs of
adaptation related to riverine flood protection, which the other studies did not consider, by
analyzing the costs of protecting against river flooding in urban and agricultural areas against the
50-year monthly flood in urban areas and the 10-year monthly flood (maximum monthly runoff)
in agricultural areas (see box 15).

Table 13. Gross and net annual adaptation costs for water supply and riverine flood
protection, by region, 2010–50 ($ billions at 2005 prices, no discounting)

Type of cost
calculation and
protection
category

East
Asia

and

Pacific

Europe

and

Central

Asia

Latin

America

and

Caribbean

Middle

East

and

North

Africa

South

Asia

Sub-

Saharan

Africa

Total

National Centre for Atmospheric Research (NCAR), wettest scenario

Gross

Flood protection

0.9

1.7

1.0

0.2

1.1

0.4

5.3

Water supply

3.1

1.7

5.3

0.5

1.8

6.2

18.6

Total

4.0

3.4

6.3

0.7

2.9

6.6

23.9

Net

Flood protection

0.8

1.4

0.3

–0.2

1.0

0.3

3.6

Water supply

0.3

0.9

5.2

0.0

–2.3

5.9

10.0

Total

1.1

2.3

5.5

–0.2

–1.3

6.2

13.3

Commonwealth Scientific and Industrial Research Organization (CSIRO), driest scenario

Gross

Flood protection

1.6

0.9

2.0

0.6

1.7

0.2

7.0

Water supply

2.1

0.5

2.9

0.2

5.9

7.6

19.2

Total

3.7

1.4

4.9

0.8

7.6

7.8

26.2

Net

Flood protection

1.6

0.6

1.7

0.5

1.6

–0.2

5.8

Water supply

0.6

–0.3

1.5

–0.4

2.4

7.3

11.1

Total

2.2

0.3

3.2

0.1

4.0

7.1

16.9

Note: Gross costs set negative values to zero for sector protection in any country with negative costs. Net
costs are the pooled costs without restrictions on pooling across country borders (positive and negative
values are treated symmetrically).

Source: Economics of Adaptation to Climate Change study team.

Adaptation costs for industrial and municipal raw water supply are higher for the CSIRO
simulations, with its drier global mean conditions, than for the NCAR simulations, with its wetter
conditions (table 13 and map 3), because more reservoir storage capacity is required to provide
the same yield. The adaptation costs for riverine flood protection are also greater for the CSIRO
scenario because the model simulates a larger increase in the magnitude of the 10-year and 50-
year monthly flood events than does the NCAR scenario, despite relatively drier mean conditions.
The highest costs are in Sub-Saharan Africa under both climate scenarios. Latin America and the
Caribbean also sustain high costs under both models, and South Asia sustains high costs under
CSIRO because these regions experience the largest percentage decline in mean runoff (see map
3). The gross costs of adaptation are significantly greater than the net costs, especially for water
supply: $23.9 billion gross and $13.3 billion net annual cost under NCAR and $26.2 billion gross
and $16.9 billion net annual cost under CSIRO. These differences are driven mainly by the
decreased need for storage capacity in South Asia and East Asia and Pacific under both scenarios
because of increased mean runoff (see map 3).

Map 3. Change in mean water runoff under the Commonwealth Scientific and Industrial
Research Organization and National Centre for Atmospheric Research global climate
scenarios, 2000–50

Commonwealth Scientific and Industrial

Research Organization (CSIRO), driest

scenario

National Centre for Atmospheric Research

(NCAR), wettest scenario

Note: The Economics of Adaptation to Climate Change study team acknowledges the Program for Climate
Model Diagnosis and Intercomparison and the World Climate Research Programme's (WCRP) Working
Group on Coupled Modelling for their roles in making available the WCRP’s Coupled Model
Intercomparison Project phase 3 (CMIP3) multimodel dataset. Support of this dataset is provided by the
Office of Science, U.S. Department of Energy.

