RETHINKING POVERTY IN NIGERIA: THE DEMOGRAPHICS AND HEALTH OF HOUSEHOLDS WITH THREATENED LIVELIHOODS
Esther Omowumi Lamidi
A Dissertation
Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
August 2016
Committee:
Wendy D. Manning, Advisor
John Liederbach Graduate Faculty Representative
Kelly S. Balistreri
Karen B. Guzzo
Kara Joyner
© 2016
Esther O. Lamidi
All Rights Reserved iii ABSTRACT
Wendy D. Manning, Advisor
Many studies have reported staggering levels of food insecurity and highlighted important socioeconomic correlates of food poverty in Nigeria. Yet, a more nuanced assessment is required to determine how food insecurity relates to other aspects of household socioeconomic wellbeing and how traditionally vulnerable groups avoid food insecurity while others with high levels of resources experience food insecurity. Further, food insecurity is often treated as a static indicator but may be episodic. Accordingly, further attention to the correlates and implications of persistent and transitory food insecurity is warranted. The present study draws on relevant theoretical frameworks to advance our knowledge about household food insecurity and wellbeing in Nigeria. Using nationally representative and panel data from the Nigeria General Household
Survey (n= ~4700 households), the study: 1) analyzes the sociodemographic correlates (including wealth index) of persistent and transitory household food insecurity in Nigeria; 2) examines three mechanisms by which the presence of two vulnerable populations (children and older adults with a disability) in the household relate to household food insecurity; and 3) investigates the association between household food insecurity and child malnutrition and how the relationship is modified by community-level characteristics. The findings reveal that household food insecurity presents a very different picture of household socioeconomic wellbeing than many other measures of household socioeconomic status. The results reaffirmed the episodic nature of poverty, including food insecurity. Whereas only half of Nigerian households were food insecure in 2010, the majority had experienced food insecurity by 2013. Persistently food secure households were more economically advantaged than persistently food insecure households. Yet, iv chronically food insecure households were not predominantly uneducated, unemployed, or without support. Inadequate access to financial support does not explain the gap in the experiences of food insecurity between households with vulnerable populations and those without vulnerable populations. But, the longer time spent collecting cooking fuel by households with children partially account for their vulnerability to severe food insecurity. Household food insecurity was significantly related to under-five malnutrition but the association is complex. The findings of this dissertation provide new insights into processes underlying household food insecurity and their implications for child wellbeing. v
This dissertation is dedicated to my father, Pa Ayankoso vi ACKNOWLEDGMENTS
My single most important advice to doctoral students is, find someone who believes in you! I found mine at Bowling Green State University. Wendy Manning is, to many people, a mentor, a teacher, an advisor, and many more, but to me, she is my encourager. When the going got rough, she gave me reasons to persevere. She is seldom thought of as a motivational speaker but I’m convinced she will make a great second career speaking to young people after retiring.
Professor Manning, I am very grateful for your supports. My heartfelt gratitude also goes to my dissertation committee members – Dr. Karen Guzzo, Professor Kara Joyner, and Dr. Kelly
Balistreri - for their unquantifiable contributions to my graduate education. Dr. Karen, you have set a very high standard for me in terms of prioritizing the needs of others and I really hope I can be a great mentor to my students as you have been to me. I pride myself in the fact that I was taught by one of the best in the field of Demography. Professor Joyner, I really appreciate your investments in me. Thank you Dr. Balistreri for motivating me to be the best I can. You were extremely accommodating and I greatly appreciate the opportunity to learn from you. Thinking about the statistical support from Hsueh-Sheng Wu, motivations for healthy living from Ms.
Lesley Wadsworth, professional guidance from Dr. Krista Payne, and the therapeutic laughing sessions with Kasey, Sue, Matthew, Gwen, Vanessa, Lindsey, Lydia, and others, I realized how blessed I was to be part of the amazing NCFMR/CFDR/TARS team. You make the ‘cubicles’ feel like living rooms and I will miss you all. I also thank all faculty and staff members of the
Department of Sociology at BGSU for their supports and encouragements. I appreciate all former teachers, friends, and relations who were there for me during my graduate studies. Words cannot express how grateful I am to my two amazing mothers - Mrs. Lamidi and Mrs. Paula Rhodes – and my beloved friend, Michael. May God preserve you to reap the fruit of your labor. Finally, I give all glory to my source, my anchor, and the lifter of my head - Jesus Christ. vii
TABLE OF CONTENTS
Page
CHAPTER 1: INTRODUCTION………………………………………………………...... 1
Chapter 2: Rethinking Poverty in Nigeria: Food Insecurity and
Household Wealth ………………………………………………………...... 4
Chapter 3: Household Composition and Experiences of Transient and
Persistent Food Insecurity in Nigeria: The Role of Social Capital, Education,
and Time use ………………………………………………………...... 5
Chapter 4: Multilevel Analysis of Household Food Insecurity and Child
Malnutrition in Nigeria ………………………………………………………...... 5
CHAPTER 2: RETHINKING POVERTY IN NIGERIA: FOOD INSECURITY AND
HOUSEHOLD WEALTH ………………………………………………………...... 7
Introduction………………………………………………………...... 7
The Study Context………………………………………………………...... 9
Measuring Household Welfare…………………………………………………...... 11
Household Livelihoods Based on Food Security………………………………...... 15
The Present Study………………………………………………………...... 18
Data and Methods………………………………………………………...... 22
Sample………………………………………………………...... 24
Analyses of the relationship between household wealth and
household food insecurity………………………………………...... 24
Analyses of persistent and transitory household food
insecurity……………………………………………...... 25 viii
Measures………………………………………………………...... 26
Outcome variables……………………………………………...... 26
Household wealth……………………………………………...... 27
Other household sociodemographic variables……………………. .. 28
Analytic strategy………………………………………………………...... 29
Results………………………………………………………...... 30
Findings on transitions into and out of household food insecurity……...... 37
Discussion………………………………………………………...... 40
CHAPTER 3: HOUSEHOLD COMPOSITION AND EXPERIENCES OF
TRANSIENT AND PERSISTENT FOOD INSECURITY IN NIGERIA:
THE ROLE OF SOCIAL CAPITAL, EDUCATION, AND TIME USE………………...... 55
Introduction………………………………………………………...... …… 55
Household Structure and Food Insecurity………………………………………...... 59
The Role of Social Capital in the Experience of Food Insecurity among
Vulnerable Populations …………………………………………………...... 63
Alternative Explanations …………………………………...... 69
Food management skills (education)……………………………………...... 69
Time use………………………………………………………...... 70
Current Investigation………………………………………………………...... 72
Data and Methods………………………………………………………...... 74
Sample………………………………………………………...... 75
Cross-sectional sample…………………………………………...... 75
Longitudinal sample……………………………………………...... 76 ix
Measures………………………………………………………...... 77
Outcome variable……………………………………………...... 78
Focal predictor……………………………………………...... 79
Focal variables……………………………………………...... 79
Social capital……………………………………………...... 79
Educational attainment of household head………………..... 80
Time spent collecting cooking fuel………………………..... 80
Control variables……………………………………………...... 80
Analytic strategy………………………………………………………...... 82
Results………………………………………………………...... 83
Household composition and transitions in and out of food insecurity ...... 91
Transitions into food insecurity by number of children and presence of
older adults with a disability in the household………………………...... 92
Transitions out of household food insecurity by number of children
an d presence of older adults with a disability in the household………… ..... 96
Discussion………………………………………………………...... 101
CHAPTER 4: MULTILEVEL ANALYSIS OF HOUSEHOLD FOOD INSECURITY
AND CHILD MALNUTRITION IN NIGERIA………………………………………...... 120
Introduction………………………………………………………...... 120
Background………………………………………………………...... 123
The paradox of food security and poor child health…………………...... 126
Current Investigation………………………………………………………...... 128
Data and Methods………………………………………………………...... 131 x
Sample………………………………………………………...... 132
Stunting………………………………………...... 133
Wasting………………………………………...... 133
Underweight………………………………………...... 134
Measures………………………………………………………...... 134
Outcome variables……………………………………………...... 134
Focal predictor: Food security…………………………………...... 135
Focal variables: Community characteristics………………………. .. 135
Sociodemographic measures……………………………………...... 136
Analytic strategy………………………………………………………...... 138
Results………………………………………………………...... 139
Discussion………………………………………………………...... 144
CHAPTER 5: CONCLUSIONS………………………………………………………...... 156
Key Findings………………………………………………………...... 157
Limitations………………………………………………………...... 160
Contributions………………………………………………………...... 162
Summary………………………………………………………...... 163
Future Research………………………………………………………...... 165
REFERENCES………………………………………………………...... 168
APPENDIX A………………………………………………………...... 190
APPENDIX B………………………………………………………...... 191 xi
LIST OF TABLES
Table Page
2.1 Food Security and Household Wealth in Nigeria, 2010: Weighted
Descriptive Statistics by Household Ranking Based on Ownership of Assets ...... 45
2.2 Food Security and Household Wealth in Nigeria, 2010: Weighted
Descriptive Statistics by Household Ranking Based on Values of Assets ...... 46
2.3 Household Food Security and Household Wealth Based on Ownership of Assets
in Nigeria in 2010, Multinomial Logistic Regression Relative Risk Ratios...... 47
2.4 Household Food Security and Household Wealth Based on Estimated Values of
Assets in Nigeria in 2010, Multinomial Logistic Regression Relative Risk Ratios .. 49
2.5 Transitions Into and Out of Household Food Insecurity in Nigeria:
Weighted Descriptive Statistics by Initial Food Security Status ...... 51
2.6 Odds Ratios, Logistic Regression Predicting First Transitions into Household
Food Insecurity in Nigeria ...... 53
2.7 Odds Ratios, Logistic Regression Predicting Transitions Out of Household
Food Insecurity in Nigeria ...... 54
3.1 Weighted Descriptive Statistics by Number of Children and Presence of Older
Adults with a Disability in the Household ...... 106
3.2 Experiences of Food Insecurity by Household Composition in Nigeria,
Multinomial Logistic Regression Relative Risk Ratios ...... 108
3.3 Persistence and Transitory Household Food Insecurity among Households
with Children and Older Adults with a Disability in Nigeria ...... 111 xii
3.4 Transitions into Food Insecurity Status among Vulnerable Populations in Nigeria:
Weighted Descriptive Statistics ...... 112
3.5 Odds Ratios, Logistic Regression Predicting Transitions into Household
Food Insecurity in Nigeria by Household Composition ...... 114
3.6 Transitions Out of Food Insecurity Status among Vulnerable Populations in Nigeria:
Weighted Descriptive Statistics ...... 116
3.7 Odds Ratios, Logistic Regression Predicting Transitions Out of Household Food
Insecurity in Nigeria by Household Composition ...... 118
4.1 Descriptive Statistics for Variables in Cross sectional Analyses of Stunting among
Nigerian Children Aged 0-60 Months by Household Food Security Status ...... 150
4.2 Multilevel Logistic Regression Models of Under-five Stunting in Nigeria ...... 151
4.3 Descriptive Statistics for Variables in Cross sectional Analyses of Wasting among
Nigerian Children Aged 0-60 Months by Household Food Security Status ...... 152
4.4 Multilevel Logistic Regression Models of Under-five Wasting in Nigeria ...... 153
4.5 Descriptive Statistics for Variables in Cross sectional Analyses of
Underweight among Nigerian Children Aged 0-60 Months by Household
Food Security Status ...... 154
4.6 Multilevel Logistic Regression Models of Under-five Underweight in Nigeria ...... 155
xiii
LIST OF FIGURES
Figure Page
2.1 Nigeria's Economy and Growth, 2010-2014 ...... 10
3.1 Predicted Probabilities of Severe Household Food Insecurity by Number of
children in the Household and Time Spent Collecting Wood (Fuel)...... 110
3.2 Predicted Probabilities of Severe Household food Insecurity by Number of
Children in the Household and Occupation of Household Head ...... 110
1
CHAPTER 1: INTRODUCTION
Access to adequate food and freedom from hunger is a basic human right that people, everywhere and at all times, should enjoy (Adebayo and Ojo 2012; The World Bank 2006). But malnutrition remains a formidable health problem confronting one out of every eight persons on the planet earth (FAO, IFAD, and WFP 2013). Globally, food insecurity - unreliable and or insufficient quality and quantity of nutritionally adequate food (Hadley 2014; Sirotin et al. 2014)
- poses a serious challenge to the health and overall wellbeing of human population (Alaimo et al. 2001; Nord 2014; Olson 1999; Vozoris and Tarasuk 2003). Simulation studies show that as many as five million to 170 million additional people are at risk of hunger by 2080
(Schmidhuber and Tubiello 2007). Although many countries achieved the Millennium
Development Goal (MDG) of halving the proportion of people living on less than one dollar a day between 1990 and 2015, less than one-fourth achieved the MDG goal of alleviating non- income poverty - hunger (The World Bank 2006).
The majority of the approximately one billion people currently suffering from chronic hunger reside in the developing nations, largely in Africa (Food and Agriculture Organization
2014). Sub-Saharan Africa has the highest incidence and depth of poverty and remains the only region of the world being threatened by widespread persistent food insecurity and famine
(Devereux and Maxwell 2001). Unsurprisingly, between 40 percent and 60 percent of the residents of the most populous country in the region – Nigeria- are food insecure (Ajani et al.
2006; Amaza et al. 2006; Babatunde et al. 2007). As such, malnutrition, especially child undernutrition, is highly prevalent in the country (Ajani et al. 2006; Ajao et al. 2010; Akinyele
2009; FAO, IFAD, and WFP 2013; Adebayo and Ojo 2012; Omonona and Adetokunbo 2007;
Ozughalu and Ogwumike 2015). 2
Nigeria ranks high on national income, relative to other developing countries, and the nation’s GDP per capita has risen substantially since the 1970s oil boom. Yet, the majority of
Nigeria’s population live in abject poverty (The World Bank 2014a). Food accounts for the large share of household expenditure in Nigeria (Omonona and Adetokunbo 2007), and more often than not, poverty is closely linked to experiences of hunger and malnutrition in the country
(Ozughalu and Ogwumike 2015). More than 12 million Nigerians were undernourished in 2011-
2013 (FAO, IFAD, and WFP 2013); about two out of every five Nigerian children under the age of five are stunted; 21 percent are severely stunted; and 18 percent are wasted (National
Population Commission and ICF International 2014). Further, even though Nigeria ranks seventh in population size around the world, more mothers and children die in Nigeria than in any other country, except India (UNICEF 2011), with many of the deaths linked to malnutrition. Given
Nigeria’s rich resource endowment, the level of poverty in the country is concerning.
The high prevalence rates of poverty and food insecurity in Nigeria and other sub-
Saharan African countries has attracted research attention. While the existing literature on household food insecurity and health provides estimates of basic correlates, there remain important gaps in the literature. First, most of the studies of food insecurity and wellbeing in
Nigeria have very limited spatial coverage, thereby offering limited insights into the enormous spatial and sociodemographic heterogeneity across the country. Second, household food insecurity is episodic in nature (RTI International 2014), but previous studies relying almost exclusively on cross-sectional data have been unable to analyze both transient and chronic household food insecurity status, especially in relation to health and wellbeing. Third, even though poverty is multidimensional (Bollen et al. 2002; Carter and May 1999; Howe et al. 2008), and scholars have emphasized the need to understand how the different experiences of poverty 3 compare (Falkingham and Namazie 2002), there are limited studies on the relationship between household food insecurity and other aspects of household socioeconomic conditions. Fourth, multilevel analyses of food insecurity and child malnutrition are sparse. Multilevel modeling techniques are essential to unraveling the correlates of child malnutrition at multiple contextual levels. The few studies of household food insecurity and child malnutrition in Nigeria are mostly descriptive in nature (Ajani et al. 2006; Ijarotimi and Oyeneyin 2005).
Finally, and most importantly, there are two paradoxes emerging from recent research on food insecurity in Nigeria and elsewhere that need to be explained. First, in spite of the strong link between food insecurity and child nutritional status (Ajao et al. 2010; Ali et al. 2013; Baig-
Ansari et al. 2006), there is a high prevalence rate of malnutrition among children residing in food secure households (Atoloye et al. 2015; Ijarotimi and Oyeneyin 2005; Kahsay et al. 2015).
Second, households with vulnerable populations (e.g. children) are notoriously more vulnerable to food insecurity than those with no such populations, but many households with vulnerable populations manage to remain food secure while those with no vulnerable populations experience food insecurity (Balistreri 2012).
The present study contributes to the existing literature on household food insecurity and child health by: 1) examining food insecurity among nationally representative samples of
Nigerian households and children; 2) capitalizing on the richness and the panel nature of the recently collected Nigeria General Household Survey (NGHS) data to analyze both transitory and chronic household food insecurity; 3) investigating the association between household food insecurity and a wide range of socioeconomic and demographic factors at the individual, household, and community levels, including one of the most widely used measure of household socioeconomic condition – wealth index; and 4) drawing on relevant theoretical frameworks (e.g. 4 resource perspectives, nutrition transition theory, the principle of family adaptive strategy, and the concept of social capital) to understand the experiences of household food insecurity, especially in light of the two food security paradoxes highlighted above.
Although the study is organized into three chapters, each tackling distinct research questions, the three overarching issues addressed in this research are: 1) what are the important individual-level, household-level, and community-level socioeconomic and demographic correlates of persistent and transitory household food insecurity in Nigeria? 2) what factors modify the experiences of transient and chronic household food insecurity? and 3) how does household food insecurity relate to child malnutrition (stunting, wasting, and underweight)?
Below, I briefly summarize how each chapter addresses the above questions.
Chapter 2: Rethinking Poverty in Nigeria: Food Insecurity and Household Wealth
This chapter seeks to answer two major questions: 1) To what extent does the widely used measure of household socioeconomic condition in the developing countries – household wealth – capture the experiences of food insecurity? and 2) what are the patterns and correlates of persistent and transitory household food insecurity in Nigeria? The chapter examines the relationships between food insecurity and two measures of household wealth (ownership of assets and actual values of assets). It also presents a profile of transitorily and persistently food insecure households in Nigeria based on the established sociodemographic characteristics associated with poverty and food insecurity in previous research (e.g. Belachew et al. 2012;
Bigsten et al. 2003; Garrett and Ruel 1999; Hanmer et al. 1999; National Population Commission
2014; Sumarto et al. 2007).
5
Chapter 3: Household Composition and Experiences of Transient and Persistent Food
Insecurity in Nigeria: The Role of Social Capital, Education, and Time Use
An emerging theme in food insecurity research is a form of food security paradox whereby many households with vulnerable populations manage to remain food secure while those with no vulnerable populations are food insecure (Balistreri 2012). The paradox suggests a buffering effect in the experiences of food insecurity among vulnerable households. The purpose of this chapter is to analyze three mechanisms by which presence of two vulnerable populations
(children and older adults with disability) in the household relates to household food insecurity.
These are: access to social capital, education, and time use. The chapter examines the mediating and moderating roles of access to social capital, different levels of education of household head, and variations in the amount of time households spend collecting or fetching cooking fuel on the experiences of food insecurity.
