INTRA - HOUSEHOLD HUMAN CAPITAL MEASURES AND CHILD AND MATERNAL HEALTH : EVIDENCE FROM ZAMBIA By Simone Margaret Faas A THESIS Submitted to Michigan State University in partial fulfillment of the requirement for the degree of Agricultural, Fo od and Resource Economics Master of Science 20 20 ABSTRACT INTRA - HOUSEHOLD HUMAN CAPITAL MEASURES AND CHILD AND MATERNAL HEALTH : EVIDENCE FROM ZAMBIA By Simone Margaret Faas Zambia has one of the highest rates of childhood stunting in the world . Traditio nal health production functions model that good health quality for young children i s dependent on the nece ssary inputs of parental influences, including parental health, parental education, and household wealth. Using data from a Feed the Future survey fro m rural Zambia and the severa l measurements of spousal human capital and the health outcomes of young child ren and women of child - bearing age . I find th e ability to read and write of both spouses is highly correlated with positive changes in . Literacy and education campaigns which target both boys and girls should be heavily emphasized among rural and disadvantaged communities in southern Afri ca, both important to improving future health outcomes for children and adults. iii ACKNOWLEDGEMENT S I would like to thank my parents, Michael and Carol Faas, for their love and constant encoura geme nt. I would like to thank my friends for all their help, patience and support. I would like to tha nk the Graduate School at Michigan State University for providing the funding y so urce of funding for my M I would like to thank the Indaba Agricultural Policy Resea rch Institute in Lusaka, Zambia , especially Dr. Rhoda Mofya - Mukuka , for allowing me to use their data for this study. I would like to thank my advisor, D r. M aria Porter for her guidance. I would like to thank my thesis committee, including Dr. Nicole Mason - Wardell and Dr. Rob ert Richardson, for their helpful and insightful comments and suggestions. Thank you , AFRE , for your patience and understanding. T hank you, Peace Corps , for the once - in - a - lifetime experience and the best government - issued family I could have ever imagined . Thank you , Chankhonzi , for taking me in , giving me a home, a family, and an entire life in one of the most beautiful corners of t he w orld. iv TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ......................... vi LIST OF FIGURES ................................ ................................ ................................ ...................... vii KEY TO ABBREVIATIONS ................................ ................................ ................................ ...... viii CHAPTER 1. INTRODUCTION ................................ ................................ ................................ ... 1 CHAPTER 2. BACKGROUND AND COUNTRY CONTEXT ................................ ................... 5 ................................ ............................. 5 2.2 The Context of Zambia ................................ ................................ ................................ ......... 7 Zambia ................................ ................................ .......................... 9 2.2.b Maternal Health in Zambia ................................ ................................ .......................... 10 2.2.c Education and Litera cy in Zambia ................................ ................................ ............... 11 ................................ ................................ ..... 13 CHAPTER 3. ESTIMATION STRATEGY ................................ ................................ ................. 14 3 .1 Estimation Specifications ................................ ................................ ................................ ... 14 3.2 Description of Specifications ................................ ................................ .............................. 15 CHAPTER 4. DATA ................................ ................................ ................................ .................... 17 4.1 Outcome Vari ables ................................ ................................ ................................ ............. 21 4.1.a Child Anthropometry - children under 5 years old: ................................ ..................... 21 - women 15 - 49 years old: ................................ ............... 23 4.2 Literacy and Education Variables ................................ ................................ ....................... 24 4.2.a Literacy ................................ ................................ ................................ ........................ 24 4.2.b Education ................................ ................................ ................................ ..................... 24 4.3 ................................ ................................ ... 26 4.3.a Empowerment Measures ................................ ................................ .............................. 26 4.4 Control Variables ................................ ................................ ................................ ................ 30 ................................ ................................ ................... 30 4.4.b General Regression Controls ................................ ................................ ....................... 30 CHAPTER 5. RESULTS ................................ ................................ ................................ .............. 32 ................................ ................................ ............................... 32 5.1.a Parental Literacy ................................ ................................ ................................ .......... 32 ................................ ................................ ................................ 35 5.2.a Spousal Literacy ................................ ................................ ................................ ........... 35 CHAPTER 6. DISCUSSI ON ................................ ................................ ................................ ........ 38 CHAPTER 7. CONCLUSION ................................ ................................ ................................ ...... 41 7.1 Limitations and Future Research ................................ ................................ ........................ 41 v APPENDICES ................................ ................................ ................................ .............................. 44 APPENDIX A. ROBUSTNESS CHECKS ................................ ................................ .......... 45 APPENDIX B. CORRELATIONMA TRIX OF KEY INDEPENDENT VARIABLES ..... 50 ................. 52 BIBLIOGRAPHY ................................ ................................ ................................ ......................... 54 vi LIST OF TABLES Table 1: Su mmary Statistics. ................................ ................................ ................................ ........ 20 Tabl e 2: Parental Literacy as Predictors of Child Health Outcomes. ................................ ........... 34 Table 3: Spousal Literacy as Predictors of Women's Dietary Diversity. ................................ ...... 36 Table 4: Parental Education as Predictors of Child Health Outcomes. ................................ ........ 46 ................................ .. 47 - Ma ss - Index. ................................ .... 48 Table 7: Parental Empowerment as Predictors of Child Health Outcomes. ................................ . 49 Table 8: Correlation Matrix of Key Independent Variables. ................................ ........................ 51 Table 9: The Domains o 53 vii LIST OF FIGURES Figure 1: Map of Africa with Zambia emphasized. ................................ ................................ ........ 8 Figure 2: Map of the Zone of Influence in Eastern Province, Zambia. ................................ ........ 18 Figure 3: Distribution of Children's Height - for - Age Z - scores. ................................ .................... 22 Figure 4: Distribution of Children's Weight - for - Age Z - scores. ................................ ................... 22 Figure 5: Number of Food Groups Consumed by Women in Previous 24 Hours. ....................... 23 Figure 6: Distribution of Men's Years of Education. ................................ ................................ .... 25 Figure 7: Distribution of Women's Years of Education. ................................ .............................. 25 ................................ ......................... 28 ................................ ............................... 28 Figure 10: Contribution of each Domain to Women's Dise mpowerment. ................................ ... 29 ................................ . 29 Figure 12: Distribution of Height - for - Age Z - scores by Parental Literacy Status. The red reference line on each chart indicates the mean value of the HAZ of the children. ..................... 