Climate Change Economics, Vol. 6, No.

1 (2015) 1550003 (17 pages)

© World Scientific Publishing Company
DOI: 10.1142/S2010007815500037

AN ECONOMY-WIDE ANALYSIS OF CLIMATE CHANGE IMPACTS

ON AGRICULTURE AND FOOD SECURITY IN BANGLADESH

ONIL BANERJEE*,¶,‡‡, MOOGDHO MAHZAB†,||, SELIM RAIHAN‡,**

and NABIUL ISLAM§,††
*CSIRO Ecosystem Sciences and CSIRO Sustainable Agriculture Flagship
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

Dutton Park, Queensland, Australia

†
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

Bangladesh Institute of Development Studies

E-17 Agargaon, Sher-e-Bangla Nagar, Dhaka 1207, Bangladesh
‡
University of Dhaka and South Asian Network on Economic Modelling (SANEM)
SANEM Dhaka Secretariat:
Flat # K-5, House # 1/B, Road # 35, Gulshan 2, Dhaka 1212, Bangladesh
§
Bangladesh Institute of Development Studies
E-17 Agargaon, Sher-e-Bangla Nagar, Dhaka 1207, Bangladesh
¶
[email protected]
||
[email protected]
**[email protected]
††
[email protected]

Accepted 22 January 2015

Published 13 March 2015

Bangladesh is one of the most vulnerable countries to climate change (CC) with higher tem-
peratures reducing crop yields and sea level rise decreasing arable land supply. The Government
of Bangladesh aspires to offer its people a comparable standard of living to that of middle-
income countries by 2021. Bangladesh’s population will reach 247 million by 2050 and GDP is
projected to grow annually by 7.9%. With increasing population density, greater demand for
resources, and CC impacts, adaptation and mitigation strategies will be required for agricultural
output to meet growing food demand. We develop a dynamic computable general equilibrium
model linked with a food security module to explore CC impacts on agriculture and food
security. Although CC impacts had a relatively small effect on GDP, reducing it by $29,925
million Taka (
0.11%) by 2030, agricultural sector impacts were felt more acutely, reducing
output by
1.23%, increasing imports by 1.52%, and reducing total caloric consumption by
17%, with some households remaining underfed due to inequitable food distribution. Evidence
generated here can guide policy to ensure economic growth contributes to meeting national
development and food security targets.

Keywords: Climate change; food security; food self-sufficiency; agriculture; computable gen-
eral; equilibrium model; Bangladesh.

¶ Corresponding author.
‡‡Current address: Inter-American Development Bank, 1300 New York Avenue NW, Washington, DC 20577, USA.

1550003-1
 O. Banerjee et al.

Highlights
. Climate change impacts on the Bangladesh economy will reduce GDP by 29,925
million Taka (
0.11%) by 2030.
. Agricultural output will fall by
1.23% and domestic food consumption will be
more dependent on imports, increasing by 1.52% by 2030.
. Total caloric consumption will be reduced by 17%, the nutritional composition of
some households’ diet will worsen and some will still go hungry due to distribu-
tional issues.
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

1. Introduction
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

Bangladesh’s climate is subtropical monsoonal exhibiting high seasonal variation in

precipitation and temperature (Ali, 2002). Summers are hot and wet, while winters can
be quite dry, resulting in drought in some regions (Yu et al., 2010). During the
monsoon season, up to two-thirds of the country may be inundated, and cyclones and
storm surges occur frequently. In a country where water would seem to be abundant,
water resources are under great stress (Shahid and Hazarika, 2010).
Sources of water for drinking, irrigated agriculture, and industry are under real
threat from saline intrusion in the southwestern parts of the country, the ingress of
polluted surface waters, and arsenic contamination of shallow groundwater
(Chowdhury, 2010). Major urban centers are already challenged to meet demand while
population growth and economic development will put additional strain on water
resources.
Exacerbating these challenges, Bangladesh is one of the most vulnerable and ex-
posed countries to climate variability and change. Temperatures are projected to rise
implying greater crop evapotranspiration (Yu et al., 2010). Impacts on agricultural
output will be determined by how precipitation may compensate for increased
evapotranspiration, the carbon dioxide fertilization effect, the potential to increase
irrigation of affected crops and bring new areas into agricultural production (Cline,
2007). Sea level rise is also projected to reduce the supply of arable land, increase
storm surges, and affect settlement patterns, fisheries, and tourism (IWM and CEGIS,
2007). Increased surface water inflows into Bangladesh and potentially greater mon-
soonal precipitation will increase flooding risk during the wet season.
The Government of Bangladesh aspires to offer its people a comparable standard of
living to that of middle- and high-income countries by 2021 (Planning Commission,
2010). The Government has well-defined targets and strategies to achieve this aim and
places significant emphasis on achieving food security and reducing poverty. Ban-
gladesh’s population is increasing at an Average Annual Growth Rate (AAGR) of
1.29% and will reach 247 million by 2050 (BIDS, 2013b). Gross domestic product
(GDP) is projected to grow on average at 7.9% until 2050 (BIDS, 2013a). This
projection is good news since the World Bank has assessed that Bangladesh would

