1. Food and Nutrition Security: What’s
the Role of Agricultural Policy in Asia?
Siem Reap, Cambodia; September 25-27, 2013
International Conference on “Agricultural Transformation in Asia: Policy Options for Food
and Nutrition Security”
H.E. Srun Darith (moderator); Dr. Akhter Ahmed,
Dr. Kamiljon Akramov, Dr. Olivier Ecker, Dr. Yanyan Liu

2. Introduction
H.E. Srun Darith, Dr. Olivier Ecker

3. Motivation
 Malnutrition slows economic growth (2-3% GDP lost) and deepens poverty
through productivity losses (10% of lifetimes earnings) from poor physical
performance and cognitive capacity as well as increased health care costs.
 Poverty and malnutrition are closely associated and highly prevalent in rural
areas.
 Although agricultural growth has been shown to have high poverty reduction
effects (Christiaensen et al. 2011; Diao et al. 2010; World Bank 2007), empirical
evidence on its nutrition impact is limited and inconclusive (Pinstrup-Anderson
2013; Berti et al. 2004; Masset et al. 2011).
 Nonetheless, development and agricultural policy is often based on the
assumption that agricultural growth—particularly among smallholder
farmers—improves household food security and thereby reduces malnutrition.

4. Pathways from Agricultural Transformation and
Growth to Food and Nutrition Security
Agricultural transformation
and growth driven by
 Demand increase
 Productivity growth due to
policy
reform, investment, techno
logical progress
Purchasing power
increase from
 Income growth
among farmers
 Food price reduction
Change in food self-
sufficiency dependence
among subsistence farmers
Household food
and nutrition
security (in terms
of food quantity
and dietary quality)
Nutrition outcomes
Intra-household resource
allocation, care, education,
health environment

5. Dietary Quality-Growth Relationship
Cambodia
Global trend
GDP per capita
(constant 2005 US$)
Share of calorie supply from staples (%)
Bangladesh
Nepal
Tajikistan
Source: O. Ecker based on data from FAO’s FSI and World Bank’s WDI, complemented with IMF’s WEO and UNSTAT data.

6. Undernutrition-Growth Relationship
Source: O. Ecker based on data from World Bank’s WDI, complemented with IMF’s WEO, UNSTAT, and recent country survey data.
Cambodia
Global trend
GDP per capita
(constant 2005 US$)
Prevalence of underweight among children under 5 years (%)
Bangladesh
Nepal
Tajikistan

7. Dietary Diversity as FNS Indicator
 Dietary quality contributes to an individual’s nutrition and health status and
thereby to people’s economic productivity.
 Households will only diversify their diets into higher-value micronutrient-rich
foods when they have satisfied their basic calorie needs. For the poor, these
foods are often unavailable or unaffordable.
 Dietary diversity is a strong predictor of dietary quality in terms of
(micro)nutrient intake and adequacy (Ruel et al. 2013).
 Household dietary diversity is strongly correlated with per capita calorie
consumption and dietary energy adequacy and is correlated with nutrition
outcome indicators such as anthropometrics (Ruel 2003; Ruel et al. 2013).
 Dietary diversity is responsive to welfare trends and sensitive to shocks and
seasonality, indicating high inter-temporal validity (Headey & Ecker 2013).
 Dietary diversity is measured as a count of different foods or food groups
consumed over a specified reference period.
 All country cases studies use 12-scale or 16-scale household Dietary Diversity
Scores (DDS) as indicator of household food and nutrition security (FNS).

8. Evidence from 4 Country Case Studies
1. Cambodia: Does agricultural transformation slow progress
toward achieving food and nutrition security?
Presented by Dr. Olivier Ecker
2. Tajikistan: Agricultural biodiversity, dietary diversity and
nutritional outcomes
Presented by Dr. Kamiljon Akramov
3. Nepal: Nutritional Intake, Agricultural Production, and Conflict
Presented by Dr. Yanyan Liu
4. Bangladesh: Pathways of impact of agriculture on nutrition
Presented by Dr. Akhter Ahmed

9. Does Agricultural Transformation Slow
Progress toward Achieving Food and
Nutrition Security in Cambodia?
Coauthor: Jean-Francois Trinh Tan
Financial support: United States Agency for International Development (USAID)
Dr. Olivier Ecker