39 viii KEY TO ABBREVIATIONS BMI: Body - Mass - Inde x CSO: Central Statistic al Office DDS: Dietary Diversity Score DHS: Demographics and Health Survey FAO: Food and Agriculture Organization FtF: Feed the Future HAZ: Height - for - Age Z - Score IFPRI: International Food Policy Research Institute IYCF: Infant and Young Child Feeding OPHI : Oxford Poverty and Human Deve lopment Initiative RALS: Rural Agricultural Liv elihoods Survey UNICEF USAID : United States Agency for International Development WAZ : Weight - for - Age Z - score WEAI s Empowerment in Agriculture Index WHO : World Health Org anization ZOI : Zone of Influence 1 CHAPTER 1. INTROD UCTION While Zambia is known to be a peaceful country, with relatively mild climates, good soil, and low population density , it is also highly foo d insecure. Of all Zambian households, 46% are undernou rished, a rate similar to that of war - torn and climate - ravaged countries such as Yemen, Chad , and South Sudan (FAO et al., 2017; Grebmer et al., 2018) . Zambia has one of the highest rates of childhoo d stunting in the world , with over 40% of children in Z ambia under the age of five being moderately or severely stunted and 15% underweight (UNICEF, 2017 ; CSO et al., 201 5 ) . - term health outcomes are determined by a myriad of familial factor s and environmental influences, and primary interventio and well - being often focus on improving resources and education for caretakers of young children, especially pregnant mothers, as some long - term health out co mes are partly influenced by conditions experienced i n - utero (WHO, 2007; UNICEF, 2017). This thesis provides some explanation for such surprisingly high rates of stunted and underweight Zambian children. I estimate a series of linear regressions with Ord ina ry Least Squares (OLS) models using survey data from a Feed the Future (Ft F) Zone of Influence survey conducted in Eastern P rovince, Zambia, which includes Agriculture Index (WEAI) module tailored specifically to the Zambian con text. I f ocus on the sub - sample of rural, male - headed househol ds with children under 5 years of age. I find that in Zambia, the primary factors predicting child health outcomes include parental education and empowerme nt in household decision - making related to agricultur e have a moderately positive relationship with , 2 Zambian households . This is the ca se even though I use several different measures that have been developed to formulate the widely used WEAI . Because maternal health is often closely tied to the long - term health of young children ( Rosenzweig and Schultz, 1982; Thoma s et al., 1990 ; de Onis and Branca, 2016 ) , I also analyze alth may be associated with their own measures of human capital and empowerment, as well as the human capital characteristics of their husband. In regressions on I find th at for women in Zambia, some measures of the degree t o which they are empowered to make agricultural and financial household decisions may indicate improved health outcomes. I also find that the other key factors influencing their health include m easures o f their own education or literacy levels, and that of their husbands . - term is due in part nts, thei r education, and household wealth (Rosenzweig and Sch ultz, 1982; Thomas et al., 1990 ; Smith and Haddad, 2000; WHO, 2007 ; ) . The child health production function has been adapted from the generic health production function to include inputs related to parental human capital, which is often related to income, asset ownership (including land), and the probability of seeking health services (Barrera, 1990; Thomas et al., 1990 ; Cui et al., 2019 ). Maternal education is very often associated with medical and nutritional resources more efficiently (B arrera, 1990 ; Behrman, 1997; Smith and Haddad, 2000 ; Fafcha mps and Shilpi, 2013 ). In Zambia, a very limited number of studies have examined the dominating factors which ried to tand its relationship to child and 3 bargaining power of women to increase the d dietary diversity (Mofya - Mukuka and Sambo, 2018 ). This thesis makes a valuable contribution to the literature. Rather than focusing on only one measure differen c es in the relative importance of parental education, litera cy, health, and literacy in order to examine the possibility for potential differences in the influenc e of individual parents. In doing so, I find that the strong est predictor of improved long - term health measures of young children is for the father and mother to both be able to read and write. That is, when both the husband and wife are literate , their ho u sehold is more likely to have healthier children in compari son to households where only one spouse (either man or woman) can read and write , especially compared to households where neither spouse can read and write . These findings have important policy im p lications, and show that the combined influence of having e ach parent being literate is particularly predictive of positive health outcomes among young children. Thus, literacy campaigns should target both men and women, and boys and girls, particularly i n disadvantaged communities in southern Africa. This paper i s presented as follow s : Chapter 2 health determinants and parental influence, then presents the context of Zambia and relevant demographic inf o rmation . Chapter 3 presents the empirical strategy and expl ains the analysis and specifications applied in the estimations. Chapter 4 introduces the data used in the analysis and describes the key variables. Chapter 5 describes the regression results for c health outcomes, and the interpretat ions of the coefficients. Chapter 6 discuss es the 4 interpretations of the results as they relate to the existing literature. Chapter 7 concludes the paper with a broad interpretation of the results , and discusses some potential limitations as well as the potential of future research . 5 CHAPTER 2. BACKGROUND AND COUNTRY CONTEX T lth outcomes are determined, one must c onsider what factors contribute to the health of an individual, and particularly to a young child. Empirical literature ife stages (Rosenzweig and Schultz, 198 2; Barrera, 1990; Thomas et al., 1990; ) . Smith and Haddad (2000) provide an overview of the relationships between determinants of child malnutrition in developing countries, and suggest that the primary d eterminants fit into three levels of influence: 1) immediate; 2) underlying; 3 ) basic. The first level, immediate causes of malnutrition, include dietary intake and health status. These are related to the underlying factors of food insecurity, care for mot hers and children, and the quality of the health environment. Basic influences , such as the economic resource availability and the political environment, strongly affect the two previous levels. They offer a reduced form for household maximization of child (1) where household; is the health environm ent, includi ng the availability of sanitation and health services; environment, including the agroclimatic potential, soil fertility and water stress level; is the cultural norms affecting cari ng practices; and , which 6 are national and househ old the nutritional provisioning process, a 1 In their analysis of 63 countries, Smith and Haddad (2000) has th e largest influence o n child malnutrition . T his finding is supported by many other studies of lude characteristics of parental human capital (Barrera, 1990; Thomas et al., 1990; Thomas, 1994; ) . Maternal education is c onsistently identified as a determinant in several measures of improved child outcomes, and a possible reason could be that better - educated mothers are more efficient in their use of health inputs and can more easily un derstand health information to provid e better care for their children (Barrera, 1990; G ü ne , 2015). to the marriage market - matching between spouses of similar levels of education, where better edu cated women marry well - educated men who are more likely to earn higher incomes than men with less education ( Chiap p or i et al., 2009; Fafchamps and Shilpi, 2013), and that related research should be more inclusive of con trols which may be endogenous to moth ( Be hr man and Wolfe, 1987; Behrman, 1997). Nonetheless, spouses often make different decisions about the allocation of household resources depending on their own backgrounds and relative bargaining power ( Chiap p ori, 1988; Browning et al., 1994; Haddad et al. , 1994; Hoddinott and Haddad, 1995; Udry et al., 1995) , and i n households where disempowered women gain increased control and decision - making ability over some assets and income , there is sometimes an ob served shift in the distribution o f household resources in favor of improved 1 See Chapter 2 of Smith and Haddad (2000) for a full description of contributing functi o ns. 7 ( Thomas, 1994; Lundberg, Pollak, and Wales, 1997; Rubalcava and Thomas, 2000; Mofya - Mukuka and Sambo, 2018) . 