IFPRI Country Representative for Bangladesh Dr. Akhter Ahmed presents panel data from IFPRI's Bangladesh Integrated Household Survey (BIHS) in New Delhi, India.

1. Using the BIHS Data to support Agriculture,
Nutrition, and Social Protection Policies in
Bangladesh
Akhter Ahmed
IFPRI Country Representative for Bangladesh
PHND Retreat | Alwar, Rajasthan, India | 16 March 2018

2. Created a comprehensive database
 IFPRI-PRSSP’s Bangladesh Integrated Household Survey
(BIHS): most comprehensive, nationally representative rural
household survey to date. Largest panel survey.
 4 unique features of data collection:
1. plot-level agricultural production
2. individual food intakes of all HH members
3. anthropometry measurements of all HH members
4. data to measure women’s empowerment in
agriculture index (WEAI)
 BIHS sampling is statistically representative
 nationally of rural Bangladesh
 rural areas for each of the 7 administrative divisions
 USAID-supported Feed the Future Zone in southern
Bangladesh
1

3. BIHS: Big data, big impact
 Downloads of 2011 BIHS dataset: 600 (2013)  20,000+ (now)
 Downloads of 2015 BIHS dataset: 149,248 (now)
 Diverse users across 6 continents
2

4. What factors affect farmers’ income?
Using random effects panel regression and IFPRI BIHS data, results
show that farmers’ income tends to increase if:
 HH male head and female spouse have at least secondary school
education
 HH male head and female spouse have access to commercial loans
 Women are more empowered (measured by WEAI)
 Non-farm income share increases
 Have access to electricity (solar panel or national grid) and own cell
phone
 Domestic and international remittances increase
Farmers’ income tends to decrease when:
 Share of cropped land under rice cultivation increases

5. Most farmers grow one crop – Rice
4
54.4
20.1
12.5
5.9
3.9
51.0
0
10
20
30
40
50
60
1 2 3 4 5 Only rice
Percentageoffarmers
Number of crops grown in 2011
Source: IFPRI 2012 Bangladesh Integrated Household Survey (BIHS)

6. Overwhelming dominance of rice in diet
Share of nutrient from rice of total nutrient intake
71
57
62
44
78
67
70
52
63
46
52
36
0
10
20
30
40
50
60
70
80
90
Food energy (calorie) Protein Zinc Iron
Percentoftotalnutrientintake
All Poorest 20% Richest 20%
Source: IFPRI BIHS 2011/12 5

7. Crop diversity increased: Simpson
diversification index, by division
0.19
0.18
0.19
0.27 0.27
0.19
0.05
0.21
0.29
0.18
0.23
0.30
0.28
0.22
0.04
0.24
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Barisal Chittagong Dhaka Khulna Rajshahi Rangpur Sylhet Bangladesh
SimpsonIndex
Axis Title
2012 2015
The Simpson diversification index is calculated as 𝑆𝐷𝐼 = 1 − 𝑖=1
𝑛
𝑃𝑖
2
, where Pi is the
proportionate area of the ith crop in gross cropped area.
6

8. 7
Household diet quality improved
(Estimated WFP’s Food Consumption Score: 0-112)
Average FCS Percentage of households with low FCS
(<42)
23.1
8.4
0
5
10
15
20
25
2011/12 2015
Percent
56.4
66.7
0
10
20
30
40
50
60
70
80
2011/12 2015
Foodconsumptionscore
Source: IFPRI BIHS 2011/12 and 2015

9. Proportion of people who have not consumed
food groups in past 7 days declined
Average food consumption frequency hides severity
0.0 0.4
43.8
52.0
65.2
4.5
0.8
41.2
0.0 0.0
29.3
23.0
47.4
2.4
0.2
23.9
0
10
20
30
40
50
60
70
Percentageofpeoplewithoutconsumption
offoodgroupinpast7days
2011/12 2015
Source: IFPRI BIHS 2011/12 and 2015 8

10. What factors affect diet diversity?
Using random effects panel regression, we estimated
determinants of diet diversity using two different measures –
FCS and HDDS. We obtain similar results from both regressions.
We find that diet diversity improves if:
 Household male head and female spouse have higher levels of
education
 Agricultural diversity increases: Household grew higher
number of non rice food crops last year
 Women are more empowered (measured by WEAI)
 Rice price increases
 Household have higher number of milking cows and are
engaged in fisheries
 Non-farm income share increases; have access to electricity,
own cell phone and have higher asset holdings.
9

11. A paradox: stunting is highest in regions of
lowest poverty, and vice versa
10
Child stunting Poverty
Source: WFP 2012

12.  Sylhet Division: lowest
women’s empowerment,
second highest income
 Barisal Division: highest
women’s empowerment,
second lowest income
Paradox is partly explained by regional
difference in women’s empowerment
11Source: IFPRI BIHS 2015
30
28 27
20 20
12 11
23
0
10
20
30
40
Percentageofwomen

13. IFPRI BIHS data show:
 58% of girls in rural areas get
married before age 18
 Adolescent girls aged <19
account for 36% of all child
births in rural Bangladesh
Early marriage Early
pregnancy  Low
birthweight Stunting
High rate of adolescent pregnancies is associated
with stunting in Bangladesh
58.3
40.8 39.8
34.2
0
10
20
30
40
50
60
<15 15-16 17-18 19-20
Stuntingprevalence(%)
Age groups (years)
Age at child birth and
rate of stunting
Source: IFPRI BIHS 2015 Source: DHS 2014