10. Motivation and Research Questions
Cambodia’s Rectangular Strategy (2009-2013) aims at achieving food and
nutrition security through agricultural transformation and growth (p. 13):
 The first Strategic Rectangle to promote broad-based economic growth is the
“enhancement of the agricultural sector”, “especially in the high-potential
agricultural and agro-industrial sectors”.
 “The agricultural policy of the Royal Government is to improve agricultural
productivity and diversification […] to bolster economic growth, create
employment and generate income in the rural areas, thus ensuring
nutritional improvement, food security and increased agricultural exports.”
 This requires “shifting the direction from «expansionary» or «extensive»
agriculture to «deepening» or «intensive» agriculture, especially by
increasing the yields using the existing land through intensification”.
 Does agricultural transformation and growth translate into improved food
and nutrition security (FNS)?
 What are the policy-relevant variables enabling this transmission?

11. Dietary Quality-Growth Relationship
Cambodia
Global trend
GDP per capita
(constant 2005 US$)
Share of calorie supply from staples (%)
1992
2005
2009 SGDP=600 = -0.09
Annual change in
share of calorie
supply (%-points)
Annual GDP
per capita
growth (%)
Arc semi-
elasticity
1992-2005 -0.54 5.30 -0.10
2005-2009 -0.50 5.33 -0.09
Source: Own estimation based on data from FAO’s FSI and World Bank’s WDI, complemented with IMF’s WEO and UNSTAT data.

12. Undernutrition-Growth Relationship
Source: Own estimation based on data from World Bank’s WDI, complemented with IMF’s WEO, UNSTAT, and recent country survey data.
Cambodia
Global trend
GDP per capita
(constant 2005 US$)
Prevalence of underweight among children under 5 years (%)
1996
2005
2010
SGDP=600 = -0.17
Annual change in
child underweight
(%-points)
Annual GDP
per capita
growth (%)
Arc semi-
elasticity
1996-2005 -1.58 6.43 -0.25
2005-2010 0.12 5.13 0.02

13. Agricultural Transformation and Malnutrition
20
25
30
35
40
45
200
300
400
500
600
700
1996 1998 2000 2002 2004 2006 2008 2010
Constant 2005 US$ (PPP) Percent of children (<5 years)
Child underweight
GDP per capita
Agriculture value
added per worker
Agricultural Transformation
Source: Own estimation based on World Bank’s WDI data.

14. Measuring Agricultural Transformation
 Agricultural transformation is characterized by at least four interlinked
developments:
1. Commercialization: From subsistence to market-oriented production
2. Intensification: Total factor productivity (TFP) growth
3. Specialization: Reduction of production diversity on profitable activities
4. Deagrarianization: Moving-out of agriculture
 The effects of agricultural transformation on FNS are likely to substantially
differ across farm households, depending on their characteristics and stage in
the agricultural transformation process.
 Household surveys provide variables indicating farmers’ stage in
transformation including the share of food sales on total food
production, agricultural income, agricultural production diversity, and the
share of off-farm income on total income.

15. Data and Methodology
Data: Cambodia Socio-Economic Survey (CSES) 2009
Methodology: Linear regressions to explore (cross-sectional) correlation between
agricultural transformation indicators and dietary diversity and child nutrition
 Grouping of (farm) households acc. to their agricultural transformation stage
 Dep. var:
 Household Dietary Diversity Score, with a maximum of 16 food groups
 Child weight-for-age z-score, measuring underweight
 Indep. var.:
 Per capita household expenditure; market distance
 Ag. trans. var.: Share of food consumption from purchases, share of non-
farm income, food crop diversity, livestock diversity
 Household characteristics: Household size, gender of household
head, female/mother’s education (primary, secondary), [child sex, age]
 Controlling for district- and month-fixed effects

16. Household Classification
Criterion:
All
10,157 (100%)
Farm
7,930 (78%)
Non-farm
2,227 (22%)
Subsistence
farmer
1,379 (14%)
Source: CSES 2009 data.
Commercial
farmer
2,378 (23%)
Part-time
farmer
4,173 (41%)
Full-time
farmer
3,757 (37%)
Agricultural
production
Share of non-farm
income >|< 50%
Share of food
consumption from
purchases >|< 50%

