Women’s Empowerment in Agriculture and Dietary Quality Across the Life Course: Evidence from Bangladesh

Abstract

Using nationally-representative survey data from rural Bangladesh, we examine the relationship between women’s empowerment in agriculture and indicators of individual dietary quality. Our findings suggest that women’s empowerment is associated with better dietary quality of individuals within the household, but the strength of this association varies across the life course. Women’s empowerment is correlated with more diverse diets of children under five, but empowerment measures are not consistently associated with increases in nutrient intake for this age group. Rather, maternal schooling and household socio-economic status play a more important role for younger children. Women’s empowerment is positively and significantly associated with adult men’s and women’s dietary diversity and nutrient intakes. Empowerment does not benefit all individuals within the household equally, with gender bias emerging in adolescence. Variations in the strength of the association between women’s empowerment and different individuals’ dietary quality across the life course has implications for the design and targeting of interventions to improve dietary quality, particularly of women, children, and adolescent girls.

1. INTRODUCTION

In Bangladesh, despite increases in food production and reduction of child stunting in the past two decades, the rate of wasting (14%) remains high (NIPORT 2015), and deficiencies in micronutrients such as iron, zinc, and vitamin A are prevalent among young children and women of reproductive age. Evidence supporting the strong link between dietary diversity and nutritional outcomes in women and children (Rah et al., 2010; Arimond et al., 2010) has motivated efforts to increase dietary diversity through the design and implementation of nutrition-sensitive agricultural programs.

Explicit recognition of women’s role in their families’ nutrition (Ruel & Alderman, 2013) is central to nutrition-sensitive agricultural programs. These programs often assume that empowering women results in gender-equitable outcomes, but this assumption has rarely been tested. Previous studies on Bangladesh find that women’s education improves household protein intake and dietary diversity (Rashid et al., 2011) and that women’s empowerment in agriculture is associated with greater availability and diversity of food consumed at the household level (Sraboni et al., 2014). However, the relationship between women’s empowerment and dietary quality of individuals has not yet been explored; moreover, whether these gains from women’s empowerment accrue to specific individuals is unknown.

A related question is the age at which gender bias emerges. Son preference is well documented in South Asia, possibly owing to sons’ providing old age support and the need to provide dowries for daughters. Do empowered women favor sons, and when does this bias begin? Using nationally representative data this paper examines the relationship between women’s empowerment in agriculture and dietary quality of individuals by age group and gender in rural Bangladesh. Although studies have investigated determinants of dietary quality at the household level (Rashid, Smith & Rahman 2011; Sraboni et al., 2014) and under-5 child and maternal dietary diversity in Bangladesh (Nguyen et al., 2013), this paper is a first attempt to link women’s empowerment to individual dietary quality across the life course. Our indicators of dietary quality are the dietary diversity score (the number of food groups consumed) and individual intakes of macronutrients (calories and protein) and micronutrients (iron, zinc, and vitamin A). Our measures of women’s empowerment in agriculture are constructed from the Women’s Empowerment in Agriculture Index (WEAI).

Overall, we find that women’s empowerment is differentially associated with dietary diversity and nutrient intakes across the life course. Women’s empowerment appears to be weakly correlated with measures of dietary quality in the youngest children age 6–59 months, but strongly correlated with those of older children, adolescents, and adults. Women’s empowerment has a differential association with males and females within the household; the number of groups to which the primary female belongs has a differential positive correlation with the dietary diversity of girls age 5–10 and nutrient intakes of adult women themselves. Empowerment does not benefit all individuals within the household; gender bias emerges in adolescence. Variations in the strength of the association between women’s empowerment and different individuals’ dietary quality has implications for the design and targeting of interventions to improve dietary quality, particularly of women, children, and adolescent girls.

2. CONCEPTUAL FRAMEWORK

(a). Women’s empowerment in agriculture and nutritional outcomes

To link empowerment in agriculture to nutrition, it is important to understand the pathways between agriculture and nutrition, and the role that women play along those pathways. Ruel and Alderman (2013) identify three pathways through which women’s roles in agriculture impact of women’s participation in agriculture on their time allocation, and ii) the impact of women’s participation in agriculture on their own health and nutritional status. Kadiyala et al. (2014) also highlight the role of the gender division of labor in agriculture, the intrahousehold allocation of food, which affects women’s nutritional status with its intergenerational effects on nutrition outcomes, and women’s power in decisionmaking, which influences whether gains in income translate into nutritional improvements.

Although linkages between resources controlled by women and nutrition are well established, the challenge of measuring empowerment has made linkages between women’s empowerment and nutrition more difficult to quantify. Reflecting the multiple experiences and views of empowerment, many definitions of empowerment exist (see Ibrahim & Alkire (2007) for a comprehensive review). Kabeer (1999) defines empowerment as expanding people’s ability to make strategic life choices, particularly in contexts in which this ability had been denied to them. In this definition, the ability to exercise choice encompasses three dimensions: resources (not only access but also income and future claims to material, human, and social resources), agency (processes of decisionmaking, negotiation, etc.), and achievements (well-being outcomes, educational levels). The WEAI focuses on “agency”, which is far less studied than resources or achievements because it directly addresses the issue of choice or decisionmaking.

As an aggregate indicator, the WEAI is composed of two sub-indexes: (1) the five domains of women’s empowerment [5DE] and (2) gender parity (the Gender Parity Index [GPI]). The 5DE subindex measures the roles and extent of women’s engagement in the agricultural sector in five domains, which are constructed using 10 indicators, as shown in Table 1 (Alkire et al., 2013). In this paper, we use the individual-level women’s empowerment score and its component indicators related to control over physical and social capital, as well as the intrahousehold inequality score (men’s empowerment score minus women’s empowerment score), as measures of women’s empowerment.

Table 1.

The 5 domains of empowerment in the WEAI

Domain	Indicator	Definition of Indicator	Weight
Production	Input in productive decisions	Sole or joint decision making over food and cash-crop farming, livestock, and fisheries	1/10
	Autonomy in production	Autonomy in agricultural production (e.g. what inputs to buy, crops to grow, what livestock to raise, etc.). Reflects the extent to which the respondent’s motivation for decision making reflects his/her values	1/10
Resources	Ownership of assets	Sole or joint ownership of major household assets	1/15
	Purchase, sale, or transfer of assets	Whether respondent participates in decision to buy, sell or transfer his/ her owned assets	1/15
	Access to and decisions on credit	Access to and participation in decision making concerning credit	1/15
Income	Control over use of income	Sole or joint control over income and expenditures	1/5
Leadership	Group member	Whether respondent is an active member in at least one economic or social group	1/10
	Speaking in public	Whether the respondent is comfortable speaking in public concerning various issues such as intervening in a family dispute, ensure proper payment of wages for public work programs, etc.	1/10
Time	Workload	Allocation of time to productive and domestic tasks	1/10
	Leisure	Satisfaction with the available time for leisure activities	1/10

Source: Alkire et al. 2013.

(b). Women’s empowerment, intrahousehold allocation, and dietary quality

The pathway working through women’s empowerment and its effect on women’s access to and control over resources can be linked to women’s power in decisionmaking and the intrahousehold allocation of food (Kadiyala et al. 2014). Drawing on a collective model of the household, we assume that the resources a woman controls affects her bargaining power within the household. Her relative bargaining power affects her ability to decide what and how to produce, and how to allocate food, health, and other goods within the household.

Empirical work has often used women’s control over different forms of assets as a proxy for her bargaining power: this includes human capital (education), physical capital (assets), and social capital (membership in groups) (Quisumbing and Maluccio 2003; Doss 1999). Using direct measures of empowerment, such as the WEAI, makes it possible to capture specific domains of empowerment; measuring the difference in empowerment between the primary male and female in the same household permits a direct test of the bargaining model of the household. Several studies using the WEAI (Malapit et al., 2015, Malapit and Quisumbing 2015) have also suggested that the domains of empowerment relevant to women and children’s diet and nutrition outcomes may not always overlap, especially in different contexts.

Women’s preferences in allocating nutritious and/or high-status food to household members may also be influenced by social norms as well as the costs and benefits of having specific individuals in the household, which could vary by sex and life-cycle stage. Norms also affect preferences. These could be gender norms, or norms associated with childcare practices and old-age caregiving which could intersect with gender norms. In societies with son preference, which could arise from higher labor market returns to male labor as well as sons’ responsibility for old age support, women may prefer to allocate more nutritious food to sons (Choe et al., 1997). In Bangladesh, as sons and daughters enter adolescence, sons may be favored in intrahousehold food allocation owing to the pressure to provide dowries to daughters and to marry them off early, which may make them a net cost to the household. Young adult females, who could be daughters-in-law, have lower status within the household and could be disadvantaged with respect to intrahousehold food distribution and the dietary quality. A mother-in-law dependent on her son for old age support and who was the primary decisionmaker in the household may choose to invest in her son’s well-being over those of other household members.