2.2 The Context of Zambia Za mbia is a landlocked country in ce ntral southern Africa with a population of over 17.3 million people (see Figure 1) (World Bank , 2019a). Zambia was classified as a lower - middle income country in 2011 after a decade of r obust economic growth, largely than k s to its exportation of copper (W orld Bank, 2019b) . Despite these recent years of national economic growth and decades of foreign aid, Zambia is considered one of the poorest and most food - insecure countries in the world. Zambia experienced several years of increasing political and financial instability and crippling droughts. Currently, a bout 56.5% of the improvements has not improved overal l poverty levels , but rather incre inequality (World Bank , 2019a). Over 76% of rural Zambians live below the poverty line of US D 1.90 (2011 PPP) 2 , and the rural poor account for 82% of all the poor in Zambia (World Bank, 201 9c) , whic h indicates that the bur den of poverty disproportionately affects the rural communities in Zambia. Nearly half (45.9%) of the population in Zambia is considered undernourished (FAO et al., 2017), and the Global Health Index finds the prevalence o f hunger levels, grouping it with countries which are experiencing extreme climatic crises and sectarian violence, such as Yemen and Chad (Grebmer et al., 2018, pg. 5) . Food insecurity disproportionally affects rural poor populat ions, esp ecially young children. Zambia has one of 2 International Poverty line: 6.4 Zambian Kwacha (2015) or US D 1.90 (2011 PPP) (World Bank, 2019c). 8 the highest under - 5 mortality rates in the world at 75 deaths per 1000 live births, and a large part of this is due to chronic malnutrition leaving children vulnerable to infections and disease (CSO et al. , 201 5 ; F AO et al., 2017 ) . A comm on result of chronic malnutrition is stunting, the delayed or unachieved physical growth of a child defined as having height - for - age which is 2 standard deviations (or more) below the median height - for - age established by th e W orld H ealth Child Growth Standards (WHO Multicentre Growth Reference Study Group, 2006) . Childhood s tunting is associated with severe long - term health concerns, including an increased risk of premature deaths from pneumonia and diarrhea, as well a s bein g more susceptible to fatal infections, such as sepsis and meningitis (de Onis and Branca, 2016) . Figure 1 : Map of Africa with Zambia emphasized . Many countries in the world, including Zambia, experience a regula r period of hunger each year after rural households use most of their monetary assets to purchase agricultural inputs and plant their crops, and before the first crops are available to harve st. In Zambia, this period begins in November and usually lasts un til late February, when fresh maize is beginning to be harvested 9 (FEWS NET, 2013). Seasonal short - term hunger contributes to a periodically higher prevalence of underweight children and adul ts. rebmer et al., 2018), Zambia has not experienced violent conflicts or severe climatic crises (though, the regional rain patterns have become more erratic in recent years) . Low agricultural p roduction and a lack of infrastructure in rural areas continues to slow pro gress. Zambia is a very large country, and has a low population density, and when co nsidering the relatively mild yet diverse climate (Zambia hosts three agro - ecological zones and m any large freshwater resources), there is enormous potential for i ncreased agricultural production to alleviate poverty and hunger . 2. 2 . a lopment of a country or region. The health status of children unde r 5 years old is often determined using anthropometric measures, including the height and weight of a child and how it compares to the standardized scale for a given age. These measures can indicate if a child is stunted, wasted or underweight. In Zambia, 40% of c hildren under 5 years old are stunted (too short for their age), 6% are wasted (too thin for their height), and 15% are underweight (too thin for their age). Only about 11% of child ren ages 6 23 months consume an appropriate diet, as outlined by recommen ded infant and young child feeding (IYCF) practices (WHO, 2008) . The above - mentioned under - 5 mortality rate of 75 deaths per 1000 live births was estimated in 2014, and translates t o one in every 13 children born in Zambia do not survive to their fifth bir thday. In Eastern P rovince, where my survey data was collected , approximately 43.3% of children are stunted and the under - 5 mortality rate is estimated at 115 deaths per 1000 live b irths. Overall, while stunting rates have only moderately decrease d in the previous two decades, the mortality 10 rate has improved dramatically, as the national under - 5 mortality rate was approximately 191 deaths per 1000 live births in 1992 (CSO et al., 201 5 ). Longer spacing between births is associated with a lower preva lence of several negative health rate of stunting is approximately 34%, as compared to the rate of 48% of children born less than 24 months since the previous b irth. Chi ldren born within two years of the previous birth are also more likely to die before age five than children who are born more than 3 years after the previous birth , and ch ildren who are born as the seventh or more birth of the mother also have a h igher cha n c e of dying before age five compared to the second to sixth child (CSO et al., 201 5 ). Zambia. Children are more likely to be stunted when their mother is underweig ht or if she has never attended school, and they are 4 times more likely to be underweight (5% versus 20%) and 2.5 times as likely to die before their fifth birthday if their mot her has no education compared to children whose mother has more than a seconda ry education (109 deaths per 1000 live births versus 43 deaths per 1000 live births) (CSO et al., 201 5 ). 2. 2.b Maternal Health in Zambia Beginning before they are even born, youn g children in Zambia face a multitude of obstacles to achieve good he alth and can play an important role in determining the health outcomes of her child. To understand pathways for impro lth of wo men of child - bearing age. In Zambia, approximately 10% of women ages 15 - 49 are underweight (Body - Mass - Index less than 18.5). Women in Eastern P rovince are less likely to be under weight than women in other provinces, as only 7.8% are underweight. W omen who have 11 more years of education are less likely to be underweight, as about 12% of women who have no education or only primary educations are underweight, while about 8% of women wh o attended secondary school are underweight (CSO et al., 201 5 ) 2. 2.c Education and Literacy in Zambia The Zambian education system consists of primary education from grade 1 through grade 7, then secondary education from grade 8 through grade 12, then tert iary education which consists of university and post - secondary profes sional tr To increase the primary school enrollment rates, the Zambian government has maintained a regulation of free primary educatio n in government - run schools since 2002. Secondary education is not fr ee, but d oes have a maximum cap on the tuition a school can charge for grades 8 - 12 (CSO et al., 201 5 ) . rienced increased primary school attendance: the rate of approximatel y 80% of primary - school - aged children attending school remained the same over the period from 2007 to 2014. Most u nfortunately, the high enrollment rate for primary school drops precipito usly for secondary school grade levels, with only about 40% of studen ts contin uing their education at the secondary - increasing from 35% in 2007 to 41% in 2014. Nati onally, approximately 13% of women have completed secondary school or higher, and just over 21% of men have completed secondary school or higher. Eastern P rovince has some of the lowest rates of attendance and secondary school completion, with 70% primary school attendance and 22% secondary school attendance, and only 5.4% of adult women and 9.8% of adult men having completed secondary school or 12 higher. Women in Eastern province have attended school for an average of 4.8 years, and men attended for an avera ge of 5.8 years (CSO et al., 201 5 ) . There are several contributing f actors to the relatively low and stagnant attendance rates, notably a lack of secondary schools in rural areas as well as steeply increased school fees beginning in Grade 8, which ultimat ely result in approximately fewer than 17% of all Zambian adults comp leting Gr ade 12 . People living in rural areas are at a particular disadvantage to furthering their educations, because most rural communities can only access schools offering primary education. Women in rural areas have an average of 5.5 years of education , versus the 8.2 years of women in urban areas, and only 4.3% of rural women complete secondary school or higher, compared to the nearly 24% of urban - dwelling women who have completed secondary school or higher (CSO et al., 201 5 ) . Additionally, the policie s of free primary education and price - capped secondary education are often not followed or enforced, or other indirect fees are applied, and therefore these policies to expand education have likely not been as beneficial as expected (RTE, 2012). The Zambia n Demogra phic and Health Survey (DHS) for 2013/2014 conducted a literacy exam among male and female respondents ages 15 - 49, and found that 83% of men and 68% of women are literate in at least one of seven major Zambian language groups. Younger people (ages 15 - 24) o f both s exes are more likely to be literate than people in older age groups, and literacy also increases with higher wealth quintiles. There is also a large difference in literacy rates between urban and rural populations: 83% of women in urban ar eas can r ead and write, compared to the 54% of women in rural areas who can read and write. Similar to the underperforming education rates, Eastern Province has some of the lowest literacy rates in the country, as about 66.3% of men and 49.3% of women can read and write. I t is not surprising that Eastern would have such 13 low literacy rates, given the low education rates and the strong relationship between education and literacy. (CSO et al., 201 5 ). 2. 2.d In rural communi ties in Z ambia, w omen and girls are actively involved in the agricultural activities of their household, and women constitute over half of the agricultural workers in Zambia (CSO, 2015). Typical of many regions of Africa, there is some division of agricult ural labo r, gende red crop production and home garden work, but agricultural households in Zambia are often more integrated than most areas in West Africa regarding the joint decision - making ite high levels o f engagement in agricultural productivity, and even in decision - making regarding agricultural activities, many women in Zambia are rarely documented as land owners (SIDA, 2008). Th control over resources is particularly un fortunate, given recent research which has found that as women in rural Zambia have increased access to and control over agricultural resources in the household, there is an observed increase in the household dietary diversity (Mofya - Mukaka and Sambo, 2018 ) . 14 CHAPT ER 3. ESTIMATION STRATEGY 3 .1 Estimation Specifications quality health: ( 2 ) where is a given health outcome for child i in household j , namely their height - for - age (HAZ) and weig ht - for - age (WAZ) z - scores, which determine the status of being stunted or underweight; represents the different parental empowerme nt measures which are estimated for their empo werment scores, and the ; is a vector of binary v ariables indicating the literacy status of the spouses ; is the body - mass - index for the wife of house hold j ; is a vector of household characteristics, including the number of household members, the dependency ratio, a gender parity indicator 3 and the ages of the spouses; is a vector of child characteristics including age in months and ge nder; is a vector of district indicator s ; the are parameters to be estimate d ; and is an error term. The key parameters of interest are and , as these represent the parameters for the parental inputs and other inputs used section 3 .2. 3 Gender Parity is defined in section 4.4. 15 ns as used for aracteristics: ( 3 ) where represents a given health outcome of the wife in household j , namely her own dietary diversity and body mass index. These estimates a re included in the analysis of understanding s health production function ( Barrera, 1990; Thomas et al., 199 0 ), therefore it is valuable to establish how variou s human capital inputs are related to improved health outcomes for women of child - bearing ages ( 15 - 49 years old) . All primary female respon dents ages 15 - 49 are included in this analysis, and m ost, though not all, of the women included in the analysis had children under 5 years old, but they are included because they are in the child - bearing age range. 3 . 2 Description of Specifications E quat ions (2 ) and (3 ) are each estimated using five specifications to determine the relationship between the parental/sp ousal human capital inputs and the health outcomes of children under five years of age and women of child - bearing age . The specifications are denoted by above in equations (2) and (3) : Literacy - focused specifications: 1) Indicator variables representing if both spouses can read and write, the woman cannot read and write but the man can, or the woman can read and write but the man cannot (on ly one of these indicators can equal 1 for each household); 2) The same variables as above, with the inclusions of maternal BMI (BMI is excluded in equation ( 3 ) ); 16 3) All variables from the seco nd specification, as well as the individual empowerment scores of both the husband and wife; 4) All variables from the second specification, as well as a dummy indicating if the household got a loan, and empowerment measures of ecisions to borrow credit and how to use the credit, respectively : 5) decisions. Following the established methods of studies using the WEAI measurements ( Sraboni et al., 2014; M alapit and Quisumbing, 2015; Malapit et al, 2015; Malapit et al., 2019), the decision input variables for specifications 4 and 5 are included to examine the relationships of female empowerment measures with outcomes. These specifi c variables are included because they are derived from the empowerment indicators which most contribute to Robust standard errors are obtained by clustering at the hou sehold level for the ch regressions, because there are several households with more than one child under the age of five. In Appendix A, there are several robustness checks estimating how health outcomes are related to the independent variables of parental and spousal ed empowerment scores. Due to the likely endogeneity of the empowerment measures and scor es , all results should be interpreted as associative rather than causal. 17 CHAPTER 4. DATA I use cross - sectional survey data from Zambi a. The survey was co nducted as the baseline instrument for Initiative - Zone of Influence (ZOI) : Zambia population - based survey , 4 and was implemented between N ovember 19 and December 6, 2012 . T he geographic area targeted by the Feed the Future interventions (i.e. , consists of five districts within Eastern P rovince: Lundazi, Chipata, Katete, Petauke, and Nyimba ( see Figure 2 ). Across the ZO I, a total of 1,640 ho us eholds were interviewed during this round of data collection. All the sampled households had been interviewed as part of the Rural Agricultural Livelihoods Survey, which was implemented by the Lusaka - based Indaba Agricultural Policy Research Institute du ri ng June and July 2012 (FEEDBACK, 2013). The ZOI survey collected information on 1) household characteristics: household member demographics, dwelling characteristics, and a household hunger scale; 2) physical characteristics of women and young children: an thropometric measures of women ages 15 - 49 and children under 5 years old; 3) Dietary Consumption: a 24 - hour dietary recall for women ages 15 - 49 and children under 2 ye the WEAI module , which was administered to both th e male and female household heads, whenever possible. About three - fourths (~76%) of the nearly 1 , 400 female respondents were between 15 and 49 years old, but the survey i ncludes women above that age range, as well. 4 Researchers at the International Food Policy Research Institute (IFPRI) designed the survey, and Rockville , Maryland - based Westat implemented it in partnership with TANGO international and IFPRI. T he Zambia Central Statistic al Office (CSO) conducted the survey with help from the National Food and Nutrition Commission. 18 Figure 2 : Map of t he Zone of Influence in Eastern P rovince , Zambia. 5 My analysis is restricted to rural households. I further restrict the sample to households which have at least one child under 5 years old and from which a male and female adult respondent have both com pleted the survey. While the survey is administered t o both male and female adults of households, there are fewer men who responded to the survey: more than 300 female respondents did not have a male co - respondent despite most of them living in male - he ade d household. This absence of male respondents resulted in a smaller useable sample size, since only houses with both male and female respondents could be analyzed due to the estimation of 5 This map reflects the district borders a s they were defined in 2012, when the data was collected. In recent years, Zambia has redefined the borders of many districts. 