Chapter 4: Multilevel Analysis of Household Food Insecurity and Child Malnutrition in
Nigeria
The goal of this chapter is to address the following three research questions. First, I investigate the association between food insecurity and child malnutrition (stunting, wasting, and underweight). I examine the effects of recent, distal, and chronic experiences of food insecurity on child malnutrition. Second, I consider the food security paradox that many children in food secure households experience high levels of malnutrition. Drawing on nutrition transition theory and using multilevel logistic regression modeling technique, I evaluate how social context, measured in terms of community level infrastructural development and change in community socioeconomic development: 1) relate to child malnutrition, and 2) modifies the experiences of stunting, wasting, and underweight among children living in food secure households. Third, I 6 conduct cross-level interactions between food insecurity status at the household level, and community level infrastructural development and change in community socioeconomic development (community level variables) in order to determine the effects of the community variables on stunting, wasting, and underweight among food secure children, relative to their counterparts in food insecure households.
I conclude the dissertation with a chapter (Chapter 5) that summarizes the findings from all the previous three chapters (Chapters 2-4), highlights the implications of the results for policy and research, and discusses next steps in future research on household food insecurity and child health. 7
CHAPTER 2: RETHINKING POVERTY IN NIGERIA: FOOD INSECURITY AND
HOUSEHOLD WEALTH
Introduction
The literature on household welfare is permeated with definitions and measurements of household economic status and poverty (Bollen et al. 2002). Although there is no ubiquitous conceptualization of poverty, the Committee on Economic, Social and Cultural Rights at the
United Nations, in 2001, broadly defined poverty as including any “human condition characterized by sustained or chronic deprivation of the resources, capabilities, choices, security and power necessary for the enjoyment of an adequate standard of living and other civil, cultural, economic, political and social rights” (5). People can be poor in absolute terms, when judged against a minimum standard of living, or poor, when their living conditions are lower, relative to the living standards of another segment of the society (Falkingham and Namazie 2002; Foster
1998). In poor countries like Nigeria, where the majority of the population struggles with basic subsistence, absolute poverty better portrays household living conditions (Falkingham and
Namazie 2002).
Understanding people’s living conditions is essential in every human society because poverty influences nearly all demographic and health outcomes, including life expectancy, maternal and child mortality and morbidity, human capital development, fertility, contraceptive use, and use of healthcare facilities, among others (Adeyemi et al. 2009; Bollen et al. 2002;
Filmer and Pritchett 2001; González et al. 2010; Gwatkin et al. 2007; Houweling et al. 2003;
Montgomery et al. 2000). Poverty estimates are particularly crucial to policies and programs aimed at alleviating poverty in the poor regions of the world (Hanmer et al. 1999). But the determination of who is poor and who is not largely depends on operationalization of poverty in 8 surveys and is very sensitive to measurement error (Falkingham and Namazie 2002; Hanmer et al. 1999). Empirically, the choice of indicators of living standards not only influences the relationship between economic status and health-related outcomes (Lindelow 2006), but it also shapes the associations between many sociodemographic variables and health outcomes (Bollen et al. 2002; Houweling et al. 2003).
In view of the salience of the definition and measurement of poverty, actual measures of household consumption of food and other goods and services are the most preferred indicators of household welfare. However, more often than not, income and expenditure serve as proxies for household consumption patterns (Falkingham and Namazie 2002). In the absence of the aforementioned measures of household welfare – household consumption, income, and expenditure – studies utilizing the Demographic and Health Survey data have assessed household welfare in the developing world based on household ownership of certain wealth indicators.
More so, the dearth of panel data in many developing countries precludes the analysis of transitory and persistence household poverty experiences (Carter and May 1999). But effective monitoring of stability and change in household food conditions is crucial to policies and programs aimed at improving the living standards of the people. Both persistent and transitory food (in)security also have important implications for health and wellbeing (Howard 2011;
Metallinos-Katsaras et al. 2012; Ryu and Bartfeld 2012). This study takes advantage of a recent longitudinal data collection in Nigeria to: 1) investigate the extent to which one of the most widely used measures of household socioeconomic condition – wealth index – captures the experiences of food insecurity in Nigerian households, and 2) examine the patterns and correlates of persistent and transitory food insecurity. Food is a basic need for survival, inadequacy of which suggests serious deprivation or extreme poverty (Hanmer et al. 1999). In Nigeria, food 9 accounts for the bulk of household expenditure (Omonona and Adetokunbo 2007) and the seasonality of income in the country means that food security is, arguably, a better indicator of livelihood than many money-metric measures employed in previous studies (Carter and May
1999). The dominant mode of subsistence in Nigeria is agriculture with considerable seasonal variations in food production based on weather conditions, local, national, and international food market situations, and socio-political factors. Also, non-agricultural household members employed in the professional sector of the economy are often subjected to momentary impoverishment through irregular payments of wages and salary. This research utilizes more recent survey data - the 2010 and 2012 Nigeria General Household Survey (NGHS) - thereby allowing analysis of poverty experiences in more recent socio-historical time in the country.
The Study Context
Nigeria is the most populous African country. With its nearly 179 million people, the nation accounts for one-sixth of Africa’s entire population (United Nations 2014). Nigeria’s rapidly growing population is expected to become the third largest in the world, surpassing the U.S. population, by 2050 (United Nations Press Release 2013). Based on its over 3000 dollars GDP per capita (The World Bank 2014b), Nigeria ranks high on national income, relative to other developing countries. In addition, Nigeria’s GDP per capita has risen substantially since the
1970s oil boom. Figure 2. 1 displays the upward trends in Nigeria’s GDP per capita between
2010 and 2014. 10
Figure 2. 1. Nigeria's Economy and Growth, 2010-2014
GDP per capita (current US$) Agriculture, value added (% of GDP) GDP Agriculture 3,500 25
3,000 24
2,500 23
2,000 22
1,500 21
1,000 20
500 19
0 18 2010 2011 2012 2013 2014 Source: The World Bank World Development Indicators, 2015
However, the majority of the population of the self-acclaimed largest economy in Africa, worth $510 billion which is far more than South Africa’s $352 billion (African Development
Bank Group 2014), live in abject poverty. Using the World Bank’s poverty headcount ratio of one dollar twenty five cents a day, 68 percent of Nigerians lived in poverty in 2010, up from 63 percent in 2004. The vast majority of the population (85%) survived on less than two dollars a day in 2010 (The World Bank 2014a). The seemingly mismatch between Nigeria’s economic growth and changes in living standards of its people suggests unequal distribution of the nation’s growing wealth. Also, the booming economy foreshadows the weakening agricultural sector in
Nigeria. As shown in Figure 2. 1, there has been a continuous decline in the net output of the agricultural sector (value added) over the past four years.
The manifestations of poverty cut across nearly all sectors of Nigeria’s economy. But the health impacts of poor living conditions in the country have received the most attention. At least one out of every four children under the age of five in Nigeria was underweight in 2012 (The
World Bank 2014a). Also, even though Nigeria ranks seventh in population size around the world, more mothers and children die in Nigeria than in any other country, except India 11
(UNICEF 2011). Given Nigeria’s rich resource endowment, the level of poverty in the country is concerning. But a large body of literature has emphasized how unlike many other resource-rich nations like Saudi Arabia, Nigeria’s rich natural resource endowment has been a curse rather than a blessing to the majority of its populace (Hodler 2006).
Poverty is multidimensional (Bollen et al. 2002; Carter and May 1999; Howe et al. 2008) and to improve the health and wellbeing of poor people in Nigeria and elsewhere, it is important to study poverty in its different forms (Bevan and Joireman 1997; Falkingham and Namazie
2002). Evaluating the achievement of the Millennium Development Goal of eradicating extreme poverty and hunger by 2015 requires an assessment of access to food. Food insecurity, particularly in urban parts of the globe, though of lower political priority, influences nearly all aspects of life and other social problems that top government agenda (Maxwell 1999).
Measuring Household Welfare
Due to its clearer policy implications and the complexity involved in other conceptualizations of household welfare, income is the most preferred measure of living standard (Montgomery et al.
2000; Howe et al. 2008). But the variability in sources of income, its seasonal nature in agrarian societies, and the paucity of income data resulting from difficulty involved in collecting accurate information about earnings, make income a less preferred measure in resource-poor countries like Nigeria (Howe et al. 2008; Sahn and Stifel 2003). Also, assessment of income does not provide a clear consideration of the causes of poverty, thereby making it challenging to establish effective interventions (Carter and May 1999; Hanmer et al. 1999). Therefore, there is a growing demand for alternative measures of household welfare and poverty in the poor countries like
Nigeria (Falkingham and Namazie 2002; Sahn and Stifel 2003). 12
Economists have widely employed consumption expenditures, a money-metric alternative measurement of standard of living, in many developing countries (Booysen et al.
2008; Falkingham and Namazie 2002; Sahn and Stifel 2003). Adjusting household consumption expenditure according to household size produces household per capita expenditure on consumption, which is a more preferred measure of household welfare than overall household consumption expenditure (e.g. Chen and Ravallion 2004; 2007; Montgomery et al. 2000;
Ravallion and Chen 1997). However, collection of consumption expenditure data is plagued with reliability issues and the data gathering is cumbersome (Sumarto et al. 2007). More so, the widely available Demographic and Health Survey (DHS) and other social survey data, from many sub-Saharan African countries, lack information on consumption and expenditure patterns
(McKenzie 2005). Thus, researchers have adopted alternative measures of socioeconomic wellbeing in the region, notably the wealth or asset index. While the asset index and indicator of household consumption expenditure are both imperfect proxies for household welfare, they differ in their estimates of household poverty (Lindelow 2006; Sahn and Stifel 2003).
The asset index method provides a relative assessment of living conditions which involves ranking households (e.g. in quintiles) based on their ownership of an array of assets
(Booysen et al. 2008). The asset or wealth index has been used in the study of a range of health outcomes including HIV prevalence, gender issues, fertility, child mortality, contraceptive use, and women’s reproductive health (Afifi 2009; Falkingham and Namazie 2002; Mishra et al.
2007). The adoption of the asset-based measure is premised on the idea that a goal of most poverty alleviation programs is to increase ability of individuals and households to acquire income generating assets (Sahn and Stifel 2003). Yet, assets are not only discrete in nature, but they are also durable (Booysen et al. 2008). This means that the asset index provides an 13 ambiguous indicator of long term household economic status that is less responsive to short-term economic shocks (Boosyen et al. 2008; Falkingham and Namazie 2002; González et al. 2010;
Howe et al. 2008). In times of economic difficulty, individuals may prefer to lessen their food and other household expenditure while preserving their household assets. At the same time, changes in assets may not necessarily mean changes in household welfare (Falkingham and
Namazie 2002). Asset index is, therefore, limited in monitoring people’s experiences of poverty over time.
Many studies do not adjust for variation in household size and composition when constructing the asset index (Howe et al. 2008). Therefore, relatively large and poor households may be ranked high on the wealth quintiles when they report more assets, simply due to number of family members. Clumping ensues when there are inadequate number of assets to distinguish the poor from the very poor households at the lower end of the asset scale (Booysen et al. 2008;
McKenzie 2005), and very low access to public facilities or many durable goods among the poorest households can result in the problem of truncation when using wealth indices (Howe et al. 2008; McKenzie 2005).
Another limitation of the asset index is that it is derived from aggregate country-level data which raises the issue of spatial validity of the measurement, particularly in rural areas
(Booysen et al. 2008; Falkingham and Namazie 2002) and across regions with striking socioeconomic variations (Lamidi 2015). Asset measures sometimes reflect mere geographic proximity to public infrastructure (Booysen et al. 2008; Falkingham and Namazie 2002) and many assets used to estimate asset poverty, such as electricity, pipe borne water, flush toilets, and cement flooring, are more urban than rural in nature (Booysen et al. 2008; Filmer and
Pritchett 2001; Lindelow 2006). It is therefore not surprising that assets predict poverty better in 14 rural areas than in the urban centers (Sumarto et al. 2007). Previous studies have also reported cross-country, regional, and state variations in the relative importance of assets (e.g. Filmer and
Pritchett 2001). For example, household goods like television, radio, and refrigerator distinguish wealth classes in Jamaica, while access to pipe-borne water assesses wealth in Madagascar (Sahn and Stifel 2003).
The choice of asset indicators is limited to those available in surveys, particularly the
Demographic and Health Survey (DHS) (Falkingham and Namazie 2002). The DHS not only has limited number of durable goods with no information about their quantity or quality, but it also fails to include many assets commonly owned by the poor and rural dwellers (Houweling et al.
2003). This may be because the assets in DHS were not originally intended to measure household socioeconomic status (Falkingham and Namazie 2002; Howe et al. 2008). Although the World Bank developed a measure of relative economic status based on housing characteristics, durable consumer goods, water and sanitary facilities, among other amenities
(Gwatkin et al. 2007), different studies construct wealth indices using different sets of wealth indicators despite variations in household ranking based on the choice of assets (Houweling et al.
2003; Montgomery et al. 2000) and how the scale is constructed (Booysen et al. 2008).
Moreover, the wealth index is constructed from several variables, many of which may confound the associations between sociodemographic characteristics and health outcomes (Howe et al. 2008; Montgomery et al. 2000). The components of the wealth index affect and are affected, by different socioeconomic and health factors (Houweling et al. 2003). There is, therefore, no theoretical underpinning as to what specific aspect(s) of household socioeconomic status the wealth index assesses (Filmer and Pritchett 2001; Howe et al. 2008). In fact, the asset index is not based on actual experiences of poverty. Rather, it assumes close relation between 15 material possessions and other aspects of wellbeing (Hanmer et al. 1999). However, socioeconomic wellbeing is multidimensional (Falkingham and Namazie 2002) and highly complex (Bevan and Joireman 1997; Howe et al. 2008). It requires approaches beyond material consumption or resources (Falkingham and Namazie 2002). The relative strength of each dimension of household welfare and the interactions of the different measures of socioeconomic status are important to intervention programs and policies (González et al. 2010).
Although studies relying on the Demographic and Health Survey data have depended on the binary indicators of ownership of the different household assets, the numerical values and/or the quality of the assets may be equally important. Therefore, in addition to the conventional asset index, the present study examines the role of quantity and quality of household assets, measured as values of assets, in understanding household socioeconomic status, particularly in relation to the experiences of food insecurity. Food is a basic need for survival, inadequacy of which suggests extreme or hardcore poverty (Hanmer et al. 1999). Questions about food security are relatively simple to ask, and food insufficiency has been found to be a powerful correlate of money-metric poverty (Falkingham, 2000; McKenzie 2005).
Household Livelihoods Based on Food Security
A society is food insecure when some or all of its people, at some point in time, do not have access (physical, social, and economic) to sufficient, safe, and nutritious food that they need or that they prefer to have in order to lead active and healthy lives (Ivers and Cullen 2011). Food insecurity at the household level involves unreliable and or insufficient quality and quantity of nutritionally adequate food (Hadley 2014; Sirotin et al. 2014). The theoretical understanding of the problem of food insecurity has evolved over time, from simply a failure of agricultural production, to “a failure of livelihoods to guarantee access to sufficient food at the household 16 level” (Devereux and Maxwell 2001: 1). But except in rare instances of a sudden surge in food prices or problems with food supply, household food insecurity is highly invisible to the public gaze; it is a problem often relegated to the household level (Maxwell 1999). Nevertheless, household food security, being an aspect of poverty (Hoddinott and Yohannes 2002; Kalichman et al. 2012), is associated with maternal and child wellbeing, human capital development, contraceptive use, risky sexual behavior, risks of contracting STIs including HIV, suicidality, chronic diseases, among other outcomes (Cook et al. 2006; Dewing et al. 2013; Diamond-Smith et al. 2015; Ivers and Cullen 2011; Kalichman et al. 2012; Oyefara 2007; Ramsey et al. 2011;
Sirotin et al. 2014; Tsai and Weiser 2014; Vozoris and Tarasuk 2003). Globally, simulation studies show that as many as five million to 170 million additional people will be at risk of hunger by 2080 (Schmidhuber and Tubiello 2007)
Sub-Saharan Africa has the highest incidence and depth of poverty and remains the only region of the world being threatened by widespread persistent food insecurity and famine
(Devereux and Maxwell 2001). It replaced East Asia as the region with the highest incidence of extreme poverty in the world between 1981 and 2001 (Chen and Ravallion 2004). The high fertility countries in the region are particularly prone to the problem of food insufficiency because rapid population growth is closely related to the problem of household livelihood; it elevates levels of food demands, increases competition for land and water resources, as well as poses environmental challenges (Godfray et al. 2010). Unsurprisingly, over 40 percent of the residents of the most populous country in Sub-Saharan Africa – Nigeria- are food insecure
(Amaza et al. 2006). Some estimates are larger than 60 percent (Ajani et al. 2006; Babatunde et al. 2007). 17
Although problems with household livelihoods are more pronounced in the developing world, there are limited studies of household welfare, based on food access, in resource-poor settings (Joshi et al. 2010; Nord 2014). Also, due to the reciprocal relationship between
HIV/AIDS and food insecurity (Anema et al. 2009), the existing analyses of food insufficiency in developing countries have been predominantly conducted in relation to HIV/AIDS. However, in South Africa, and perhaps elsewhere, HIV, unlike other diseases, is more prevalent in the wealthier segment of the population (Mishra et al. 2007).
In Nigeria, the few studies of food insecurity are based on limited geographic coverage – mostly one city. Oyefara (2007) studied food insecurity among the female commercial sex workers in Lagos metropolis; Amaza et al. (2006) examined food insecurity in Borno state;
Omonona and Adetokunbo (2007) in an urban part of Lagos; Babatunde et al. (2007) sampled 94 farming households in Kwara state; Ayantoye et al. (2011) presented estimates of transitions in and out of food insecurity in rural households in Ondo and Ekiti states; and Ajani and his colleagues (2006) analyzed food insecurity among households headed by primary and secondary school teachers in Lagos and Ibadan cities. Building on the existing understanding of poverty in
Nigeria, the present study utilizes a recent, nationally representative, and panel survey data to examine the experiences of household food insecurity in contemporary Nigeria. Given the large share of Nigeria’s population living in abject poverty, food insufficiency, an absolute measure of poverty, is an appropriate indicator of household welfare for the context of this study.
Food insecurity is less susceptible to many of the issues arising from studies of household socioeconomic status using other measures, including validity and reliability issues, problems with underreporting, and difficulty deriving monetary or use values of consumption. Like consumption and expenditure measures, food insecurity estimates are derived from retrospective 18 recollection of food consumption in the past seven days. However, unlike consumption expenditure, eating patterns are less sporadic and experiences of hunger should loom longer in people’s minds, especially in households that are affected over a relatively long period of time.
Compared to other aspects of socioeconomic wellbeing, the meaning and impacts of food insufficiency may be less restrictive by time and by geographic boundaries. Nonetheless, I present food insecurity not as a superior measure of poverty, but one that allows direct examination of a crucial manifestation of hardship.
The Present Study
The aims of this chapter are twofold: First, it investigates the extent to which the widely used measure of household socioeconomic condition in the developing countries – wealth index – captures the experiences of food insecurity in Nigerian households. Second, it examines the patterns and the correlates of persistent and transitory household food insecurity in Nigeria. A unique feature of this study is the ability to determine the extent to which wealth index correlates with food insecurity. That is, to test the adequacy of the wealth index in identifying households experiencing hardcore poverty. The NGHS has an added advantage in this regard because respondents reported not just possession of each of the asset, but also the number and the estimated value of each asset present in their households. Therefore, I am able to test the relevance of asset quantities and qualities to measures of socioeconomic status.