3. DATA, EMPIRICAL SPECIFICATION AND VARIABLES

Data

We use data from the 2012 Bangladesh Integrated Household Survey (BIHS), which is nationally representative of rural Bangladesh¹. The BIHS interviewed 1,608 nonfarm and 3,895 farm households. We restrict our analysis to households that are involved in one or more agricultural activities—growing crops or raising/engaging with livestock/poultry/fisheries—so that individuals are not misclassified as disempowered when they do not participate in any agricultural activity.² The final estimation sample consists of 2,896 households with 7506 adults, 1786 children aged 11–17 years, 2015 children aged 5–10 years and 1024 children under 5 years of age. Analyses of the empowerment gap between men and women in the same households are conducted for a subsample of 2851 dual-adult households.

Empirical specification

To analyze the relationship between measures of individual dietary quality and women’s empowerment, we estimate the following equation:

$$\mathit{D}=a_0+a_{1}female+a_{2}empowerment+{\mathbf{a}}_{\mathbf{3}}\mathit{I}+{\mathbf{a}}_{\mathbf{4}}\mathit{H}+\epsilon ,$$ (1)

where D is a vector of outcomes that capture dietary quality; female is a dummy variable indicating whether the individual is female; empowerment is a measure of empowerment derived from the WEAI or one of its component indicators, I is a vector of individual characteristics, H is a vector of household characteristics, a_i are the parameters to be estimated, and ε is an error term. The key coefficient of interest is a₂, which captures the correlation between the primary female’s empowerment and dietary quality of each individual, having controlled for a set of observable individual and household characteristics.

To test whether the coefficient a₂ differs for males and females, we include a dummy variable for the individual’s sex (= 1 if female) and interact it with the empowerment variable as well as other characteristics. The resulting equation to be estimated for individual dietary quality (D_d) is:

$${\mathit{D}}_d=b_0+b_{1}female+b_{2}empowerment+b_3(empowerment\times \mathrm{female})+{\mathbf{b}}_{\mathbf{4}}\mathit{I}+{\mathbf{b}}_{\mathbf{5}}\mathit{H}+{\mathbf{b}}_{\mathbf{6}}\mathit{H}*\mathbf{female}+\mathit{v},$$ (2)

where b_i are the parameters to be estimated, and ν is an error term. For males, the relationship between women’s empowerment and dietary quality is given by b₂. For females, the relationship is given by (b₂ + b₃). If b₃, the coefficient on the interaction term between empowerment and the female dummy is significantly different from zero, this suggests that women’s empowerment is differentially associated with male and female members in the household. The reasoning is analogous for the coefficients on the other household characteristics, b₅ and their interaction terms b₆.

One possible source of bias in our analysis is the endogeneity of the empowerment measures, which may be affected by the same factors that influence dietary quality. Owing to a lack of suitable instruments, we cannot implement an instrumental variables technique to correct for endogeneity. Instead, we estimate Equation (2) using OLS and interpret our results as associations rather than causal relationships. We also estimate (2) with household fixed effects (FE)³, which control for household-level unobservables that may confound the impact of empowerment on dietary quality outcomes.

(c). Outcome variables

Individual food consumption data, collected using a combination of 24-hour food recall and food-weighing methods, was used to construct the dietary diversity score and individual nutrient intakes. Conversion factors (Darnton-Hill et al., 2012) were used to calculate quantities of nutrients contained in foods eaten by individual household members.

Dietary diversity for four age groups

Dietary diversity is an attractive outcome measure because it is responsive to current empowerment status of women, unlike anthropometric measures like height, which could be highly correlated with early childhood nutritional status (Alderman et al., 2006) and the empowerment status of the caregiver during childhood. We use individual diet diversity scores, that have been validated for several age and sex groups as proxies for macro and micronutrient adequacies (Kennedy et al., 2007; Arimond et al., 2010), although not completely validated for Bangladesh. Individual age-specific dietary diversity scores are based on WHO and FAO (2016) guidelines, with a 7 food group score for children age 6–59 months and 5–10 years, and 9 food group score for children age 11–17 years and adults age 18 and above. The two different sets of diet diversity scores reflect the differences in micronutrient requirements across the lifecourse (Ruel, Deitchler, & Arimond, 2010; Arimond et al., 2010).

1. 7 food groups for children aged 5–10 years and 6–59 months: number of food groups consumed during the last 24 hours out of - 1. Cereals and tubers, 2. Legumes and nuts, 3. Dairy products 4. Flesh foods, 5. Eggs 6. Vitamin A-rich fruits and vegetables and 7. Other fruits and vegetables.

2. 9 food groups for adults (aged 18 years and above) and children aged 11–17 years: number of food groups consumed during the last 24 hours: 1. Starchy staples; 2. Green leafy vegetables; 3. Other vitamin A-rich fruits and vegetables; 4. Other fruits and vegetables; 5. Organ meat; 6. Meat and fish; 7. Eggs; 8. Legumes and nuts; 9. Milk and milk products (Kennedy, Ballard & Dop, 2011).

Nutrient intake

Individual nutrient intake measures include: (a) food energy (kcal); (b) proteins (grams); (c) iron (mg); (d) zinc (mg); and (e) vitamin A (retinol equivalents). Summary statistics for the outcome variables, presented separately for males and females, are in Table 2. The descriptive statistics indicate the low dietary diversity of individuals, and significant gender differences in nutrient intakes at older ages.

Table 2.

Summary Statistics for Individual Outcomes and Characteristics

	Males					Females
	Obs	Mean	SD	Min	Max	Obs	Mean	SD	Min	Max	Test of difference between means (p-value)
Under-5 child outcomes
Dietary diversity (out of 7 groups)	497	3.608	1.135	1	7	527	3.567	1.155	1	7	0.574
Calorie intake (kcal/day)	497	875.748	471.067	18.615	3480.646	527	842.964	451.351	34.090	2818.860	0.256
Protein intake (grams/day)	497	21.956	12.612	0.096	66.657	527	21.132	12.247	0.508	82.692	0.289
Iron intake (mg/day)	497	3.88	2.625	0.06	19.41605	527	3.653	2.33	0.030	14.030	0.143
Zinc intake (mg/day)	497	2.775	1.693	0.013	11.187	527	2.659	1.593	0.057	10.368	0.257
Vitamin intake (RAE/day)	497	106.876	232.801	0	2427.449	527	83.392	181.034	0	2887.84	0.071
Characteristics
Age (in months)	497	33.943	14.326	6.247	60.658	527	33.703	14.594	6.115	59.441	0.791
Child of primary female (=1 if child of head, 0 otherwise)	497	0.867	0.339	0	1	527	0.837	0.369	0	1	0.172
Mother’s age	497	27.893	6.454	18	57	527	27.757	6.549	17	50	0.738
Mother’s education	497	4.541	3.618	0	15	527	4.622	3.572	0	14	0.718
5–10 year old child outcomes
Dietary diversity (out of 7 groups)	998	3.859	1.056	1	7	1,017	3.872	1.046	1	7	0.774
Calorie intake (kcal/day)	998	1690.906	515.663	349.515	4075.029	1,017	1653.514	500.959	365.101	3457.789	0.099
Protein intake (grams/day)	998	41.49	15.773	8.800	146.697	1,017	40.006	14.623	6.860	126.269	0.029
Iron intake (mg/day)	998	7.304	3.049	1.600	23.295	1,017	7.049	2.805	1.294	24.696	0.051
Zinc intake (mg/day)	998	5.261	2.098	1.195	16.622	1,017	5.036	1.901	0.913	17.561	0.012
Vitamin intake (RAE/day)	998	148.687	254.291	0.000	2421.128	1,017	153.965	261.698	0.000	2421.128	0.646
Characteristics
Age (in years)	998	7.61	1.769	5	10	1,017	7.689	1.779	5	10	0.318
Child of primary female (=1 if child of head, 0 otherwise)	998	0.897	0.304	0	1	1,017	0.905	0.294	0	1	0.557
11–17 year old child outcomes
Dietary diversity (out of 9 groups)	917	4.229	1.199	1	8	869	4.227	1.211	1	9	0.968
Calorie intake (kcal/day)	917	2414.094	667.404	166.650	5302.883	869	2173.566	545.101	461.258	4651.198	0
Protein intake (grams/day)	917	58.713	19.969	3.384	150.276	869	53.531	18.003	10.133	166.265	0
Iron intake (mg/day)	917	10.321	3.972	0.488	31.589	869	9.363	3.352	1.833	25.626	0
Zinc intake (mg/day)	917	7.472	2.754	0.359	23.043	869	6.731	2.235	1.299	18.208	0
Vitamin intake (RAE/day)	917	190.977	283.687	1.424	2788.890	869	195.011	386.438	1.839	5630.061	0.801
Characteristics
Age (in years)	917	13.712	1.867	11	17	869	13.596	1.812	11	17	0.183
Child of primary female (=1 if child of head, 0 otherwise)	917	0.932	0.251	0	1	869	0.925	0.273	0	1	0.555
Adult outcomes
Dietary diversity (out of 9 groups)	3,725	4.182	1.229	1	9	3,781	4.099	1.204	1	8	0.003
Calorie intake (kcal/day)	3,725	2805.918	723.451	788.026	7765.345	3,781	2387.824	609.323	485.975	6297.162	0
Protein intake (grams/day)	3,725	68.646	23.013	14.04822	276.1021	3,781	58.208	19.397	6.728	230.489	0
Iron intake (mg/day)	3,725	12.332	5.084	2.754094	83.61058	3,781	10.433	4.261	1.984	113.206	0
Zinc intake (mg/day)	3,725	8.745	3.113	2.067821	26.69963	3,781	7.377	2.484	1.752	21.938	0
Vitamin intake (RAE/day)	3,725	254.708	451.238	1.347923	6264.363	3,781	217.264	354.852	0.000	3649.804	0
Characteristics
Age (in years)	3,725	41.759	16.029	18	110	3,781	37.938	14.901	18	100	0
Years of schooling	3,725	3.908	4.157	0	16	3,781	3.395	3.716	0	16	0