1550003-2
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

need to grow between 7% and 8% per annum if it is to achieve its development

ambitions (World Bank, 2011).
Considering projected economic and population growth, we develop a recursive
dynamic computable general equilibrium (DCGE) model to evaluate how climate
change (CC) may affect agriculture and Bangladesh’s economy overall. Focusing on
the access pillar of food security,1 we consider CC impacts on food production and
households’ ability to purchase food. Following the direct caloric intake approach and
the national food poverty line of 2122 kcal, we develop a food security module and use
outputs of the DCGE to evaluate if this food security target is met for various
household categories.
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

DCGE models provide advantages over other analytical frameworks offering a

consistent theoretical lens through which socioeconomic and environmental trade-offs
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

are analyzed. DCGE models have a high degree of explanatory power where inter-
sectoral linkages and resource constraints may be binding (Banerjee and Alavalapati,
2010). Static one-period models can inform on orders of magnitude and direction of
effect of a policy or environmental shock in the short or long-run while dynamic
models enable a more precise specification of time, illustrating economic transition
paths, and short and long-term costs and gains (Cattaneo, 1999).
This study builds on the work of Yu et al. (2010) by considering how CC impacts
on the agricultural sector could affect household food security. Furthermore, this
analysis customizes the most recent social accounting matrix (SAM) for Bangladesh,
developed by Khondker and Raihan (2011) reflecting recent structural changes in
Bangladesh’s economy. Two core scenarios are implemented. In the first scenario, the
baseline forecast, Bangladesh’s economy is projected from the base year of 2006/2007
to 2030 imposing growth projections based on historical tendencies and in the absence
of CC. In the second scenario, the CC scenario, we impose assumptions on the model
that reflect current understanding of CC impacts (Huq, 1999; IWM and CEGIS, 2007;
Yu et al., 2010). The impacts we simulate are (i) sea level rise and increased annual
flooding reducing arable land supply, and; (ii) increased crop evapotranspiration
resulting from higher than average temperatures. The differences in macroeconomic
indicators, output, and income between the baseline forecast and the CC scenario
reflect the effect of CC on Bangladesh’s economy.
This paper is organized as follows. Following this introduction, the research
methods are presented, providing an overview of the DCGE approach, a preliminary
exploration of the core data source, the SAM, and the DCGE workflow design.
Section 3 describes the CC scenarios and Sec. 4 presents the results of the scenario
analysis. Section 5 develops a food security module to evaluate CC impacts on food

1Food security was defined in 1996 at the World Food Summit as a situation “when all people at all times, have
physical, social and economic access to sufficient, safe and nutritious food to meet their dietary needs and food
preferences for an active and healthy life” (FAO, 1996). Building on this definition at the World Summit of Food
Security in 1999 the four pillars of food security were established as availability, access, utilization and stability, with
nutrition an integral dimension of the system (Ecker and Breisinger, 2012).

1550003-3
 O. Banerjee et al.

security. The final section provides a discussion on the policy implications of the
findings.

2. Methods
2.1. A DCGE approach
The model developed here is based on the International Food Policy Research Insti-
tute’s (IFPRI) Standard CGE Model and Robinson and Thurlow’s dynamic extension
to the model (Robinson and Thurlow, 2004). This model is implemented in the General
Algebraic Modeling System (GAMS) and solved as a mixed complimentary problem
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

using the PATH solver. This model was developed by IFPRI to facilitate the use of
CGE models in developing countries (Lofgren et al., 2002). Although this model is
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

thoroughly documented in Lofgren et al. (2002) and Robinson and Thurlow (2004),
the basic model structure is presented here.
The DCGE model describes the behavior of agents in their economic environment;
it is a system of equations describing the utility maximizing behavior of consumers,
profit maximizing behavior of producers, and the equilibrium conditions and con-
straints imposed by the macroeconomic environment. Agent behavior is represented by
linear and nonlinear first-order optimality conditions and the economic environment is
described as a series of equilibrium constraints for factors, commodities, savings and
investment, the government, and rest of the world (RoW) accounts (Lofgren et al.,
2002). The model may be broken into a series of blocks, namely: production, factor
markets, institutions, commodity markets, and macroeconomic balances. These model
blocks are discussed in turn.
The model’s structure enables a given activity to produce more than one com-
modity, while any one commodity may be produced by more than one activity. Firms
are price takers and minimize costs subject to nested technological constraints. At the
top of the nest, sectoral output is determined by combining value-added with inter-
mediate consumption through a fixed share, Leontief production function. Composite
labor is a constant elasticity of substitution (CES) function of various types of labor
indicating imperfect substitution between types of labor. Composite capital and land
are also formed in this way. Value-added is created by a CES function of factors (labor,
capital, and land) where firms employ factors until the value of the factor’s marginal
product is equal to the factor price. Goods and services consumed as intermediate
inputs are treated as perfectly complimentary with aggregate intermediate consumption
formed by a Leontief function.
Households receive income from labor, capital, land, and transfers from other
agents including remittances from abroad. Factor income is apportioned to households
in fixed shares while income from transfers is the sum of all transfers for each
household category. Households pay direct taxes and make transfers to the govern-
ment. Disposable household income is equal to household income net of transfers,
taxes, and savings. Household savings are a linear function of disposable income.