It is plausible that accumulation of consumer assets is a better measure of relative poverty in affluent, rather than poor, social contexts (Zeller et al. 2001). Previous studies have shown very weak (Montgomery et al. 2000; Sahn and Stifel 2003) to modest (Howe et al. 2008) agreement between wealth index and consumption expenditure. The correlation coefficient between asset indices and household expenditure mostly ranges from 0.20 to 0.42 (The World 19
Bank 2008; Booysen et al. 2008; Falkingham and Namazie 2002; Lindelow 2006), with high correlation of 0.71 reported only in Peru and South Africa (Sahn and Stifel 2003). It is possible that the wealth index is more highly correlated with consumption expenditures than extreme poverty. Some studies have shown negative relationship between household wealth and food insecurity (see Knueppel et al. 2010; Szabo et al. 2015), but these studies rely on a single measure of household wealth – ownership of assets. In view of the negative relationship between food insecurity and household socioeconomic status, I expect household wealth based on both ownership and values of assets to be negatively related to household food insecurity. Also, considering the weak association between the wealth index and consumption expenditures, I hypothesize that the wealth index based on values of assets will more strongly predict household food insecurity than the wealth index based on ownership of assets.
Whereas previous studies relied on static measures of poverty (e.g. Ajani et al. 2006;
Amaza et al. 2006; Oyefara 2007; Omonona and Adetokunbo 2007), the panel nature of the
NGHS makes it possible to monitor improvements (or deteriorations) in household food conditions over time and at the national and regional levels. The episodic nature of poverty is well established in the literature (see Bigsten and Shimeles 2008; Bigsten et al. 2003; Stevens
1994). Increasing evidence also point to the transient, as opposed to persistent, nature of household food insecurity. The occasional nature of food insecurity is evident in the work of
Coleman-Jensen et al. (2015) which shows that as many as one-fourth of food insecure households in the U.S. experienced food insecurity in only one or two months in 2014. On average, households were food insecure for about 7 months in a year (Coleman-Jensen et al.
2015). This means that at any given point, a relatively small segment of households experience food insecurity but a larger share of households had been food insecure at some point in their 20 lives. The share of food insecure households in the Early Childhood Longitudinal Study-
Kindergarten (ECLS-K) at any given point ranges from 6% to 11% depending on the measure of food insecurity (6-7% in Bhargava et al. 2008; 9–11% in Howard 2011; 7-10% in Ryu et al.
2012). However, only 79-81% of households (or fewer if estimated based on Bhargava et al.
2008) were persistently food secure.
In spite of the transient nature of food insecurity, few studies have analyzed transitions in and out of food insecurity in Africa using longitudinal data (Aliber 2009; Ayantoye et al. 2011;
Belachew et al. 2012; Cole and Tembo 2011 are a few exceptions). Evidence from limited longitudinal studies of household food insecurity suggests that food insecurity in the developing contexts may be more transient than persistent. Belachew et al. (2012) showed that nearly all
Ethiopian households in their study experienced transient food insecurity within two years.
Similarly, in their study of food insecurity among rural households in Ondo and Ekiti states in
Nigeria, Ayantoye et al. (2011) found that about 72% of food secure households transitioned into food insecurity while 13% of food insecure households became food secure within six months.
Given the transient nature of food security and considering the recent economic and demographic trends in Nigeria, I expect significant changes in household food insecurity over time.
I examine the relationships between food insecurity and the established sociodemographic characteristics associated with poverty and food insecurity in previous research (e.g. Ayantoye et al. 2011; Belachew et al. 2012; Bigsten et al. 2003; Garrett and Ruel
1999; Hadley et al. 2008; Hanmer et al. 1999; National Population Commission 2014; Sumarto et al. 2007). These sociodemographic indicators include the following: education of household head, employment status of household head, household size, household dependency ratio (the 21 ratio of the number of dependents aged 0-14 and 65 and above to the number of working-age adults aged 15-64 in a household), age of household head, gender of household head, marital status and religion of household head, urban residence, and region of residence. Prior studies have not established a full sociodemographic profile of food insecurity in Nigeria.
Based on prior studies (Ajani et al. 2006; Amaza et al. 2006; Belachew et al. 2011; 2012;
Hanmer et al. 1999; Sumarto et al. 2007), I expect that households headed by less educated persons will be more likely to be food insecure than those headed by highly educated individuals.
But the living standard-enhancing power of education is subject to availability of employment opportunities for maximal productivity among educated individuals (Hanmer et al. 1999). Given the high rate of unemployment in Nigeria and continued population growth that further depresses wages (Hanmer et al. 1999; Omonona and Adetokunbo 2007), it is important to also consider the role of employment of household head in the experiences of household food insecurity. Labor force participation by the household head is associated with lower risks of poverty, particularly in the rural areas (Sumarto et al. 2007). More so, the type occupation that the household head engages in is a strong predictor of per capita expenditure (Bigsten et al. 2003). Therefore, I examine not just the effect of being economically engaged versus being out of the labor market, but also differentiate households headed by unemployed adults from those headed by individuals engaged in agriculture, sales and services, professional jobs, and other activities.
I include a series of demographic indicators of the households in the analyses.
Household size is related to poverty with larger households experiencing higher levels of poverty
(Ajani et al. 2006; Amaza et al. 2006; Bigsten et al. 2003). The dependency ratio is important as the number of producers versus consumers is related to poverty (Belachew et al. 2012; Bigsten et al. 2003). Age of the household head is negatively related to poverty (Bigsten et al. 2003; 22
Hanmer et al. 1999; Sumarto et al. 2007; Omonona and Adetokunbo 2007). Gender of the household head is critical indicator of poverty as households headed by women are more likely to be poor than those headed by men, though this effect varies across countries (Amaza et al.
2006; Belachew et al. 2011; 2012; Hanmer et al. 1999; Joshi et al. 2010). Religion permeates nearly every aspect of life in Nigeria (Orubuloye et al. 1993) and Christians may be more likely to experience food insecurity than Muslims (Hadley et al. 2008). Family/household composition is a strong predictor of household economic wellbeing (Coleman-Jensen et al. 2015; Manning and Brown 2006) and polygyny is related to household resource allocation (Tertilt 2005).
Scholars have highlighted the importance of incorporating urban–rural distinctions into poverty estimates (Booysen et al. 2008; Chen and Ravallion 2007; Montgomery et al. 2000;
Sumarto et al. 2007). Livelihoods and lifestyles vary across urban and rural households (Garrett and Ruel 1999). Urban poverty is easily linked to food insecurity (Maxwell 1999) because food prices are higher in urban than in rural areas and as such, city residents spend more on food, consume less, and purchase most of their food (Garrett and Ruel 1999). Nonetheless, poverty is mostly concentrated in the rural areas (Booysen et al. 2008; González et al. 2010; Hanmer et al.
1999). Lastly, there is a huge spatial variation in poverty levels across the geopolitical regions in
Nigeria, with the northern regions being generally poorer than the southern zones based on asset indices (National Population Commission 2014; Ogwumike 2001). The poverty rate in the south- west for instance, is more than thrice the rate in the north-east region (The World Bank 2014a).
Data and Methods
I utilized panel data from the Nigeria General Household Survey (NGHS). The NGHS is a nationally representative annual survey of 22,000 households conducted as part of the Living
Standards Measurement Study-Integrated Surveys on Agriculture (LSMS-ISA). In 2010, the 23 survey was expanded to include a panel component that sampled 5,000 households out of the
22,000 core sample of the NGHS. Unlike its cross-sectional counterpart, the panel survey is biennial by design but the panel households are interviewed twice per wave of data collection; the two rounds of interview correspond to the post-planting and the post-harvest periods in each year. The NGHS has a response rate of greater than 99%. Although the NGHS primarily aims to collect household-level agricultural-related statistics, the survey also collects extensive information about household welfare and social behavior which could aid in the analysis of household socio-demographic characteristics in relation to health and wellbeing.
The panel NGHS is the first panel survey implemented by the Nigeria National Bureau of
Statistics and one of the first few panel surveys in Nigeria. The LSMS-ISA team in the World
Bank’s Development Research Group provides technical guidance in the design and implementation of the NGHS survey as well as assist with the analysis of the data. The survey was supported by various organizations including the Nigeria Federal Ministry of Agriculture and Rural Development, the National Food Reserve Agency, the Bill and Melinda Gates
Foundation, and the World Bank. There are a total of four rounds of interview conducted annually between 2010 and 2013. Two rounds of interview (2010 and 2012) correspond to the post-planting season while data were collected in two post-harvest seasons (2011 and 2013). The post-planting surveys were conducted between August and October while the post-harvest interviews were carried out between February and April. My first set of analyses rely on the
2010 data (first round) and the second set use all four rounds of data. The design, implementation, and coverage of the NGHS have been detailed elsewhere (National Bureau of
Statistics 20151).
1 The data are also available for download through the World Bank’s Living Standard Measurement Study website (http://go.worldbank.org/BY4SLL0380). 24
Unlike the Nigeria Demographic and Health Survey data which have been previously utilized in the analyses of household socioeconomic and nutritional status, the NGHS not only collects information about the quantity and quality of household durable assets, but it also recorded the occurrences of household food insecurity within the week preceding the survey.
Thus, in addition to being one of the first few panel data in Nigeria, the NGHS presents a unique opportunity to analyze the relationship between two measures of household socioeconomic status
– household food insecurity and household wealth. More so, the NGHS allows researchers to account for the effects of a wide range of household sociodemographic characteristics such as education, employment, household size, age, gender, marital status, religion, place of residence, among others, in the analyses of household socioeconomic wellbeing.
Sample
Analyses of the relationship between household wealth and household food insecurity
Although the NGHS originally sampled 5,000 households, 4,997 households completed part or all of the questionnaire at wave one. At the initial round (2010), only 141 households had missing information on three or fewer of the nine items used to measure household food insecurity. The majority (75%) of these households lacked data on the measures of less severe household food insecurity and only 26 of them failed to supply information on more than one occurrence of household food insecurity. Therefore, I retained them in my analysis. I replaced the missing data on each household food insecurity question with the mean score on the item by the rest of the sample. An additional 222 households with missing information on four or more food insecurity items were excluded from my analysis. Twenty three household heads did not report their educational attainment at wave one but they did provide the information at wave 2. I substituted their missing education at wave1 for their reported educational attainment at wave 2. 25
A comparison of the educational status of household heads at waves one and two suggests some reporting errors. In these cases I used the lower level of reported education. I excluded four heads of households who did not report their educational attainment at both waves from the sample. Lastly, I dropped from the sample nine households with missing information on the age of household head and another 41 with no estimated values of household assets. Thus, there were
4,721 households in my analyses of the association between household wealth and food insecurity.
Analyses of persistent and transitory household food insecurity
I conducted two sets of analyses of persistent and transitory household food insecurity. I analyzed first transitions into food insecurity among households that were initially food secure and first transitions out of food insecurity among previously food insecure households. The analytic sample for transitions into food insecurity is based on 3,647 households that reported food security status at two consecutive rounds of interview and were food secure at one or more round. I excluded one household with missing information on age of household head, two households with unknown education of household head, and 26 households with incomplete information about the values of their household assets leaving a total sample size of 3,618 households. The second set of analyses examine transitions out of food insecurity among 3,423 households that were observed at two or more rounds, had food security status at two consecutive rounds, and were food insecure at least once. I dropped from the analyses a total of
32 households with missing information on age of household head (1), education of household head (2), and values of household assets (29). Therefore, I analyzed transitions out of household food insecurity among 3,391 households.
26
Measures
All the variables in my analyses of the association between household wealth and food insecurity were assessed at round one (post-planting interview). For my analyses of transitions into and out of food insecurity, I used lagged predictors of household food insecurity, measured at the round preceding the time of the expected transitions (when food secure for transitions into and when food insecure for transitions out).
Outcome variables
The main dependent variable in this study is household food insecurity. I utilized all the nine items in the refined Household Food Insecurity Access Scale (HFIAS) which assesses the access component of household food security (Coates et al. 2007). Two of the items were slightly different from those in the HFIAS questionnaire. The first of these two items asked about a coping strategy used by households with children (restriction of adults’ meals to accommodate children’s nutritional demands) and the other one inquired about help-seeking behavior among households experiencing food insecurity (reliance on friends and relatives for food). Also, rather than asking about the frequency-of-occurrence of food insecurity separately from incidence, as in the HFIAS questionnaire, the NGHS combines both incidence and frequency of household food insecurity in a series of questions about the number of days during which households recorded certain occurrences of food insecurity. Nevertheless, the food insecurity questions in the NGHS were very similar to the ones in the HFIAS questionnaire, especially when used to categorize households along a continuum of severity of food insecurity.
I examine household food insecurity in both cross-sectional and longitudinal analyses.
First, I analyze the relationship between household wealth and food insecurity in 2010. I compare the experiences of food insecurity among three major categories of households. These 27 are: food secure households, moderately food insecure households, and severely food insecure households. The details of the reports of food insecurity among households in each of the three categories are presented in Appendix A. My classification of Nigerian households into the above three categories closely mirrors the official and well tested classification adopted by the Food and Nutrition Technical Assistance III Project (FANTA) team in the US (see Coates et al. 2007).
Second, the longitudinal measure established stability and change in the experiences of household food insecurity over the study period (2010-2013). In this part of my analyses, I examine the patterns of both chronic and transient household food insecurity as well as factors predicting transitions into and out of food insecurity in Nigeria. Along this line, I combined the two categories of food insecurity (moderate and severe) to create a binary indicator of food insecurity that is coded 1 if a household was food insecure and 0 if otherwise. Households that were food secure at some point between 2010 and 2012 were observed to see if they transitioned into food insecurity at the following round of data collection. Similarly, food insecure households within the study period were examined to see whether they were persistently food insecure or whether they transitioned out of food insecurity.
Household wealth
In the NGHS, the household head provided information about ownership of a series of assets such as radio, television set, generating set, fridge, etc., by his/her household. Similar to previous studies (e.g. Bollen et al. 2002; Booysen et al. 2008; Filmer and Pritchett 2001; Howe et al. 2008), I constructed the wealth index from the dichotomized indicators of ownership of assets. Each component of the wealth index was assigned weights generated from the first component of the principal components analysis (PCA). The PCA is a data-reduction statistical procedure that allows large numbers of correlated variables to be represented by few 28 uncorrelated components. The procedures for using the PCA to construct asset indices have been widely documented in previous analyses (e.g. Bollen et al. 2002; Booysen et al. 2008; Howe et al. 2008). The asset indicators used in this study and their weights are presented in Appendix B.
The NGHS collected additional information about the number and the estimated value of each asset reported in the household. Therefore, I was able to create an alternative measure of household wealth – the total values of assets in each household measured in naira (Nigerian currency). I constructed two wealth indices, one that is based on ownership of assets and another that is based on actual values of household assets. Both categorize Nigerian households into wealth quintiles. I expect that values of assets which account for the ownership, quantity, and quality of household asset will be more strongly correlated with household food insecurity and other household socioeconomic indicators than the wealth index based on ownership of assets.
Thus, whereas my analyses of food insecurity and household wealth were based on both ownership and values of assets in quintiles, I used the log of wealth values in continuous naira in my subsequent analyses.
Other household sociodemographic variables
Education is based on a categorical measure that is coded as a series of dummy variables: no education or forms of education other than formal education, primary education (reference category), secondary education, and higher levels of education. Employment status of the household head is based on reports of employment activities within seven days preceding the survey. The questions asked whether or not the head of households: 1) worked for someone who was not a member of their households, 2) worked on a farm owned or rented by a member of their households, and 3) worked on their own account or in a business enterprise belonging to them or someone in their households. Household heads who reported engaging in any of the 29 three work categories were considered employed and were compared to their unemployed counterparts in their experiences of household food insecurity. In a follow up question, respondents were asked to report the sector of their primary occupations. I combined the reports of employment status and the specific occupations to create the following employment categories: unemployed, agriculture (reference category), sales and services, professional jobs, and others.
Household size is a measure of the total number of individuals residing in each household. The respondents, including household heads, reported their ages in years. Household dependency ratio is the ratio of the total number of dependents (children under the age of 15 and adults sixty five years and above) to the total number of working age adults (age 15-64) in each household. Gender of the household head is coded as one for males and zero females. I identified a household as either married polygynous, married monogamous (reference category), or unmarried, based on the marital status of the household head. A household could also be a
Christian household, Muslim household (reference) or “Other household” depending on the religious affiliation of its head. Urban status is based on location of a household in an urban enumeration area (EA) as defined by the Nigeria Census. I coded one for urban zero for rural.
There are six geopolitical regions in the analyses and they are coded as a series of dummy variables: North central, North east, North west, South south, South east and South west, with
South-west as the reference category. The survey round refers to the time of interview.
Analytic strategy
First, I described the food security status and the sociodemographic characteristics of all
Nigerian households and presented results according to wealth quintile in 2010, using two measures of household wealth – ownership of assets and values of assets. Next, I predicted the 30 risks of being moderately or severely food insecure, relative to being food secure by household wealth (based on ownership of assets and actual values of assets) in three multinomial logistic regression models. The first model includes just the indicators of wealth. The next model includes other measures of household socioeconomic status (education and employment of household head). The final model controlled for the full roster of sociodemographic characteristics.
My second research question focuses on stability and change in household food insecurity. At the bivariate level, I examined the share of households that were persistently food secure, persistently food insecure, transitioned into food insecurity, transitioned out of food insecurity and those that experienced both kinds of transitions. I also described the socioeconomic and demographic features of households that were at risks of transitioning – originally food secure and food insecure households. In the multivariate analyses, I used the log of household wealth (to avoid a skewed distribution) based on actual values of assets, education of household head, employment of household head, household size, age of household head, household dependency ratio, gender of household head, marital status and religion of household head, urban residence, and region of residence to predict the risks of transitioning into and out of food insecurity. I estimated zero-order models, a model with household socioeconomic characteristics (Model 1) and another model with all the sociodemographic variables.
Results
The first goal of this study is to analyze the relationship between household food insecurity (a measure of economic deprivation) and a widely-used measure of household socioeconomic status - wealth index. I examined the patterns of food insecurity among households in five wealth quintiles derived from: 1) principal component analysis of reports of ownership of assets (have 31 versus have not), and 2) the actual values of the assets measured in naira. Table 2. 1 presents the distribution of all Nigerian households and households in the five wealth quintiles (based on ownership of assets) on food insecurity and other sociodemographic characteristics. As shown on the table, about half of all households in Nigeria reported a level of food insecurity in 2010.
More than half (53%) of those households experiencing food insecurity were severely food insecure. Using ownership of assets, more households in the middle (57%), richer (55%), and the richest (52%) wealth quintiles reported food insecurity than in the poor quintiles (41-46%). At least one-quarter of the richest households in Nigeria was severely food insecure in 2010.
More than one-third of Nigerian households were headed by individuals with no education in 2010. Only a minority of household heads (15%) had post-secondary education.
There seems to be an inverse, yet imperfect, relationship between education and wealth. Whereas the ‘poorest’ households were mostly headed by individuals with no formal education, the majority (74%) of the ‘richest’ households were headed by persons with secondary or higher education. One in ten households were headed by unemployed adults. Agriculture was the primary mode of subsistence in Nigeria; 44% of all household heads were farmers. Based on ownership of assets, poor Nigerian households were largely agricultural households and heads of the richest households mostly held sales and services and professional jobs. There was, however, not much variation in the share of households in each wealth quintile that was headed by unemployed persons. Compared to 12% of rich households, 7% of poor households had unemployed heads. Given its relatively high fertility rate, Nigerian households were unsurprisingly large.