17. Regression Results: Dietary Diversity
Dep. var.: Household Dietary
Diversity Score
All Farm
Full-time
farmers
Subsistence
farmers
Commercial
farmers
Part-time
farmers
Non-farm
Per capita expenditure (log) 0.425*** 0.391*** 0.401*** 0.353*** 0.434*** 0.395*** 0.433***
Market distance (log) -0.163*** -0.166*** -0.140*** -0.085* -0.183*** -0.180*** -0.041
Farm household (=1) 0.241***
Share of non-farm income 0.156*** 0.358*** 0.513** 0.207 0.198
Share of food consumption from
purchases
0.333*** 0.616*** 1.944*** -0.054 0.111
Food crop diversity (log) 0.008 -0.058 -0.179 0.017 0.005
Livestock diversity (log) 0.113*** 0.160*** 0.138* 0.177*** 0.086*
Household size (log) 0.897*** 0.789*** 0.730*** 0.674*** 0.747*** 0.835*** 1.098***
Female-headed household (=1) 0.125*** 0.131*** 0.091** -0.058 0.140** 0.164*** 0.102
Female education, primary (=1) 0.060** 0.052** -0.009 -0.046 0.012 0.097** -0.004
Female education, secondary (=1) 0.007 0.035 -0.007 -0.009 0.016 0.079 -0.079
Constant 1.931*** 2.101*** 1.694*** 1.957*** 2.006*** 2.064*** 1.900***
Observations 10,157 7,930 3,757 1,379 2,378 4,173 2,227
Adjusted R-squared 0.380 0.371 0.410 0.482 0.376 0.353 0.416
Note: ***,**,* Coefficient is statistically significant at the 1%, 5%, and 10% level, respectively.
Source: Own estimation based on CSES 2009 data.

18. Regression Results: Child Nutrition
Dep. var.: Child weight-for-age
z-score
All Farm
Full-time
farmers
Subsistence
farmers
Commercial
farmers
Part-time
farmers
Non-farm
Per capita expenditure (log) 0.147*** 0.109** 0.113 0.172 -0.042 0.102* 0.166**
Market distance (log) -0.059* -0.041 -0.037 0.097 -0.121 -0.093* 0.024
Farm household (=1) 0.093
Share of non-farm income -0.160* 0.273 -0.315 0.765** -0.158
Share of food consumption from
purchases
0.160 0.234 0.312 0.511 -0.497**
Food crop diversity (log) 0.014 0.120 0.253 0.138 0.008
Livestock diversity (log) 0.085 0.116 0.213 0.077 0.047
Age (log) -0.412*** -0.400*** -0.346*** -0.345*** -0.342*** -0.444*** -0.467***
Female (=1) 0.138*** 0.160*** 0.259*** 0.221** 0.278*** 0.056 0.051
Household size (log) 0.129** 0.163** 0.014 -0.091 0.084 0.255*** -0.014
Female-headed household (=1) -0.001 -0.014 0.085 -0.134 0.270* -0.145† 0.020
Mother's education, primary (=1) 0.096** 0.089* 0.040 0.163 0.017 0.112* 0.139
Mother's education, secondary (=1) 0.059 0.067 0.148 0.199 0.101 0.032 0.071
Constant -1.234** -0.782 -1.457 -2.771* -0.446 -0.190 -1.663
Observations 4,653 3,724 1,713 702 1,011 2 929
Adjusted R-squared 0.128 0.119 0.116 0.163 0.120 0.125 0.141
Note: ***,**,* Coefficient is statistically significant at the 1%, 5%, and 10% level, respectively.
Source: Own estimation based on CSES 2009 data.