1550003-4
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

Firms receive income from returns to capital and transfers from other agents. Firms pay
income tax and also save. The government receives income from income tax paid by
firms and households, indirect taxes on goods and services, capital taxes, import
duties, production taxes on industries, payroll taxes from labor, export taxes, and
income from capital and transfers.
Income taxes for firms and households are a linear function of their total income.
The RoW receives income from the sale of imports, returns to capital, and transfers
while foreign spending consists of export purchases and transfers to agents in the
domestic economy. Transfers to households and firms are treated as proportional to
their disposable income while household transfers to government are treated as a linear
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

function of total income.

Goods and services are demanded by households, the government, investment,
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

and as transport and trade margins. Households have a Stone–Geary utility function,
with a linear expenditure system (LES) describing household consumption. In a
LES, households use their income to first consume a minimum level of subsistence
goods and services. With the supernumerary income remaining, households purchase
goods and services according to a linear relationship between income and consump-
tion. LES differ from CES functions in that LES functions have nonunitary income
elasticities between all pairs of goods (Annabi et al., 2006) enabling flexibility with
regards to substitution possibilities in response to changes in relative prices (Decaluwé
et al., 2010).
Investment demand is composed of gross fixed capital formation (GFCF) and
changes in inventories. GFCF is endogenous with total investment expenditure bal-
anced by the savings and investment constraint where savings is endogenous. Inven-
tory changes are exogenous in the model and fixed in volume. Investment in goods and
services occurs in fixed shares where investment in a commodity is inversely related to
its price. Government expenditures for a given budget also follow this logic.
The small country assumption is made and import and export prices are exogenous.
A constant elasticity of transformation (CET) function describes how industry output
responds to changes in prices. This functional form implies that an industry may
reorganize production in response to changes in prices, though they cannot perfectly
switch from the production of one commodity to another. Industries allocate output to
domestic and foreign markets based on the assumption that the goods destined to one
market are different from those destined to another market. This assumption is oper-
ationalized through a CET function. Domestic producers are able to capture greater
foreign market share by reducing their price relative to the exogenous world price for
that good or service. This specification enables changes in world demand to be sim-
ulated. To reflect heterogeneity in goods and services with regards to their origin,
goods, and services consumed domestically are aggregate goods composed of do-
mestically produced and imported goods.
Model dynamics specify adjustments to account for endogenous investment and
exogenous population and labor force growth, depreciation, and changes in total factor

1550003-5
 O. Banerjee et al.

productivity. Capital is sectorally allocated as a function of the rate of capital depre-

ciation and the differential in profits between sectors from the previous period
(Robinson and Thurlow, 2004). Endogenous adjustments to account for capital ac-
cumulation and exogenous adjustments to population, labor force, and total factor
productivity are discussed in turn.
Capital supply is based on the previous period’s capital stock and allocation of
investment spending. Investment is carried out in proportion to a sector’s share in
economy-wide capital income and is adjusted by the ratio of a sector’s rate of profits
and the economy-wide average rate of profit. This specification implies that a sector
with higher than average profits will receive a larger share of investment than its
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

average share in aggregate capital income.

Population growth has a direct and positive impact on household consumption
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

expenditure. The quantity of income-independent consumption increases at the same

rate as population growth. The level of minimum household consumption expenditure
also increases proportionally with population growth. Growth affects average rather
than marginal consumption demand implying that new consumers share the same
preferences as existing consumers. The subsistence consumption of commodities is
adjusted upwards by the rate of population growth.
With flexible nominal wages and full employment, the between period levels
of labor supply increase according to the rate of labor force growth, which grows
at the same rate as the population. Total factor productivity is exogenous and
determined by a technological parameter introduced in the model equations for
the calculation of the quantity of aggregate value-added. The efficiency parameter
in the CES value-added function is shifted upwards by total factor productivity
growth.
There are three macroeconomic balances in the model: the government current
account balance, the current account of the balance of payments, and the savings and
investment balance. Decisions regarding macroeconomic balances in the model are
known as closure rules and are necessary to maintain a balanced economic environ-
ment. The choice of closure rules may have a significant impact on model behavior
(Dewatripont and Michel, 1987).
The balanced closure used in this study is recognized to provide a more accurate
representation of how real economies have behaved (Lofgren et al., 2002). In this
closure, modeling experiments are conducted in a macroeconomic environment
where investment and government consumption shares are fixed while their quan-
tities are flexible. Nominal absorption shares of investment and government con-
sumption are fixed at their base year levels. For factor closures, labor, capital, and
land are fully employed and mobile between sectors. A flexible real exchange rate is
chosen for the RoW closure while the government closure fixes direct tax rates
enabling flexible government savings. The consumer price index is chosen as the
numeraire.