19 variables based on the male partner. 6 For nearly every household inc luded in the sample, the adult male and female survey respondents are reported as being husband and wife, so I often refer to them as spouses . Similarly, the vast majority of primary respondents were recorded as the parent of any child under 5 years old , s o the spouses are both presumed to be the parents of children in the household. I analyze individual - level data of 1 , 094 children under 5 years old . These children come from a total of 943 households. Table 1 presents the descriptive statistics of the v aria bles used in the analysis. - score var and their empowerment measuremen ts . The correlation matrix of key variables can be found in Appendix B . I perform linear regressions using OLS to analyze the relationships between the parental human capitol measures, and I analyze predictors of S us ing Ordered Logit . 6 Female - headed households (FHHs) were excluded from this analysis due to the pervasiv e differences in household characteristics exhibited in FHHs, such as less land and asset ownership, well as women often being the sole - decision - maker (Asian Development Bank, 2013), which does not reflect the differences in spousal empowerment I examine i n this analysis. 20 Table 1 : Summary Statistics . VARIABLE Obs Mean Std. Dev. Min Max A. Child Outcomes and characteristics, 0 - 59 months Height for Age Z - score (HAZ) 1072 - 1.851 1.423 - 5.89 5.24 Weight for Age Z - score (WA Z) 1 092 - 0.904 1.06 - 4.79 3.49 Girl 1094 0.518 0.5 00 0 1 Age in months 1094 31.8 14.6 6 59 B. Women's Outcomes and Characteristics Dietary Diversity - Number of Food Groups Consumed (out of 9 food groups) 943 3.97 1.17 0 8 Body Mass Ind ex ( BMI) 874 22.6 3.4 0 10.7 45. 9 C. Education and Literacy Indicators Man c an r ead and write 941 0. 61 0.5 00 0 1 Woman c an r ead and w rite 940 0. 34 0.492 0 1 Both spouses can read and write 943 0. 26 0.494 0 1 Man can, Woman cannot read and writ e 943 0. 35 0.388 0 1 Man cannot, Woman can read and write 943 0. 06 0.171 0 1 Neither spouse can read or write 943 0. 32 0.477 0 1 n umber of y ears of s chooling 910 6.42 2.82 0 12 Woman's n umber of y ears of s chooling 700 5. 42 2.59 0 12 D. Empowerment Indicators Female Empowerment Score, = 1 if empowered in all indicators 943 0.735 0.141 0.067 1 Male Empowerment Score 886 0.81 5 0.11 0 0.333 1 Percent difference of Parity Gap 886 0.08 8 0.233 - 1.7 0 0.926 Parity, =1 if woman is equ al t o or greater than 886 0. 361 0.445 0 1 Household Access Credit 943 0.402 0.49 0 1 943 0.199 0.4 0 1 943 0.23 0 0.421 0 1 Number of Agricultural Pro duction Decisions Woman is involved in 943 5.4 0 2.5 0 0 11 E. Household Characteristics Household head age in years 943 43.53 15.1 17 93 age in years 943 37.6 15 1 5 93 Household size 943 7.06 2.79 2 25 Dependency Ratio 943 1. 4 2 0.80 2 0 3.5 Annual Per Capita Gross Income (USD) 943 4 73.3 775.9 7.2 13105.9 21 4 . 1 Outcome Variables 4 . 1 .a Child Anthropometry - children under 5 years old: 7 Height - for - Age Z - score / Stunting : A child is considered stunted if his or her height - for - age meas urement (HA Z) is two or more standard deviations below the median of the reference group. The HAZ can be used as an indication of chronic malnutrition. The children in this sample have an aver age HAZ of - 1.8 (Figure 3 ). 8 Almost half (44.7%) of the children under 5 ye ars old in this sample are severely or moderately stunted, as they have an HAZ which is less than - 2.0 . Weight - for - Age Z - score / Underweight: A child is considered underweight if his or her weight - for - age measurement (WAZ) is two or more standar d deviation s below the median of the reference group. A low WAZ can indicate a combination of both chronic and acute malnutrition. The average WAZ for ch ildren in this study is - 0.9 (Figure 4 ). Nearly one - in - seven (14%) of children from this sample are und erweight, a s their WAZ is less than - 2.0. The sample averages are similar to the national prevalence of stunted and underweight children under 5 years o f age, which are 40% and 15%, respectively, according to the 2013 - 2014 Zambia Demographic Household Sur vey (CSO et al., 2015) . 7 C alculated using the 2 0 06 WHO Child Growth Standards 8 Z - scores beyond ± 6.0 were not included in the analysis, as values beyond those limits are considered measurement errors (WHO Multicentre Growth Reference Study Group, 2006). 22 Figure 3 : Distribution of Children's Height - for - Age Z - scores. Figure 4 : Distribution of Children's Weight - for - Age Z - scores. 23 4 . 1 .b - women 15 - 49 year s old: Diet ary Diversity Score (DDS) : DDS is defined as t he number of food groups consumed in the past 24 hours, out of nine food groups: 1) grains, roots, and tubers; 2) legumes and nuts; 3) dairy products; 4) organ meat; 5) eggs; 6) flesh foods (i.e. mea t, fish, in sects); 7) vitamin - A - rich dark, leafy greens; 8) Other vitamin - rich fruits and vegetables (ex., pumpkin or orange - flesh sweet potatoes); and 9) other fruits and vegetables (FAO, 2011) . The women in this sample consumed an average of about four f ood groups during the day before the survey (Figure 5 ). Figure 5 : Number of Food Groups Consumed by Women in Previous 24 Hours . 24 4 . 2 Literacy and Education Variables 4.2.a Literacy The literacy status of spouses, whether the p rimary male and female respondents are able to read and write, is used to estimate the relationship between spousal human capital and the health outcomes of young children or women, respectively. Of the adult survey respondents, 61.4% of men are able to re ad and writ e, and 3 4.9 % of women c an read and write. 4 . 2 . b Education The number of years of education completed by the individual spouses are used as in dependent variables to estimate the relationship between the human capital of spouses and the health out comes of yo ung children or women of child - bearing age. The average amount of education completed by male respondents is about 6.4 years (Figure 6 ). The a verage amount of education completed by female respondents is about 5.4 years (Figure 7 ) . 25 Figure 6 : Distribution of Men's Years of Education. Figure 7 : Distribution of Women's Years of Education. 26 4 . 3 To incorporate the possibility for the potential influence of nutritional outcomes, I also inc lude empowerment indicators in regressions for ealth outcomes, using data from the WEAI. The WEAI is an instrument developed by the Oxford Poverty and Human Developm ent Initiativ e (OPHI) and IFPRI as a cross - several dimensions related to household, social and agricultural decisions in low - and middle - income countries (Alkire et al., 2013a). The index was initially develo ped to measur e changes in I nitiative programs, but other institutions and governments have embraced WEAI to better understand relationships 9 T he WEAI modul e is administered to men and women to measure their own individual levels of empowerment. The WEAI aggregates data from doz ens of variables to calculate an individual WEAI score, which ranges from 0.066 to 1.0, based on the five domains of emp owerment. The se five domains are created from a total of ten indicators, and each domain is weighted equally. (See Appendix C for a tabl e of the domains and indicators ). 4 . 3 . a Empowerment Measures My analysis uses 9 of th e 10 indicators , and the domains w ere re - weight ed accordingly. 10 , 11 Within my analysis , of the 943 female respondents whose WEAI score is used , the average 9 See Alkire et al., (2013a) for more details ab out methodology, piloting and validation of the W EAI. 10 To analyze the WEAI module, I follow the instructions provided in the Instructional Guide on the WEAI (Alkire et al., 2013b). 11 The time - use indicator, 5.1, was not available for m y dataset. 27 empowerment score is 0.735 (Figure 8 ). The male respondents had an average WEAI score of 0.81 (Figure 9 ). Following the methods intro duced by Srab oni et al. (2013), I use the indicators which are most important in the disempowerment of women in this sample to create models to test for contributors t o dis empowerm ent are used because these are considered the areas in which women are most vulnerable . Figure 10 presents the values of how much each of the five domain s contributes to the disempowerment of the women surveyed. The domains of Production and Resources, re spectively, contribute 37.1% and 28.9% to the disempowerment of women in rural Zambia. Further, looking at a breakdown of the indicators of disempowerment, Autonomy in Agriculture and Access to and Decisions on Credit contribute the most in th eir respectiv e 1 1 ). 28 Figure 8 Figure 9 s. 29 Figure 10 : Contribution of each Domain to Women's Disempowerment. Figure 11 . 