Source: IFPRI Bangladesh Integrated Household Survey, 2011–2012.

^*RAE: retinol activity equivalent.

(d). Key Independent Variables

Women’s Empowerment in Agriculture Index:

We measure empowerment using the WEAI, an aggregate index computed using individual-level data from primary male and female respondents within the same households. The index reflects a person’s achievement in five domains in agriculture based on ten indicators with their corresponding weights (Table 1). An overall empowerment score is constructed using all ten indicators, and the WEAI’s decomposability property is used to identify the domains and indicators in which women have the least adequate achievements. A decomposition of the WEAI indicators (Sraboni, Quisumbing & Ahmed 2014) shows that the leadership (35.1%) and resources (21.6%) domains contribute most to women’s disempowerment in rural Bangladesh. In terms of the individual indicators, group membership (in the leadership domain) and access to and decisions on credit (in the resources domain) are the areas in which Bangladeshi women are the most inadequate. However, the credit indicator may be problematic because the survey questions do not distinguish between non-borrowers who are truly credit constrained from those who have sufficient liquidity and therefore choose not to borrow (Sraboni et al., 2014). To avoid ambiguity, we use the second indicator (in the resources domain) in which women have the least adequate achievement: the extent to which women can make their own decisions regarding purchase, sale or transfer of assets. In this paper, we use the following measures of empowerment, which are analogous to overall empowerment, control of social and physical capital, and intrahousehold bargaining power discussed in the conceptual framework.

Model 1: Aggregate empowerment score of primary female respondent:

the weighted average of achievements in the ten indicators that comprise the five domains of empowerment in agriculture. This measure is increasing in empowerment, ranges from 0 to 1, and reflects overall empowerment in agriculture.

Model 2: (Leadership domain, Group membership indicator) Number of groups in which woman is an active member:

total number of groups in which the female respondent reports being an active member. This is a measure of social capital.

Model 3: (Resources domain, Rights over assets indicator) Number of sole/joint decisions, concerning purchase/sale/transfer of assets, taken by woman:

total number of decisions made solely or jointly by the female respondent, summed over all asset types. For each asset type, the survey asks who can decide whether to sell, give away, mortgage/rent, and purchase the asset. This captures rights over physical capital.

Model 4: Intrahousehold inequality score:

Considerable evidence exists that paying attention to intrahousehold gender inequality is important to attain development objectives (Alderman et al. 1995). Therefore, the fourth model employs an indicator of women’s relative empowerment within the household, the intrahousehold inequality score (IIS). The IIS is zero if the household enjoys gender parity, which is when the female empowerment score is greater than or equal to the male empowerment score⁴. Otherwise, it equals the difference in the male and female aggregate empowerment scores. Higher numbers indicate a larger gap between male and female empowerment favoring males. Model 4, a direct test of an intrahousehold bargaining model, is estimated only for dual-adult households with valid responses from the primary male and female adults.

(e). Other Independent Variables

Household characteristics include: household size, dependency ratio (the ratio of the number of members aged between one and 10 years and over 64 years to those aged between 10 and 64 years), amount of cultivable land owned by household, number of dairy cows owned, access to electricity, sanitary latrine and hand tubewell ownership, and the price of various food items (rice, pulses, chicken, small and large fish).

Individual characteristics include: age, age-squared, sex (=1 if female), years of schooling for adults, and maternal education for children. We cannot link adolescents and children aged 5–10 to their biological mothers, unless they are children of the household head. For these two age groups, the primary female respondent’s age and education are proxies for their biological mother’s characteristics. To control for any differences between children whose parents are more accurately measured (the child of the household head) compared with other children, we use a dummy variable which equals 1 if the child is the offspring of the household head. We also control for location specific effects. Summary statistics of the household characteristics are presented in Table 3.

Table 3.

Summary Statistics for Household Characteristics

Variable	Obs.	Mean	Std. dev.	Min	Max
Household level characteristics
Empowerment variables
Empowerment score of primary female	2,896	0.683	0.227	0.067	1
Number of groups primary female is an active member of	2,896	0.31	0.486	0	3
Number of self/joint decisions over purchase, sale or transfer of assets made by primary female	2,896	12.042	9.445	0	48
Difference in empowerment scores (primary male-primary female)	2,851	0.159	0.195	0	0.895
Primary female characteristics
Age	2896	37.521	11.475	17	80
Education level	2896	3.069	3.444	0	16
Household demographics
Household size	2,896	4.541	1.662	2	17
Dependency ratio	2,896	0.787	0.612	0	6
Household wealth
Number of dairy cows owned	2,896	0.759	1.202	0	9
Ln (owned cultivable land+1)	2,896	0.782	1.629	0.000	6.982
Access to electricity (=1, 0 otherwise)	2,896	0.464	0.499	0	1
HH owns sanitary latrine (=1, 0 otherwise)	2,896	0.251	0.434	0	1
HH owns hand tubewell (=1, 0 otherwise)	2,896	0.268	0.443	0	1
Food prices (in Taka)
Rice	2,896	30.205	3.287	20	50
Chicken	2,896	130.322	11.706	80	250
Pulse	2,896	99.896	9.504	50	140
Large fish	2,896	85.133	28.179	20	300
Small fish	2,896	104.888	39.24	30	400
Divisions
Barisal	2,896	0.075	0.264	0	1
Chittagong	2,896	0.099	0.298	0	1
Dhaka	2,896	0.331	0.471	0	1
Khulna	2,896	0.122	0.328	0	1
Rangpur	2,896	0.142	0.349	0	1
Rajshahi	2,896	0.111	0.314	0	1
Sylhet	2,896	0.119	0.324	0	1

Source: IFPRI Bangladesh Integrated Household Survey, 2011–2012.

4. RESULTS

(a). Women’s empowerment and dietary quality

Selected coefficients from levels and fixed effects (FE) estimates for dietary quality outcomes of the four age groups are presented in Tables 4, 5, 6, 7 and 8 (full results are available on request). Each table presents results for one age group, moving from the youngest (children 6–59 months) to the oldest (adults). We present both levels and FE estimates because they provide different insights. The FE models can be interpreted as comparisons among individuals within the same household, averaged over the total number of households. Fixed effects estimation is a more robust way to examine the differential association of empowerment with the diet quality of males and females within the same household. The levels results are used to interpret the association of empowerment with individual outcomes across all households since empowerment, being a household level variable, is dropped from the fixed-effects model.

Table 4.