1550003-6
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

2.2. SAM for Bangladesh (2006/2007)

The core data source for a DCGE is the SAM. A SAM is a statistical representation of
an economy which describes payments and receipts between economic agents, factors,
and intermediate and final goods and services. We customize the most recent SAM
available for Bangladesh. Developed by Khondker and Raihan (2011), this SAM was
used to assess the economic implications of Bangladesh’s 6th five year plan for de-
velopment (Khondker and Raihan, 2011). The new SAM is based on an input–output
table for 2000, as well as the most up to date information available on production and
consumption. The base year of 2006/2007 was chosen since it is the most recent year
for which comprehensive data on Bangladesh’s economy was available. Furthermore,
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

it was considered a relatively normal year, unaffected by the recent global economic
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

crisis and other shocks such as severe weather events. The SAM construction proce-
dures are documented in detail in Khondker and Raihan (2011).
We customize the SAM to focus on the agricultural sector. The economy is ag-
gregated to 27 sectors/commodities, 14 of which are related to agriculture and food
production (Table 1).
The SAM contains skilled (  class level of 10) and unskilled labor (class 0 to 9)
categories, capital, and land as factors of production. There are 11 institutions, eight of
which are households, six rural and two urban. Households are disaggregated
according to the Household Income and Expenditure Survey (HIES) classification
system. The HIES has been conducted in Bangladesh since 1973–1974, collecting data
on income, expenditure, consumption, savings, housing, education, employment,
and many other indicators. Table 2 details the characteristics of the households in
the model. Rural agricultural households are described according to their land
endowments; nonagricultural households are distinguished by whether or not they are

Table 1. Sectors in the 2006/2007 Bangladesh SAM.

Agricultural sectors Nonagricultural sectors

1. Rice 15. Forestry

2. Wheat 16. Oil
3. Other grains 17. Water and electricity
4. Potato 18. Housing
5. Vegetables 19. Health
6. Pulses 20. Education
7. Other crops 21. Public administration
8. Fruit 22. Manufactured goods
9. Livestock 23. Construction
10. Poultry 24. Mining and gas
11. Fish 25. Trade
12. Milled rice 26. Transport
13. Milled grain 27. Services
14. Processed food

1550003-7
 O. Banerjee et al.

Table 2. Household endowment categories.

Household type

1. Landless (0 ha)
2. Marginal (  0.198 ha)
3. Small agricultural (0.202 to 1.008 ha)
4. Large agricultural (agricultural > 1:012)
5. Rural nonagricultural poor
6. Rural nonagricultural nonpoor
7. Urban educated
8. Urban less educated
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

poor. The two urban households types disaggregated by level of education with less
educated households possessing less than or equal to 8th class education and higher
educated households possessing greater than 8th class education. The remaining three
institutions are the government, firms, and the RoW. The final two accounts in the
SAM are public and private investment, and inventories.

3. Scenario Design
The first scenario is the baseline scenario which modeled Bangladesh’s economy from
the base year of 2006/2007 to 2030 assuming historical tendencies and in the absence
of CC. Such factors, productivity, yield, and the overall economy followed a balanced
growth path. The second scenario simulated projected CC impacts and was informed
by future climate scenarios and agricultural crop modeling conducted by Yu et al.
(2010). Estimates of sea level rise were based on IWM and CEGIS (2007).
Estimates for future temperature and precipitation changes from Yu et al. (2010)
were based on analysis of 16 global circulation models for A1B, A2, and B1 emissions
scenarios. Results indicated positive temperature changes for every experiment and
every month with a median warming of 1.1 C by 2030. With regards to precipitation
by 2030, it was not possible to discern clear trends, though by 2050, some models
predicted a trend of increased annual and wet season precipitation. Median estimates
predicted precipitation may increase up to 4% over the baseline by 2050 (Thurlow
et al., 2011).
Yu et al. (2010) used the Crop Environment Resource Synthesis (CERES) modeling
system to estimate CC impacts on crop output. The authors reported the joint impact
on crop output due to changes in temperature and precipitation, coastal and inland
flooding, and a carbon dioxide enrichment effect on crop yields. In the CC scenario,
we used an output-weighted average of median production changes for the A2 and B1
emissions scenarios and applied these linearly over the period of analysis. Median
production percentage changes were estimated as
0.25%,
0.37%,
3.06%, and
2.05% for aus, aman, and boro rice varieties and for wheat, respectively (Yu et al.,

1550003-8
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

2010). This weighted average for rice output amounted to a
1.8% change in pro-
duction by 2030.
Estimates of sea level rise were provided by IWM and CEGIS (2007). Sea level rise
of 15 and 27 cm were estimated to result in a loss in agricultural land of 1.5% and
2.5%, respectively (IWM and CEGIS, 2007). Following Thurlow et al. (2011) and
considering an average rate of expansion of agricultural land of 1.0% per year, we
implement an evenly distributed loss of 1% of agricultural land over the period of
analysis.

4. Results
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

Table 3 provides an overview of Bangladesh’s economy in the base year. GDP in

Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

2006/2007 was over 4.7 trillion Bangladesh Taka (BDT). The investment share of
GDP accounted for over 25% of GDP while imports surpassed exports by more than
327 billion BDT.
Rural landless households spend 47% of their food budget on milled rice while
urban educated households spend 26%. Expenditure shares on other crops show that
wealthier households spend a smaller proportion of their income on food staples such
as milled rice and instead consume more meat protein and a considerably greater share
of processed food.
To facilitate analysis of the relative impacts on macroeconomic indicators, results
reported in Table 4 are the differences in AAGR between the CC scenarios and the
baseline scenario. All macroeconomic indicators, with the exception of government
consumption, grew faster in the baseline. Introducing the climate shock, the AAGR of
GDP grew slower by
0.0047% or by
0.11% in 2030. Real GDP at market prices
(gross value-added less intermediate consumption) in the base year was 4.70 trillion
BDT and grew to 27.55 trillion BDT by 2030. With CC, GDP in 2030 was 29,925
million BDT less, equivalent to 6313 million BDT in net present value terms (7% rate
of discount). The CC impact on GDP was equivalent to approximately 11% of total

Table 3. Bangladesh macroeconomic aggregates

(2006/2007).