30 4.4 Control Variables 4.4 . a R egression C ontrols These control variabl es are includ the maternal health analysis. Nearly 52% of children under 5 years old in this sample are female, and about 4 8 % are male children. : The children incl uded in this study are a maximum of 59 months old, and the youngest child who is inc luded in the analysis is 6 months old (there were very few observations for children 0 - 6 months included in the data collection). The average age in months is nearly 32 mon ths of age, w hich is 2 years and 8 months old. Body - Mass - Index (BMI) : BMI is defined as t he ratio of weight (in kgs) to the square of height (in meters) (kg/m 2 ). A woman is considered underweight if her BMI is below 18.5 . The average BMI of the women in this sample is 22.6 , and 6.4% of the women are underweigh t. The proportion of underweight women in this sample is less than the 10% national average found during the 2013 - 2014 Zambia - DHS . T he average in Eastern Province was lower , at 7.8% (CSO et al., 2015). 4.4.b General Regression Controls These variables are includ outcomes. a ge : The average age of the male household head in this sample is about 43 .5 years old. The young est male prim ary respondent is 17 while the oldest included in the analys is is 93 . 31 a ge : The average age of the female respondent in this sample is about 37 .5 years old. The youngest female primary respondent is 15 while the oldest woman included i n the analysi s is 93 ; however , anthropometric measures are only available for women 15 - 49 . Household size : The average household in this sample has about 7 people living in their house or immediate vicinity with whom the primary respondents share their r esources. The largest household has 25 people, while the smallest househo ld only has 2 people. Gender Parity : The Gender Parity Index captures the relative empowerment of women, as their empowerment score is compared to the empowerment score of the male W EAI responden t in their household. The Gender Parity Gap refers to the difference in empowerment between the male and female WEAI respondents in a household. Dual - adult households are considered to WEAI score is the same as o r greater than her parity, as measured by their WEAI scores. Dependency r atio : The average household in this sample has a dependency ratio of 1.4, indicating that for every adu lt, there are 1.4 minors in the household, or about 3 minors for every 2 adults, on average. The smallest dependency ratio is 0, indicating no minors in the home, and the largest ratio is 6, indicating there are 6 minors for each adult in the household. P er c apita g ross i ncome (USD) : The average gross per capita income from household s in this sample is 473.25 USD. The lowest per capita income in a household is 7.20 USD, while the largest per capita income is 13, 105.90 USD. 32 CHAPTER 5. RESU LTS I estima te the regr essions via OLS. I estimate the DDS using Ordered Logit, as it is a count variable, rather than a continuous variable. To facilitate interpretation, the estimated parameters a re divided by the mean value of the outcom e variable to find the average percentage change, rather than describing the changes as point - value changes. This interpretation is intended to clarify the estimates as relative changes to the outcome variable sinc e the outcome variables are standardized a s z - score values. 5.1.a Parental Literacy health outcomes (Table 2), when both parents can read and write, there is a very strong a nd positive relationship with households where both parents can read and write are expected to have a HAZ that is an average of 34% higher than children from homes where both parents are illiterate, and a WAZ that is, on average, about 32% greater. These values suggest a very large difference among children, especially when considering that the average HAZ is - 1.8 and the average WAZ is - 0.9, and 34% and 32% increases, respectively, could potentially bring many mod erately stunted and underweight children above the threshold of being stunted and underweight. In cases when the father is literate but the mother is not, children in the household could have a WAZ that is, on average, 20% higher than that of children whos e male household head is illiterate, but these values were only statistically significant at the 10% level for specifications 2 and 5. 33 indica tes that an i ncrease of 0.1 This 0.10 increase is the approximate equivalent of a woman being considered adequately empowered in one more indic ator of the WEAI index. 12 Looking at the regressions for HAZ, there is a large negative correlation related to households that have gotten a loan 0 .30 indicates an average de crease of 16% in the HAZ o f children from households wh ich borrow credit from households wh ich do not borrow credit. 12 See Alk ire et al. (2013a) and Appendix C for further explanati on and breakdown of WEAI indicators. 34 Table 2 : Parental Literacy as Predictors of Child Health Outcomes . Height - for - Age Z - score (mean = - 1.8) Weight - for - Age Z - sc ore (me an = - 0.9) VARIABLES 1 2 3 4 5 1 2 3 4 5 Both Spouses can Read and Write 0.523*** 0.631*** 0.608*** 0.607*** 0.631*** 0.290*** 0.295** 0.280** 0.301** 0.292** (0.142) (0.156) (0.153) (0.156) (0.156) (0.109) (0.122) (0.121) (0.124) (0.121) Woma n can, Man cannot Read and Write 0.075 0.119 0.110 0.098 0.116 0.035 0.108 0.102 0.095 0.122 (0.185) (0.206) (0.200) (0.203) (0.208) (0.150) (0.169) (0.167) (0.170) (0.169) Woman cannot, Man can Read and Write 0.118 0.207 0.201 0.179 0.206 0.146 0.180* 0.173 0.181 0.183* (0.122) (0.132) (0.132) (0.132) (0.132) (0.098) (0.109) (0.107) (0.110) (0.108) Maternal Body Mass Index 0.001 - 0.001 0.001 0.001 0.017 0.014 0.017 0.017 (0.017) (0.017) (0.017) (0.017) (0.013) (0.013) (0.013) (0 .013) Woman' s Empowerment Score 0.012** 0.010** (0.006) (0.004) Man's Empowerment Score - 0.006 - 0.002 (0.007) (0.004) Household Borrows Credit - 0.300** - 0.034 (0.136) (0.108) Woman's Input on Dec ision to Borr ow Credit 0.172 - 0.138 (0.208) (0.163) Woman's Input on Decision to Use Credit 0.009 0.097 (0.194) (0.154) Woman's Input in Agricultural Decisions - 0.007 0.028 (0.026) (0.020) Observa tions 855 706 706 706 706 869 713 713 713 713 R - squared 0.054 0.074 0.079 0.081 0.074 0.047 0.051 0.059 0.053 0.054 Robust standard errors in parentheses . *** p<0.01, ** p<0.05, * p<0.1 Additional controls include: Number of househo ld members, d ependency ration, gender parity indicator, 35 The results values presented are the Odds Ratios from the Ordered Logit regression s: the differ ence from 1 indicates the percentage likelihood of a change from one discrete value to the next. 5. 2 . a Spousal Literacy h ere both spo uses can read and write are more l ikely to have women with higher dietary diversity scores. Table 3 health is simultaneous literacy of spouses . When both spous es can read and write, women are 4 0% more likel y to have consumed an additional food group in the previous 24 hours than women who come from households where both spouses are illiterate. The average woman in this sample had consumed four food groups the da y before, and this indicates that the average w oman who is literate and has a literate husband is more likely to have consumed five food groups. This result is robust across all specifications at the 5% significance level. 36 Table 3 : Spousal Literacy as Predictor s of Women's Dietary Diversity . Dietary Diversity Score VARIABLES 1 2 3 4 Both Spouses can Read and Write 1.472** 1.466** 1.435** 1.459** (0.249) (0.249) (0.245) (0.247) Woman can, Man cannot Read and Write 1.200 1.194 1.220 1.203 (0.329) (0.327 ) (0.335) (0. 329) Woman cannot, Man can Read and Write 1.123 1.129 1.118 1.116 (0.178) (0.179) (0.177) (0.177) Woman's Empowerment Score 1.008 (0.007) Man's Empowerment Score 1.000 (0.007) Household Borrows Credit 1.062 (0. 177) Woman' s Input on Decision to Borrow Credit 1.346 (0.323) Woman's Input on Decision to Use Credit 0.890 (0.208) Woman's Input in Agricultural Decisions 1.047 (0.031) Observations 812 812 812 812 Values expressed as odds ratios. Robu st standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 Additional controls include: Number of household members, dependency ratio, gender parity indicator, 37 The lack of statist ically signif Appendix A, Table 6) potentially demonstrate that wo the same series of specifications, I can eliminate the threat of multi collin fications use d to thin (Thomas et al., 1990; Thomas, 1994; Malapit and Quisumbing, 2015). 