Women’s empowerment and dietary quality of children under 5

	Diet Diversity (7 groups)		Calorie Intake (Kcal/day)		Protein Intake (grams/day)		Iron Intake (mg/day)		Zinc Intake (mg/day)		Vitamin Intake (RAE)
	Levels	Fixedeffects	Levels	Fixedeffects	Levels	Fixedeffects	Levels	Fixedeffects	Levels	Fixedeffects	Levels	Fixedeffects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Model 1: Aggregate empowerment
(1.a) girl	−0.041	−2.482	−605.500	1459.000	−20.860	57.880	0.906	21.130	−1.829	11.210	−405.400	−864.000
	(1.767)	(5.093)	(635.1)	(2,148)	(17.76)	(48.65)	(3.552)	(14.47)	(2.378)	(7.738)	(352.5)	(509.9)
(1.b) empowerment score	0.451**		122.800		2.887		0.593		0.344		−65.47
	(0.225)		(80.70)		(2.257)		(0.451)		(0.302)		(44.79)
(1.c) empowerment score × female	−0.101	0.424	−36.920	−176.000	−1.742	−2.508	0.182	−0.998	0.021	−0.330	84.780	−153.4**
	(0.307)	(0.682)	(110.4)	(287.7)	(3.088)	(6.516)	(0.618)	(1.938)	(0.413)	(1.036)	(61.29)	(68.29)
Levels estimates:
Association of empowerment with females: (1.b) + (1.c)	0.350		85.880		1.145		0.775		0.365		19.310
p-value of F-test: (1.b) + (1.c) = 0	0.097*		0.257		0.589		0.068*		0.198		0.646
N	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102
R-squared	0.212	0.602	0.373	0.822	0.325	0.846	0.321	0.730	0.307	0.799	0.051	0.820
Model 2: Leadership domain
(2.a) female	−0.052	−2.251	−595.200	1120.000	−21.340	41.660	1.140	16.260	−1.818	8.511	−372.400	−958.400
	(1.776)	(5.452)	(637.4)	(2,290)	(17.81)	(50.97)	(3.566)	(15.19)	(2.383)	(8.088)	(353.3)	(588.3)
(2.b) number of groups	0.063		6.464		0.685		0.154		0.190		−24.440
	(0.107)		(38.23)		(1.068)		(0.214)		(0.143)		(21.19)
(2.c) number of groups × female	−0.066	0.062	0.561	−75.710	−0.484	−3.423	−0.025	−1.037	−0.127	−0.566	16.410	−24.560
	(0.152)	(0.382)	(54.44)	(160.3)	(1.521)	(3.567)	(0.305)	(1.063)	(0.204)	(0.566)	(30.18)	(41.18)
Levels estimates:
Association of empowerment with females: (2.b) + (2.c)	−0.003		7.025		0.201		0.129		0.063		−8.030
p-value of F-test: (2.b) + (2.c) = 0	0.976		0.856		0.853		0.553		0.666		0.709
N	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102
R-squared	0.207	0.596	0.371	0.821	0.324	0.851	0.318	0.737	0.306	0.806	0.05	0.788
Model 3: Resources domain
(3.a) female	−0.029	−1.561	598.500	1458.000	−21.150	57.760	1.101	16.730	−1.767	9.762	381.400	918.300
	(1.759)	(5.466)	(634.5)	(2,316)	(17.74)	(52.23)	(3.554)	(15.37)	(2.377)	(8.260)	(352.2)	(594.9)
(3.b) number of asset decisions	0.019***		−2.190		0.049		−0.008		0.002		0.816
	(0.00566)		(2.041)		(0.0571)		(0.0114)		(0.00765)		(1.133)
(3.c) number of assets decisions × female	−0.0177**	0.011	0.619	−0.340	−0.042	−0.006	0.004	−0.052	−0.006	−0.017	−0.683	−0.908
	(0.00803)	(0.0224)	(2.897)	(9.502)	(0.0810)	(0.214)	(0.0162)	(0.0630)	(0.0108)	(0.0339)	(1.608)	(2.441)
Levels estimates:
Association of empowerment with females: (3.b) + (3.c)	0.002		−1.571		0.007		−0.004		−0.004		0.133
p-value of F-test: (3.b) + (3.c) = 0	0.737		0.445		0.910		0.743		0.616		0.907
N	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102	1,024	102
R-squared	0.216	0.600	0.372	0.819	0.324	0.845	0.318	0.734	0.305	0.800	0.049	0.787
Model 4: Intrahousehold inequality
(4.a) female	0.149	−2.260	489.100	1,588	−18.530	63.55	1.914	21.75	−1.401	11.93	−356.000	−971.5*
	(1.785)	(5.139)	(633.7)	(2,219)	(17.79)	(49.09)	(3.567)	(14.80)	(2.386)	(7.904)	(356.5)	(500.3)
(4.b) intrahousehold inequality	−0.260		213.5**		−4.138		−0.324		−0.399		65.70
	(0.257)		(91.19)		(2.558)		(0.513)		(0.343)		(51.23)
(4.c) intrahousehold inequality × female	−0.042	−0.635	−156.3	64.910	2.468	−9.079	−0.369	−0.316	−0.0355	−0.970	−86.73	233.4**
	(0.348)	(1.026)	(443.0)	(170.5)	(3.470)	(9.800)	(0.695)	(2.954)	(0.465)	(1.578)	(69.47)	(99.87)
Levels estimates:
Association of empowerment with females: (4.b) + (4.c)	−0.302		−135.64		−1.67		−0.693		−0.4345		−21.03
p-value of F-test: (4.b) + (4.c) = 0	0.1979		0.7562		0.4759		0.1399		0.2478		0.6540
N	1,008	100	1,008	100	1,008	100	1,008	100	1,008	100	1,008	100
R-squared	0.205	0.617	0.379	0.816	0.329	0.850	0.324	0.729	0.309	0.800	0.052	0.835

Source: Estimated using data from the IFPRI Bangladesh Integrated Household Survey, 2011–2012. Other covariates (not reported) control for individual and household characteristics and location-specific effects.

Note: Standard errors are in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1.

Table 5.

Women’s empowerment and dietary quality of children aged 5–10 years

	Diet Diversity (7 groups)		Calorie Intake (Kcal/day)		Protein Intake (grams/day)		Iron Intake (mg/day)		Zinc Intake (mg/day)		Vitamin Intake (RAE)
	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Model 1: Aggregate empowerment
(1.a) female	1.547	0.264	−671.500	−186.400	−3.984	−6.603	−0.200	3.619	−1.302	0.068	49.880	125.100
	(1.046)	(0.494)	(500.0)	(529.1)	(15.22)	(12.59)	(3.012)	(2.455)	(2.023)	(1.753)	(278.8)	(153.7)
(1.b) empowerment score	0.418***		140.300**		5.186**		1.146***		0.874***		23.760
	(0.140)		(66.95)		(2.038)		(0.403)		(0.271)		(37.33)
(1.c) empowerment score × female	0.053	0.059	−20.230	−56.500	−1.483	−0.218	−0.297	−0.198	−0.343	−0.162	−22.580	19.990
	(0.202)	(0.0902)	(96.52)	(96.51)	(2.938)	(2.296)	(0.581)	(0.448)	(0.391)	(0.320)	(53.81)	(28.04)
Levels estimates:
Association of empowerment with females: (1.b) + (1.c)	0.471		120.070		3.703		0.849		0.531		46.340
p-value of F-test: (1.b) + (1.c) = 0	0.0013***		0.0844*		0.0804*		0.04		0.0591*		0.232
N	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589
R-squared	0.172	0.072	0.194	0.548	0.166	0.544	0.119	0.490	0.149	0.493	0.026	0.160
Model 2: Leadership domain
(2.a) female	1.486	0.356	−743.300	−187.100	−6.566	−5.977	−0.669	3.689	−1.719	0.133	47.600	138.400
	(1.047)	(0.492)	(498.9)	(531.0)	(15.20)	(12.62)	(3.010)	(2.464)	(2.023)	(1.759)	(278.0)	(154.0)
(2.b) number of groups	0.122*		−38.200		−1.195		−0.009		−0.091		2.302
	(0.0678)		(32.28)		(0.983)		(0.195)		(0.131)		(17.99)
(2.c) number of groups × female	−0.060	0.097**	84.870*	−17.550	2.037	0.470	0.309	0.001	0.243	0.007	14.290	17.400
	(0.0945)	(0.0436)	(45.00)	(47.01)	(1.371)	(1.118)	(0.272)	(0.218)	(0.183)	(0.156)	(25.07)	(13.63)
Levels estimates:
Association of empowerment with females: (2.b) + (2.c)	0.062		46.670		0.842		0.300		0.152		16.592
p-value of F-test: (2.b) + (2.c) = 0	0.350		0.137		0.379		0.113		0.233		0.343
N	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589
R-squared	0.166	0.086	0.192	0.548	0.163	0.544	0.114	0.489	0.144	0.492	0.026	0.163
Model 3: Resources domain
(3.a) female	1.550	0.267	−708.900	−211.000	−5.028	−6.777	−0.562	3.536	−1.633	−0.040	66.340	137.200
	(1.039)	(0.496)	(499.3)	(530.3)	(15.19)	(12.62)	(3.010)	(2.462)	(2.023)	(1.756)	(277.7)	(153.9)
(3.b) number of asset decisions	0.014***		−0.270		0.005		0.014		0.008		−1.524
	(0.00348)		(1.674)		(0.0509)		(0.0101)		(0.00678)		(0.931)
(3.c) number of assets decisions × female	0.001	0.001	1.208	−1.965	0.088	−0.011	−0.004	−0.007	−0.004	−0.007	2.982**	0.844
	(0.00489)	(0.00218)	(2.351)	(2.326)	(0.0715)	(0.0554)	(0.0142)	(0.0108)	(0.00953)	(0.00770)	(1.307)	(0.675)
Levels estimates:
Association of empowerment with females: (3.b) + (3.c)	0.016		0.938		0.092		0.009		0.004		1.458
p-value of F-test: (3.b) + (3.c) = 0	0***		0.570		0.0662*		0.340		0.526		0.112
N	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589	2,015	589
R-squared	0.180	0.072	0.191	0.549	0.163	0.544	0.114	0.490	0.144	0.494	0.028	0.163
Model 4: Intrahousehold inequality
(4.a) female	1.303	0.312	−653.500	−142.100	−6.353	−4.226	−0.893	3.885	−1.754	0.221	129.300	155.300
	(1.057)	(0.515)	(501.3)	(545.8)	(15.32)	(12.99)	(3.036)	(2.540)	(2.043)	(1.812)	(281.5)	(159.8)
(4.b) intrahousehold inequality	−0.443***		−167.8**		−5.405**		−1.108**		−0.913***		5.449
	(0.162)		(76.71)		(2.344)		(0.465)		(0.313)		(43.07)
(4.c) intrahousehold inequality × female	0.050	−0.032	57.020	−19.520	3.162	−1.814	0.570	−0.148	0.518	−0.101	−62.780	−28.500
	(0.236)	(0.103)	(111.9)	(108.7)	(3.420)	(2.587)	(0.678)	(0.506)	(0.456)	(0.361)	(62.86)	(31.84)
Levels estimates:
Association of empowerment with females: (4.b) + (4.c)	−0.393		−110.780		−2.243		−0.538		−0.395		−57.331
p-value of F-test: (4.b) + (4.c) = 0	0.0224**		0.174		0.368		0.276		0.235		0.211
N	1,983	583	1,983	583	1,983	583	1,983	583	1,983	583	1,983	583
R-squared	0.170	0.072	0.197	0.549	0.164	0.545	0.114	0.489	0.148	0.493	0.025	0.161