14. 13
Males have higher calorie intake for all age groups
1558
2244
2832
2740
2390
1535
2067
2416
2297
1910
0
500
1000
1500
2000
2500
3000
5-<10 10-<18 18-<40 40-<65 ≥65
Percapitadailycalorieintake(kcal/day)
Age group
Male Female
Source: IFPRI BIHS 2011/12

15. 14
But females have slightly higher calorie adequacy
(taking individual’s activity levels into account)
84.4
93.7
85.3 86.6
90.190.3
93.8
86.7 87.4
95.8
0
10
20
30
40
50
60
70
80
90
100
5-<10 years 10-<18 years 18-<40 years 40-<65 years ≥65 years
Calorieadequacy(%)
Age group
Male Female
Source: IFPRI BIHS 2011/12

16. 15
33.6
35.6
25.1
27.4
46.0
31.2
26.3
24.4
31.1
49.2
0
5
10
15
20
25
30
35
40
45
50
55
5-<10 10-<18 18-<40 40-<65 ≥65
Prevalenceofunderweight(%)
Age group
Male Female
Source: IFPRI BIHS 2011/12
Prevalence of underweight is higher among
male, except for older adults
(measured by BMI with age- and gender-specific cutoffs)

17. 16
Prevalence of overweight in adults increases
as income increases
(measured by BMI age-specific cutoffs)
2.6 3.2
5.2
8.5
10.8
6.0
4.7
8.6
7.9
11.2
18.1
9.6
0
5
10
15
20
1st Quintile
(Poorest)
2nd Quintile 3rd Quintile 4th Quintile 5th Quintile
(Richest)
All
Prevalenceofoverweight(%)
Per capita household expenditure quintile (proxy for income)
Male Female
Source: IFPRI BIHS 2011/12

18. Targeting effectiveness of major safety net programs: World Bank
used the results for revamping safety net program
IFPRI 2011/12 BIHS data
17
31
24
22
16
8
18
19
26
22
15
27
24
22
19
8
31
26
25
14
4
35
23
25
13
3
37
23
16
20
5
37
24
18
14
7
49
23
21
6
2
0
10
20
30
40
50
60
1 (poorest) 2 3 4 5 (richest)
Percentofhouseholds
Primary School Stipend Secondary education stipend Old Age Allowance
GR OMS VGD
VGF EGPP

19. 18
Policy Considerations

20. Build synergies across 3 components of food
security
Food Security
Utilization
Availability
Access
Nutrition-Sensitive
Agriculture
Nutrition-Sensitive
Social Protection

21. Policy considerations: Nutrition-sensitive
agriculture (1 of 2)
 Promote agricultural diversity:
 Reduce risk of high-value, high nutritive value food
production via contract farming, agricultural credit, etc.
 Create an enabling policy environment for the private
sector for agricultural value chains development.
 Invest in research on productivity of rice, non-rice crops,
livestock, and fisheries.
 Promote rice intensification and agricultural
diversification via agricultural extension.
20

22. Policy considerations: Nutrition-sensitive
agriculture (2 of 2)
 Promote biofortified crop production
 HarvestPlus and BRRI developed 4 varieties of zinc-fortified
rice.
 HarvestPlus is also working with BARI on biofortified lentil
and sweet potatoes.
 Ministry of Food through its Public Food Distribution System
(PFDS) can procure zinc-fortified rice from farmers at a
premium price, which can incentivize farmers to grow
biofortified rice.
 Promote distribution of biofortified crops in safety nets
 PFDS can create institutional demand for biofortification by
distributing zinc-fortified rice to safety net programs as PFDS
outlets.
21

23. Policy considerations:
Nutrition-sensitive social protection (1 of 2)
 Revamp existing safety net programs:
 Add nutrition BCC to social protection.
 Distribute fortified rice (pushti chal) through the Vulnerable
Group Development (VGD) program and the new Khaddya
Bandhob (Food Friendly) Program.
 Introduce school feeding program in secondary schools:
 School feeding program in Bangladesh currently operates
only in primary schools.
 Introducing school feeding program in secondary schools is a
promising platform to target nutrition interventions to reach
adolescents, particularly girls.
22

24. Policy Considerations:
Nutrition-sensitive social protection (2 of 2)
 Improve targeting:
 Target the poor
 Target geographically: coastal belts, urban slums, low income
areas.
 Target vulnerable population groups: using a lifecycle
approach (adolescent girls, pregnant and lactating women).
 Looking forward, design evidence-based safety nets to improve
nutrition outcomes:
 GoB has committed to introduce a national Child Benefit
Program under the National Social Security Strategy (NSSS).
IFPRI and WFP are currently discussing with the Government
on the design, which may include high quality nutrition BCC
and targeting adolescent girls, and pregnant and lactating
women. 23

25.  Prevent early marriage of girls
 Pregnancy in girls who are still growing leads to competition
between the mother and the fetus for access to nutrients, a
battle which the fetus invariably loses, which leads to low
birthweight.
 Low birthweight is strongly associated with child stunting.
 IFPRI’s qualitative results found that girls marry early due to
(1) harassment by male youth, and (2) avoid paying higher
dowry.
A massive social campaign is needed to increase the age
of marriage, as well as to postpone pregnancy of those
girls who do marry early.
To delay pregnancy after marriage, promote effective
family planning.
24
Policy considerations: Prevent early marriage