38 CHAPTER 6. DISCUSSION The results provide an overarching indication that the joint literacy of parents is very important to positive health outcomes of young children . These results are supported by a large body of existing literature w hich find that increased parental education is associated with healthier children ( Rosenzweig and Schultz, 1982; Ba rrera, 1990; Thomas et al., 1990; Smith and Haddad, 2000; Fafchamps and Shilpi, 2013). Additionally, the results support researc h which find s that the - education (B ehrman, 1997), perhaps because marriage market matching of better educated women marrying better educated men, and the education al attainment of both parents is important (Chiappori, Iyigun, and Weiss, 2009; Fafchamps and Shilpi, 2013). The parental liter acy - focused results (Table 2 ) find a very large correlation between households where both parents are literate and positive outc omes for chil about 26% of households in this sample have parents who are both literate, and in a bout 32 % of households neither parent is literate, these results suggest that there could be very real differences in these grou ps of househo lds. Indeed, the average HAZ for children from households with parents who are both able to read and write is appro ximately - 1. 47 , while the average HAZ for children from households with illiterate parents is - 2. 11 . See Figure 1 2 for the distr ibution of ch 39 Figure 12 : Distribution of Height - for - Age Z - scores by Parental Literacy Status. The red reference line on each chart indicates the mean value of the HAZ of the chi ldren. The e suggests that there i s a positive relationship between this empowerment measur e health ural decision s and activities . However , t he other empowerment measures derived from the WEAI disempowerment, do not indicate strong relationsh ips with the . These results are aligned with previous studies which fail to identify strong r elationships between the , despite the identified positive correlation with 40 the general em powerment of the mother (Malapit et al., 2015; Malapit and Quisumbing, 2015; Malapit et al., 2019). These findings may b e partially due to the construction of the survey and its focus on productive activities, rather than asking more specifically about beh aviors that m ight y. T he analyses for determining predictors of health outcomes revealed s ome consistent es , implying that when a literate woman is married to a literate man, she is likely to have a slightly more diverse diet than a woman who is not literate or whose husband is not literate. indicate any statistic ally significant relationship between the intra - household health outcomes. This may imply that there are many unobserved influences that contribute to an an implica tion which is supported by research which suggests that unobserved influences, such as childhood health endowments , are health (Behrman and Wolfe, 1987). 41 CHAPTER 7. CONCLUSION The results indicate a robust positive relatio nship between parent al literacy h . This finding is compatible with associations that better educated household heads are often better able to provide for their households. The results of the l iteracy specifications imply that while the individual li teracy of a parent might positively contribute somewhat to child health outcomes, the combination of both household heads being able to read and write ha s a consistently large, positive relationship. In rural Zambia, where so few women comple te secondary school (about 4%), policies and on are important for closing the gender gap in education and literacy. for the sake of improving and children - being is important, it is essential for both parents to achieve literacy and w ork towards higher levels of education. One parent cannot be left behind in this context of educational parity, as their combined contributions create an env ironment whic h provides their children with better odds of a healthy present and future. 7.1 Limit ations and Future Research Few works of research are ideal, and this analysis was limited in several ways. Because this study is based on cross - sectional dat a from a most ly self - reported survey, rather than experimental data from an intervention evaluation or panel data , it is not possible to make any causal inferences. The results present statistically significant positive associations between parental litera cy and childr literacy a ffects child 42 The Feed the Future survey collected data on many household and individual characteristics, but did not include a few indicators of c future well - being. These indicators inc lude recent illnesses, such as diarrhea and malaria, as well as birth order and distance to health services . Using this same data set, further analysis could be performed g the types of water and sanitation access the households use. While these hygiene indicators may not provide direct information about the recent poor health incidents of individuals in the household, they can provide some context of household conditions t hat may inform researchers An additional limitation is that t he WEAI module so licits respon agricultural activities and decision making, but it does not ask specific for information related to the types of crops planted, nutritional knowledge, food accessibility, or dietary preferences . While the surveye d households had recently participated in the RALS, which collects extensive information about the income sources, that data set is not easily linked to this consumption component. The RALS surv ey was collected as panel data in 2012, 2015 and 2019, and there are additional rounds of FtF surveys, but the 2015 survey interviewed different households from RALS than the 2012 survey used in this study , so a panel analysis is not possible. If the RALS data were properly linked to the FtF data, an analysis on agricultural diversity would be interesting to see how it may relate to the foods consumed by members of the household. This information could help to clarify if the observed dietary co nsumption of women and children is a choice or based on limited options. 43 Beyond the scope of this d ata set, a nalysis of countries in Africa which have documented marked hic changes a re aligned with the shift in health outcomes. Many organizations have collected rich d ata sets from this region which could possibly be used to better understand the related pathways and relationships connecting policy changes or household beh avior and chi ld health outcomes. Zambia has some of the highest rates of stunting in the world, esp ecially in its rural areas, and research which emphasizes improved agricultural production could be analyzed in the context of Because stunt ing results from chronic malnutrition, potential interventions would require longer ti and health care follows children over their life course. 44 APPENDICES 45 APPENDIX A. ROBUSTNESS CH ECKS 46 Table 4 : Parental Education as Predictors of Child Health Outcomes. Height - for - Age Z - score (mean = - 1.8) Weight - for - Age Z - score (mean = - 0.9) VARIABLES 1 2 3 4 5 1 2 3 4 5 Man's Years of Education 0.058*** 0.068*** 0.070* ** 0.073*** 0 .068*** 0.039*** 0.035** 0.035** 0.037** 0.035** (0.015) (0.017) (0.017) (0.017) (0.017) (0.011) (0.015) (0.015) (0.015) (0.015) Woman's Years of Education 0.026 0.03 0.027 0.025 0.029 0.015 0.007 0.006 0.006 0.005 (0.020) (0.027) (0.027) (0.027) (0.0 27) (0.016) (0.021) (0.021) (0.021) (0.021) Maternal Body Mass Index 0.001 0.000 - 0.001 0.000 0.026* 0.024* 0.024 0.026* (0.019) (0.020) (0.019) (0.019) (0.015) (0.014) (0.015) (0.014) Woman's Empowerment Score 0.011 0.011** (0.007) (0.005) Man's Empowerment Score - 0.009 - 0.005 (0.007) (0.004) Household Borrows Credit - 0.277* - 0.095 (0.152) (0.122) Woman's Input on Decision to Borrow Credit 0.116 - 0.131 (0.222) (0 .179) Woman 's Input on Decision to Use Credit - 0.085 0.135 (0.219) (0.178) Woman's Input in Agricultural Decisions 0.010 0.027 (0.031) (0.024) Observations 792 536 536 536 536 807 542 542 542 542 R - squared 0.060 0.0 84 0.090 0.09 2 0.084 0.044 0.059 0.068 0.062 0.062 Robust standard errors in parentheses . *** p<0.01, ** p<0.05, * p<0.1 Additional controls include: Number of household members, dependency ration, gender parity indicator, 47 T able 5 : Dietary Diversity Score VARIABLES 1 2 3 4 Man's Years of Education 1.013 1.01 4 1.011 1.013 (0.030) (0.030) (0.030) (0.030) Women's Years of Education 1.082** 1.082** 1.083** 1.082** (0.039) (0.039) (0.039) (0.039) Woman's Empowerment Score 1.007 (0.009) Man's Empowerment Score 1.000 (0.008) Household Bor rows Credit 1.085 (0.219) Woman's Input on Decision to Borrow Credit 1.181 (0.338) Woman's Input on Decision to Use Credit 0.928 (0.260) Woman's Input in Agricultural Decisions 0.995 (0.036) Observations 537 537 537 537 Values ex pressed as odds ratios. Standard errors in parentheses . *** p<0.01, ** p<0.05, * p<0.1 The values presented are odds ratios of the ordered logit coefficients. Additional controls include: Number of household members, dependency ratio, gender parity indicator, indicator s. 48 Table 6 : y - Mass - Index. Body - Mass - Index VARIABLES 1 2 3 4 Both Spouses can Read and Write 0.509 0.534 0.551 0.481 (0.3 59) (0.361) (0.356) (0.359) Woman can, Man cannot Read and Write - 0.155 - 0.187 - 0.150 - 0.150 (0.484) (0.476) (0.488) (0.478) Woman cannot, Man can Read and Write - 0.291 - 0.283 - 0.274 - 0.302 (0.326) (0.326) (0.325) (0.326) Woman's Emp owerment Score 0.023 (0.014) Man's Empowerment Score 0.018 (0.013) Household Borrows Credit 0.131 (0.336) Woman's Input on Decision to Borrow Credit - 0.518 (0.446) Woman's Input on Decision to Use Credit 0.331 (0.452) Woman's Input in Agricultural Decisions 0.067 (0.055) Observations 729 729 729 729 R - squared 0.052 0.062 0.053 0.053 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Additiona l controls include: Number of household members, dependency ratio, gender parity indicator, 49 Table 7 : Parental Empowerment as Predictors of Child Health Outcomes . Additiona l controls inc lude: Number of household members, dependency ration, gender parity indicator, Height - for - Age Z - sco re Weight - for - Age Z - score VARIABLES 1 2 3 4 5 1 2 3 4 5 Woman's Empowerment Score 0.004 0.013** 0.012** 0.014** 0.014** 0.004 0.010** 0.010** 0.012*** 0.009** (0.003) (0.006) (0.006) (0.006) (0.006) (0.003) (0.004) (0.004) (0.004) (0.004 ) Man's Empow erment Score - 0.005 - 0.008 - 0.006 - 0.006 - 0.008 0.001 - 0.003 - 0.002 - 0.003 - 0.003 (0.005) (0.007) (0.007) (0.007) (0.007) (0.004) (0.004) (0.004) (0.004) (0.004) Maternal Body Mass Index 0.003 - 0.001 0.003 0.004 0.016 0.014 0.015 0.01 6 (0.018) (0.017) (0.018) (0.018) (0.013) (0.013) (0.013) (0.013) Both Spouses can Read and Write 0.608*** 0.280** (0.153) (0.121) Woman can, Man cannot Read and Write 0.110 0.102 (0.200) (0.167) Woman cannot, Man can Read and Write 0.201 0.173 (0.132) (0.107) Household Borrows Credit 0.214 - 0.047 (0.208) (0.109) Woman's Input on Decision to Borrow Credit - 0.089 - 0.131 (0.201) (0.158) Woman's Input on Decision to Use Credit - 0.305** 0.028 (0.139) (0.155) Woman's Input in Agricultural Decisions - 0.015 0.020 (0.02 8) (0. 021) Observations 873 706 706 706 706 885 713 713 713 713 R - squared 0.040 0.055 0.079 0.064 0.055 0.035 0.049 0.059 0.051 0.050 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 50 APPENDIX B . CORRELATIONMATRIX OF KEY INDEPENDENT VARIABLES 51 Table 8 : Correlation Matrix of Key Independent Variables. VARIABLES Both Spouses can Read and Write Woman can, Man cannot Read and Write Woman cannot, Man can Read and Write Man's Years of Education Woman's Years of Ed ucation Maternal Body Mass I ndex Woman's Input on Decision to Borrow Credit Woman's Input on Decision to Use Credit Household Borrows Credit Woman's Input in Agricultural Decisions Both Spouses can Read and Write 1 Woman can, Man cannot Read and Write - 0.222*** 1 Woman canno t, Man can Read and Write - 0.532*** - 0.232*** 1 Man's Years of Education 0.402*** - 0.169*** 0.0812 1 Woman's Years of Education 0.592*** 0.104* - 0.374*** 0.421*** 1 Maternal Body Mass Index 0.132** - 0.0204 - 0.0257 0.069 5 0.0445 1 Woman's Input on Decision to Borrow Credit 0.0819 - 0.0751 - 0.0196 0.133** 0.0442 - 0.00611 1 Woman's Input on Decision to Use Credit 0.0200 - 0.0227 - 0.0220 0.108* 0.0216 0.0673 0.684*** 1 Household Borrows Credit 0.00828 - 0.0590 - 0 .0239 0.120** - 0.0238 - 0.0346 0.489*** 0.596*** 1 Woman's Input in Agricultural Decisions 0.105* - 0.0749 - 0.0477 0.0654 0.129** 0.0734 0.141*** 0.167*** 0.0693 1 52 APPENDIX C . WOMEN S EMPOWERMENT IN AGRI CULTURE INDEX 53 Table 9 : The Domains Domain Indicator Definition of Indicator Weight 1. Production 1.1 Input in Productive Decisions Sole or joint decision - making over food and cash - crop farming, livestock, and fisherie s 1/10 1. 2 Autonomy in Production Autonomy in agricultural production (what inputs to buy, what livestock to raise, etc.) 1/ 10 2. Resources 2.1 Ownership of Assets Sole or joint ownership of major household assets 1/15 2.2 Purchase, Sale, or Transfer of Assets Wh ether respondent participates in decision to buy, sell, or transfer own assets 1/15 2.3 Access to and Decisions on Credit Access to and participation in decision - making concerning credit 1/15 3. Income 3.1 Control Over use of Income Sole or joi nt control o ver use of income and expenditures 1/5 4. Leadership 4.1 Group Membership Whether respondent is an active member in at least one economic or social group (e.g., agricultural marketing, credit, water users' groups) 1/10 4.2 Speaking in Publi c Whether th e respondent is comfortable speaking in public about various issues such as intervening in family disputes, ensuring proper payments of wages for public work programs, etc. 1/10 5. Time 5.1 Workload Allocation of time to productive and domesti c tasks 1/10 5.2 Leisure Satisfaction with the available time for leisure activities 1/10 Source: Alkire et al. (2013 a ) . 54 BIBLIOGRAPHY 55 BIBLIOGRAPHY Alkire, S., R. Meinzen - Dick, A. Peterman, A. Quisumbing, G. Seymour, and A. Vaz. The ent in Agric ulture Index (No. 58). OPHI, Oxford, UK. 2013a. Alkire, S., H. Malapit, R. Meinzen - Dick, A. Peterman, A. Quisumbing, G. Seymour, and A. Vaz. Instructional G uide on the A (WEAI) . OPHI, Oxford, U K. 2013 b. Ba R ole of M aternal S chooling and it s I nteraction with P ublic H ealth P rograms in C hild H ealth P Journal of Development Economics (1990). PIER Working Pap er 97 025 ( 1997 ) Behrman , J. and B. Wolfe Journal of Econometrics 36 ( 1987 ): 185 - 204. Browning, M., F. Bourguignon, P. Chiappori, and V. Lechen nd Outcomes: A Structural Model of Intrahousehol Journal of Political Economy 102, 6 (1994): 1067 1096. CSO ( Central Statistical Office ) . Zambia Labor Force Survey Report 2014 . Lusaka, Zambia. 2015. CSO ( Central Statistical Office ) , MOH ( Mini stry of Health ), and ICF International. Zambia Demographic and Health Survey 2013 - 2014 . Rockville, Maryland, US A. 2015 Econometrica 56, 1 (1988): 63 90. Chiappori, P., M. Iyigun and Y. Weiss vestment in Schooling and th American Economic Review 99, 5 ( 2009 ): 1689 - 1713. Maternal and Child Nutrition 12(2016):12 2 6. Fafchamps, M., and F. and Economic Development and Cultural Change, University of Chicago Press, 63 (2014): 73 - 115. FAO. Guidelines for M easuring H ousehold and I ndividual D ietary D iversity . Rome, Italy. 2011. FAO, IFAD, UNICEF, WFP, and WHO. The State of Foo d Security and Nutrition in the World 2017. Building R esilience for P eace and F ood S ecurity. (FAO, Ed.). Rome, Italy. 2 017. FEEDB ACK, F. T. F. Feed the Future Zambia Zone of Influence Baseline Report . Rockville, MD. 2013. 56 Grebmer, K. von, J. Bernstein, F. Fritschel. 2018 Global Hunger Index: Forced Migration and Hunger . Dublin, Ireland; Bonn, Germany. 2018. R ole of M aternal E ducation in C hild H ealth: Evidence from a C ompulsory S chooling L Economics of Education Review (2015) Haddad, L., J. Hoddinott, and H. Alderman. Intrahousehold Resource Allocat ion: An Overview (No. 1255). Policy Rese arch Working Paper . Washington, DC. 1994. Oxford Bulletin of Economics and Statistics 57, 1 (199 5): 77 96. Lundberg, S. J., R. A. Pollak The Journal of Human Resources 32, 3 (1997): 463 480. Malapit, H. J. L., S. Kadiyala, A. R. Quisumbing Empowerment Mitigates the Negative Effects of Low Production Diversity on Maternal Journal of Development Studies 51, 8 (2015): 1097 1123. D imensions of W E mpowerment in A g riculture M atter for N Food Policy 52 (2015): 54 63. E mpowerment G aps in A griculture and C W ell - B Development Policy Review 37 , 2 (2019): 176 203. Mofya - Mukuka, R., and J. Sambo. Household Dietary Diversity Impact of Women Control over Income from Agriculture in Zambia (No. 136). Lusaka, Zambia. 2018. Right t o Education (RTE) Project . Country Factsheet: Zambia . August 2012. Rosen as Inputs to Child Health: The Econo mic Aspects of Health . University of Chicago Press. 1982. Rubalcava, L., and D. Thomas Working Papers . World Development 61 (2014): 11 52. Smith, L., and J. Haddad. Explaining Child Malnutrition in Developing Countr ies: A Cross - Country Analysis. International Food Policy Research Institute, Washington, D.C. 2000. 57 Parental Resources and Child Jour nal of Human Resources 29, 4 ( 1994 ): 950 - 988. Thomas, D., J. Strauss, and M. Henriques. Journal of Development Economics 33 (1990): 197 234. Udry, C., J. Hoddinott, H. Alderman, and D ifferentials in F arm P roductivity: I mplications for H ousehold E fficiency and A gricultural P Food Policy 20, 5 (1995): 407 423. UNICEF. World Bank Group. W orld Data: Zambia ( https://data.world bank.org/country/zambia ). 2019a. World Bank Group. Zambia Overvie w . ( www.worldbank.org/en/country/zambia/overview) . 2019b . World Bank Group. Poverty & Equity Brief: Zambia . 2019 c . World Health Organization (WHO) Multic entre Growth Reference Study Group. WHO Child Growth Standards: Length/ H eight - for - A ge, W eight - for - A ge, W eight - for - L ength, W eight - for - H eight and B ody M ass I ndex - for - A ge: Methods and Development . W orld H ealth O rganizatio n Press. Geneva, Switzerland. 2006. Wo rld Health Organization (WHO). A ssessing I nfant and Y oung C hild F eeding P World Health Organization Press. Geneva, Switzerland. 2008.