GDP component Millions of Taka

1. Private consumption (C) 3,561,043

2. Public consumption (G) 261,056
3. Investment (I) 1,229,898
Private 939,567
Public 257,258
Change in stock 33,073
4. Exports (X) 934,403
5. Imports (M) 1,261,628
GDP at market prices 4,724,772

1550003-9
 O. Banerjee et al.

Table 4. Decomposed climate shock percent deviation from baseline in AAGR.

Macroeconomic indicator Sea level rise (%) Paddy yield (%) Wheat yield (%) Joint impact (%)

Private consumption
0.0036
0.0130 0.0001
0.0165

Fixed investment 0.0057 0.0071 0.0001 0.0129
Government consumption 0.0000 0.0000 0.0000 0.0000
Exports 0.0125 0.0168
0.0001 0.0293
Imports 0.0120 0.0162
0.0001 0.0282
GDP 0.0001
0.0049 0.0001
0.0047

government expenditure in 2006/2007. Private consumption grew slower by
62,804

by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

million BDT (
0.0165%) by 2030. Fixed investment grew faster by 32,879 million
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

BDT (0.0129%) by 2030, as did exports and imports (0.0293% and 0.0282%,
respectively, each increasing by 65,800 million BDT by 2030).
Equation (1) is the GDP equation, where C is private consumption, I is investment,
G is government expenditure, X are exports and M are imports. In the CC simulation,
the positive impact of faster export and investment growth was insufficient to offset
slowed growth in consumption and increased import growth, resulting in slightly
slower GDP growth.
GDP# ¼ C# þ I" þ G þ X"
M": (1)
Table 5 provides an overview of the joint impact of climate shocks on trade, output,
and prices. Figures in bold indicate slowed growth in the baseline. Domestic agri-
cultural output in the baseline contracted for wheat, other grains, vegetables, pulses,
and fruit. The CC impact slowed growth in paddy, milled rice, and potato output while

Table 5. Deviation in AAGR from baseline due to climate shock for trade, output, and prices.

Commodity Imports Exports Domestic Composite Composite

(%) (%) output (%) output (%) price (%)

Wheat
0.0122 0.1344 0.0913
0.0069
0.0696

Other grains 0.0385
0.2021
0.1314
0.0185
0.0488
Potato 0.1105
0.1584
0.0793
0.0354
0.0192
Vegetables 0.0233
0.2508
0.1702
0.0164
0.0546
Pulses 0.0362
0.2511
0.1666
0.0180
0.0497
Fruit 0.0348
0.3171
0.2136
0.0147
0.0513
Other crops 0.0378
0.1326
0.0825 0.0005
0.0554
Livestock 0.1355
0.1158
0.0419
0.0408
0.0090
Poultry 0.0441
0.0305
0.0085
0.0074
0.0506
Fish
0.0156
0.0037
0.0072
0.0072
0.0706
Milled rice 0.3015
0.3118
0.1314
0.0760 0.0580
Milled grain
0.0178
0.0015
0.0063
0.0064
0.0716
Processed food
0.0066
0.0226
0.0113
0.0110
0.0643

1550003-10
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

Table 6. Deviation in AAGR from baseline due to climate shock for household income.

Household income Sea level rise (%) Paddy yield (%) Wheat yield (%) Joint impact (%)

Landless
0.0209
0.0486 0.0002
0.0694

Marginal
0.0179
0.0442 0.0003
0.0619
Small farmers
0.0156
0.0402 0.0004
0.0557
Large farmers
0.0115
0.0332 0.0005
0.0445
Rural nonagricultural poor
0.0190
0.0453 0.0003
0.0642
Rural nonagricultural not poor
0.0171
0.0451 0.0003
0.0621
Urban less educated
0.0226
0.0482 0.0003
0.0707
Urban educated
0.0181
0.0468 0.0003
0.0648
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

output for other commodities contracted further. Overall, domestic agricultural output
declined by
1.23% by 2030. Composite output for all agricultural commodities
(a composite of domestically produced and imported agricultural goods) grew in the
baseline. CC resulted in slower growth, with the exception of other crops. Agricultural
prices in the baseline fell for wheat, livestock, poultry, fish, milled grain, and processed
food. The climate shock further depressed prices for wheat, livestock, poultry, fish, milled
grain, and processed food, and slowed growth in the case of other grains, potato, vegetables,
pulses, fruit, and other crops. The price of paddy and milled rice grew faster with CC.
All agricultural imports grew in the baseline. CC caused faster import growth for
most commodities, with the exception of wheat, fish, milled grain, and processed food.
In 2030, agricultural imports were 1.52% higher with CC. Relative to the increased
rates of growth of other agricultural commodities, the impact on milled rice imports
was the greatest (0.3015%). In the baseline, export growth of wheat, other grains,
potatoes, vegetables, pulses, fruit, other crops, and milled rice slowed. With CC,
exports contracted further across the board with the exception of wheat and in 2030,
agricultural exports were
0.28% lower.
For all household categories, income grew in the baseline; CC impacts slowed this
growth somewhat (Table 6). Large farmers appeared to be the most insulated from
climate shocks, with the urban less educated the most vulnerable.
With regards to factors, as land became scarcer as a function of sea level rise and
increased demand for agricultural commodities, the joint impact of the climate shock
led to a 0.0231% increase in the AAGR of land rentals over the baseline. For labor and
capital, the climate shock slowed AAGR by between
0.0718% and
0.759% for
skilled and unskilled labor, respectively and by
0.0743% in the case of capital.