19. Conclusions and Policy Implications
 Economic growth is good but is not enough for reducing (child) malnutrition.
 Agricultural transformation may slow down progress toward achieving food
and nutrition security, depending on the patterns of transformation and the
adaptation capacity of the food and nutrition insecure farm households.
 Market expansion benefits FNS overall.
 Farm households tend to be more food and nutrition secure than non-farm
households, while FNS—among subsistence farmers—increases with growing
non-farm income.
 Subsistence farmers’ FNS increases with higher shares of food consumption
from purchases, whereas there is no evidence for positive effects from food
production diversification—but, from livestock diversification.
 Unlike for FNS, there is no evidence for positive effects of agricultural
transformation on child nutrition.
 To make agricultural transformation more nutrition-sensitive, complementary
nutrition-specific interventions are needed.

20. Tajikistan: Agricultural Biodiversity, Dietary
Diversity, and Nutritional Outcomes
Dr. Kamiljon T. Akramov
Coauthor: Mehrab Malek
Financial support: United States Agency for International Development (USAID)

21. Motivation
• Despite recent improvements, malnutrition in Tajikistan remains very
high: stunting among children under 5 is about 30%
• The current strategy of national government and development
partners is to promote agricultural growth and diversification to
ensure food security and nutritional outcomes
– Agrarian policy concept, Food Security Strategy and Agricultural Investment
Plan adopted by government in 2011
– USAID’s FFP and FTF programs and World Bank managed Global Agriculture
and Food Security Program
• These interventions could be very beneficial given the fact that
Tajikistan has less diversified agricultural production system
– About 75% of sown area is allocated to wheat and cotton
• However, there is little evidence regarding the linkages between
agricultural diversity, dietary diversity and nutrition in Central Asian
context

22. Household diets are dominated by cereals (wheat)
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Percent
Composition of calorie intake in Tajikistan, 1992-2007
Other
Animal products
Sugar
Vegetable oil
Vegetables
Fruits
Cereals
Page 22

23. Research questions
• This study aims to provide empirical evidence on
agriculture-nutrition linkages in Tajikistan by investigating
three interrelated questions
– How is agricultural diversity associated with household dietary
diversity?
– Does allocation of more land to cotton and wheat affect dietary
diversity?
– How is dietary diversity correlated with nutritional outcomes?
• Assumption: Agricultural biodiversity influences nutritional
outcomes mainly by improving dietary diversity of
households and individuals
• What are the policy implications of main findings of the
study?

24. Measuring agricultural and dietary diversity
• Dietary diversity
– Count based household DD score was developed using
FAO’s (2011) guidelines (12 food groups)
– Calorie intake and food expenditure weighted Berry
indexes capture richness and evenness
– Calorie intake and food expenditure based Log-abundance
indexes captures richness and abundance
• Agricultural diversity
– Count based household level agricultural diversity score
– Land allocation based and population-weighted log
abundance diversity scores were calculated at the district
level

25. Data and Methodology
• Data sources
– Tajikistan Living Standards Survey (TLSS) 2007 and 2009
– District level population and land allocation data (Regions of Tajikistan
database, National Agency on Statistics, 2011)
• Methodology: Multilevel mixed effects and control function
models to examine relationships between agricultural
diversity and dietary diversity and nutritional outcomes
– Dependent variables: HH dietary diversity scores and child
stunting, measuring chronic malnutrition
– Key independent variables: agricultural diversity at HH and district levels
and share of cotton and wheat in total land area; HH dietary diversity
score, with a maximum of 12 food groups
– Control variables: child, HH and community characteristics, and region
fixed effects

