Presentation by Agnes Quisumbing and Hazel Malapit (IFPRI) at "A Learning Event for the Women's Empowerment in Agriculture Index," held November 21, 2013 in Washington DC.
Session 2a - Quisumbing and Malapit - Using the WEAI for analysis in different socio-cultural contexts: Ghana, Bangladesh and Nepal
1. Using the WEAI for analysis in
different socio-cultural contexts:
Ghana, Bangladesh and Nepal
Hazel Malapit and Agnes Quisumbing
Poverty, Health and Nutrition Division
International Food Policy Research Institute
Supported by the US Agency for International Development through the
Bangladesh Policy Research and Strategy Support Program and the WEAI
2. Introduction
• Global report has provided suggestive evidence that
women’s empowerment is strongly correlated with
several outcomes (hunger, diet diversity), but not
with others (nutritional status)
• How can we use the WEAI
– To diagnose patterns of disempowerment and identify
areas for policy intervention?
– To understand the relationship between empowerment
and desired outcomes in different socio-cultural contexts?
• Two neat features of the WEAI:
– Decomposable into its component domains and
indicators
– Based on extremely detailed individual- and
household-level data
3. What this presentation tries to do:
1. Use the WEAI to diagnose areas where gaps in
empowerment exist for women in Bangladesh,
Nepal, and Ghana, three very different sociocultural contexts
2. See how outcomes related to food security
and nutrition are correlated with indicators
that contribute most to disempowerment
using regression analysis
3. Learn from similarities and differences in the
results to hypothesize how empowerment
“works” in different social and cultural
contexts
4. Main messages (spoiler alert!)
• Patterns of (dis)empowerment vary across
country and context
• Indicators and policy instruments will
therefore vary
• Domains of empowerment are not equally
important in determining different outcomes
at the household, mother, and child level
5. Data
• Bangladesh: Bangladesh Integrated Household
Survey (BIHS) 2011-2012; nationally representative
of rural Bangladesh (http://www.ifpri.org/dataset/bangladeshintegrated-household-survey-bihs-2011-2012)
• Nepal: Nepal Suaahara Baseline Survey, 2012; survey
included 8 intervention districts where Suaahara
planned to implement programs, and 8 matched
comparison districts
• Ghana: Feed the Future’s Population-Based Survey,
Baseline 2012; statistically representative of FTF’s
zone of influence (http://agrilinks.org/library/feed-future-ghanabaseline-survey-dataset)
9. Regression analysis
We estimate the following:
f = β0 + β1 empowerment + β2 h + β 3 c + ε
where:
f = vector of outcomes
empowerment = measures of empowerment
β0 , β1, β2 , β3 = coefficients to be estimated
h = vector of individual & HH characteristics
c = vector of community characteristics
ε = error term
Where possible, we use instrumental variables
methods to deal with endogeneity of empowerment
10. Empowerment measures
Bangladesh
WEAI
Nepal
Ghana
Women’s 5DE score
Women’s 5DE score
Women’s 5DE score
Relative Autonomy
Index (RAI) score
# of production
decisions
Production
Resources
# of credit decisions
# assets with sole/joint
ownership
#decisions over
purchase/sale of assets
Income
Leadership
Time
# of credit decisions
# ag and nonag
activities in which she
has input in income
decisions or feels she
can make decisions
# of groups in which she
is an active member
# of groups in which
she is an active
member
# hours worked
# hours worked
11. Does the aggregate women’s 5DE
score tell us anything meaningful?
Bangladesh
Household level
outcomes
Maternal outcomes
Child outcomes
Nepal
Ghana
HH per capita
calorie adequacy
(+)
HH per adult
equivalent
adequacy (+)
Dietary diversity (+)
Maternal dietary
diversity (+)
Girls: dietary diversity
(-), min. acceptable
diet (-), min. diet
diversity (-), wasted
(+), underweight (+)
13. BANGLADESH
Summary of Results
• Empowerment gaps are greatest in terms of leadership in the
community and control and access to resources
• Women’s 5DE score, the number of groups in which women actively
participate, women’s control of assets are positively associated with
calorie availability and dietary diversity.