Source: Estimated using data from the IFPRI Bangladesh Integrated Household Survey, 2011–2012. Other covariates (not reported) control for individual and household characteristics and location-specific effects.

Note: Standard errors are in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1.

Table 6.

Women’s empowerment and dietary quality of children aged 11–17 years

	Diet Diversity (9 groups)		Calorie Intake (Kcal/day)		Protein Intake (grams/day)		Iron Intake (mg/day)		Zinc Intake (mg/day)		Vitamin Intake (RAE)
	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixedeffects	Levels	Fixedeffects	Levels	Fixedeffects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Model 1: Aggregate empowerment
(1.a) female	0.443	0.168	−824.900	−473.100	−12.590	−20.040	1.586	−2.467	−0.227	−2.344	−198.900	167.000
	(1.398)	(0.885)	(723.2)	(919.3)	(22.90)	(23.94)	(4.556)	(4.308)	(3.072)	(3.199)	(430.0)	(295.9)
(1.b) empowerment score	0.159		230.000***		5.007*		1.685***		1.190***		51.290
	(0.169)		(87.22)		(2.762)		(0.549)		(0.370)		(51.86)
(1.c) empowerment score × female	0.084	0.038	−182.600	164.600	−6.022	1.992	−1.400*	0.419	−0.850	0.407	−108.500	−65.810
	(0.241)	(0.132)	(124.9)	(137.0)	(3.954)	(3.568)	(0.787)	(0.642)	(0.530)	(0.477)	(74.24)	(44.09)
Levels estimates:
Association of empowerment with females: (1.b) + (1.c)	0.243		47.400		−1.015		0.285		0.340		−57.210
p-value of F-test: (1.b) + (1.c) = 0	0.161		0.595		0.720		0.613		0.371		0.281
N	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531
R-squared	0.210	0.121	0.208	0.389	0.166	0.347	0.117	0.321	0.143	0.330	0.048	0.066
Model 2: Leadership domain
(2.a) female	0.490	0.098	−885.800	−512.400	−14.360	−21.420	1.309	−2.482	−0.470	−2.380	−224.900	156.200
	(1.397)	(0.885)	(724.1)	(923.5)	(22.90)	(24.00)	(4.560)	(4.319)	(3.078)	(3.209)	(429.7)	(296.7)
(2.b) number of groups	0.109		16.660		0.860		0.432*		0.165		29.970
	(0.0749)		(38.81)		(1.227)		(0.244)		(0.165)		(23.03)
(2.c) number of groups × female	−0.076	−0.047	−23.820	15.920	−1.389	−0.551	−0.243	0.110	−0.129	0.089	−44.900	−28.000
	(0.109)	(0.0601)	(56.27)	(62.66)	(1.780)	(1.629)	(0.354)	(0.293)	(0.239)	(0.218)	(33.39)	(20.13)
Levels estimates:
Association of empowerment with females: (2.b) + (2.c)	0.033		−7.160		−0.529		0.189		0.036		−14.930
p-value of F-test: (2.b) + (2.c) = 0	0.681		0.860		0.612		0.462		0.835		0.537
N	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531
R-squared	0.209	0.122	0.208	0.386	0.165	0.347	0.114	0.320	0.138	0.329	0.048	0.065
Model 3: Resources domain
(3.a) female	0.578	0.176	−824.600	−535.600	−13.330	−21.630	1.454	−2.782	−0.329	−2.581	−213.500	182.500
	(1.399)	(0.884)	(724.3)	(921.3)	(22.92)	(23.92)	(4.563)	(4.303)	(3.079)	(3.200)	(430.4)	(296.8)
(3.b) number of asset decisions	0.006		4.414**		0.135**		0.034***		0.022**		1.838
	(0.00385)		(1.992)		(0.0630)		(0.0125)		(0.00847)		(1.184)
(3.c) number of assets decisions × female	−0.002	0.002	−2.874	−0.298	−0.129	−0.071	−0.029	−0.013	−0.019	−0.007	−1.778	−0.645
	(0.00574)	(0.00311)	(2.971)	(3.241)	(0.0940)	(0.0841)	(0.0187)	(0.0151)	(0.0126)	(0.0113)	(1.765)	(1.044)
Levels estimates:
Association of empowerment with females: (3.b) + (3.c)	0.004		1.540		0.006		0.005		0.002		0.060
p-value of F-test: (3.b) + (3.c) = 0	0.305		0.485		0.934		0.729		0.803		0.936
N	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531	1,786	531
R-squared	0.210	0.121	0.205	0.386	0.167	0.348	0.116	0.322	0.141	0.329	0.048	0.060
Model 4: Intrahousehold inequality
(4.a) female	0.377	0.474	−1035.000	−481.700	−19.440	−22.660	−0.260	−3.289	−1.264	−2.799	−321.300	−16.710
	(1.411)	(0.907)	(732.4)	(943.9)	(23.16)	(24.45)	(4.576)	(4.299)	(3.083)	(3.181)	(434.9)	(301.3)
(4.b) intrahousehold inequality	−0.016		−236.300**		−5.186		−1.562**		−1.213***		−34.600
	(0.198)		(102.7)		(3.247)		(0.642)		(0.432)		(60.98)
(4.c) intrahousehold inequality × female	−0.186	−0.210	146.400	−140.000	8.771*	−1.545	1.937**	−0.248	1.097*	−0.308	125.800	137.100**
	(0.288)	(0.161)	(149.6)	(167.6)	(4.729)	(4.341)	(0.935)	(0.763)	(0.630)	(0.565)	(88.81)	(53.50)
Levels estimates:
Association of empowerment with females: (4.b) + (4.c)	−0.202		−89.900		3.585		0.375		−0.116		91.200
p-value of F-test: (4.b) + (4.c) = 0	0.335		0.408		0.297		0.581		0.801		0.157
N	1,753	524	1,753	524	1,753	524	1,753	524	1753	524	1753	524
R-squared	0.210	0.121	0.208	0.385	0.163	0.340	0.112	0.316	0.142	0.326	0.045	0.086

Source: Estimated using data from the IFPRI Bangladesh Integrated Household Survey, 2011–2012. Other covariates (not reported) control for individual and household characteristics and location-specific effects.

Note: Standard errors are in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1.

Table 7.