5. Food Security and Poverty

5.1. Analytical approach
In this analysis, we applied the direct caloric intake method to evaluating food
security using a food poverty line of 2122 kcal/day/person and a food security

1550003-11
 O. Banerjee et al.

module. This food poverty line has been in use for many years in Bangladesh and is
the standard used by the Bangladesh Bureau of Statistics. The food poverty line
corresponds to the minimal nutritional requirements to maintain a healthy life. A key
advantage of this approach is that it controls for differences in household purchasing
power through time (Ravallion and Sen, 1996). Our analysis utilized the latest HIES
data for 2010.
The first step of the analysis involved specifying a fixed food bundle which satisfied
the caloric requirement of the poverty line. Similar to Ravallion and Sen (1996), the
food bundle was specified to contain 10 food items: rice, potato, vegetables, pulses,
fruit, livestock, fish, grains, poultry, and processed food (Ravallion and Sen, 1996).
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

Next, for each household category from the SAM, expenditures on the 10 food bundle
items in the base year were extracted and divided by their price per kilogram to obtain
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

the quantity of each food item consumed.

A caloric conversion table was used to determine the caloric consumption per
household category. This caloric intake was compared to the minimum requirement for
each household category, which was calculated as the product of the 2122 kcal/person/
day food poverty line and the population of each household category. Comparing these
two figures, whether or not a household category had a caloric deficit or surplus was
deduced. Calculations were performed in the same manner for the results from the
baseline forecast and the CC scenario for the year 2030.
For these calculations, we assumed households had equal access to food within
household categories (i.e., all households within a household category consume the
same number of kcal), and the proportion of the population in each household category
remained constant over the period of analysis. Furthermore, the landless and marginal
households were aggregated into a landless/marginal category and the rural nonagri-
cultural poor and rural nonagricultural nonpoor were aggregated into a nonagricultural
rural category.
For purposes of comparison, household consumption for a subset of the food bundle
in the base year of 2007 was compared with the desirable caloric composition as
determined by the World Food Programme. Taking urban less educated households as
a test case, the actual and desirable caloric composition is reported in Table 7. This
household category is food insecure in the base year.
This urban less educated household category consumed a greater proportion of rice
(63% versus 62%), grains (17% versus 5%), and potatoes (7% versus 3%) than de-
sirable. More costly food items such as vegetables, pluses, fruits, and meat were
estimated to be consumed at a proportion less than what a balanced diet demands.

5.2. Food security results

The landless/marginal, small, and large farming households, and both urban educated
and urban less educated households had a caloric deficit in the baseline. Nonagri-
cultural rural households had a caloric surplus. In the absence of the CC shock, by

1550003-12
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

Table 7. Food intake in the base year, 2030 with CC and desirable food intake.

Food item Actual kcal % 2030 kcal intake % Desired % of

intakea total with CCa total kcal intakeb desired

Rice 532 63 898 51 1,247 62

Grain 147 17 487 28 105 5
Potato 62 7 92 5 59 3
Vegetables 24 3 48 3 165 8
Pulses 23 3 42 2 136 7
Fruit 9 1 14 1 42 2
Animal-based food 30 4 92 5 237 12
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

Processed food 23 3 77 4 34 2
Total 850 100 1,750 100 2,025 100
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

a Model results.
b Follows the World Food Programme’s food composition recommendations.

2030 the situation was much improved as a result of sustained economic growth, with
all household categories meeting the minimum caloric requirements. When the CC
shock was imposed, however, urban less educated households became calorie defi-
cient. Considering all household categories together, in the base year there was a
caloric deficit of 46 billion kcal. In 2030 in the baseline forecast, there was a surplus of
over 516 billion kcal. With CC, the surplus was reduced to 355 billion kcal.
Figure 1 presents the percentage deficit or surplus with regards to the minimum
caloric requirement of 2122 kcal. In the baseline in 2007, nonagricultural rural house-
holds were the most food secure with an intake of 63% above the minimum while the
urban less educated were the most food insecure with intake 60% below the minimum.
Next were the small farmers, urban educated and landless/marginal households falling
short of the minimum by
36%,
33%, and
32%, respectively. In the baseline in 2030,
urban less educated households just met the minimum while the rest surpassed it.

Figure 1. Modeled household food security.

1550003-13
 O. Banerjee et al.