As do most sectoral studies of global adaptation costs, this study focuses on hard adaptation
measures, which are easier to cost than behavioral measures. There is no implication that these
are the best measures for adaptation. Ideally, adaptation options to ensure water supply during
average and drought conditions should integrate strategies on both demand and supply sides.
While demand-side adaptations are not explicitly costed in this study (demand projections already
account for some increase in efficiencies over time, so this could lead to double counting), there
is wide scope for economizing on water consumption (see, for example, Zhou and Tol 2005).
Adaptation options for flood protection can reduce either the probability of flood events or their

magnitude (reducing flood hazard) or the impacts of floods. In both cases, adaptation should
consider structural and nonstructural measures that address both flood probability and impact.

Agriculture

The analysis of agriculture brings together, for the first time, detailed biophysical modeling of
crop growth under climate change with the world’s most detailed global partial equilibrium
agricultural model to estimate the costs of adaptation for returning the number of malnourished
children to pre-climate change levels. One of the few earlier estimates of adaptation costs for
agriculture takes a simpler approach, assuming that an arbitrary 10 percent increase in research
and extension funding and a 2 percent increase in capital infrastructure costs are needed by 2030
to adapt to climate change (UNFCCC 2007). Also, the UNFCCC estimate includes no explicit
link to climate impacts or any accounting for autonomous (personal) adaptation.

The analysis of agricultural adaptation costs uses the International Food Policy Research
Institute’s (IFPRI) International Model for Policy Analysis of Agricultural Commodities and
Trade (IMPACT) to incorporate the direct impacts of climate change on agricultural production
(yields and crop area) and the indirect effects through food prices and trade on calorie availability
and the number of malnourished children (see box 16). IMPACT includes 32 crops and livestock
commodities, including cereals, soybeans, roots and tubers, meats, milk, eggs, oilseeds, oilcakes
and meals, sugar, and fruits and vegetables. Changes in the number of malnourished children
between 2050 and 2000 without climate change are compared to changes with climate change to
determine costs of adaptation.

Box 16. Agriculture sector methodology

Climate change affects agriculture through changes in yields and in areas planted. Farmers respond
by changing their management practices. The resulting production effects work their way through
agricultural markets, affecting prices. Consumers respond by changing consumption patterns.
When prices rise, consumption falls and the number of malnourished children rises. Adaptation
expenditures on productivity enhancing investments can offset these impacts of climate change.

The biological effects of climate change are modeled with the Decision Support System for
Agrotechnology Transfer (DSSAT) crop modeling program, assessing yield and area effects for
five major commodities at 0.5 degree resolution. The DSSAT model includes a carbon dioxide
fertilization effect of 369 parts per million (ppm) atmospheric concentration, reflecting recent
research suggesting that fertilization effects are much weaker in the field than in the laboratory.
Using a 532 ppm value reduces the costs of adaptation by less than 10 percent.

The productivity effects of climate change are aggregated to 32 crops and 281 food production
units of the International Food Policy Research Institute’s International Model for Policy Analysis
of Agricultural Commodities and Trade. Growth in crop production in each country is determined
by crop and input prices, exogenous rates of productivity growth and area expansion, investment in
irrigation, and water availability. Demand is modeled as a function of prices, income, and
population growth and has four components: food, feed, biofuels feedstock, and other uses. The
model links national agricultural markets through international trade. World agricultural
commodity prices are determined annually at levels that clear international markets.

Costs of adaptation are measured against the human well-being measure of malnutrition in
preschool children, a highly vulnerable group. The number of malnourished children is determined
in part by per calorie availability but also by access to clean drinking water and maternal
education. Investments in agricultural research, roads, and irrigation increase agricultural
productivity under climate change, increasing calorie availability and reducing child malnutrition
estimates.