Women’s empowerment and dietary quality of adults

	Diet Diversity (9 groups)		Calorie Intake (Kcal/day)		Protein Intake (grams/day)		Iron Intake (mg/day)		Zinc Intake (mg/day)		Vitamin Intake (RAE)
	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Model 1: Aggregate empowerment
(l.a) female	0.110	−0.222	−472.2*	−339.8*	−4.609	−3.131	−0.354	0.304	−1.138	−0.563	90.180	17.550
	(0.485)	(0.167)	(280.9)	(174.1)	(8.988)	(4.661)	(2.035)	(0.980)	(1.208)	(0.639)	(179.2)	(55.95)
(l.b) empowerment score	0.247***		292.700***		6.225***		1.948***		1.417***		−37.030
	(0.0827)		(47.83)		(1.530)		(0.346)		(0.206)		(30.52)
(l.c) empowerment score × female	−0.006	0.058	−110.100	−64.420	−2.906	−0.975	−0.606	−0.105	−0.563*	−0.244	28.850	22.430*
	(0.116)	(0.0397)	(67.16)	(41.52)	(2.149)	(1.112)	(0.486)	(0.234)	(0.289)	(0.152)	(42.86)	(13.34)
Levels estimates:
Association of empowerment with females: (1.b) + (1.c)	0.241		182.600		3.319		1.342		0.854		−8.180
p-value of F-test: (1.b) + (1.c) = 0	0.003***		0.000***		0.028**		0.000***		0.000***		0.786
N	7,506	7,433	7,506	7,433	7,506	7,433	7506	7,433	7506	7,433	7506	7,433
R-squared	0.195	0.030	0.186	0.373	0.148	0.341	0.085	0.276	0.119	0.315	0.014	0.052
Model 2: Leadership domain
(2.a) woman	0.128	−0.185	−542.9*	−391.0**	−6.595	−4.172	−0.762	0.162	−1.489	−0.752	98.33	26.91
	(0.482)	(0.165)	(279.3)	(172.5)	(8.918)	(4.616)	(2.023)	(0.971)	(1.203)	(0.633)	(177.6)	(55.47)
(2.b) number of groups	0.093**		−44.14**		−1.830**		−0.126		−0.125		32.90**
	(0.0388)		(22.52)		(0.719)		(0.163)		(0.0969)		(14.32)
(2.c) number of groups × woman	−0.063	−0.014	41.660	52.370***	1.430	1.734***	0.314	0.291***	0.174	0.185***	27.630	11.560*
	(0.0546)	(0.0186)	(31.66)	(19.47)	(1.011)	(0.521)	(0.229)	(0.110)	(0.136)	(0.0715)	(20.13)	(6.261)
Levels estimates:
Association of empowerment with females: (2.b) + (2.c)	0.030		−2.480		−0.400		0.188		0.049		−5.270
p-value of F-test (2.b) + (2.c) = 0	0.432		0.911		0.574		0.244		0.614		0.710
N	7,506	7,433	7,506	7,433	7,506	7,433	7,506	7,433	7506	7,433	7506	7,433
R-squared	0.194	0.029	0.181	0.374	0.146	0.343	0.079	0.277	0.111	0.315	0.015	0.052
Model 3: Resources domain
(3.a) woman	0.082	−0.208	−521.8*	−380.2**	−5.943	−3.497	−0.580	0.291	−1.375	−0.649	112.000	32.800
	(0.480)	(0.166)	(278.8)	(173.1)	(8.904)	(4.633)	(2.020)	(0.974)	(1.199)	(0.635)	(178.1)	(55.63)
(3.b) number of asset decisions	0.012***		7.445***		0.233***		0.051***		0.035***		1.107
	(0.002)		(1.152)		(0.0368)		(0.0083)		(0.0049)		(0.736)
(3.c) number of assets decisions × woman	0.001	0.001	−1.366	0.299	−0.040	−0.014	−0.011	−0.003	−0.008	−0.004	−0.517	−0.194
	(0.0028)	(0.0009)	(1.624)	(1.010)	(0.0519)	(0.0270)	(0.0118)	(0.006)	(0.007)	(0.004)	(1.037)	(0.324)
Levels estimates:
Association of empowerment with females: (3.b) + (3.c)	0.013		6.079		0.193		0.039		0.027		0.590
p-value of F-test (3.b) + (3.c) = 0	0.000***		0.000***		0.000***		0.000***		0.000***		0.421
N	7,506	7,433	7,506	7,433	7,506	7,433	7506	7,433	7506	7,433	7506	7,433
R-squared	0.202	0.030	0.188	0.373	0.153	0.341	0.086	0.276	0.121	0.315	0.014	0.051
Model 4: Intrahousehold inequality
(4.a) woman	0.136	−0.177	−565.2**	−418.3**	−7.209	−4.601	−0.870	0.070	−1.522	−0.833	96.890	29.600
	(0.486)	(0.166)	(281.4)	(174.2)	(9.009)	(4.660)	(2.037)	(0.980)	(1.207)	(0.637)	(180.1)	(56.39)
(4.b) intrahousehold inequality	−0.220**		−335.900***		−6.626***		−1.908***		1.329***		9.556
	(0.0958)		(55.51)		(1.777)		−0.402		(0.238)		(35.52)
(4.c) intrahousehold inequality × woman	−0.0427	−0.0768*	159.9**	125.1***	4.384*	2.952**	0.941*	0.444	0.648*	0.449**	9.189	−4.592
	(0.135)	(0.0459)	(77.93)	(48.02)	(2.494)	(1.285)	(0.564)	(0.270)	(0.334)	(0.176)	(49.87)	(15.55)
Levels estimates:
Association of empowerment with females: (4.b) + (4.c)	−0.263		−176.000		−2.242		−0.967		−0.681		18.745
p-value of F-test: (4.b) + (4.c) = 0	0.005***		0.001***		0.200		0.0145**		0.004**		0.592
N	7,389	7,317	7,389	7,317	7,389	7,317	7389	7,317	7389	7,317	7389	7,317
R-squared	0.193	0.030	0.185	0.376	0.147	0.344	0.082	0.278	0.117	0.318	0.014	0.051

Source: Estimated using data from the IFPRI Bangladesh Integrated Household Survey, 2011–2012. Other covariates (not reported) control for individual and household characteristics and location-specific effects.

Note: Standard errors are in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1.

Table 8:

Results summary for education, landownership and dairy cow ownership from Model 1 (aggregate empowerment score)