26. Regression Results: Dietary diversity
Count-based DD Calorie-weighted
DD
Expenditure-
weighted DD
Calorie-based log-
abundance DD
Expen.-based log-
abundance DD
Agricultural diversity (HH) 0.0500*** 0.0178 0.0235** 0.171*** 0.125***
(0.0127) (0.0119) (0.0118) (0.0623) (0.0411)
Agricultural diversity (district) 0.0107* 0.0207*** 0.0238*** 0.0971*** 0.0902***
(0.0062) (0.0061) (0.0062) (0.0318) (0.0227)
HH expenditure (log) 0.0901*** 0.0447*** 0.0544*** 0.732*** 0.499***
(0.00730) (0.00619) (0.00640) (0.0491) (0.0337)
Poor -0.0178*** -0.00895* -0.0119** -0.140*** -0.122***
(0.00580) (0.00522) (0.00512) (0.0329) (0.0223)
Location 0.0199** 0.00996* 0.0138** 0.0881** 0.0111
(0.00787) (0.00591) (0.00656) (0.0381) (0.0251)
HH size 0.00756*** 0.00150 0.000866 0.000773 0.0989***
(0.00106) (0.000963) (0.000919) (0.00506) (0.00345)
No of children under 14 0.00331** -0.000484 0.000925 0.0121 0.00303
(0.00159) (0.00139) (0.00137) (0.00756) (0.00501)
HH head's gender 0.00336 -0.0107** -0.00617 0.00425 0.00916
(0.00600) (0.00494) (0.00510) (0.0286) (0.0177)
Altitude (log) -0.00559 -0.00505 -0.00971** -0.0331 -0.0244*
(0.00469) (0.00420) (0.00418) (0.0227) (0.0148)
Distance to oblast center -0.00730*** -0.00263** -0.00591*** -0.0278*** -0.0113**
(0.00168) (0.00134) (0.00138) (0.00814) (0.00520)
Grain and Cotton share -0.0385** -0.004 0.0144 -0.340*** -0.263***
(0.0162) (0.0128) (0.0130) (0.0804) (0.0522)
Constant 0.279*** 0.481*** 0.483*** -0.353 -1.326***
(0.0541) (0.0487) (0.0488) (0.321) (0.222)
Observations 2,991 2,991 2,991 2,991 2,991
F-test 29.38 10.85 15.6 63.88 114.55
R-squared 0.248 0.127 0.138 0.421 0.587
Note: Robust standard errors in parentheses; All specifications control for HH and community characteristics, region fixed effects
*** p<0.01, ** p<0.05, * p<0.1
Dependent variable

27. Regression Results: Child Nutrition
Count-based DD Calorie-weighted
DD
Expenditure-
weighted DD
Calorie-based log-
abundance DD
Expen.-based log-
abundance DD
Dietary diversity -0.337 0.731 0.598 -0.0477 -0.0350
(0.320) (0.455) (0.499) (0.0595) (0.0804)
Child age in months (log) 0.0458 0.261*** 0.262** 0.0819 0.0775
(0.0803) (0.0979) (0.113) (0.0608) (0.0472)
Interaction of DD with children age 0.0495 -0.249* -0.237 0.00110 0.00401
(0.101) (0.138) (0.151) (0.0188) (0.0238)
Child's gender 0.0302 0.0299 0.0295 0.0299 0.0299
(0.0195) (0.0196) (0.0196) (0.0196) (0.0196)
HH expenditure (log) 0.00580 -0.0147 -0.00733 0.0214 -0.00287
(0.0426) (0.0408) (0.0418) (0.0456) (0.0481)
Poor 0.00768 0.00615 0.00681 0.00413 0.00493
(0.0346) (0.0347) (0.0346) (0.0345) (0.0344)
Location -0.0626* -0.0646** -0.0626** -0.0625* -0.0634**
(0.0322) (0.0318) (0.0317) (0.0321) (0.0320)
Altitude (log) 0.0264 0.0295 0.0273 0.0252 0.0270
(0.0318) (0.0313) (0.0314) (0.0319) (0.0316)
Distance to oblast center 0.0294*** 0.0292*** 0.0290*** 0.0292*** 0.0295***
(0.00989) (0.00986) (0.00985) (0.00993) (0.00989)
Chi-squared#
50.08 52.26 52.14 50.23 48.13
p-value 0.0000 0.0000 0.0000 0.0000 0.0000
Log-likelihood -1540.299 -1539.7904 -1539.5953 -1540.001 -1541.4897
Wald test (full model) 432.91 422.89 405.14 440.38 435.57
p-value 0.0000 0.0000 0.0000 0.0000 0.0000
Observations 2,291 2,291 2,291 2,291 2,291
Note: Robust standard errors in parentheses; All specifications control for HH and community characteristics, region fixed effects
#
Chi-squared tests for joint significance of dietary diversity, age of child (log), and their interactions
*** p<0.01, ** p<0.05, * p<0.1
Key independent variable