In the CC scenario, urban less educated households fell below the minimum caloric
requirement (
17%) and were food insecure. Furthermore, Table 7 shows that the
nutritional composition of this household category’s consumption worsened, with
greater dependency on grains and no change in the consumption of vegetables and
fruits. This household category’s consumption of nutrient rich and animal-based foods
remained well below the desired intake with CC in 2030. Landless marginal, small
farming, urban educated, large farmers, and nonagricultural rural households all sur-
passed the caloric minimum by 45%, 50%, 60%, 91%, and 224%, respectively. On the
whole, CC reduced total household caloric consumption by 17%.
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

6. Discussion and Policy Implications

Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

Economic growth, population dynamics, and changes in preferences towards more

input-demanding, higher value crops are driving demand for agricultural commodities.
Global food stocks have been on the decline, increasing the exposure of developing
and developed countries to shortfalls due to climate variability, risk, and speculator
behavior. CC will further exacerbate the challenges Bangladesh faces in meeting its
growth and food security targets. Rising temperatures will increase crop evapotrans-
piration reducing crop yields while sea level rise will make cultivable land scarcer. To
counteract these forces, some exogenous in nature, Bangladesh will have to make
more effective use of increasingly scarce resources. In this paper, we developed a
DCGE model linked to a food security module to explore the degree to which CC may
impact the agricultural sector and the economy overall. In the face of CC, we used
modeling results to shed light on the country’s trajectory toward a more food secure
future.
We found that CC imposed on the agricultural sector had a relatively small effect on
GDP, reducing it by 29,925 million BDT in 2030 (
0.11%), or by 6313 million BDT
in present value terms. For the agricultural sector, the CC impact resulted in con-
traction of some subsectors or slower growth in others, slower export growth and
higher rates of growth in agricultural imports, particularly for rice. In 2030, agricultural
sector output and exports were
1.23% and
0.28% lower, respectively with CC,
while agricultural sector imports were 1.52% higher. Prices for most goods and ser-
vices grew slower than in the baseline or contracted even further. Household income
grew slower as a result of CC impacts.
Although the CC impacts simulated showed a relatively small effect on overall
economic growth at least in the short run until 2030, results of our food security
analysis reveal that CC will pose real challenges to achieving food security. In the
absence of CC, all household categories became food secure by 2030. With CC,
although the food security of urban less educated households improved, they remained
food insecure. In addition, the nutritional composition of this household category’s
consumption worsened, with a greater dependency on grains and a less than desirable
share of calories derived from fruit, vegetables, and animal-based foods. Overall, CC

1550003-14
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

reduced Bangladesh’s total caloric consumption by 17% compared with the baseline
projection.
The fact that there is considerable food insecurity in the initial years of the modeling
exercise, improving with the prospect of enhanced growth, is of policy relevance. The
trajectory towards food security differs both temporally and spatially, indicating the
need for policies that can leverage the positive impact of growth and maximize
the number of households that move towards a food secure future at the least cost.
While data availability limited the possibility of a spatially disaggregated SAM, it is
reasonable to expect that CC impacts will be felt more acutely in the most vulnerable
regions such as coastal and drought prone areas. The spatial dynamics (rural/urban,
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

coastal/inland, upland/lowland) of food security will require the development of

targeted policies and programs.
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

Findings indicated that rice imports will grow faster as a result of CC. For
years, Bangladesh’s government has maintained a policy of national food self-
sufficiency, using rice imports primarily as a mechanism for maintaining domestic
price stability. Although the economic model developed here allows for endogenous
growth of imports based on domestic to foreign price ratios, in reality this may not
occur. From a political perspective, a policy of meeting food security targets through
greater dependence on foreign markets is unlikely to be politically acceptable, given
Bangladesh’s experience with recurring famine (Banerjee et al., 2014). The impli-
cations of this are that future caloric deficits may be greater than those estimated
here. To compensate for this gap, domestic agricultural output would have to
grow faster.
Crop yield impacts and sea level rise are only two potential CC impacts that were
simulated in this analysis. Other consequences of CC may also have a significant
impact on the economy and households, for example the potential health impacts and
their implications for many aspects of development including labor productivity. Also
not evaluated was the potential for a trend of increasing intensity and frequency of
extreme weather events. These events could include cyclonic storms, severe flooding
and severe drought. With sea level rise, saline intrusion is also a concern where
saltwater pushes its way upstream, contaminating groundwater supplies and rendering
once cultivable areas unmanageable. The inclusion of these and other potential CC
impacts would exacerbate the economy-wide and household-level food security
responses presented here.
Bangladesh has been adapting to climate variability and change for decades. In-
creasing waterlogged areas, for example, has prompted research into crop varieties
tolerant of such conditions. With groundwater extraction for irrigation in some areas of
the country’s northwest beginning to show signs of stress, various policies and pro-
grams have been encouraging the agricultural development of coastal regions. Salt-
tolerant crops and appropriate management strategies are being pursued to overcome
the challenges characteristic of coastal regions. The development of higher yielding
varieties and other technological innovations will also help meet growing demand.

1550003-15
 O. Banerjee et al.

Nonetheless, to keep pace, investment in research and development will be critical to

continue to encourage this form of innovation.
This study demonstrated that although CC as modeled here, had a relatively small
impact on overall growth in the short run until 2030, it does present challenges for
achieving household-level food security. Overall, Bangladesh is likely to produce
enough food to satisfy and even surpass its caloric requirements by 2030. Despite
surplus production, however, some will still go hungry and undernourished. This
phenomena relates to both inter- and intra-household distribution of food and nutrition.
The results presented here show that underfed individuals persist in some household
categories and the nutritional composition of their diets worsens. Improving food
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

distribution is a key policy imperative if Bangladesh is to achieve its vision of offering

its people a comparable standard of living to that of middle-income countries in the
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

near term. Once considered a “test case of development”, with its advances in social
welfare and economic growth in the last decade, Bangladesh has certainly demon-
strated it is up for the challenge.