	Diet Diversity (9 groups)		Calorie Intake (Kcal/day)		Protein Intake (grams/day)		Iron Intake (mg/day)		Zinc Intake (mg/day)		Vitamin Intake (RAE)
	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects	Levels	Fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Children 6–59 months
Schooling
(1) Maternal schooling	0.042***		3.060		0.169		0.0295		−0.004		−0.802
	(0.0154)		(5.524)		(0.154)		(0.0309)		(0.0207)		(3.066)
(2) Maternal schooling × female	0.011	0.058	3.992	−14.600	0.121	−0.213	0.032	−0.070	0.052*	−0.061	0.905	−10.080
	(0.0218)	(0.0618)	(7.820)	(26.04)	(0.219)	(0.590)	(0.0437)	(0.175)	(0.0293)	(0.0938)	(4.340)	(6.182)
Levels estimates:
Effect ofmaternal schooling on females: (1) + (2)	0.053		7.052		0.290		0.062		0.048		0.103
p-value of F-test: (1.b) + (1.c) = 0	0.001***		0.207		0.064*		0.050*		0.020**		0.973
Landownership
(1) Size of land owned	0.063*		14.130		0.142		0.063		0.009		0.842
	(0.0344)		(12.38)		(0.346)		(0.0692)		(0.0463)		(6.870)
(2) Size of land owned × female	0.000	−0.150	0.433	16.96	0.009	0.154	0.002	−0.028	0.001	0.065	0.067	−2.599
	(0.00111)	(0.106)	(0.398)	(44.88)	(0.0111)	(1.016)	(0.00223)	(0.302)	(0.00149)	(0.162)	(0.221)	(10.65)
Levels estimates:
Effect of landownership on females: (1) + (2)	0.063		14.563		0.151		0.065		0.010		0.909
p-value of F-test: (1.b) + (1.c) = 0	0.069*		0.124		0.454		0.345		0.669		0.560
Dairy cow ownership
(1) Dairy cow	−0.034		12.480		0.306		−0.031		0.035		−6.883
	(0.0465)		(16.70)		(0.467)		(0.0934)		(0.0625)		(9.269)
(2) Dairy cow × female	0.000	−0.165	−0.588	−15.360	−0.013	−1.195	−0.001	0.068	0.001	0.069	0.256	22.620
	(0.00168)	(0.169)	(0.604)	(71.29)	(0.0169)	(1.615)	(0.00338)	(0.480)	(0.00226)	(0.257)	(0.335)	(16.92)
Levels estimates:
Effect of dairy cow on females: (1) + (2)	−0.034		11.892		0.293		−0.032		0.036		−6.627
p-value of F-test: (1.b) + (1.c) = 0	0.399		0.983		0.713		0.990		0.347		0.055*
Children 5–10 years
Schooling
(1) Schooling of primary female	0.048***		4.744		0.271*		0.053*		0.021		2.805
	(0.0105)		(5.037)		(0.153)		(0.0303)		(0.0204)		(2.809)
(2) Schooling of primary female × female	−0.001	0.002	−1.133	−3.030	−0.036	0.017	0.020	0.022	−0.001	0.002	−1.956	−2.692
	(0.0148)	(0.00677)	(7.091)	(7.245)	(0.216)	(0.172)	(0.0427)	(0.0336)	(0.0287)	(0.0240)	(3.954)	(2.105)
Levels estimates:
Effect of schooling on females: (1) + (2)	0.047		3.611		0.235		0.073		0.020		0.849
p-value of F-test: (1.b) + (1.c) = 0	0.000***		0.470		0.121		0.015**		0.326		0.761
Landownership
(1) Size of land owned	0.049**		20.420**		0.533*		0.116*		0.054		−1.291
	(0.0212)		(10.14)		(0.309)		(0.0611)		(0.0410)		(5.652)
(2) Size of land owned × female	0.007	−0.002	−18.010	6.492	−0.529	0.178	−0.102	0.038	−0.095	0.023	3.470	−4.470
	(0.0309)	(0.0154)	(14.76)	(17.94)	(0.449)	(0.427)	(0.0889)	(0.0832)	(0.0597)	(0.0594)	(8.229)	(5.212)
Levels estimates:
Effect of landownership on females: (1) + (2)	0.056		2.410		0.004		0.014		−0.041		2.179
p-value of F-test: (1.b) + (1.c) = 0	0.013**		0.823		0.991		0.832		0.349		0.716
Dairy cow ownership
(1) Dairy cow	0.087***		16.460		0.133		0.049		0.084		−10.220
	(0.0282)		(13.48)		(0.410)		(0.0812)		(0.0546)		(7.517)
(2) Dairy cow × female	−0.055	0.029*	10.130	−21.260	0.07	−0.593	0.021	−0.124	−0.001	−0.092*	9.283	−5.024
	(0.0386)	(0.0168)	(18.43)	(16.43)	(0.561)	(0.391)	(0.111)	(0.0762)	(0.0746)	(0.0544)	(10.28)	(4.774)
Levels estimates:
Effect of dairy cow on females: (1) + (2)	0.032		26.590		0.203		0.070		0.083		−0.937
p-value of F-test: (1.b) + (1.c) = 0	0.230		0.034**		0.594		0.347		0.104		0.893
Adolescents age 11–17
Schooling
(1) Schooling of primary female	0.061***		8.350		0.304		0.067		0.042		−0.855
	(0.0133)		(6.898)		(0.218)		(0.0435)		(0.0293)		(4.101)
(2) Schooling of primary female × female	−0.041**	−0.001	−3.925	10.090	0.064	0.202	−0.006	0.042	−0.008	0.037	−0.958	−5.128
	(0.0188)	(0.0114)	(9.752)	(11.87)	(0.309)	(0.309)	(0.0614)	(0.0556)	(0.0414)	(0.0413)	(5.798)	(3.820)
Levels estimates:
Effect of schooling on females: (1) + (2)	0.021		4.425		0.368		0.061		0.034		−1.813
p-value of F-test: (1.b) + (1.c) = 0	0.119		0.522		0.092*		0.159		0.252		0.659
Landownership
(1) Size of land owned	0.038*		31.860***		0.977***		0.228***		0.104**		14.010**
	(0.0224)		(11.61)		(0.368)		(0.0732)		(0.0493)		(6.906)
(2) Size of land owned × female	−0.008	0.025	−16.170	13.100	−0.634	0.283	−0.224**	−0.018	−0.101	0.043	1.200	−1.312
	(0.0340)	(0.0207)	(17.60)	(21.51)	(0.557)	(0.560)	(0.111)	(0.101)	(0.0748)	(0.0749)	(10.47)	(6.923)
Levels estimates:
Effect of landownership on females: (1) + (2)	0.030		15.690		0.343		0.004		0.003		15.210
p-value of F-test: (1.b) + (1.c) = 0	0.245		0.235		0.412		0.955		0.965		0.053*
Dairy cow ownership
(1) Dairy cow	0.037		42.420***		0.568		0.039		0.075		−11.500
	(0.0292)		(15.08)		(0.478)		(0.0950)		(0.0641)		(8.968)
(2) Dairy cow × female	0.044	0.002	18.950	−5.578	0.751	−0.592	0.066	0.016	0.075	0.002	7.577	−3.738
	(0.0411)	(0.0229)	(21.26)	(23.79)	(0.673)	(0.620)	(0.134)	(0.111)	(0.0903)	(0.0828)	(12.64)	(7.658)
Levels estimates:
Effect of dairy cow on females: (1) + (2)	0.081		61.370		1.319		0.105		0.150		−3.923
p-value of F-test: (1.b) + (1.c) = 0	0.005***		0.000***		0.006***		0.263		0.019**		0.660
Adults
Schooling
Own schooling	0.043***	0.004*	2.251	−1.446	0.279***	0.006	0.069***	0.005	0.025***	−0.004	−0.784	0.199
	(0.00387)	(0.00210)	(2.238)	(2.197)	(0.0716)	(0.0588)	(0.0162)	(0.0124)	(0.00962)	(0.00807)	(1.428)	(0.706)
Landownership
(1) Size of land owned	0.033***		7.386		0.320		0.108**		0.017		4.810
	(0.0113)		(6.532)		(0.209)		(0.0473)		(0.0281)		(4.168)
(2) Size of land owned × woman	0.01	0.002	−12.280	19.50***	−0.278	0.462***	−0.044	0.104***	−0.0310	0.073***	−4.326	−3.661**
	(0.0156)	(0.00542)	(9.040)	(5.661)	(0.289)	(0.152)	(0.0655)	(0.0319)	(0.0389)	(0.0208)	(5.768)	(1.820)
Levels estimates:
Effect of landownership on women: (1) + (2)	0.043		−4.894
p-value of F-test: (1.b) + (1.c) = 0	0.000***		0.444		0.838		0.163		0.616		0.905
Dairy cow ownership
(1) Dairy cow	0.054***		41.820***		0.761***		0.023		0.091**		3.866
	(0.0145)		(8.382)		(0.268)		(0.0607)		(0.0360)		(5.348)
(2) Dairy cow × woman	−0.017	−0.007	−5.557	−9.956	−0.098	−0.357*	0.022	−0.014	0.002	−0.032	−6.003	−3.365
	(0.0208)	(0.00715)	(12.04)	(7.474)	(0.385)	(0.200)	(0.0872)	(0.0421)	(0.0518)	(0.0274)	(7.682)	(2.402)
Levels estimates:
Effect of dairy cow on women: (1) + (2)	0.037		36.263		0.663		0.045		0.093		−2.137
p-value of F-test: (1.b) + (1.c) = 0	0.015**		0.000***		0.017**		0.474		0.013**		0.699

Source: Estimated using data from the IFPRI Bangladesh Integrated Household Survey, 2011–2012. Other covariates (not reported) control for individual and household characteristics and location-specific effects.

Note: Standard errors are in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1.

Table 4 presents selected results from levels and FE estimates on women’s empowerment and dietary quality for children aged 6–59 months. The results mostly indicate that there is no significant disadvantage for girls aged 6–59 months with respect to various dietary quality indicators, and that the aggregate empowerment score is significantly associated with improved dietary diversity of boys (p<.05) and only weakly with improved dietary diversity of girls. The number of asset decisions is significantly associated with improvements in boys’ dietary quality, but not of girls. The coefficient on the intrahousehold inequality score indicates that a reduction in intrahousehold inequality is associated with improvement in the calorie intakes of boys. Intrahousehold differences in the correlations of empowerment by sex are mostly insignificant. However, the interaction coefficients from the FE regression indicates that an improvement in the empowerment score and reduction in intrahousehold inequality are differentially associated with a decrease in the vitamin A intake of girls.

Table 5 presents selected results from levels and FE estimates for children aged 5–10 years. The levels estimates indicate that women’s overall empowerment is significantly associated with greater dietary diversity for both boys and girls (Model 1, Col 1), and improvements in a range of intake indicators- calories (Model 1, Col 3), protein (Model 1, Col 5), iron (Model 1, Col 7), and zinc (Model 1, Col 9), although some of the associations are weaker for girls. The primary female’s participation in groups is significantly correlated with a higher dietary diversity score for boys (Model 2, Col 1). The number of asset decisions made also has a significant positive correlation with both boys’ and girls’ dietary diversity (Model 3, Col 1), as well as girls’ protein intake (Model 3, Col 5). Finally, a reduction in the intrahousehold inequality score is associated with improvements in a range of boys’ dietary quality indicators- dietary diversity (Model 4, Col 1), calorie (Model 4, Col 3), protein (Model 4, Col 5), iron (Model 4, Col 7), and zinc intakes (Model 4, Col 9), but is positively correlated only with girls’ overall dietary diversity score (Model 4, Col 1). Across all households, therefore, the indicators for women’s empowerment are significantly associated with multiple outcomes for both girls and boys. In terms of differential associations by sex, the only significant interaction term in the FE estimates suggests that the primary female belonging to more groups has a significant correlation with improved dietary diversity for girls (Model 2, Col 2).