28. Summary of Findings
• Key empirical results suggest that
– Agricultural diversity is positively associated with dietary diversity, and
– Dietary diversity is in turn correlated with child nutritional outcomes
and this relationship depends on child’s age
• Findings also suggest that there is a negative association
between household dietary diversity and share of land
allocation to cotton and wheat
• Households in communities located further away from urban
centers tend to have lower dietary diversity
• These results are robust
– Across alternative measures of household dietary diversity
– Changes in estimation techniques
– Controls for key child, household and community characteristics

29. Policy implications
• Further promotion of agricultural diversity may be
necessary by allocating more land to horticulture
and feed crops
– In some districts up to 85% of arable land is still
allocated to cotton and wheat
• Investment in infrastructure is important to
promote market integration across different regions
of the country
• Regional cooperation and trade is crucial to ensure
food and nutritional security in the country

30. Nutritional Intake, Agricultural
Production, and Conflict in Nepal
Dr. Yanyan Liu
Financial support: United States Agency for International Development (USAID)

31. Nepal: A Diverse Country
• 3 ecological region
– Mountain
– Hill
– Terai
DOLPA
MUGU
JUMLA
KAILALI
BARDIYA
HUMLA
DOTI
SURKHET
NAWAL
PARASI
KAPIL-
BASTU RUPAN-
DEHI
DANG
BANKE
ACHHAM KALIKOT
JHAPA
MORANG
SIRAHA
SAPTARI
DARCHULA
BAJHANG
BAITADI
DADEL-
DHURA
KANCHAN-
PUR
BAJURA
PARSA
BARA RAUT-
AHAT
DHANUSA
MAHO-
TARI
SUNSARI
SARLAHI
DHADING
MAKAWAN-
PUR
CHITWAN
KASKI
TANAHU
PALPA
SYANGJA
PARBAT
ARGHAK
HACHI
GULMI
UDAYAPUR
SINDHULI
ILAM
BHOJ-
PUR
DHAN-
KUTA
TAPLEJUNG
OKHAL-
DHUNGA
TERHA-
THUM
KHOTANG
LALIT
BHAK
KATHM
SULUK-
HUMBU
DOLAKHA
SANKHUWA-
SABA
NUWAKOT
SINDHU-
PALCHOK
KAVRE
RASUWA
LAMJUN
G
GORKHA
PYUT-
HAN
ROLPA
SALYAN
MYAGDI
DAILEKH
JAJARKOT
RUKUM
MUSTANG
MANANG
CHINA
INDIA
N
Far
western
Midwestern
Western
Central
Eastern

32. Motivation: Food Security in Nepal
• Food insecurity remains a severe problem
– 2% annual population growth rate
– Stagnant arable land
– Recovering from the 1996-2006 civil which
caused 15,000 deaths
• 44% population under poverty line (UNDP)
• 30% underweight for children under 5

33. Nutritional Intake (NLSS)
1996 2003 2011
Total daily energy intake (Kcal p.c.) 2112 2118 2376
% energy intake from cereals,
roots, or tubers 0.841 0.802 0.728
% energy intake from eggs, milk,
or meat 0.058 0.076 0.096
% energy intake from vegetable or
fruits 0.012 0.016 0.027

34. Research Questions
1. How is nutritional intake associated with
income and agricultural production?
(agriculture-nutrition relation)
2. How is children’s anthropometry associated
with nutritional intake (nutrition-health
relation)
3. How did the civil war affect household’s
nutritional intake?

35. Q1: Agriculture-Nutrition Relation
• Data: Nepal Living Standard Survey (NLSS)
2010/2011
• Dependent variables
– Log total daily energy intake (Kcal p.c.)
– % energy intake from cereals, roots, and tubers
– % energy intake from eggs, milk, and meat
– % energy intake from vegetable and fruits

36. Q1: Agriculture-Nutrition Relation (2)
• Explanatory variables
– Log total expenditure p. c.
– Total amount of livestock p. c.
– Whether produce cereals, tubers, or roots
– Whether produce vegetable or fruits
– Household characteristics including
education, household size, female headship, age of
head, number of female adults, and number of male
adults
– Community fixed effects to control for market
price, access, etc.