Compared to only 3.5 people in her neighboring Ghanaian households (2014 DHS report), on average, there were more than five people in each household in Nigeria. Household 32 size varied little across wealth quintiles. The households were headed by persons averaging 50 years of age. Heads of the ‘wealthy’ households appeared to be relatively younger (about 48 years) than those of middle and poorest households who averaged 52 years of age. For every 100 working-age adults in Nigeria in 2010, there were 96 dependents. However, the proportion of dependents per working-age population decreases across the wealth quintiles from lowest to the highest. Five percent of Nigerian households had no working-age adults (missing dependency ratio) and there were more of such households in the poorest and the middle quintiles than in other wealth quintiles. Nigerian households were predominantly patriarchal; only 16% were headed by women. Based on ownership of assets, the middle quintile had the largest share of female-headed households and the richest quintile had the lowest. The vast majority of the households were headed by married individuals, mostly in monogamous relationships. Fewer wealthy households, than the middle and poor households, were polygynous. Nigerian households were fairly evenly divided between Christianity (55%) and Islam (43%) – the two dominant religions in the country. Judging by ownership of assets, Muslims were more represented among heads of poor households while the heads of the rich households were predominantly Christians. More rural than urban households were included in the sample. Nearly all of the households categorized as poor based on wealth ownership were based in the rural areas. The NGHS sampled similar shares of households from all the six geopolitical zones in the country, with slightly greater shares from the North west and the South west. More than half of all ‘poor’ households in Nigeria were concentrated in two geopolitical regions – the North east and the North west.
Given the counterintuitive findings of higher levels food insecurity in ‘wealthy’ households based on ownership of assets, I compared the ranking of Nigerian households based 33 on actual values of assets to the previous ranking that was based on ownership of assets. Table 2.
2 describes the food security status and other sociodemographic characteristics of the same households presented in Table 2. 1 but based on values of assets. Unlike the previous classification (based on ownership of asset) of more wealthy households as being food insecure than poor households, I found the expected relationship between food security and household wealth using values of assets. The richest had the smallest share of food insecure households and the poorest had the largest share. Nonetheless, the relationship between food security and wealth was far from the ideal – most rich households being food secure and most poor households being food insecure. More than one-fifth of rich Nigerian households were severely food insecure and nearly half of poor households in the country were food secure in 2010.
The educational gap between the rich and the poor was smaller when households were ranked based on values of assets than when using mere ownership of assets. This suggests that even though households headed by more educated persons reported more assets (numerically) or assets that ranked more highly than the poor households, those assets were not as highly valued as they ranked. The occupational distribution of household heads in each wealth quintile also differed by measure of wealth. When the values of their assets were taken into consideration, households headed by agriculturists were less likely to be classified as being poor than when ranked based on mere ownership of assets. In contrast, households headed by persons employed in sales and services appeared poorer when the values of their assets were considered than when ranked based on ownership of assets. Unlike wealth ownership, household size increased with increasing values of wealth. Heads of the poorest households (based on values of assets) were older than heads of households in the higher wealth quintiles. Dependency ratio appeared to be related to ownership of assets but not wealth values. Female-headed households appeared to be 34 randomly distributed across wealth quintiles based on ownership of assets. But the higher the values of wealth in a household, the lower the likelihood that it was headed by a female.
The share of polygynous households in each wealth quintile changed, from being higher at lower wealth quintiles based on ownership of assets, to being higher at higher wealth quintiles based on values of assets. Households headed by unmarried adults also appear poorer based on values of their possessions than ownership of assets. Except for the slightly smaller share of
Muslims among the richest households, religion seems unrelated to wealth measured in terms of values of assets. The socioeconomic status of rural households was more diverse when assessed based on values of assets than ownership of assets; 70% of the poorest households were based in the rural areas using the former measure compared to 91% using the latter measure. Relative poverty appeared to be exaggerated in the northern regions, particularly the North east and the
North west, when measured using ownership of assets. Conversely, in relation to the northern regions, households in the southern regions were not as wealthy as they appeared when I evaluated their asset values.
In a series of multinomial logistic regression, I examined the relationship between the two measures of household wealth and food security. Table 2. 3 presents the results of multinomial logistic regression models predicting the risks of experiencing moderate and severe food insecurity, relative to being food secure, among households in the five wealth quintiles based on ownership of assets in Nigeria. The results in Model 1 showed significantly higher risks of moderate food insecurity at higher wealth quintiles than among the ‘poorest’ households that persisted even after controlling for other measures of socioeconomic status (education and employment) in Model 2. Similarly, severe food insecurity was significantly more prevalent among the ‘middle’, the ‘richer’, and the ‘richest’ households than among the ‘poorest’ 35 households. Even after accounting for other covariates of food security and household wealth, the risks of moderate food insecurity remained significantly higher among the ‘middle’ households than among the ‘poorest’ households (Model 3). Although the ‘richer’ and the
‘richest’ households had significantly lower risks of being severely food insecure after controlling for other predictors, the risks of severe food insecurity were comparable among the
‘poorest’, the ‘poorer’, and the ‘middle’ households. The above findings partially supported my hypothesis of a negative but weak association between household wealth and food insecurity.
Other significant predictors of food insecurity in Model 3 are: education and occupation of household head, household size, polygyny, urban residence, and region of residence.
Households headed by persons with no formal education and postsecondary education had lower risks of food insecurity than whose heads had only primary education. Compared to those whose heads practiced agriculture, households headed by professionals were 24% less likely to be severely food insecure. However, moderate food insecurity appeared to be significantly lower in households headed by unemployed individuals than those headed by agriculturists. Every additional member to the households increased their chances of severe food insecurity by 8%.
Polygynous households were significantly less likely to be severely food insecure than their monogamous counterparts. Households practicing “other” forms of religion (a minority of households in Nigeria) were significantly more food insecure than Muslim households. The problem of household food insecurity was significantly more common in the urban areas than in the rural areas; urban households were nearly twice as likely to be severely food insecure as rural ones. Compared to those in the southern part of the country, households based in northern
Nigeria were more protected against the risks of food insecurity. The three geopolitical regions in the north were less likely to report moderate and severe food insecurity than the South west. 36
Compared to the other two regions in the south – the South east and the South south – the southwestern region had significantly lower risks of household food insecurity.
In Table 2. 4, I presented the results of my multinomial logistic regression analyses of the relationship between household wealth based on values of assets and food insecurity. As shown on the table, the results were notably different from the results presented in Table 2. 3 (analyses based on ownership of wealth). In Model 1, there was no significant association between household wealth and moderate food insecurity but once I accounted for differences in education and occupation of household heads, the richest households became significantly less likely to be moderately food insecure than the poorest households. There was a consistent negative association between value of assets and food insecurity. The risks of severe food insecurity significantly decreased with increasing household wealth both before and after controlling for other predictors of household socioeconomic wellbeing.
The major difference in the experiences of household food security by education was between households whose heads had no formal education and those whose heads had primary education. Households headed by highly educated persons had lower chances of both moderate and severe food insecurity but the effects were nonsignificant net of other covariates. The older the head of a household, the higher the risks of severe household food insecurity. The effects of occupation of household head, religion, and urban residence on the risks of food insecurity in
Table 2. 4 mirrored those reported in Table 2. 3. Larger household sizes predicted significantly higher risks of not only severe but also moderate household food insecurity. The regional variations in the risks of household food security were less pronounced in the analyses of wealth based on values of assets. For instance, households in the North east compared to those in the
South west in their risks of moderate food insecurity. Also, among the three northern regions, 37 only the North west had significantly lower risks of severe household food insecurity than the
South west.
Findings on transitions into and out of household food insecurity
In the second part of my analyses, I examine first transitions into and out of food insecurity episodes between 2010 and 2013. My preliminary analyses (results not shown) showed that experiences of food security was more transient than persistent in Nigeria. Fewer than half (45%) of households were persistently food secure (24%) or persistently food insecure
(21%). The larger share (55%) of households was transitory in the experience of food insecurity.
Considering the relatively short period of time between the initial and the final rounds of the
NGHS (approximately three years) and for the purpose of clarity, I focus on the first transitions into and out of food insecurity in my next set of analyses. The sample for my analyses of transitions into food insecurity comprises 3,618 households that were food secure at the initial survey round and reported their food security statuses at the following round of data collection.
In the same vein, I analyzed transitions out of food insecurity among 3,391 households that were originally food insecure and had valid reports of food security at the following interview. Table
2. 5 describes both samples in greater details. The results reaffirm the transient nature of food security in Nigeria. Nearly two-fifths of food secure households and about half of food insecure households were observed only twice. More than half of food secure households transitioned into food security over time. But transitions out of food insecurity were equally common in the country; only 44% of food insecure households were persistent in their statuses. Transitions in and out of food insecurity seem to occur fairly quickly.
Previously food secure households had assets worth a little over ₦200,000 compared to
₦137,000 among previously food insecure households. However, given the larger spread of their 38 household wealth, there might be greater inequality among previously food insecure households than among previously food secure households. Previously food insecure households in Nigeria were not predominantly uneducated households, more than two-thirds (68%) were headed by persons with primary or higher education. The occupation statuses of heads of households in both samples were similar – 10% were unemployed and more than two-fifths were engaged in agriculture. The households (both originally food secure and originally food insecure) averaged six in size and were headed by individuals of about 50 years of age. The average dependency ratio was 98% for initially food secure households and 96% for initially food insecure households. Female-headed households account for smaller share (13%) of initially food secure households than initially food insecure ones (18%). More initially food insecure than food secure households were headed by unmarried adults. Half of initially food secure households were
Christians compared to nearly two-thirds of originally food insecure households. Larger shares of households in both food security categories were based in the rural areas (55-61%) than in the urban centers. The greatest share of households in the two samples were drawn from the
Southwestern region. Northwest also accounts for a large share of the food secure households.
In the logistic regression models presented on Table 2. 6, I predicted the odds of transitioning into food insecurity among households that were previously food secure. Before and after controlling for other household sociodemographic characteristics, I found significant protective effects of household wealth and higher levels of education on the risks of food insecurity. However, unemployment of household head was a major driver of the move from food security to food insecurity, net of other predictors. The significantly higher odds of sliding into food insecurity among households headed by sales and service agents, relative to those in agriculture, were mostly explained by other household characteristics. Larger households had 39 significantly lower odds of transitioning into food insecurity in the zero order model. But once I accounted for other sociodemographic differences across households, every additional member to a food secure household elevated the chances of movement into food insecurity by four percent.
Compared to their male-headed counterparts, female-headed households had 51% higher odds of sliding into food insecurity but mostly due to their limited assets (₦96,000 versus
₦217,000 in male-headed households) and their concentration in the more food insecure regions of the country – southern regions. The significant differences in the odds of transitioning into food insecurity among married polygynous, married monogamous, and unmarried households in
Nigeria were due to other household characteristics. Urban residence increased the odds of becoming food insecure by 31%, net of other covariates. Households in all the three northern regions were significantly less likely to become food insecure than households in the South west zone. On the other hand, households in the Southeastern region had nearly thrice the odds of falling into food insecurity as those in the South west. The South west and the South south were similarly susceptible to transitions into food insecurity. The risk of transitioning into food insecurity increased from the second round of interview in 2011 to the third round in 2012.
However, households that were food secure from the first round till the third round were significantly less likely to transition into food insecurity than those that were observed only twice. The significant increase in the risks of moving into food insecurity between the second and the third round could result from the increase in the risks of food insecurity in Nigeria between 2011 and 2012 (period effect) rather than the duration of time since food security.
The predictors of transitions out of household food insecurity are presented in Table 2. 7.
Higher levels of education and wealth were associated with movement out of food insecurity. 40
Households headed by agriculturists were significantly more equipped to escape food insecurity than those headed by sales and service agents. Conversely, households with unemployed heads were significantly more vulnerable to persistent food insecurity. The larger the size of a food insecure household, the higher its chances of remaining food insecure, net of other predictors of food insecurity. Age of household head, household dependency ratio, gender, marital status, and religion of household head were all significantly related to transitions out of food insecurity in the zero-order models. However, their effects were rendered non-significant by other variables in the analyses. Urban households were significantly less likely to exit food insecure episodes than rural households. Households in the regions of Nigeria that were more prone to transitioning into food insecurity (the South east, South south, and the South west) were also significantly less likely to move out of food insecurity. The longer a household stayed food insecure, the lower its chances of escaping food insecurity.
Discussion
It is well established in extant literature that the measurement of living standards is crucial to health and poverty alleviation efforts. In poor countries like Nigeria, absolute poverty better portrays household living conditions than relative measures of socioeconomic wellbeing
(Falkingham and Namazie 2002). However, until recently, there has been a paucity of data on income, consumption, and expenditure in resource-poor countries like Nigeria. The widely available Demographic and Health Survey (DHS) and other social survey data from many sub-
Saharan African countries lack information on consumption and expenditure (McKenzie 2005).
Thus, researchers have relied on alternative measures of socioeconomic wellbeing in the region, notably the wealth or asset index. Although scholars have broadly highlighted the imperfections inherent in using the asset index to assess household welfare (e.g. Lindelow 2006; Sahn and 41
Stifel 2003), only a few (e.g. Knueppel et al. 2010) have explicitly examined the association between food insecurity and household wealth and none in Nigeria.
Using two measures of household wealth (ownership of assets and actual values of assets), the present study analyzes the experiences of household food insecurity across wealth quintiles in Nigeria. Regardless of how it is measured, the wealth indices present a very different picture of household socioeconomic wellbeing than food insecurity in Nigeria. However, actual values of assets provided significant improvements over the traditional wealth index (based on ownership of assets) in understanding household living conditions. For instance, using wealth ownership, there was no significant difference in household food insecurity between the middle and the poorest wealth quintiles. But, I found the expected negative association between food insecurity and household wealth based on actual values of assets. Nonetheless, the relationship between food security and wealth based on values of assets was far from the ideal – rich households being food secure, poor households being food insecure. More than one-fifth of rich
Nigerian households were severely food insecure and nearly half of poor households in the country were food secure in 2010. More so, the relationship between household food insecurity and other sociodemographic characteristics, particularly household size, polygyny, and region of residence, was also slightly different depending on the measure of household food insecurity.
Further, even though the episodic nature of poverty is well established in the literature, food security is often treated as being static. Yet, evidence from limited longitudinal studies of household food insecurity suggest that food insecurity in the developing contexts may be more transient that persistent. Belachew et al. (2012) showed that only about a third of Ethiopian households in their study were chronically food insecure. Half of all Nigerian households were food insecure in 2010; by 2013, majority (76%) had experienced food insecurity (results not 42 shown). The findings reaffirmed the episodic nature of poverty, including food insecurity.
Persistently food secure households were more economically advantaged than persistently food insecure households. Yet, chronically food insecure households in Nigeria were not predominantly uneducated or unemployed households. Education and wealth had significant protective effects against the risks of transitioning into food insecurity. They also significantly aided the escape from food insecurity among food insecure households. On the other hand, unemployment of household head, larger household size, and urban residence aggravated the risks of transitioning into food insecurity and significantly reduced the chances of moving out of food insecurity.
The results of my analyses have important implications for research, policies, and programs aimed at poverty alleviation in Nigeria and elsewhere. This study compounds the list of important limitations of the wealth index in assessing household welfare. At the same time, it illuminates an important avenue for improving this conventional measure – by considering the values of household assets. More importantly, with increasing availability of data, studies of household socioeconomic wellbeing should endeavor to focus more on measures of absolute deprivation such as food security rather than relative socioeconomic wellbeing. Secondly, the high levels of food insecurity and the worsening food conditions in Nigerian households controvert the recent acclamations of the nation’s growing economy (African Development Bank
Group 2014). Perhaps more concerning is the high level of food insecurity among educated households in the country. It may suffice to say that increased employment and increased returns to formal education hold the answer to the deteriorating food conditions in the country. Also, given that half of Nigeria’s population lack adequate access to a basic need for survival – food - 43 the need for social support and programs designed to supplement household sustenance in the short- and in the long-term is imminent.
While this study provides new insights into food insecurity, there are a few important limitations. First, while I included a wide range of assets in my analyses (33 in number), the
NGHS did not provide monetary estimates of housing conditions (e.g. floor and roofing materials). I also did not include access to and use of facilities that could be public or private
(e.g. tap water and use of electricity) and certain types of cooking fuel (e.g. dung, straw, shrubs, and grass) in my computations of the wealth index. Second, food security status was defined based on experiences of households within seven days of the survey rather than throughout the year. However, the food security measure used in this study strongly correlates with the reports of food security in the past year based on a single item in the NGHS (results not shown). Third, like most longitudinal data, the NGHS is affected by the problem of attrition. However, the attrition rates at the four NGHS rounds were 5% or lower. My sensitivity analyses showed relatively high prevalence of food insecurity among households that dropped out of the sample after the initial interview but they changed the sociodemographic distribution of my sample very little. Finally, future efforts should provide a more comprehensive assessment of issues associated with deprivation.
In spite of the above limitations, this study contributes immensely to food insecurity research. Understanding poverty experiences in poor countries like Nigeria is expedient but onerous (Bevan and Joireman 1997) because “as a rule, the poorer a country, the more difficult it is to know just how poor its people are and whether their living standards are improving over time” (Ravallion and Chen 1997: 357). Using a longitudinal study design, I revisited the validity of household wealth as a measure of household socioeconomic status. In particular, I empirically 44 established an important improvement to the classic wealth index. I also examined factors driving transitions into and out of food insecurity between 2010 and 2013. This research has set the stage for future analyses of changes in food insecurity in developing economies like Nigeria.