Acknowledgments
This work was funded through the AusAID-CSIRO Research for Development Alli-
ance. Thanks to Mac Kirby and Christian Roth of CSIRO and the joint University of
Queensland and CSIRO Livelihoods and Resources Forum (LARF) for the valuable
discussions.

References
Ali, L (2002). An Integrated Approach for the Improvement of Flood Control and Drainage
Schemes in the Coastal Belt of Bangladesh. The Netherlands: A. A. Balkem Publishers.
Annabi, N, J Cockburn and B Decaluwé (2006). Functional forms and parameterization of
CGE models. Available at http://papers.ssrn.com/sol3/papers.cfm?abstract id¼897758.
Banerjee, O and J Alavalapati (2010). Illicit exploitation of natural resources: The forest
concessions in Brazil. Journal of Policy Modeling, 32(4), 488–504.
Banerjee, O, T Darbas, PR Brown and CH Roth (2014). Historical divergence in public
management of foodgrain systems in India and Bangladesh: Opportunities to enhance food
security. Global Food Security, 3, 159–166.
Bangladesh Institute of Development Studies (BIDS) (2013a). Gross domestic product pro-
jections, Bangladesh. Bangladesh Institute of Development Studies, Dhaka.
Bangladesh Institute of Development Studies (BIDS) (2013b). Population projections, Ban-
gladesh. Bangladesh Institute of Development Studies, Dhaka.
Cattaneo, A (1999). Technology, Migration and the Last Fronteir: A General Equilibrium
Analysis of Environmental Feedback Effects on Land Use Patterns in the Brazilian Amazon.
Baltimore: John Hopkins University.
Chowdhury, NT (2010). Water management in Bangladesh: An analytical review. Water Policy,
12(1), 32–51.
Cline, WR (2007). Global Warming and Agriculture: Impact Estimates by Country. Washington,
DC: Center for Global Development: Peterson Institute for International Economics.

1550003-16
 An Economy-Wide Analysis of CC Impacts on Agriculture and Food Security in Bangladesh

Decaluwé, B, A Lemelin, V Robichaud and H Maisonnave (2010). The PEP standard

computable general equilibrium model single-country, recursive dynamic version,
PEP-1-t. Available at http://www.pep-net.org/programs/mpia/pep-standard-cge-models/
pep-1-t-single-country-recursive-dynamic-version/.
Dewatripont, M and G Michel (1987). On closure rules, homogeneity and dynamics in applied
general equilibrium models. Journal of Development Economics, 26(1), 65–76.
Ecker, O and C Breisinger (2012). The food security system: A new conceptual framework.
IFPRI Discussion Papers (1166).
FAO (1996). Rome declaration on world food security and world food summit plan of action.
FAO, Rome.
Huq, S (1999). Vulnerability and Adaptation to Climate Change for Bangladesh. Dordrecht,
Boston: Kluwer Academic Publishers.
by NANYANG TECHNOLOGICAL UNIVERSITY on 04/28/15. For personal use only.

IWM and CEGIS (2007). Investigating the impact of relative sea level rise on coastal com-
munities and their livelihoods in Bangladesh. Institute of Water Modeling and Center for
Clim. Change Econ. 2015.06. Downloaded from www.worldscientific.com

Environmental and Geographic Information Services, Dhaka.

Khondker, BH and S Raihan (eds.) (2011). A social accounting for Bangladesh 2006/07:
Methodology and results. Model Structures, Simulation Results and Related Studies, Vol. 2,
Sixth Five Year Plan of Bangladesh 2011–2015, Dhaka.
Lofgren, H, RL Harris, S Robinson, M Thomas and M El-Said (2002). A standard computable
general equilibrium (CGE) model in GAMS. IFPRI, Washington, DC. Available at http://
www.ifpri.org/pubs/microcom/5/mc5.pdf.
Planning Commission (2010). Outline perspective plan of Bangladesh 2010–2021. Making
vision 2021 a reality, Dhaka. Available at www.unpei.org/. . ./doc. . ./169-pei-bangladesh-
outline-plan-2010.html.
Ravallion, M and B Sen (1996). When method matters: Monitoring poverty in Bangladesh.
Economic Development and Cultural Change, 44(4), 761–792.
Robinson, S and J Thurlow (2004). A recursive dynamic computable general equilibrium
(CGE) model: Extending the standard IFPRI static model. Unpublished manuscript.
Shahid, S and M Hazarika (2010). Groundwater drought in the northwestern districts of
Bangladesh. Water Resources Management, 24(10), 1989–2006.
Thurlow, J, P Dorosh and WH Yu (2011). A stochastic simulation approach to estimating the
economic impacts of climate change in Bangladesh. UNU-WIDER. Available at http://www.
wider.unu.edu/publications/working-papers/2011/en GB/wp2011-086/.
World Bank (2011). Growth in Bangladesh. Available at http://go.worldbank.org/
BBLP60TK71.
Yu, WH, M Alam, A Hassan, AS Khan, AC Ruane, C Rosenzweig, DC Major and J Thurlow
(2010). Climate Change Risks and Food Security in Bangladesh. London; New York:
Earthscan.

1550003-17