37. Findings: Agriculture-Nutrition Relation
Log energy
intake
% energy intake from
Cereals, roots,
tubers
Milk, egg,
meat
Vegetable,
fruits
Log expend p.c. 0.102*** -0.0474*** 0.0257*** 0.00477***
Heads of livestock
p.c. 0.0158*** -0.00195** 0.00182*** 0.000145
If produce
cereals, roots,
tubers 0.0856*** 0.00892 -0.00101 0.00155
If produce
vegetable, fruits 0.00998 -0.0139*** 0.00837*** 0.00227***
• Livestock ownership and agricultural production
have direct effects on nutritional intake

38. Q2: Nutrition-Health Relation
• Data: NLSS 2010/11
• Dependent variables
– Length/Height-for-age Z score (<5 yrs)
– Weight-for-age Z score (<5 yrs)
• Explanatory variables
– Log expenditure p.c.
– Total energy intake p.c.
– % energy intake from cereals, roots, tubers
– Household characteristics

39. Findings: Nutrition-Health Relation
Z score
Height-for-age Weight-for-age
Log expenditure p.c. 0.137** 0.101**
Log total energy intake -0.0251 -0.0927
% energy intake from
cereal, roots, tubers -1.191*** -0.617**
• Higher dietary diversity contributes to higher Z
scores

40. Q3: Effect of Civil War on Nutrition
• Data:
– NLSS R1 (1995/96) and R2 (2002/03)
– Number of people killed by year and district
• Possible pathways through adverse effects on
– agricultural production
– agricultural assets (such as livestock)
– market and transportation infrastructure

41. Q3: Effect of Civil War on Nutrition
• Difference-in-difference (DID) method
controlling for ecological region-year specific
effects
• Findings: the effects of the civil war
– Total energy intake (no effect)
– % energy intake from cereals, roots, or tubers (+)
– % energy intake from eggs, milk, or meat (-)
– % energy intake from vegetable or fruits (-)

42. Pathways: Effect of Civil War on Nutrition
• Effects of civil war on
– Total expenditure p.c. (no effects)
– Total amount of livestock p.c. (no effects)
– Total income from crops (no effects)
• The pathway is not likely to be through the
effects on agricultural production or assets.
• Income elasticity on nutritional intake is lower in
more conflict-intensive areas, indicating that
markets being damaged may be the pathway.

43. Summary of findings
• Dietary diversity is positively associated with
livestock ownership and production of
vegetable and fruits
• Children’s length/height-for-age Z score and
weight-for-age Z score are positively
associated with household’s dietary diversity.
• The civil war decreased households’ dietary
diversity, likely through the disruption of the
markets.

44. Pathways of impact of agriculture on
nutrition: Evidence from Bangladesh
Coauthor: Esha Sraboni
Financial support: United States Agency for International Development (USAID)
Dr. Akhter Ahmed

45. Pathways of impacts of agriculture on nutrition
At household level:
 Income
 Education
 Agricultural diversity
 Dietary diversity
 Gender equity
At national level:
 Agriculture research (biofortification, productivity increase)
 Relative food prices
 Policies

46. ISSUES

48. Overwhelming dominance of rice in diet: Share
of rice in total nutrient intakes of Bangladeshis
Source: IFPRI 2011-12 Bangladesh Integrated Household Survey

49. Rice-centric agriculture in Bangladesh:
Share of crops on total cropped land
Source: IFPRI 2011-12 Bangladesh Integrated Household Survey

50. Relative prices and diet quality
 Real (inflation-adjusted) prices of rice has fallen by 36% over the past two
decades. This has helped the rural landless and the urban poor who
purchase the rice they consume
 Despite a falling real price of rice, 17% of the population—about 27 million
ultra poor—remain seriously underfed
 Therefore, the level of technology and institutional innovations that
made this price decline possible must be maintained
 However, the real prices of some foods that are rich in micronutrients
(fruits, vegetables, pulses, animal sourced foods) are increasing
 If policies are not undertaken to increase supply of non-cereal, nutrient-
rich foods, then their prices will continue to increase in the face of income
and population growth
 Consequently, the diet quality and nutritional status of the poor are
likely to deteriorate further