To further the understanding of transitory food insecurity, scholars must further explore the health implications of chronic and transient household food insecurity. 45
Table 2. 1. Food Security and Household Wealth in Nigeria, 2010: Weighted Descriptive Statistics by Household Ranking Based on Ownership of Assets Variables Nigeria Poorest Poorer Middle Richer Richest Household food security Food secure 49.2 58.6 54.0 43.3 45.0 47.8 Moderately food insecure 23.9 17.4 21.6 26.5 24.5 27.5 Severely food insecure 27.0 24.1 24.5 30.2 30.6 24.8 Education of household head No or other education 36.2 76.1 57.6 41.0 18.3 4.7 Primary education 25.8 16.5 26.0 30.4 33.2 21.6 Secondary education 22.6 6.4 13.2 20.4 32.3 34.1 Higher education 15.4 1.0 3.3 8.1 16.3 39.6 Occupation of household head Unemployed 9.6 5.7 7.3 10.5 11.5 11.6 Agriculture 43.9 79.3 71.5 50.1 23.8 12.2 Sales and services 25.2 9.1 13.3 26.4 36.5 33.9 Professional jobs 13.3 2.3 5.1 7.6 17.2 28.5 others 8.0 3.8 2.9 5.5 11.0 13.8 Household size 5.4 5.7 5.6 5.2 5.1 5.3 (3.0) (2.3) (3.2) (3.1) (2.9) (2.7) Age of household head 49.8 51.7 50.2 51.8 48.2 48.1 (15.5) (16.1) (15.9) (16.7) (15.4) (13.1) Dependency ratio 0.96 1.19 1.10 0.93 0.89 0.81 (0.91) (0.94) (1.00) (0.91) (0.85) (0.80) Missing dependency ratio No 95.3 93.4 94.5 92.0 97.0 98.6 Yes 4.7 6.6 5.5 8.0 3.0 1.4 Gender of household head Male 84.2 87.3 84.2 75.7 84.4 88.9 Female 15.8 12.7 15.8 24.3 15.6 11.1 Marital status of household head Married polygynous 17.1 20.8 22.2 20.0 14.3 11.2 Married monogamous 61.5 62.7 56.5 51.0 60.9 73.4 Unmarried 21.4 16.6 21.4 29.0 24.8 15.4 Religion of household head Muslim 43.2 69.9 54.2 41.5 36.0 24.0 Christian 55.2 27.6 42.8 56.6 63.5 75.3 Others 1.6 2.5 3.0 1.9 0.5 0.7 Urban residence Rural 59.0 92.4 85.1 65.7 40.2 28.1 Urban 41.0 7.6 14.9 34.3 59.8 72.0 Region of residence North central 12.5 11.5 15.7 14.6 14.4 7.5 North east 10.0 20.2 15.1 9.1 6.6 3.0 North west 21.2 45.3 33.0 20.8 10.5 5.5 South east 14.7 8.8 13.8 16.7 16.7 16.2 South south 14.8 4.6 8.8 15.7 17.4 23.4 South west 26.8 9.5 13.6 23.3 34.4 44.5 Unweighted n 4,721 988 898 953 948 934 Source: Nigeria General Household Survey 2010
46
Table 2. 2. Food Security and Household Wealth in Nigeria, 2010: Weighted Descriptive Statistics by Household Ranking Based on Values of Assets Variables All households Poorest Poorer Middle Richer Richest Household food security Food secure 49.2 44.5 47.7 48.3 49.8 58.1 Moderately food insecure 23.9 22.0 25.9 22.2 26.0 24.0 Severely food insecure 27.0 33.6 26.4 29.6 24.3 17.9 Education of household head No or other education 36.2 59.6 38.4 32.3 27.3 13.2 Primary education 25.8 23.7 28.9 30.9 25.0 20.0 Secondary education 22.6 13.6 26.1 24.3 27.8 24.2 Higher education 15.4 3.0 6.7 12.5 20.0 42.7 Occupation of household head Unemployed 9.6 14.8 7.6 7.4 6.4 10.5 Agriculture 43.9 54.1 48.7 45.4 41.5 23.7 Sales and services 25.2 23.1 24.0 24.7 26.5 29.0 Professional jobs 13.3 4.3 10.6 11.7 17.1 28.0 others 8.0 3.7 9.1 10.8 8.6 8.8 Household size 5.4 4.1 5.1 5.5 6.2 6.6 (3.0) (2.6) (2.7) (2.9) (2.9) (3.3) Age of household head 49.8 54.5 48.2 47.3 48.4 49.4 (15.5) (17.4) (15.7) (14.6) (14.3) (12.7) Dependency ratio 0.96 0.88 1.01 1.01 1.00 1.96 (0.91) (0.94) (0.95) (0.95) (0.85) (0.85) Missing dependency ratio No 95.3 85.8 97.8 98.4 98.6 99.3 Yes 4.7 14.2 2.2 1.6 1.4 0.7 Gender of household head Male 84.2 69.1 84.3 89.9 91.0 92.5 Female 15.8 30.9 15.8 10.1 9.0 7.5 Marital status of household head Married polygynous 17.1 7.8 14.7 20.0 23.7 23.7 Married monogamous 61.5 54.3 62.6 62.0 63.2 68.0 Unmarried 21.4 37.9 22.7 18.0 13.1 8.3 Religion of household head Muslim 43.2 43.5 46.0 45.1 44.1 36.0 Christian 55.2 54.2 52.8 52.9 54.8 62.9 Others 1.6 2.3 1.2 2.0 1.1 1.1 Urban residence Rural 59.0 70.4 61.1 57.7 58.4 42.0 Urban 41.0 29.6 38.9 42.3 41.6 58.0 Region of residence North central 12.5 10.4 12.1 13.3 15.8 11.9 North east 10.0 8.0 10.3 10.5 11.9 10.1 North west 21.2 21.4 24.8 23.4 20.5 14.5 South east 14.7 17.9 14.1 15.2 12.8 12.1 South south 14.8 10.9 11.4 14.1 16.8 23.5 South west 26.8 31.4 27.3 23.5 22.2 27.8 Unweighted n 4,721 1,218 990 915 825 773 Source: Nigeria General Household Survey 2010
47
Table 2. 3. Household Food Security and Household Wealth Based on Ownership of Assets in Nigeria in 2010, Multinomial Logistic Regression Relative Risk Ratios (n = 4,721) Model 1 Model 2 Model 3 Moderately Severely Moderately Severely Moderately Severely food insecure food insecure food insecure food insecure food insecure food insecure vs. food vs. food vs. food vs. food vs. food vs. food Predictors secure secure secure secure secure secure Household socioeconomic characteristics Household wealth based on ownership of assets (0 = Poorest) Poorer 1.35* 1.10 1.28† 1.00 1.12 0.85 Middle 2.07*** 1.70*** 1.89*** 1.40* 1.40* 0.98 Richer 1.83*** 1.65*** 1.60** 1.23 1.05 0.70* Richest 1.94*** 1.26† 1.77*** 1.01 0.97 0.45*** Education of household head (0 = Primary education) No or other forms of education 0.68*** 0.54*** 0.81† 0.71** Secondary education 0.82 0.84 0.98 1.00 Higher education 0.63** 0.51*** 0.78 0.63** Occupation of household head (0 = Agriculture) Sales and services 1.11 1.21 0.98 1.00 Professional job 1.13 0.87 1.06 0.76 Unemployed 0.89 1.66*** 0.68* 1.20 Others 0.92 1.34† 0.80 1.09 Other household sociodemographic characteristics Household size 1.03 1.08*** Age of household head 1.01 1.01 Dependency ratio 0.93 1.08 Missing dependency ratio 0.60* 0.73 Female household head 1.17 0.93 Marital status of household head (0 = Married monogamous) Married polygynous 1.09 0.76* Unmarried 1.08 1.09 Religion of household head (0 = Muslim) Christian 0.82 0.96 Others 2.16* 2.97*** 48
Urban residence 1.29* 1.87*** Region of residence (0 = South west) North central 0.64** 0.66** North east 0.88 0.62** North west 0.41*** 0.42*** South east 1.93*** 2.01*** South south 2.61*** 2.78*** Intercept 0.30*** 0.41*** 0.41*** 0.63*** 0.36*** 0.36*** 53.24 53.24 126.30 126.30 398.60 398.60 Wald Chi Square (df) (8)*** (8)*** (22)*** (22)*** (52)*** (52)*** Source: Nigeria General Household Survey 2010; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001; Relative risks relative to being food secure; df = degree of freedom
49
Table 2. 4. Household Food Security and Household Wealth Based on Estimated Values of Assets in Nigeria in 2010, Multinomial Logistic Regression Relative Risk Ratios (n = 4,721) Model 1 Model 2 Model 3 Moderately Severely Moderately Severely Moderately Severely food insecure food insecure food insecure food insecure food insecure food insecure vs. food vs. food vs. food vs. food vs. food vs. food Predictors secure secure secure secure secure secure Household socioeconomic characteristics Household wealth based on actual values of assets (0 = Poorest) Poorer 1.10 0.73* 0.98 0.64*** 1.01 0.61*** Middle 0.93 0.81† 0.80 0.69** 0.79 0.59*** Richer 1.06 0.65*** 0.90 0.55*** 0.83 0.44*** Richest 0.84 0.41*** 0.67** 0.33*** 0.52*** 0.20*** Education of household head (0 = Primary education) No or other forms of education 0.57*** 0.46*** 0.76* 0.66** Secondary education 0.86 0.89 0.99 1.02 Higher education 0.75† 0.65* 0.89 0.79 Occupation of household head (0 = Agriculture) Sales and services 1.28* 1.31* 1.00 1.00 Professional job 1.28 0.94 1.06 0.73† Unemployed 1.00 1.65*** 0.69† 1.15 Others 1.05 1.43* 0.78 1.03 Other household sociodemographic characteristics Household size 1.05* 1.11*** Age of household head 1.01 1.01† Dependency ratio 0.91 1.07 Missing dependency ratio 0.60* 0.67† Female household head 1.20 0.92 Marital status of household head (0 = Married monogamous) Married polygynous 1.16 0.86 Unmarried 1.04 1.04 Religion of household head (0 = Muslim) Christian 0.82 0.94 Others 2.31* 3.13*** 50
Urban residence 1.33* 1.87*** Region of residence (0 = South west) North central 0.67* 0.79 North east 0.92 0.80 North west 0.42*** 0.51*** South east 2.04*** 2.25*** South south 2.90*** 3.43*** Intercept 0.49*** 0.67** 1.07 0.41** 0.35*** 48.70 48.70 153.09 153.09 444.14 444.14 Wald Chi Square (df) (8)*** (8)*** (22)*** (22)*** (52)*** (52)*** Source: Nigeria General Household Survey 2010; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001; Relative risks relative to being food secure; df = degree of freedom
51
Table 2. 5. Transitions Into and Out of Household Food Insecurity in Nigeria: Weighted Descriptive Statistics by Initial Food Security Status Food secure households Food insecure households Variables (n = 3,618) (n = 3,391) Change in food security status Persistently food secure 48.5 - Transitioned into food insecurity 51.5 - Persistently food insecure - 44.2 Transitioned out of food insecurity - 55.8 Number of survey rounds since food secure 1 round 38.2 48.2 2 rounds 32.9 25.8 3 rounds 29.0 26.1 Log of household wealth based on actual values of assets 10.8 (1.6) 10.5 (1.6) Average naira value of wealth 200,624 ( 899,561) 136,949 (1,104,674) Education of household head No or other education 39.0 32.5 Primary education 23.1 29.4 Secondary education 21.0 24.5 Higher education 16.9 13.6 Occupation of household head Unemployed 10.1 10.6 Agriculture 43.9 40.6 Sales and services 24.1 27.0 Professional jobs 14.2 12.3 others 7.8 9.49 Household size 6.2 (3.3) 6.1 (3.2) Age of household head 49.5 (15.4) 50. 3 (15.3) Dependency ratio 0.98 (0.89) 0.96 (0.91) Missing dependency ratio No 95.3 95.7 Yes 4.7 4.3 Gender of household head Male 86.4 81.9 Female 13.6 18.1 Marital status of household head Married polygynous 19.0 15.0 Married monogamous 61.7 62.0 Unmarried 19.3 23.0 Religion of household head Muslim 48.6 36.7 Christian 49.9 61.3 Others 1.5 1.9 Urban residence Rural 61.0 55.1
52
Urban 39.0 44.9 Region of residence North central 14.1 10.0 North east 11.9 8.6 North west 24.0 14.9 South east 9.5 18.1 South south 12.5 18.9 South west 27.9 29.6 Source: Nigeria General Household Survey, 2010-2013; standard deviations parentheses where appropriate
53
Table 2. 6. Odds Ratios, Logistic Regression Predicting First Transitions into Household Food Insecurity in Nigeria Predictors Zero Order Model 1 Model2 Log of household wealth based on actual values of assets 0.92*** 0.90*** 0.87*** Education of household head (0 = Primary education) No or other forms of education 0.68*** 0.66*** 0.90 Secondary education 0.97 0.94 0.99 Higher education 0.56*** 0.61*** 0.61*** Occupation of household head (0 = Agriculture) Sales and services 1.41*** 1.42*** 1.09 Professional job 1.00 1.25† 1.10 Unemployed 1.46** 1.53*** 1.29* Others 1.64*** 1.66*** 1.33* Household size 0.97** 1.04** Age of household head 1.00 1.00 Dependency ratio 0.93† 0.99 Missing dependency ratio 0.73† 0.48*** Female household head 1.51*** 1.09 Marital status of household head (0 = Married monogamous) Married polygynous 0.80** 0.94 Unmarried 1.30** 0.97 Religion of household head (0 = Muslim) Christian 1.48*** 0.89 Others 1.01 0.86 Urban residence 1.38*** 1.31** Region of residence (0 = South west) North central 0.53*** 0.54*** North east 0.61*** 0.58*** North west 0.50*** 0.46*** South east 2.39*** 2.81*** South south 0.86 1.01 Number of survey rounds since food security (0 = 1 round) 2 rounds 1.15* 1.20* 1.31*** 3 rounds 0.43*** 0.46*** 0.54*** Intercept 1.25 1.95* Wald chi square (df) 158 (10)*** 377 (25)*** Source: Nigeria General Household Survey, 2010-2013; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001; n = 7,116 person-periods, 3,618 households; all zero order models control for number of rounds since food security; df = degree of freedom
54
Table 2. 7. Odds Ratios, Logistic Regression Predicting Transitions Out of Household Food Insecurity in Nigeria Predictors Zero Order Model 1 Model2 Log of household wealth based on actual values of assets 1.12*** 1.17*** 1.17*** Education of household head (0 = Primary education) No or other forms of education 1.74*** 1.88*** 1.10 Secondary education 1.36** 1.35** 1.18 Higher education 1.64*** 1.40* 1.26† Occupation of household head (0 = Agriculture) Sales and services 0.66*** 0.68*** 0.80* Professional job 1.04 0.95 1.08 Unemployed 0.53*** 0.52*** 0.64** Others 0.71* 0.73* 0.86 Household size 1.06*** 0.96* Age of household head 0.99*** 1.00 Dependency ratio 1.07† 0.98 Missing dependency ratio 0.84 1.30 Female household head 0.55*** 0.96 Marital status of household head (0 = Married monogamous) Married polygynous 1.58*** 1.06 Unmarried 0.67*** 0.97 Religion of household head (0 = Muslim) Christian 0.39*** 0.90 Others 0.58* 1.17 Urban residence 0.82** 0.78** Region of residence (0 = South west) North central 1.48** 1.27† North east 2.63*** 2.22*** North west 3.31*** 2.82*** South east 0.28*** 0.26*** South south 0.78* 0.69** Number of survey rounds since food insecurity (0 = 1 round) 2 rounds 0.44*** 0.52*** 3 rounds 0.37*** 0.49*** Intercept 0.11*** 0.17*** Wald chi square (df) 265 (10)*** 644 (25)*** Source: Nigeria General Household Survey, 2010-2013; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001; n = 6,011 person-periods, 3,391 households; all zero order models control for number of rounds since food insecurity; df = degree of freedom
55
CHAPTER 3: HOUSEHOLD COMPOSITION AND EXPERIENCES OF TRANSIENT AND
PERSISTENT FOOD INSECURITY IN NIGERIA: THE ROLE OF SOCIAL CAPITAL,
EDUCATION, AND TIME USE
Introduction
Food insecurity describes access to insufficient quantity and/or quality of nutritionally-adequate food by any segment of a population at any point in time (Hadley 2014; Ivers and Cullen 2011;
Sirotin et al. 2014). Globally, food insecurity poses a serious challenge to the health and overall wellbeing of human population (Nord 2014). Nearly a billion people are estimated to be suffering from chronic hunger, the majority of whom are residents of developing nations (Food and Agriculture Organization 2014). In addition to the magnitude of the affected population, food insecurity is associated with numerous negative social and health outcomes among children and adults (Alaimo et al. 2001; Olson 1999; Vozoris and Tarasuk 2003). Therefore, it is pertinent to understand circumstances that could potentially ameliorate or aggravate the experiences of food insecurity.
Experiences of hardships, including food insecurity, vary according to composition of household members (Snyder et al. 2006). Households may consciously or unconsciously adjust their sizes and composition either as a preventive strategy or as a means of coping with economic hardships (Pilkauskas et al. 2014). The makeup of a household may also reflect its socioeconomic standing (Manning and Brown 2006). Having certain individuals in the household may increase vulnerability to economic hardships and food insecurity. Specifically, children and elderly persons with disability are among the populations most vulnerable to the spells of poverty and food insecurity (International Food Policy Research Institute 2012;
Obayelu 2006; RTI International 2014). Yet, the associations are complex as even among
56 vulnerable populations, the experiences of food insecurity vary. Recent research documents a food security paradox (see Balistreri 2012) - the experiences of food security among many families and households with vulnerable populations and food insecurity among numerous households without vulnerable populations.
Although there is a wide range of research showing that socioeconomic resources are related to food insecurity in Nigeria (Oluwatayo 2009) and that many vulnerable populations in the country have access to very limited resources (Obayelu 2006), the interrelationship between presence of vulnerable persons in the household and socioeconomic resources in relation to food insecurity in Nigeria has not been fully explored. The different ways by which household composition contributes to the risks of household food insecurity and how access to socioeconomic resources modifies the experience of food insecurity among vulnerable populations are less well established. This is not surprising given the “preliminary and partial” nature of research on the relationship between food insecurity and its correlates as well as the underlying processes and mechanisms that mediate those relationships in low income countries
(Nord 2014: 3). The present study is one of the first attempts to analyze factors mediating and moderating the association between household food insecurity and the presence of vulnerable population in the household using nationally representative data in Nigeria.
Most developed countries have welfare programs specifically designed to combat food insecurity among their vulnerable populace. In the U.S. for instance, WIC is a special
Supplemental Nutrition Program that caters to pregnant and postpartum women, infants, and children (USDA Food and Nutrition Service 2015). Similarly, the Universal Child Care Benefit program provides monetary aid to families with children in Canada while the Old Age Security
(OAS) pension is paid monthly to most legal residents of the country aged 65 and above,
57 regardless of their employment history (Service Canada 2014). Research shows that participation in such programs not only reduces household food insecurity (Bartfeld and Ahn 2010; Nord
2012), but it also lessens its impact on the health of adults and children (Joyce et al. 2012).
However, in most resource-poor settings like Nigeria where food insecurity is highly prevalent and more severe (Nord 2014), government-provided safety nets are rare or totally absent. Only the privileged few benefit from social protection systems mostly provided by private insurance companies (Wanyama et al. 2008). Also, insufficient assets to serve as collateral and complex paperwork requirements are two out of several factors inhibiting poor people from accessing credit facilities through formal financial institutions in low income countries (Callier 1990; Miracle et al. 1980; Oleka and Eyisi 2014; Oluwatayo 2009; Yusuf et al.
2009).
The situation in Nigeria is particularly concerning; sixty eight percent of Nigerians lived on less than one dollar twenty five cents in 2010, up from 63 percent in 2004 and the vast majority of the population (85%) survives on less than two dollars a day (The World Bank
2014a). Nigeria has been and still is struggling with a series of economic crises including but not limited to soaring food prices and heavy reliance on food importation, collapse of income from oil exports, massive lay-off of workers, alarming youth unemployment and underemployment rates, removal of fuel subsidy, sociopolitical conflicts, high rate of corruption, and failing agricultural and industrial sectors, among other problems associated with rapid population growth (Babatunde and Martinetti 2011; Ogwumike 2001; Otaha 2013; Nkpoyen and Bassey
2012; Obayelu 2006; Oleka and Eyisi 2014). All of the above listed problems dampen the hope of any future social welfare programs in Nigeria.
58
Nonetheless, support to economically struggling families and individuals is not completely lacking as few informal institutions and groups, such as mutual-aid groups, informal financial institutions, and networks of families and friends at home and abroad, do provide services that research shows are highly instrumental in reducing poverty and lowering household food insecurity (Babatunde and Martinetti 2011; Ezekiel 2014; Ogwumike 2001; Oleka and Eyisi
2014; Wanyama et al. 2008; Yusuf et al. 2009). Given the central role of self-help organizations
(Birchall 2003) in Nigeria’s economy, it is important to understand how informal groups assist poor households in coping with economic hardships. In this study, I examine the association between presence of vulnerable populations in the household and food insecurity with attention to how factors related to socioeconomic wellbeing, financial support through informal groups, and time use moderate these associations.