Table 6 presents selected results from levels and FE estimates for children aged 11–17. The association between women’s empowerment indicators and dietary quality of adolescents suggest the emergence of strong preferences for adolescent boys. The levels estimates suggest that the women’s empowerment score and asset ownership are associated with higher calorie (Models 1 and 3 Col 3), protein (Models 1 and 3, Col 5), iron (Models 1 and 3, Col 7), and zinc (Models 1 and 3, Col 9) intakes of boys, the number of groups is weakly associated with higher iron intake of boys (Model 2, Col 7), and a reduction in the intrahousehold inequality score is correlated with higher intakes of calories (Model 4, Col 3), iron (Model 4, Col 7), and zinc (Model 4, Col 9) of boys. None of the empowerment indicators has a significant association with any measure of girls’ dietary quality.

Interaction terms with the sex of the individual in the FE estimates suggest that a reduction in intrahousehold inequality has a differentially positive correlation with boys’ vitamin A intakes (Model 4, Col 12)—the only estimate showing a significant differential associated by sex. If women’s empowerment is associated with improved adolescent boys’—but not girls’—dietary quality, this may jeopardize young women’s future nutrition and reproductive health, given the prevailing marriage pattern in which girls leave the household to marry during adolescence.

Table 7 presents selected results from levels and FE estimates of the relationship between measures of women’s empowerment and adult dietary quality. The strongest evidence of positive relationships between women’s empowerment and dietary quality comes from this age group. Almost all indicators of women’s empowerment are associated with improvements in both men’s and women’s dietary quality across almost all outcomes in the levels estimates. The FE estimates suggest a weak positive association between the women’s empowerment score and women’s vitamin A intakes (Model 1, Col 12). Interaction terms with the number of groups to which the primary female belongs are significant only in the FE estimates, suggesting that the participation in groups has differential positive associations with women’s nutrient intakes, across macro- and micronutrients (Model 2, Col 4,6,8,10,12). Except for a reduction in the intrahousehold inequality score that is differentially associated with an improvement in women’s dietary diversity (Model 4, Col 2), however, interaction terms with the intrahousehold inequality score suggest that men’s nutrient intakes improve more than women’s in households with lower intrahousehold inequality (Model 4, Col 4,6,10). Thus, while empowerment in terms of group membership appears to benefit women, a reduction in intrahousehold inequality may involve some tradeoffs that favor men’s nutrient intakes.

Overall, the interaction terms reveal that different indicators of empowerment have differential associations by sex. The positive association of women’s empowerment with adult women’s outcomes, but not adolescent girls’ may point to the lower status of adolescent girls. Results not reported here suggest that a woman’s own empowerment is not differentially associated with her own dietary quality.

(b). Other correlates of dietary quality

Table 8 presents selected coefficients on indicators of human and physical wealth: schooling, the size of land owned, and the number of dairy cows, presented separately for each age group (full results are available on request).

Maternal schooling has significant positive associations with dietary diversity of both boys and girls, and girls’ intakes of protein, iron, and zinc. However, because the FE coefficients are insignificant, we cannot reject the null hypothesis that maternal schooling is equally associated with boys and girls. The primary female’s schooling is positively correlated with some dietary quality indicators for boys aged 5–10 (dietary diversity, protein, and iron) as well as dietary diversity and iron intake for girls; the insignificant interaction terms indicate that schooling affects boys and girls 5–10 similarly. Interestingly, the size of land owned is associated with more diverse diets and increased calorie, protein, and iron intakes for boys, but only greater dietary diversity for girls. Household ownership of dairy cows is positively associated with dietary quality for boys and calorie intake for girls, however, these results are not robust to the inclusion of fixed effects. FE effects estimates suggest compensatory associations: ownership of dairy cows has differential positive association with girls’ dietary diversity score, but also negatively correlated with zinc intakes (albeit only at p<0.10).

Maternal schooling appears to have a positive and significant association with the dietary quality of adolescent boys, and weak correlation with protein intakes of adolescent girls. The size of land owned is positively correlated with adolescent boys’ dietary diversity and intake of nutrients, and positively but weakly associated with improved adolescent girls’ vitamin A intakes. Dairy cow ownership is associated with higher calorie intake for adolescent boys and better dietary diversity and intakes of calories and protein for adolescent girls.

Finally, in adult regressions, one’s own schooling is associated with better dietary quality outcomes (although not all coefficients are robust to the inclusion of household fixed effects). The levels estimates suggest that the size of land owned is positively correlated with men’s dietary diversity and iron intake as well as with women’s dietary diversity score. FE estimates also suggest that larger land sizes have a differential positive association with women’s nutrient intakes, except for vitamin A. Dairy cow ownership is also associated with improved dietary quality for both men and women.

5. CONCLUSION

We find that women’s empowerment in agriculture is associated with better dietary quality of individuals within the household, but these relationships vary across the life course. Although women’s empowerment is associated with better dietary diversity of children under five, various empowerment measures are not consistently associated with nutrient increases for this age group. In contrast, women’s empowerment becomes more important as a factor affecting dietary quality for older individuals. Women’s empowerment is positively and significantly associated with men’s and women’s dietary diversity and nutrient intakes. Women’s groups, in particular, appear to have a differential positive association with the dietary quality of girls and of adult women themselves. However, women’s empowerment does not benefit all individuals alike: gender bias favoring boys emerges in adolescence.

The weak link between empowerment of the primary female and the diet quality of preschool children not only implies that different empowerment domains may have different impacts on nutrition (Kabeer, 1999) but also that other characteristics, such as maternal schooling and household socio-economic status, may be more important for younger children. Infant and young child feeding practices, both of which have been linked to maternal schooling, may also have a greater role in improving younger children’s diet quality (Kabir et al., 2012; Beyene, Worku & Wassie, 2015). Alternatively, empowerment could work in a more indirect way by determining whether a woman can command the resources needed to adopt recommended IYCF practices.

Our study has several limitations. First, our analysis is based on 24-hour recall based on one visit, which may not give an accurate picture of dietary patterns compared to multiple visits. Second, because we cannot correct for endogeneity, our conclusions only indicate associations, not causation. Third, even though household fixed effects estimates control for household-level unobservables, the reduction in sample size makes our conclusions more conservative. Fourth, we did not explore the other domains beyond those identified as areas where empowerment gaps were largest.

Future research could focus on the association between other indicators comprising the WEAI and dietary quality, and on the pathways through which empowerment may influence diet quality. The importance of education in the dietary quality of young children, and the relatively greater importance of women’s empowerment for older children and adults suggest that policies designed to empower women and improve nutritional status need to be based on an understanding of which specific domains of women’s empowerment matter at different stages of the life course. The strong associations between women’s empowerment and adult outcomes suggest that nutrition interventions that only target young children, may be missing an opportunity to invest in the nutritional status of women prior to conception. The positive correlation between dairy cow ownership and women and men’s dietary quality also points to the potential of diversifying agricultural production and diets through the livestock sector. A multipronged approach consisting of appropriate women’s empowerment interventions bundled with agricultural interventions and nutrition behavior change communication may be useful in improving diet quality of all household members.

Our results also challenge the assumption that women’s empowerment automatically results in gender-equitable outcomes. Policy makers in many countries, based on a unitary model, typically directed resources to male household heads, assuming equitable intra-household sharing of resources or benefits thereof (Agarwal 1997). But assuming that women are altruistic and directing resources to them, without changing gender norms or other structures underpinning gender inequality, makes the same mistake. Even if the evidence shows that directing resources to women, rather than men, is more likely to improve household well-being, particularly in relation to health and nutrition (Yoong et al. 2012), in the South Asian context, a woman investing more in sons than in daughters may be motivated by self-interest rather than altruism, given the prevailing male advantage in labor markets and property rights, women’s need for male mediation in the community, and women’s dependence on sons in widowhood or old age (Agarwal 1997). Indeed, the possibility that different individuals within the household will have different costs and returns depending on their age and sex suggests that women may be motivated by both altruism and self-interest. Our results suggest that social norms, particularly those related to son preference and old age support, affect the extent to which women are willing to use their bargaining power for different members of the household. The recognition that both women and men may be motivated by self-interest and concerned with individual as well as family welfare, albeit in differing degrees, suggests that government and civil society interventions should not only attempt to empower women as individuals, but also address structural and societal factors to make women less economically and socially dependent on sons, husbands or brothers.

Highlights.

Women’s empowerment is associated with improvements in the dietary quality of adults Results suggest the emergence of strong preferences for adolescent boys Maternal schooling plays a more important role for younger children.

Women’s preferences in allocating nutritious food may be influenced by social norms Different domains of empowerment matter for diet quality outcomes across the life course