51. EMPIRICAL EVIDENCE

52. Data and methodology
Data: Bangladesh Integrated Household Survey (BIHS)a nationally representative
household survey conducted by IFPRI in 2011-2012
Methodology
 Instrumental variable regression to explore (cross-sectional) association
between dietary diversity and agricultural diversity
 Dep. var: Household dietary diversity score (DDS) with a maximum of 12
food groups (calculated from a 7-day food consumption recall)
 Indep. var.: Number of food crops grown (instrumented by soil types and
percentage of land irrigated), education, household size and demographic
composition, assets, electricity connection, rice price, location fixed
effects
 OLS to explore association between food crop diversity and explanatory
variables

53. Agricultural diversity leads to dietary diversity
 Found statistically significant positive association between production
diversity and dietary diversity
 Other statistically significant factors influencing dietary diversity are:
 Education of male hh head and female spouse (positive)
 HH size (positive)
 Electricity access (positive)
 Farm size (positive)
 Milk cow ownership (positive)
 Hand tubewell ownership (positive)
 Rice price (positive)

54. What factors affect crop diversity?
 Crop diversity increases if:
 Share of irrigated cropped area increases
 Agricultural extension agents visit farm household
 Household has access to loans
 Household uses power tiller for land preparation
 Crop diversity deceases if:
 Household grows winter rice

55. WHAT ARE THE OPPORTUNITIES
FOR LINKING AGRICULTURE AND
NUTRITION IN BANGLADESH?

56. CGIAR Research Program (4)
Agriculture for Nutrition and Health (A4NH)
IFPRI-led, with 11 other CG centers

57. IFPRI’s work on agriculture-nutrition linkages in
Bangladesh
 Biofortification: High-zinc rice (HarvestPlus) released in Bangladesh in
August 2013—the world’s first zinc-enriched rice variety
 Community-level integrated programs:
 Homestead food production (e.g. with HKI, BRAC)
 Fish systems, fish ponds (with World Fish and partners)
 Nutrition-sensitive value chains (fish with World Fish)
 Policy research on integration of agriculture-health-nutrition in
collaboration with:
 other new nutrition-focused programs (e.g. Transform
Nutrition, Leveraging Agriculture for Nutrition in South Asia
(LANSA))
 food security-focused policy research PRSSP
 Agricultural Policy Support Program (APSU) supported by PRSSP
with USAID funding

58. Photo: One Acre Fund
Homestead Food Production to Improve Nutrition

59. HKI’s Homestead food production in Bangladesh
Program:
•:
• Impact:
Source: Millions Fed , IFPRI, 2009; www.ifpri.org/millionsfed
Production-focused: micronutrient-rich vegetables, small
livestock production
Nutrition education to promote consumption
 Focus on women: income generation, empowerment
Nutrition objective: Improve diet diversity, micronutrient
intake
Integrating agriculture and nutrition at household and
community level
Tripled vegetable production; increased income
 73% of gardens managed by women
improved food security for 5 million people

60. Page 60
Agriculture Policies for Nutrition

61. Page 61
Agricultural policy and diet quality
The solution to poor diet quality lies in balanced consumption of nutrient-rich
foods (fruits, vegetables, fish, meat, milk) which the poor desire but cannot
afford. Policies should be undertaken to:
 Increase agricultural investments in non-staples that are high value added
and high nutrition value added
 Reduce production and market risks associated with these crops
 Support the bio-fortification of staple crops with micronutrients
 Recent agricultural research initiatives by CGIAR centers – the Bio-
fortification Challenge Program HarvestPlus – and the Bangladesh Rice
Research Institute (BRRI) have shown that it is possible to get the
plants themselves to do the work of fortification. Zinc-enriched rice
was developed and released. Bio-fortification is cost-effective and
sustainable. Rice is an excellent vehicle for bio-fortification (with iron
and zinc) in the sense that the entire population eats it.

62. Summary and Conclusions
Dr. Akhter Ahmed, H.E. Srun Darith,

63. Take-Away Messages
 We have opportunities and examples of success on how to bridge the
agriculture-nutrition divide
 Our challenge AND opportunity is to work together - cross-sectorally (how?)
 We need to do much better at documenting successes – and failures; we need
the evidence for advocacy, to stimulate investments
 …
 …