This chapter utilizes recently collected and nationally representative data to: 1) compare the patterns of household food insecurity among households with vulnerable populations
(children and older adults with disability); 2) investigate the role of informal networks, education, and time spent fetching cooking fuel on food insecurity; and 3) test whether or not differences in access to social capital, education, and time use explain variations in food insecurity across households. I seek to explain the relationship between household composition and household food insecurity by drawing on two theoretical frameworks - food insecurity as a
“managed process” (Radimer et. al.1990) and the principle of family adaptive strategy (Moen and Wethington 1992) as well as the concept of social capital (Portes 1998). The study explores three possible mechanisms by which the presence of vulnerable populations may be linked to higher risks of household insecurity, including access to financial support, food management skills (assessed with levels of education), and time use. In a setting as Nigeria, social capital,
59 measured as remittances, informal savings groups, informal lending groups, and networks of families and friends, is expected to buffer the experience of household food insecurity among households with vulnerable populations.
A further goal is to examine the episodic nature of food insecurity. In the U.S., food- insecure households experience food insecurity for an average of seven months per year (RTI
International 2014). However, even though the episodic nature of household food insecurity has been reported in other contexts (RTI International 2014), this is one of the first few studies to analyze both transient and chronic household food insecurity in Nigeria. Unlike sudden food shortages, situations involving predominantly chronic food insecurity or hunger attract less media attention (International Food Policy Research Institute 2012), but chronic and transient food insecurity may have varying implications for people’s well-being. Further, understanding of change in household food conditions is crucial to any poverty alleviation effort (Bigsten et al.
2003).
Household Structure and Food Insecurity
The International Food Policy Research Institute (IFPRI) describes vulnerability as “the extent to which households and individuals can be negatively affected by an external shock, from either the physical or socioeconomic environment.” (2012: 1). In this light, vulnerability depicts greater probability of sliding into poverty due to limited capability to absorb shocks (Birchall 2003).
Although poverty is not the same as food insecurity, heightened food insecurity is mostly linked to lower economic resources (Gundersen 2013; Oluwatayo 2009). Based on their disadvantaged positions in the labor market, their lower economic productivity, and/or physical or functional limitations, children and elderly persons with a disability are vulnerable populations (Obayelu
2006; The International Food Policy Research Institute 2012; RTI International 2014). They
60 have minimal capacity to respond to economic shocks emanating from natural and social causes
(The International Food Policy Research Institute 2012).
Households with vulnerable populations may be quick to reduce the quality and quantity of their diets in times of global food crisis (The World Bank 2011b). They may also have limited capability to alleviate food insecurity when it occurs (RTI International 2014). Triggers of food shortage may aggravate the impacts of preexisting socioeconomic disadvantage thereby precipitating or exacerbating food insecurity. Households that are most vulnerable to economic hardships may also have fewer resources to build resilience to threats to household livelihood.
Protecting vulnerable populations from the negative impacts of food insecurity requires adequate understanding of the distinct and general patterns of household food insecurity among the different groups.
Greater vulnerability to poverty and food insecurity among households with children and elderly persons with disability is best conceptualized as the resultant effect of high dependency ratio - the ratio of the total number of dependents (children under the age of 15 and adults sixty five years and above) to the total number of working age adults (age 15-64) in each household.
Along this line, research in Nigeria (Babatunde et al. 2007; Obayelu 2010; Omonona and
Adetokunbo 2007) and elsewhere (Miller et al. 2011), shows a positive relationship between household dependency ratio and poverty and food insecurity. However, research relying on household dependency ratio may have oversimplified the relationship between household composition and food insecurity. This is because, children and elderly persons may experience poverty differently. In the U.S., households with children under the age of 18 have higher rates of food insecurity while those with elderly persons are more likely to be food secure than other households in the country (Alaimo et al. 1998; Coleman-Jensen et al. 2014; Rose 1999). Also,
61 the popular notion of dependency among older adults has been recently challenged. The gains in life expectancy over the past decades and the advent of improved medical technology means that people are not only living longer but also healthier and therefore economically productive in older ages (Sanderson and Scherbov 2015). Thus, research on poverty among older people needs to focus more on the segment of the elderly population most vulnerable to hardships – older adults with a disability. Functional limitation or physical disability, lower economic productivity, and lower level of support seem to be driving the high prevalence of food insecurity among households with, or headed by, older people (Anderson et al. 2014; Quandt and Rao 1999; Woltil
2012).
More so, there is a complex association between fertility and household socioeconomic status. Even in the developed world where socioeconomic characteristics are negatively related to fertility outcomes, improvements in household wealth are significantly associated with a greater likelihood of having a child (Lovenheim and Mumford 2013). This means that in general wealthy households have the tendency to have many children, in which case large number of children, especially in contexts in which large family size is normative, will be indicative of higher socioeconomic status rather than precarious economic situations. This supports the view of large family size as a status symbol in sub-Saharan Africa (Boserup 1985). On the other hand, particularly in high fertility contexts like Nigeria, poor households may emulate the culture of large family sizes among their wealthy counterparts or increase their desired and actual family size in the hope of improving their socioeconomic standing through the economic inputs of their children. Therefore, beyond their conceptualization as dependents, it is important to understand how presence of children and older adults with a disability relate to the experiences of household
62 food insecurity. Based on the demographic transition theory, the contributions of young children to household economy is expected to be particularly robust in agrarian households.
In spite of the high prevalence rate of child labor in the developing countries
(Canagarajah and Nielsen 2001; Togunde and Carter 2006), a large body of literature still conceptualizes children as dependents (Babatunde et al. 2007; Miller et al. 2011; Muga and
Onyango-Ouma 2009; Obayelu 2006; Omonona and Adetokunbo 2007; Qvortrup 1997).
Households with children are more likely to juggle multiple economic demands including paying children’s hospital and prescription bills, purchasing nutrient-rich foods, and child care, among others, than households with no children (Joyce et al. 2012). Children also have relatively high nutrient requirements (Joyce et al. 2012; Laraia et al. 2006; The World Bank 2010; 2011).
Because the foundation for future cognitive and physical development is laid in the first few years of life, it is difficult to reverse the detrimental effects of nutritional deficiency on children
(Joyce et al. 2012). Although adults usually buffer the effects of household food inadequacy on children, children in at least half of all food-insecure households with children were considered food insecure in 2012 (RTI International 2014).
In view of the perceived dependency among children and elderly persons with a disability in the existing literature, I hypothesize that:
1. Households with children and those including older adults with a disability (two
vulnerable populations) in Nigeria will be more likely to report being moderately and
severely food insecure.
2. Households with vulnerable populations will be more likely to slide into (transient food
insecurity) and less likely to move out of food insecurity episodes (chronic food
insecurity).
63
Since the demographic transition theorists mostly postulate about child labor or economic inputs of children as incentives to having large family sizes in subsistence agriculture (Boserup
1985; Caldwell 2005), I tested for significant differences in the relationship between number of children in the household and food insecurity across three different occupational categories – agriculture, sales and services, and professional jobs.
The Role of Social Capital in the Experience of Food Insecurity among Vulnerable
Populations
Food expenditure is one out of several needs that households must juggle and often, food gets in the way of other household necessities such offsetting medical bills, energy costs, among others
(Cook et al. 2008; Nord and Kantor 2006). It is therefore not surprising that food insecurity is more prevalent among low income households than among households with high earnings
(Coleman-Jensen et al. 2014; Dean et al. 2011; Gundersen 2013; Oluwatayo 2009). The relationship between income and food security is even more important in countries like Nigeria where the bulk (70-80%) of household income is expended on food (Obayelu 2006). This means that even the slightest disruption in income flow could result in poor families and households tumbling into food insecurity. Inadequate safety nets or government programs that could raise households out of poverty in the country (Ezekiel 2014; Oluwatayo 2009) further suggests high rates of food insecurity among households with vulnerable populations in Nigeria.
However, even though vulnerable populations are at greater risks of food insecurity, not all households with these statuses experience poverty whereas many households with no such potentially dependent population report food insecurity. This phenomenon has been described as a food security paradox (see Balistreri 2012). It is plausible that access to financial support through self-help organizations (Birchall 2003) buffers households with vulnerable populations
64 from threats to their livelihoods. Therefore, I analyze the protective roles of financial support through informal savings groups, informal lending groups, and networks of families and friends at home and abroad on household food insecurity among households with children and those with elderly persons with a disability.
Food insecurity has been conceptualized as a “managed process” (Radimer et. al.1990).
This framework implies that people are not passive victims of household food shortage. Rather, they actively strategize and work out plans to manage household budgets and food situations
(Martin and Lippert 2012). This is also in line with the principle of family adaptive strategy
(Moen and Wethington 1992) which postulates that families are not subservient to stressful events and structural barriers. Instead, families and households do devise strategies “for coping with, if not overcoming, the challenges of living, and for achieving their goals in the face of structural barriers” (Moen and Wethington 1992: 234). Strategy, in this case, encompasses anything that the household does to maintain its economic wellbeing in times of hardships.
Along this line, the size and/or the composition of a household might be indicative of efforts to prevent and/or cope with adverse economic circumstances. Forward-looking parents might deliberately modify their desired and actual family size by choosing to invest their limited resources in the quality rather than number of their offspring (Caldwell 2005). On the other hand, in spite of the growing cost of children in the increasingly globalized world, some parents might maintain large families in anticipation of children’s labor inputs. Research shows high prevalence of child labor in West African countries (Canagarajah and Nielsen 2001; Togunde and Carter 2006).
Further, when unwilling or unable to modify its composition, households whose numbers and/or members make them more vulnerable to food insufficiency could manage to stay aloft of
65 food insecurity by supplementing their household resources with informal assistance. Social scientists have coined the concept of social capital to describe the network of persons that could potentially provide support to an individual and the resources embedded in such network (Portes
1998). Social capital encompasses any resource generated through the formal and informal relationships among members of a community (Oluwatayo 2009). Previous studies have identified several social networks or informal groups providing monetary and non-monetary assistance to poor Nigerians (Akinrinola and Mafimisebi 2010; Balogun and Yusuf 2011;
Okunmadewa et al. 2007; Oleka and Eyisi 2014; Oluwatayo 2009). Below, I discuss three forms of social capital that are linked to poverty experiences in Nigeria. These are: mutual-aid groups
(e.g. co-operative societies and the Rotating Savings and Credit Associations (ROSCAs)), informal financial institutions (e.g. money lenders, local savings scheme), and social networks of friends and families at home and abroad (Akinrinola and Mafimisebi 2010; Babatunde and
Martinetti 2011; Ezekiel 2014; Oleka and Eyisi 2014; Wanyama et al. 2008; Yusuf 2009).
Co-operative societies are open and voluntary associations of persons who come together for the purpose of addressing their common socioeconomic needs and thereby improving their living conditions (Committee for Promotion and advancement of Cooperatives 1999 in
Akinrinola and Mafimisebi 2010). As many as 4.3 million Nigerians were co-operative members in 2005 (Wanyama et al. 2008). Research shows that co-operative societies such as agricultural co-operatives, consumer co-operatives, food co-operatives, credit unions, among numerous others, have helped tremendously in redistributing wealth and alleviating poverty in many parts of Africa (Ezekiel 2014; Wanyama et al. 2008). These institutions mobilize and distribute financial resources, avail their members of economic opportunities, as well as provide personal development through education and skill acquisition programs. Perhaps most importantly for the
66 present analysis is the fact that co-operatives can be very instrumental in providing security in times of economic shocks resulting from shortfalls in agricultural production, health problems, loss of loved ones, household welfare demands, and other socioeconomic problems (Birchall
2003; Ezekiel 2014; Oluwatayo 2009; Wanyama et al. 2008). In spite of the role of these mutual- aid groups, co-operatives are sparsely referenced in the literature on poverty reduction (Birchall
2003) and seldom appear in studies of household food insecurity in Nigeria.
Rotating Savings and Credit Associations (ROSCAs) or rotating funds are other informal financial institutions that help to combat poverty at the micro level in many low-income countries, including Nigeria (Alufohai et al. 2012; Yusuf et al. 2009). The autonomous members of these associations pool their resources together for the common goal of improving their standard of living. They periodically (weekly or monthly) contribute a specific amount of money which aggregates into a lump sum payment that is rotated among members based on their needs, social standing in the group, age, by negotiation, or by drawing lots (Anderson and Baland 2002;
Callier 1990; Miracle et al. 1980). Capital generated through ROSCAs are often put into income- generating activities, acquisition of assets, or expended on household welfare – food, housing, health care, etc. (Miracle et al. 1980; Yusuf et al. 2009). Research also suggests that ROSCAs may be imposed on husbands by their wives in order to better manage men’s resources for the betterment of household welfare (Anderson and Baland 2002).
The local money lenders offer credit facilities to members of their communities in form of loans but usually charge lower interests than formal financial institutions (Alufohai et al.
2012; Miracle et al. 1980). Such facilities may be more easily accessible in times of sudden economic difficulties that threaten household food security. Unlike local money lenders, local savings or daily contribution scheme helps to redistribute funds among members of the same or
67 different communities. It involves house-to-house collection of daily savings by a chief collector or a mobile banker who either saves the money on behalf of the contributors or lends the money out over a short period of time before the collection date agreed upon (usually by the end of the month) by the banker and the contributors (Eyisi and Oleka 2014; Miracle et al. 1980). Saving for the rainy day through the daily contribution scheme could avert the risks of household food insecurity while credits from mobile bankers could lessen the risks of shocks to household income. Last but by no means the least, familial social capital and friends at home and abroad are important assets in the experiences of poverty and food insecurity (Babatunde and Martinetti
2011; Green-LaPierre et al. 2012), particularly in providing resources to access food (Dean et al.
2011). Large numbers of farmers who constitute the large part of Nigeria’s population
(Babatunde and Martinetti 2011) rely on credits from families and friends for their agricultural production activities (Alufohai et al. 2012).
Studies in various parts of Africa, including Nigeria, have shown positive impact of borrowing on household income, economic productivity, and household consumption patterns
(e.g. Alufohai et al. 2012; Berhane and Gardebroek 2011; Miracle et al. 1980; Nkpoyen and
Bassey 2012). Social capital has also been shown to be positively related to the overall wellbeing of households in southwestern part of Nigeria (Olowe et al. 2014; Oluwatayo 2009). Remittance- receiving households in Kwara state of Nigeria have lower risks of household food insecurity and experience better child nutritional status than non-receiving households (Babatunde and
Martinetti 2011). Studies of social capital among vulnerable populations in other contexts revealed that households with children were less likely to experience very low food security if the adults in the households had people they could rely on for financial support (Anderson et al.
2014).
68
Liverpool-Tasie et al. (2011) summarized the existing literature on household food insecurity and social capital in the rural parts of Nigeria and concluded that studies of social capital in the rural parts of the country are mostly based on the role of credit access in improving agricultural production. Some additional studies are limited to only one component of social capital and are limited to select spatial samples. For instance, Babatunde and Martinetti (2011) analyzed the influence of migrant remittances on food security and nutrition of farming households in Kwara state. Olowe and her colleagues (2014) investigated the relationship between social capital and the nutritional status of rural households in southwestern part of
Nigeria while Oluwatayo (2009) examined the risks of household food insecurity among members of co-operatives in Ekiti state. The consensus is that social capital is associated with food insecurity; however, to date no research has analyzed the relationship between multiple forms of social capital and both transient and chronic household food insecurity in a nationally representative sample of Nigerians.
Given the active role of households in combating hunger and food insecurity and the importance of financial support in coping with difficult household food situations, I hypothesize that:
3. Access to informal financial support will mediate the association between household food
insecurity and presence of vulnerable populations (children and older adults with a
disability) in the household.
Previous studies also suggest that social support programs are more crucial to food insecurity experiences among vulnerable populations than the general population (Metallinos-
Katsaras et al. 2011). I, therefore, test whether the effect of financial assistance on food insecurity varies across households with and without vulnerable persons. I posit that:
69
3b. Access to informal financial support will be more strongly associated with household
food insecurity among households with vulnerable populations than among households
without vulnerable populations.
Alternative Explanations
Food management skills (education)
Although most explanations of greater prevalence of food insecurity among vulnerable populations draw on the theory of economic disadvantage (e.g. Coleman-Jensen et al. 2014;
Oluwatayo 2009), financial hardships are not the only barrier to attaining household food security (Balistreri 2012; Devine et al. 2006; Martin and Lippert 2012; Quandt and Rao 1999;
Woltil 2012). Food management skills are equally important (Green-LaPierre et al. 2012). It is widely documented that mothers adopt a wide range of approaches to shield their families from hunger (Martin and Lippert 2012; McIntyre et al. 2003; Stevens 2010). Hunger management strategies that are important in the experiences of household food insecurity include financial management, acquisition of knowledge of dietary food intake, shopping strategies, planning of meals, prioritizing household expenditure, management of the feelings of stress and fatigue associated with hardships, and help seeking behavior, particularly from spouse (Devine et al.
2006).
I rely on a proxy measure of food management strategies, education, because of the lack of a direct measure. Higher levels of education are usually associated with higher levels of skills and greater access to information (Moen and Wethington 1992). The effectiveness of coping strategies may also vary across levels of education. Since education is inversely related to family size, households with no children may have higher levels of education than those with children.
Also, considering the gains in educational enrollment over time in Nigeria, household with older
70 persons with a disability may be more likely to be headed by uneducated persons. Therefore, controlling for education of household heads should reduce the disparity in household food insecurity by number of children and presence of adults with a disability in the household. I hypothesize that:
4. Education of household head will mediate the association between presence of
vulnerable population in the household and household food insecurity.
Further, to the extent that food management skills and the effectiveness of coping strategies vary across levels of education, households with vulnerable populations but headed by highly educated individuals would be less likely to experience household food insecurity. This logic is reflected in my fifth hypothesis which states that:
5. Educational attainment of the household head will moderate the association between
vulnerable status and food insecurity. That is, the experiences of food insecurity among
vulnerable households headed by highly educated adults will be similar to those of
households with no vulnerable populations.
Time use
Recent research seeking to explain the high prevalence of food insecurity, particularly among households with children, suggests differences in time use between food-secure households and food-insecure households. Adults in households with any experience of child food insecurity spend more time on food preparation and cleanup than their counterparts in food- secure households (Balistreri 2012). Also, time spent on other food-related activities, such as shopping for food, increases with increasing risk of very low food insecurity (Anderson et al.
2014). These findings suggest greater ease of access to food and food preparation strategies that save time among food-secure households than among food-insecure households. More so, having
71 children and elderly persons in the household increases the time spent on food preparation
(Jensen and Zhylerevskyy 2013) suggesting that households with vulnerable populations are more strained, in terms of time, in accessing food. Having many young children in the household increases the burdens of fetching cooking fuel, meaning that households with a large family may depend more on the traditional cooking fuel (wood) which are cheaper and more readily available than smaller households. In view of the above, I anticipate that time spent getting cooking fuel by households would mediate the association between presence of vulnerable persons in the household and food insecurity in Nigeria. However, the high prevalence rate of child labor in the present study context suggests that children could contribute to household socioeconomic wellbeing by helping to collect wood for cooking. Thus, I do not hypothesize about the direction of association between time spent fetching wood and food insecurity. Rather,
I postulate that:
6. Controlling for the time spent collecting cooking fuel will minimize the food security gap
between households with vulnerable household members (children and older adults with
a disability) and those without vulnerable members.
I further hypothesize that:
7. Time spent collecting fuel will be more strongly associated with household food
insecurity among households with vulnerable populations than among households
without vulnerable populations.
Although I present food management skills (education) and time use as separate
mechanisms by which the presence of children and elderly persons with a disability in the
household relate to household food insecurity, time use may also reflect the strength of
household food management skills.