The costs of adaptation for agriculture are calculated solely from the perspective of the agriculture
sector, so the starting point is investment and asset stocks in the base year (2000). Thus, the
estimates of investments in research, irrigation, and rural roads do not take account of overlaps in
spending on these activities or assets with the baseline growth or of adaptation costs for other
sectors, such as infrastructure and water resources management. This is an unavoidable
consequence of estimating the cost of adaptation for each sector separately and in parallel. For
rural roads, an attempt was made to eliminate overlapping expenditures in compiling the
consolidated estimates of the costs of adaptation for developing countries in table 24. The baseline
provision of rural roads up to 2050 used to estimate costs of adaptation is adjusted to take account
of the additional length of rural roads consistent with the baseline projections for road investment.
This adjustment reduces the investment in rural roads included in the cost of adaptation for
agriculture by about 80–85 percent for the two climate scenarios. The adjustment for these
overlaps amounts to $2.0–2.2 billion a year averaged over the full period.

Changes in temperature and precipitation in the NCAR and CSIRO climate scenarios have strong
negative effects on crop yields and production. Irrigated and rainfed wheat and irrigated rice are
especially hard hit. South Asia experiences the biggest loss in production, and developing
countries fare worse than developed countries for almost all crops under both scenarios.

These productivity impacts, even after accounting for autonomous adjustments through changes
in, say, input and crop mix (see box 17 on some private adaptation measures in agriculture in
some case study countries), lead to dramatic impacts on trade flows (another form of autonomous
adjustment). Without climate change, developed country net exports rise from 83.3 million tons
to 105.8 million tons between 2000 and 2050—a 27 percent increase. South Asia switches from a
net exporter to a net importer, and East Asia and Pacific and Sub-Saharan African imports rise
considerably (table 14 and figure 3). Developed country exports rise 28 percent under the NCAR
scenario and a dramatic 75 percent under the CSIRO scenario compared with 2000 levels (not
shown). South Asia becomes a much larger importer of food under both scenarios than under
baseline conditions of no climate change, East Asia and Pacific becomes a net exporter of food
under the NCAR scenario, and Europe and Central Asian exports and Sub-Saharan African
imports fall substantially under both scenarios. Climate change has a smaller impact on meat
trade.

Box 17. Private adaptation in agriculture and areas needing policy attention

Farmers in Sub-Saharan Africa are already adapting to increasing rainfall variability and higher
temperatures by shifting sowing dates and changing crop mix or plot location. In Ethiopia and
Ghana, farmers in focus groups reported on significant changes in the start of the rainy season
and in the length and intensity of rainfall. In Ghana, male and female farmers reported that they

had responded to the variable precipitation and higher temperatures by planting drought- and
heat-resistant crops, selecting crops with a short gestation period, planting vegetables along river
banks for easier access to water, shifting planting dates forward or backward, and sowing half the
plot later to spread the risk of early or late rains. Farmers in Bolivia also note adaptations in
agricultural practices with climate change, including using new seed varieties and turning over
pastureland to cropland in the Alturas (highlands), where temperatures have risen.

In these settings, the coping strategies of the poorest farmers are even more constrained under
conditions of climate change, leaving them less room for implementing adaptation responses. For
example, a focus group of vulnerable women in rural Ghana noted that the Nandana (poorer
people) lack collateral for loans and thus have to beg other community members for leftover
seeds to sow. They are therefore the last to sow their crops and miss crucial sowing dates.

In all the case study countries, land was identified as a policy area with an important bearing on
potential climate adaptation activities. Land tenure systems affect poverty outcomes directly. For
example, priority adaptation investments are expected to include investments in water
infrastructure (including irrigation) to cope with growing freshwater scarcity. However, the
greatest impacts of such irrigation investments on poverty reduction have been found in countries
with low levels of inequality in land holdings (Hussain 2005). Land inequity is greatest for
women. In Tetauku, Ghana, members of an EACC focus group discussion on the elderly declared
that “Women do not own land; even their own children who are boys have more inheritance
rights than th