• Reducing the empowerment gap between men and women in the
same household also contributes to increasing calorie availability
and dietary diversity
• Results concerning credit decision-making need to be interpreted
with caution (weak instruments, seeking credit is not necessarily a
sign of empowerment in this context)
• Increasing production diversity also contributes to household
calorie availability and dietary diversity
• Other non-ag dimensions of empowerment (parental education)
may be more important for child nutrition and education outcomes
14. Nepal results: summary from OLS and IV regressions
Maternal outcomes
Child outcomes
Diet
diversity
BMI
Diet
diversity
WAZ
WHZ
HAZ
Woman’s 5DE
n.s.
+
-
+ for < 2
n.s.
-
Autonomy
+
n.s.
+
+
n.s.
+
Control over
income
+ (OLS), n.s. +
(IV)
n.s.
+, + for
<2
n.s.
n.s.
Group member
n.s.
+
n.s.,
positive
for <2
+ for <2
n.s.
-, + for
<2
Workload (hours)
+
- (OLS),
n.s. (IV)
- (OLS)
+, - for <
2
+, - for <
2
+
15. NEPAL
Summary of Results
• Empowerment gaps are greatest in terms of group
membership, control over income, autonomy in production
and workload
• Empowerment measures are significantly associated with
maternal outcomes, and variable relationships with child
outcomes
• Autonomy in production is significantly associated with
improvements in maternal dietary diversity and child nutrition
• Higher workload is significantly associated with dietary
diversity for mothers and children, and children’s height-forage z-scores
• Time poverty is associated with disempowerment but actually
does improve outcomes, though sensitive to age of child—
WEAI is decomposable but not necessarily monotonic in
indicators because of context-specific gender norms
17. Ghana results: ICYF behavior and child outcomes,
summary from OLS regressions
Exclusively
breastfed
Diet
diversity
Min
Stunted
dietary
diversity
Wasted
Under
weight
n.s.
- for girl
n.s.
n.s.
n.s.
+ for girl
Agricultural n.s.
decisions
- for girl
- for girl
+ for girl
n.s.
+ for girl
Credit
decisions
n.s.
n.s.
n.s.
n.s.
n.s.
5DE
n.s.
• Women’s empowerment associated with ICYF behaviors for girls, not boys (but the
sign is contrary to expectations)
• Less diverse and lower quality diets in empowered hhs?
• Possibility that diet diversity scores are capturing poor appetite and illness among
infants, because different foods are typically offered only when children are ill or
refuse food (Davis 2000)
18. GHANA
Summary of Results
• Empowerment gaps are greatest in terms of
production decisions and control and access to
resources
• Women’s empowerment strongly associated with
good IYCF behaviors, and likelihood of wasting and
underweight for girls but not boys – however,
relationship indicates women’s empowerment does
not favor girls!
• Women’s participation in credit decisions
significantly correlated with dietary diversity,
consistent with bargaining models
19. Summary of 3-Country Results
• Different indicators of empowerment matter
for different outcomes
• Other non-agricultural dimensions of
empowerment may be more important for
some outcomes (especially for children’s
nutrition and education)
• We sometimes get unexpected results:
– Empowerment not significant for boys’ nutrition, but girls
in empowered households are worse off (Ghana)
– Time poor women have better dietary diversity and taller
children, but lower BMI (Nepal)
21. Conclusions on the use of WEAI
• WEAI is a blunt instrument, but it is like a
Swiss Army knife—individual indicators are
very revealing, so use them!
• While indicators “add up” in the computation
of the WEAI, they don’t always add up in the
same direction because gender norms are
context specific. This could lead to ambiguity
of interpretation
• Use knowledge of gender norms and context
to guide analysis and interpretation of results