72
Current Investigation
An emerging theme in food insecurity research is a form of food security paradox whereby many households with vulnerable populations manage to remain food secure while those with no vulnerable populations are food insecure (Balistreri 2012). The paradox suggests a buffering effect in the experiences of food insecurity among vulnerable households. The purpose of this study is to analyze three mechanisms by which vulnerable status relates to household food insecurity - access to social capital, education, and time use. I examine the mediating and moderating roles of access to social capital, different levels of education of household head, and variations in the amount of time households spend collecting or fetching wood or cooking fuel on the experiences of food insecurity among households with varying number of children and households with elderly persons with a disability. Further, I examined the extent to which children’s economic contributions to household welfare aid in ensuring food security among agricultural households. I tested for significant interactions between number of children in the household and occupations of household heads. I predict food insecurity among households with vulnerable populations and the effects of the mediating and moderating variables in both cross- sectional and longitudinal analyses. In the cross-sectional component of my analyses, I used multinomial logistic regression to predict the odds of moderate and severe food insecurity, relative to a household being food secure in 2010. I used logistic regression to analyze change in household food insecurity status between 2010 and 2013.
I accounted for the effects of other correlates of poverty and food insecurity reported in previous studies – occupational status of household head, household wealth, size of household, age of household head, gender, marital status, and religion of household head, urban residence, and region of residence (e.g. Ayantoye et al. 2011; Babatunde et al. 2007; Belachew et al. 2012;
73
Bigsten et al. 2003; Garrett and Ruel 1999; Hadley et al. 2008; Hanmer et al. 1999; Maxwell
1999; National Population Commission 2014; Omonona and Adetokunbo 2007; Ozughalu and
Ogwumike 2015; Sumarto et al. 2007). The employment status of the head of household is a strong predictor of per capita expenditure (Bigsten et al. 2003) and food insecurity (Babatunde et al. 2007). Asset ownership is positively associated with household consumption expenditure
(Booysen et al. 2008; Howe et al. 2008; Sahn and Stifel 2003) and food insecurity is more pronounced in households with lower economic resources than those with better economic conditions (Alaimo et al. 1998; Gundersen 2013; Oluwatayo 2009; Rose 1999).
Household size is related to poverty with larger households experiencing higher levels of poverty and food insecurity (Ajani et al. 2006; Amaza et al. 2006; Babatunde et al. 2007; Bigsten et al. 2003). Age of household head is negatively related to poverty and food insecurity
(Babatunde et al. 2007; Bigsten et al. 2003; Hanmer et al. 1999; Sumarto et al. 2007; Omonona and Adetokunbo 2007). Gender of the household head is critical indicator of poverty as households headed by women are more likely to be poor than those headed by men (Amaza et al.
2006; Belachew et al. 2011; 2012; Joshi et al. 2010). Family living arrangement is a strong predictor of household economic wellbeing (Coleman-Jensen et al. 2015; Manning and Brown
2006) and polygyny is related to household resource allocation (Tertilt 2005). Religion permeates nearly every aspect of life in Nigeria (Orubuloye et al. 1993) and Christians may be more likely to experience food insecurity than Muslims (Hadley et al. 2008). Urban poverty is easily linked to food insecurity (Maxwell 1999) because food prices are higher in the urban than in the rural areas and as such, city residents spend more on food, consume less, and purchase most of their food (Garrett and Ruel 1999). Nonetheless, poverty is mostly concentrated in the rural areas (Booysen et al. 2008; González et al. 2010; Hanmer et al. 1999). Lastly, there is a
74 huge spatial variation in poverty levels across the geopolitical regions in Nigeria (National
Population Commission 2014; Ogwumike 2001). The poverty rate in the south-west for instance, is more than three times the rate in the north-east region (The World Bank 2014a).
Data and Methods
I utilized panel data from the Nigeria General Household Survey (NGHS). The NGHS is a nationally representative annual survey of 22,000 households conducted as part of the Living
Standards Measurement Study-Integrated Surveys on Agriculture (LSMS-ISA). In 2010, the
NGHS was expanded to include a panel component that sampled 5,000 households out of the
22,000 core sample of the NGHS. Unlike its cross-sectional counterpart, the panel survey is biennial by design. However, the panel households are visited twice per wave of data collection with the two rounds corresponding to the post-planting and the post-harvest periods. Although the NGHS primarily aims to collect household-level agricultural-related statistics, the survey collects extensive information about household welfare and social behavior which could aid in the analysis of household socio-demographic characteristics in relation to health and wellbeing.
The panel NGHS is the first panel survey implemented by the Nigeria National Bureau of
Statistics. The LSMS-ISA team in the World Bank’s Development Research Group provides technical guidance in the design and implementation of the NGHS survey as well as assist with the analysis of the data. The survey was supported by various organizations including the Nigeria
Federal Ministry of Agriculture and Rural Development, the National Food Reserve Agency, the
Bill and Melinda Gates Foundation, and the World Bank. The first round of the panel survey was carried out in August-October 2010 (post-planting) and in February-April 2011 (post- harvest). The 2012 survey was administered between September and November 2012 (post- planting) and between February and April 2013 (post-harvest). More detailed information about
75 the design, implementation, and coverage of the NGHS are contained in a publication by the
Nigeria National Bureau of Statistics (National Bureau of Statistics 20152).
At all four visits to the households, the NGHS documented retrospective accounts of household food insecurity during the week preceding each interview, by the senior female or person most knowledgeable about household food consumption. In addition to its rich socio- demographic information, the NGHS questionnaire includes several questions about financial and social supports received by each member of the household. This study capitalizes on this wealth of data to analyze the role of social capital, education, and time use on the experiences of food insecurity among Nigerian households with children and elderly persons with a disability.
The panel nature of the NGHS allows for the analyses of both static and transitory experiences of household food insecurity. Compared to cross-sectional data like the Demographic and Health
Survey data, the panel NGHS proved advantageous in the analyses of household socioeconomic status in that it assessed food insecurity at four different time points (2010 post-planting, 2011 post-harvest, 2012 post-planting and 2013 post-harvest) thereby capturing any seasonal variations in sources of livelihoods within the study context.
Sample
Although the NGHS originally sampled 5,000 households, 4,997 households completed part or all of the questionnaire at round one. I further limited the sample for my cross-sectional and longitudinal analyses as described below.
Cross-sectional sample
Out of the initial 4,997 households interviewed in the 2010 NGHS, 222 households responded to five or fewer questions about their experiences of food insecurity and were
2 The data are also available for download through the World Bank’s Living Standard Measurement Study website (http://go.worldbank.org/BY4SLL0380).
76 excluded from my analytic sample. Also, I left out of my analyses nine households with missing information about age of household head, four households with no valid reports of educational attainment of household heads, and 43 households for which I could not estimate the values of their household assets. In order to maximize my sample size, I replaced missing education at wave one with the reported educational attainment at wave two for 23 heads of households and assigned a value of “zero” to 618 households with no valid time spent fetching or collecting wood. I controlled for households with missing data on time spent collecting wood with a dummy variable (1 if missing, 0 if otherwise). Following from the above, the analytic sample for my cross-sectional study of the effects of social capital, education, and time use on the relationship between household composition and household food insecurity is based on 4,719 households.
Longitudinal sample
Because the longitudinal part of my analysis seeks to examine both stability and change in the experiences of household food insecurity by household composition, I assessed food insecurity at all four NGHS interviews. Out of the 4,997 households interviewed in 2010, fifty households dropped out of the survey and were not included in my longitudinal analyses. Also, I excluded households with missing information on age (1) and education (2) of household head from the sample. For a household to be included in this part of my analyses, it needed to be observed at least two consecutive times. Therefore, I dropped 57 households that were observed only once from the analyses. The above restrictions left 4,887 households in the sample.
For my analyses of transitions into food insecurity, I further eliminated 966 households that were never food secure, 277 households that did not report their food security status at the survey round directly following their first reports of food security, and 28 households with
77 incomplete information about the values of their household assets. Similarly, the sample for my analyses of transitions out of food insecurity excludes 1,164 households that were never food insecure, 303 households with no records of food security status at the round following their first experiences of food insecurity and 30 households with unknown values of assets. Therefore, my analyses of transitions into food insecurity are based on 3,616 households and I analyzed transitions out of household food insecurity among 3,390 households.
Measures
All the variables in my cross-sectional analyses were drawn from the data collected at the first NGHS interview in 2010. In my analyses of transitions into and out of food insecurity
(longitudinal), I measured food security status at the time of survey directly following the initial food security status (first observation). The predictors were assessed at the time of the initial food security status (i.e. when a household was first food secure for transitions into food insecurity and when a household was first observed to be food insecure for transitions out of food insecurity). The only exception is social capital. The questions about use of informal assistance were asked only at the first (2010 postplanting) and third (2012 postplanting) survey rounds. Since most households in my analyses of transitions had their initial food security status observed at the first (63-68%) or third (10%) round of data collection, only a relatively small segment of my sample (21-27%) was affected by the unavailability of social capital measures at year 2. For those households, I lagged their social capital measures by one year (2010) to avoid temporal ordering problem. Remittances were also assessed at two time points but at the postharvest rounds (2011 and 2013). As in the other measures of social capital, I used lagged indicator of remittances.
78
Outcome variable
I utilized all nine items in the refined Household Food Insecurity Access Scale (HFIAS) which assesses the access component of household food security (Coates et al. 2007). Two of the items were slightly different from those in the HFIAS questionnaire. The first of these two items asked about a coping strategy used by households with children (restriction of adults’ meals to accommodate children’s nutritional demands) and the other one inquired about help-seeking behavior among households experiencing food insecurity (reliance on friends and relatives for food). Also, rather than asking about the frequency-of-occurrence of food insecurity separately from incidence of food insecurity as in the HFIAS questionnaire, the NGHS combines both incidence and frequency of household food insecurity in a series of questions asking respondents to report the number of days during which their households recorded certain occurrences of food insecurity. Nevertheless, the food insecurity questions in the NGHS were very similar to the ones in the HFIAS questionnaire, especially when used to categorize households along a continuum of severity of food insecurity as I did.
I employed two measures of household food insecurity in my analyses. The first measure captures the incidence of household food insecurity at a given point in time (2010). In the cross- sectional component of my analyses, I predicted the probability of being food secure, moderately food insecure, and severely food insecure by number of children and presence of adults with a disability in the household. My classification of Nigerian households into the above three categories of food insecurity (food secure, moderately food insecure, and severely food insecure) closely mirrors the official and well tested classification adopted by the Food and Nutrition
Technical Assistance III Project (FANTA) team in the US (see Coates et al. 2007).
79
For my longitudinal analyses, I pooled data from all the four NGHS rounds of data.
Based on respondents’ reports of household food insecurity status at the four NGHS interviews and using the classification of household food insecurity described above, I identified households as either food secure or food insecure at the time of their initial food security status. Food insecure households were either moderately or severely food insecure.
Focal predictor
My focal predictor is household composition assessed based on number of children aged
12 or younger and presence of adults aged 50 and above with a disability in the household. I examined food insecurity among households with: no children (reference), 1-2 children, 3-4 children, and 5 or more children. I also compared the food security status of households with one or more elderly persons with a disability to that of households with no such persons. I categorized individuals aged 50 and above as having disability if they reported difficulties with any one of five activities of daily living – “seeing, even if wearing glasses, hearing even if wearing hearing aid, walking or climbing steps, remembering or concentrating, and self-care.” I used aged 50 as my cut off point because of the relatively low life expectancy in Nigeria (53 years).
Focal variables
Social capital
My measure of social capital is based on access to financial support or remittances through informal social networks (networks of friends and families, informal savings groups, informal lending groups, and other informal institutions) by members of the household. The
NGHS asked respondents who were 15 years or older to report whether or not they experienced the following in the six months before they were interviewed: 1) “used any informal savings
80 groups (adashi/esusu/ajo) to save money”, 2) “used any informal groups (adashi/esusu/ajo) to borrow money”, 3) “used a cooperative, savings association or microfinance institution to save money ”, 4) “borrowed any money from friends, relatives or money lenders.” Respondents also reported whether or not within the last 12 months, they 5) “received money from relatives or friends living in or outside of Nigeria” or they 6) “received a monetary gift or an in-kind gift from abroad.” I used all six indicators of informal assistance in my cross sectional analyses. The
NGHS did not inquire about receipt of money from relatives or friends (item #5) at round 3 and
4. Therefore, I employed five measures of social capital in my longitudinal analyses. I used the items singly because they were weakly correlated (Cronbach alpha
Educational attainment of household head
Educational attainment of household head is based on a categorical measure that is coded as a series of dummy variables: no or forms of education other than formal education, primary education (reference), secondary education, and higher levels of education.
Time spent collecting cooking fuel
In the NGHS, all members of the household who were five years or older reported time spent collecting or fetching wood or cooking fuel the day before the interview. In this study, I summed the number of hours and minutes spent fetching or collecting wood in each household and used the resulting variable as a continuous predictor. Perhaps because they did not use wood for cooking, some households did not report time spent collecting wood. I included a binary variable to account for these households in my models.
Control variables
I included other covariates linked to food insecurity and poverty – socioeconomic measures (occupational status of household head and household wealth); demographics of the
81 household (household size, age, gender, marital status, and religion of household head) and context (urban residence, and region of residence).
Socioeconomic measures: Occupation status of the household head is based on reports of employment activities within seven days preceding the survey. The questions asked whether or not the head of households: 1) worked for someone who was not a member of their households,
2) worked on a farm owned or rented by a member of their households, and 3) worked on their own account or in a business enterprise belonging to them or someone in their households.
Household heads who reported engaging in any of the three work categories were considered employed and were compared to their unemployed counterparts in their experiences of household food insecurity. In a follow up question, respondents were asked to report the sector of their primary occupations. I combined the reports of employment status and the specific occupations to create the following employment categories: unemployed, agriculture (reference), sales and services, professional jobs, and others.
In the NGHS, the household head provided information about ownership of a series of assets such as radio, television set, generating set, fridge, etc., by his/her household. The survey also collected information about the number and the estimated value of each asset reported in the household. Therefore, I was able to combine information about the number and the values of all the assets in the household to create a novel measure of household wealth – the total values of assets in each household measured in continuous naira (currency). I logged the reported values of assets to reduce the high level of skewness in the variable.
Household demographics: The NGHS respondents, including household heads, reported their ages in years. Gender of the household head is coded as one for males and zero for females.
82
Context indicators: Urban status is based on location of a household in an urban enumeration area (EA) as defined by the Nigeria Census. I coded one for urban, zero for rural.
There are six geopolitical regions in the analyses and they are coded as a series of dummy variables: north-central, north-east, north-west, south-south, south-east and south-west, with south-west as the reference category.
Analytic strategy
First, I describe the different compositions of Nigerian households in terms of number of children and presence of older adults with a disability in the household, and the food security statuses of those households in 2010. Implicit in the food security paradox (Balistreri 2012) are certain factors modifying the experiences of food insecurity among households with vulnerable populations. In an attempt to understand the factors buffering the experiences of food insecurity among households with children and elderly persons with a disability, I examine the role of social capital, education, and time use on the relationship between household food insecurity and number of children and presence of adults with a disability in the household in a series of multinomial logistic regression models. The first model includes just the indicators of household composition. I then controlled for other household socioeconomic and demographic characteristics in Model 2. I tested for mediating effects of informal social networks, education, and time spent collecting cooking fuel on the association between household composition and food insecurity in Models 3, 4, and 5 respectively. The final model includes all variables in the analyses (Model 6). I further interacted measures of social capital, education, time use, and occupation of household head with the indicators of household composition and presented the results as predicted probabilities as figures. The interactions show how each of the predictors
(social capital, education, time use, and occupation of household head) modify the experiences of
83 household food insecurity among households with varying number of children and by presence of elderly persons with a disability in the household.
In the longitudinal analyses, I described the distribution of persistent and transitory food insecurity status among households with varying number of children and based on presence of older adults with a disability in the household between 2010 and 2013. I presented the share of households of different compositions that were: persistently food secure, persistently food insecure, transitioned into food insecurity, and transitioned out of food insecurity. In the multivariate analyses, I predicted the risks of transitioning into and out of food insecurity among previously food secure and previously food insecure households using logistic regression technique. First, I estimated a model with the measures of household compositions (Model 1) and then a series of models predicting transitions into food insecurity among food secure households and transitions out of food insecurity among food insecure households. Model 2 includes the indicators of household composition and other household sociodemographic characteristics. I added the measures of social capital, education, and time use in Models 3, 4, and 5 respectively. Model 6 includes all the variables in the analyses.
Results
I analyzed the relationship between presence of vulnerable populations in Nigerian households and experiences of household food insecurity using both the cross-sectional and longitudinal study designs. In the cross-sectional analyses, I examined variations in moderate and severe household food insecurity by presence of children and older adults with a disability in 2010.
Table 3. 1 describes the food security status and sociodemographic characteristics of all households and six categories of households based on number of children and presence of older adult with a disability. These are households with: no children, 1-2 children, 3-4 children, 5 or
84 more children, no adult with a disability, and any adult with a disability. I defined presence of children as having children aged 0-12 in the household. For simplicity, I hereafter describe households as having no children versus having a varying number of children. Half of all households in Nigeria were food insecure in 2010. But the rate of food insecurity varied by household composition. It appears as if households with more children were more protected against the risks of food insecurity than smaller ones. Household food insecurity was most prevalent in households with 1-2 children and least common among households with five or more children. In terms of disability, having one or more older adults with a disability in the household was associated with higher risks of severe food insecurity. Whereas only about a quarter of households with no older adults with disability reported being severely food insecure, nearly two-fifths (37%) of households with such a disabled older household member were severely food insecure.
The use of informal sources of financial capital was common in Nigeria. Members of about a third of households used informal savings, 23% received money from relatives and friends, 24% borrowed from friends, relatives, and money lenders, 14% used other informal groups to borrow money, and 10% used cooperative, savings, association or micro finance in
2010. Remittance from abroad was the least common informal source of capital, 2.5% of households received remittances from friends or relatives abroad. The share of households receiving money, gifts, and remittances declined with increasing number of children while the likelihood of borrowing money from friends, relatives, money lenders, and other informal groups increased with increasing number of children in the household. The distinction in the use of informal savings was between households with no children, which had a smaller share using informal savings group, and households with children. On the other hand, the use of cooperative,
85 savings association and micro finance was comparable across households except that it was rarely used among households with five or more children. Larger shares of households with older adults with a disability received remittances or money from relatives and friends. The use of other forms of informal financial assistance was not as common among households with older adults with a disability as among other households.
More than one-third (36%) of Nigerian households were headed by persons with no formal education and only 15% of household heads attained postsecondary levels of education.
Households having 1-2 children were headed by the most educated individuals while the heads of households with five or more children were the least educated. Also, households with older adults with a disability (56%) more often had an uneducated head than households with no adult with disability (34%). On average, each household spent about an hour and a half collecting wood or other cooking fuel the day before the interview. As expected, the larger the number of children in the household, the greater the time spent gathering cooking fuel. There was, however, little disparity in time spent obtaining cooking fuel between households that had elderly persons with a disability and those that did not. Perhaps because they did not use wood as cooking fuel, more households with four or fewer children than those with five or more children did not report time spent on wood collection.
About one in every ten households was headed by an unemployed adult. Excluding housing, the average Nigerian household had assets worth about ₦170,000 ($854) in 2010. In spite of their educational disadvantage, households with the largest number of children (5+) had the most valuables, followed by households with 1-2 children, and then households with 3-4 children. ‘Childless’ households appear to be the poorest in terms of household wealth in
Nigeria. This suggests that large family size may not be a status symbol for many households in
86
Nigeria. Households having older adults with disability were less endowed than those with no adult with a disability; they had 40% lower household wealth.
The heads of the households were about 50 years of age. Sixteen percent of households had female heads. There is a clear gender divide between households with children and those with no children. Having a female head is mostly an attribute of households without children and to some extent those with 1-2 children. Nearly all (97%) of households with five or more children had male heads. In the same vein, households with one or more older adults with a disability were twice as likely to have female heads as those with no such population.
Though Nigerian households were largely monogamous, as many as 17% of household heads were in polygynous relationships. Polygyny is linked to large household size. Compared to only 12% of households with 1-2 children, 48% of households with five or more children were polygynous. Yet, monogamy is more prevalent among households having five or more children than polygyny. In this study, ‘childless’ households were mostly unmarried households. There were slightly more Christian than Muslim households in this study. Perhaps reflecting the north- south divide in fertility decline in Nigeria (Mberu and Reed 2014), there was greater representation of Islam among households with five or more children than Christianity.
Conversely, Christianity was more widely practiced than Islam among households with no children and those with 1-2 children.
The sample had more rural than urban households. The greater the number of children present in a household, the more likely it was based in the rural area. The vast majority (76%) of households with five or more children were in the rural parts of the country. There were slightly more rural dwellers among households that reported one or more older adult with a disability than those without adults with a disability. Except for the slightly larger shares of Northwestern
87 and Southwestern households in the sample, the NGHS had a fair representation of households in all the six geopolitical zones in the country. Households with the largest number of children (5+) were mostly northern households, 83% were based in the north. In contrast, the households with no children were predominantly southern households.
Next, I examined the effects of number of children and presence of older adults with a disability in the household on household food security in a series of multinomial logistic regression. Table 3. 2 shows the relative risks of moderate and severe food insecurity, relative to being food secure, among households with varying numbers of children and one or more adult with functional limitations. As shown in Model 1, households with three or more children had significantly lower risks of moderate food insecurity but those with 1-2 children were more likely to be severely food insecure than ‘childless’ households. However, the bivariate relationship between number of children and food insecurity described above was due to the suppression effect of region of residence. Households in the northern regions of Nigeria had higher fertility rates but lower rates of food insecurity. Once I accounted for regional variations in levels of food insecurity, I found a significant positive association between number of children in the household and severe food insecurity. Controlling for differences in socioeconomic and demographic characteristics across households in Model 2, the risks of severe food insecurity significantly increased with increasing number of children in the household. Although the relationship between number of children and food insecurity varied across regions, ‘childless’ households were at significantly lower risks of severe food insecurity across regions (results not shown). Also, in four regions – North east, North west, South east, and South south – households with 1-2 children appear more food secure than those with larger number of children (results not shown). Presence of adults with a disability in the household was indeed a major risk factor for
88 food insecurity. Irrespective of their sociodemographic characteristics, households with one or more adults with a disability were significantly more prone to severe food insecurity.
Other significant predictors of food insecurity in Model 2 are: occupation of household head, household wealth, polygyny, urban residence, and region of residence. Households headed by professionals had 29% lower risks of severe food insecurity. Having an unemployed head was associated with significantly lower risks of moderately food insecurity. The larger the value of assets in the household, the lower the risks of both moderate and severe food insecurity.
Polygynous households were significantly more likely to be moderately food insecure than monogamous households but marital status of household head was not a significant predictor of severe food insecurity. Food insecurity was significantly more prevalent in the urban centers than in the rural areas. Urban households were twice as likely as their rural counterparts to be severely food insecure. Compared to the South west, households in the North central and the North west were significantly less likely to be moderately food insecure. On the other hand, both the South east and the South south were at significantly greater risks of food insecurity than the South west. Of the three northern regions, only the North west had significantly lower risks of severe food insecurity than the South west.
In Model 3, I examined the role of social capital in the experiences of food insecurity among households with and without children and elderly persons with a disability in Nigeria.
Even though the addition of social capital to Model 2 added significantly to the fit of the model
(χ2 (12) = 137.68, p = 0.0000), social capital did little to explain the heightened risks of severe food insecurity among households with children and those with older adults living with a disability. The effect of informal financial support on the risks of household food insecurity depends on the type of support. As expected, remittances were associated with lower risks of
89 food insecurity but the coefficients were not significant. Contrary to my prediction, reliance on informal groups for sustenance did not ameliorate the experiences of food insecurity. Households whose members received money from relatives or friends and those that borrowed from informal groups had higher risks of food insecurity than those that did not receive such assistance. The use of informal savings group reduced the risks of severe food security by 3% but the effect was nonsignificant.
Education, a proxy for food management strategies, similarly did not mediate the relationship between number of children and presence of adults with a disability and household food insecurity. However, education did significantly improve model fit in contrast to Model 2 (χ
2 (6) = 20.8, p = 0.002). Primary level of education predicted significantly higher risks of food insecurity than no formal education (Model 4).
Model 5 shows that time use significantly reduced the effects of having children on the risks of severe food insecurity. By adding time use to Model 2, the effect of 1-2 children on the risk of severe food insecurity, relative to no children, was reduced by 7 percentage points (p =
0.0024); the effect of having 3-4 children was reduced by 11 percentage points (p = 0.0008); and the effect of having five or more children was reduced by 17 percentage points (p = 0.0001).
However, number of children remains significantly associated with severe food insecurity. Time spent collecting fuel was associated with significant increase in food insecurity. For every additional percentage point minutes spent collecting firewood, the risks of moderate food insecurity increased by 6% and the risks of severe food insecurity increased by 10% (Model 5).
Time use did not significantly explain the heightened vulnerability to severe food insecurity among households with adults with a disability.
90
In the final model (Model 6), presence of children and older adults with functional limitations in the household remained significantly associated with higher risks of food severe food insecurity net of all the other predictors in the model. The effects of household wealth, marital status of household head, urban residence and region of residence reported in Model 2 persisted across models. Similarly, the coefficients of the indicators of social capital, education, and time use in the final model resemble those reported in Models 3, 4, and 5 respectively.
However, the lower risk of severe food insecurity among households headed by professionals was no longer statistically significant in the final model.
I examined the extent to which the experiences of household food insecurity vary by access to financial support through informal networks, education, and time use across households with varying compositions in terms of presence of children and older adults with a disability. The results of the interactions between number of children and measures of social capital, education, and time use revealed that only time spent fetching wood significantly modified the experiences of food insecurity among households with varying number of child dependents. As shown in
Figure 3. 1, irrespective of the number of children in the household, the shorter the time it took to fetch cooking fuel, the lower the likelihood of being severely food insecure. Also, it appears as if the protective effect of having smaller family size on household food insecurity is minimal in the absence of adequate access to cooking fuel. Among households that spent limited amount of time fetching cooking fuel, those with two or fewer children had lower predicted probabilities (based on Model 6 with interaction terms) of being severely food insecure relative to those with three or more children. However, as the time spent on wood collection increased, the gap in the risks of food insecurity across households with different number of children diminished, net of other household sociodemographic characteristics. Social capital, education, and time use had
91 nonsignificant moderating effects on the association between household food insecurity and presence of older adults with a disability.
I found no significant differences in the relationship between food insecurity and number of children in households across occupational categories. As shown in Figure 3. 2, ‘childless’ households headed by agriculturists, sales and service agents, and professionals were similarly less likely to be severely food insecure than their counterparts with children. However, among households with children, number of children seems to matter more among households headed by persons in sales and services as well as professional jobs than among agricultural households.
This suggests that having five or more children as opposed to 1-2 children make little difference in the experiences of food insecurity among agriculturists.
Household composition and transitions in and out of food insecurity
Next I employed the NGHS panel data collected between 2010 and 2013 to examine the patterns of stability and change in household food security status. Table 3. 3 presents the share of households that reported being persistently or transitorily food (in)secure by number of children and presence of older adults with a disability in the household. The results demonstrate the importance of longitudinal analytic techniques in understanding food insecurity in contexts like
Nigeria. Less than one-quarter of all households were persistently food secure over the three year period and only a fifth of the households were persistently food insecure. A large proportion
(55%) of households experienced transitory food security.
The notable differences in food security transitions among households with varying number of children were related to persistent rather than transitory food insecurity. Larger share of households with five or more children were persistently food secure than other households. In contrast, persistent food insecurity was more pronounced among households with four or fewer
92 children than those with five or more children. Similarly, the major disparity in the experiences of food insecurity between households with older adults having one or more disability and those without disabled elders was found in persistent rather than transient food insecurity. Households reporting adults with a disability were mostly persistently food insecure. Slightly fewer households with older adults with a disability than those without older persons with a disability transitioned into and out of food insecurity over time.
Transitions into food insecurity by number of children and presence of older adults with a disability in the household
I describe the samples for my analyses of first transitions into food insecurity among previously food secure households in Table 3. 4. As shown in the table, of all the households that were food secure at their initial observations, about half (51%) slid into food insecurity within two years. The proportion of food secure households transitioning into food insecurity declined slowly with increasing number of children in the household. However, more households including older adults with a disability (56%) than those with no such persons (51%) became food insecure over time. About three percent of initially food secure households received remittances; nearly two-thirds used informal savings group; 11% used cooperative, savings association, or micro finance institution; 14% borrowed from informal groups; and 22% borrowed from friends, relatives, and money lenders. Borrowing was more common among households with larger number of children than among those with two or fewer children. Also, the use of informal savings groups increases with increasing number of children in the household. However, receipt of remittances and the use of cooperative, savings, or micro finance were not all that common in households with large number of children (5+). Besides remittances,
93 the use of informal financial agents was less common among households with adults having a disability than among those with no adult experiencing a functional limitation.
Heads of households with the largest number of children (5+) appeared to be less educated than those having fewer children. Likewise, only two-fifths of households with vulnerable adult population, compared to two-thirds of those with no older adults with a disability, were headed by persons with formal education. On average, it took the food secure households over an hour to collect cooking fuel the day before their interviews. The larger the number of children in a household, the longer it took to collect cooking fuel. Food secure households with one or more adult with a functional limitation spent more time fetching cooking fuel than their counterparts without adults with a disability. The rate of unemployment among household heads with one or more older adults with a disability was five times the rate among those with no adult with a disability.
A typical food secure household in Nigeria had assets worth about ₦200,000 but the values of assets varied by number of children and presence of older adults with a disability in the household. I found the largest average value of asset among the largest households (5+ children) and the smallest values among households with no children and those with 3-4 children. This suggests that the relationship between household wealth and household size is not linear.
Households with no elderly persons with a disability were wealthier than those with one or more adults in need of functional assistance. As many as one in seven previously food secure households had a female head. Households with five or more children and to some extent, those with 3-4 children, were disproportionately headed by men. Relative to those with no older adult with a disability, more households with older persons with some forms of disability had female heads. Nigerian food secure households were predominantly headed by married monogamous
94 persons. However, the larger the size of a household in terms of number of children, the greater the likelihood that it was polygynous; more than half of food secure households with five or more children had polygynous heads. Nigerian households were divided between Islam and
Christianity but there were more Muslims among heads of households with five or more children and more Christian heads in households with no children. Christianity was more widely practiced in households with older adults with a disability than Islam. The households in my analyses of transitions into food insecurity were mostly based in the rural areas. The majority of households with large number of children were rural households. I found no profound rural-urban distinction among households with and without adults with a disability. The vast majority (84%) of food secure households with five or more children in Nigeria were Northerners and ‘childless’ households were mostly located in the southern regions. Twenty nine percent of the sample was observed three times before transitioning into food insecurity. The number of food secure episodes varied little with number of children and presence of older adults with a disability in the household.
In the multivariate analyses, I examined differences in the odds of transitioning into food insecurity by number of children and presence of older adults with a disability in the household.
The results of the first model presented in Table 3. 5 showed significantly lower risks of sliding into food insecurity among households with children than among ‘childless’ households.
However, food secure households with children, relative to those with no children, appeared protected against food insecurity simply due to their sociodemographic characteristics, particularly their concentration in the northern part of the country where the prevalence rate of household food insecurity was modest. As shown in Model 2, controlling for other household sociodemographic characteristics, households with 3-4 children had 7% higher odds of
95 transitioning into food insecurity and those with five or more children had up to 9% elevated odds of sliding into food insecurity than ‘childless’ households. There was no significant difference in the odds of becoming food insecure among food secure households with and without adults with a disability.
Contrary to my predictions, use of informal savings was associated with significantly higher odds of transitioning into food insecurity. But food secure households that received remittances and those that used informal groups to borrow money had lower odds of becoming food insecure (Model 3). Also, higher levels of education among household heads had significant protective effects against the risks of sliding into food insecurity (Model 4). For every minute increase in time spent fetching cooking fuel, the odds of becoming food insecure increased significantly by 4% (Model 5).
In the full model (Model 6), initially food secure households in Nigeria had similar odds of transitioning into food insecurity irrespective of the number of children they had or whether or not they included adults with a disability. Informal savings group remained significantly related to heightened odds of food insecurity while highly educated household heads had only two-thirds of the odds of food insecurity as those with primary education. The longer it took a food secure household to fetch cooking fuel, the greater its risk of sliding into food insecurity. Agricultural households had similar odds of transitioning into food insecurity as other households with employed heads but unemployment of household head predicted 25% higher chances of sliding into food insecurity. Greater values of assets in the household were linked to a significantly greater likelihood of persistent food security. Urban households were significantly more prone to transitioning into food insecurity than rural households. Households in northern parts of Nigeria were more persistently food secure than southwestern households. Food secure households in the
96
South east had nearly thrice the odds of becoming food insecure as those in the South west even after controlling for sociodemographic variations across households. South south and South west had similar odds of transitioning into food insecurity. The odds of becoming food insecure increased significantly from the second to the third observations but households that were food secure at three survey rounds were significantly less likely to transition into food insecurity by the fourth round of data collection. The significant increase in the odds of moving into food insecurity between the second and the third observations could result from the increase in food insecurity in Nigeria between 2011 and 2012 (period effect) rather than the duration of time since food security.
Transitions out of household food insecurity by number of children and presence of older adults with a disability in the household
Table 3. 6 presents the summary statistics on persistence of food insecurity and sociodemographic characteristics of initially food insecure households in Nigeria. Of all the households that were food insecure and observed at least twice between 2010 and 2013, 56% transitioned out of food insecurity. The proportion of households that escaped food insecurity increased with increasing number of children. However, households with adults having a disability less often transitioned out of food insecurity than households without elderly persons with a disability. Aside their higher propensity to borrow from friends, relatives, and money lenders, food insecure households used informal financial aids to the same degree as food secure households (Tables 3. 4 and Table 3. 6). A substantial share of food insecure households (35%) used informal savings groups and one quarter borrowed money from friends, relatives, or money lenders. But only three percent received money or gifts from friends and relatives abroad.
Remittances seem very uncommon among food insecure households with five or more children.
97
The use of informal savings group was most widespread among households with 1-2 children and least common among ‘childless’ households. The more the children in a household, the greater its reliance on borrowing from informal sources. Informal savings and informal borrowing were both more prevalent among households with no vulnerable adult population than among households having adults with a disability. But households with functionally limited adults were more than twice as likely as other households to receive remittances.
Reaffirming the fact that food insecure households in Nigeria are not uneducated households, 14% of food insecure households in the sample were headed by highly educated persons. Heads of households with five or more children were the least educated while households with 1-2 children seem to have the most educated heads in the sample. More than half of households reporting one or more older adult with a disability were headed by uneducated individuals; only 30% of other households had uneducated heads. Food insecure households in
Nigeria spent close to an hour and a half (84 minutes) per day fetching cooking fuel; average time spent collecting wood among food secure households was 76 minutes (Table 3. 4). Like in food secure households (Table 3. 4), the more children there were in the household, the longer it took to collect wood. However, unlike food secure households, food insecure households with any adults having a disability spent less time (78 minutes) collecting wood than those with no adult with a disability (85 minutes). A relatively small share of households with five or more children (3%) had missing information on time spent collecting wood. It could be that these households were more likely than households with fewer children to rely on wood for cooking.
The majority (89%) of the heads of food insecure households were not unemployed around their food insecurity episodes. Unemployment rate in ‘childless’ households (20%) was almost twice the national average but only a minority (4%) of households with five or more
98 children had unemployed heads. Unemployment was also highly prevalent among households with older adults with a disability - two-fifths were headed by unemployed persons. In general, food secure households described in Table 3. 4 were wealthier than their food insecure counterparts (Table 3. 6). The former had assets worth over ₦200,000 while the latter owned assets of about ₦137,000 on average. Food insecure households with no children were particularly poor when compared to the rest of the sample. Households having adults with a disability were worth only a fraction (64%) of the value of wealth in other households. The average age of heads of food insecure households was 50 years. A relatively large share of food insecure households (18%), compared to 14% of food secure households (Table 3. 4), were headed by women. Female heads were more common among ‘childless’ households and households with older adults experiencing functional limitations than in the country as a whole.
Fifteen percent of food insecure households were polygynous. The food insecure households in this study were predominantly Christian households. However, there were more Muslims than
Christians among food insecure households reporting five or more children. Christianity was also more widely practiced among food insecure households with adults having disability than in the nation as a whole.
More households with larger number of children were based in the rural areas than those with two or fewer children. Likewise, more food insecure households with older adults with a disability were in the rural areas than households without adults with a functional limitation.
Northcentral and Northeastern households constitute relatively small share (9-10%) of the food insecure households. Although they made up only a small segment (9-15%) of food insecure households in the sample, Northeastern and Northwestern households were the most represented among large households in Nigeria. They jointly constituted 60% of food insecure households
99 with five or more children. The major regional difference by presence of older adults with a disability is the larger share (32%) of Southeastern households among households having one or more adult with a disability, relative to their counterparts with no such persons (16%). The greater the number of children present in the households, the more quickly they transitioned out of food insecurity but it took longer to escape food insecurity when older adults with a disability were present in the household than when they were not. Nearly half (45%) of food insecure households with vulnerable adult never escaped food insecurity over the study period.
Table 3. 7 displays the results of logistic regression models predicting the odds of transitioning out of food insecurity among food insecure households in Nigeria by number of children and presence of any adult with a disability in the household. The results of Model 1 showed significantly higher chances of escaping food insecurity among households with three or more children, relative to those with no children. However, the direction of the association between food insecurity and number of children in the household was reversed once I accounted for disparities in household wealth, gender of household head, and region of residence in Models
2-6. Having one or more adult living with a disability in the household was associated with significantly lower odds of transitioning out of food insecurity (Model 2).
Among the five indicators of social capital in this analysis, only informal borrowing significantly predicted transitions out of food insecurity. Households whose members borrowed from friends, relatives, and money lenders were significantly less likely to escape food insecurity
(Model 3). In Model 4, primary education was associated with lower odds of transitioning out of food insecurity than secondary and higher education. The chances of transitioning out of food insecurity significantly reduced with increasing time spent collecting wood (Model 5).
Controlling for the longer time spent fetching cooking fuel among households with more
100 children nearly eliminated the significant differences in transitions out of food insecurity across households with varying numbers of children.
According to the results presented in the final model (Model 6), it appeared as if the significant differences in persistent versus transitory food insecurity across households in Nigeria were due to factors other than the number of children in the households. Only households with five or more children remained at marginally significantly lower risks of moving from food insecurity to food security net of other predictors. In contrast, households with one or more adult with a functional limitation in their activities of daily living were significantly more likely to remain food insecure than households without such persons. This effect was only partially mediated by socioeconomic and demographic variations across households. Given similar access to and/or use of social capital, socioeconomic status, and time spent collecting wood, as well as comparable demographic features as other households, households with older adults with a disability were still at significantly greater risks of persistent food insecurity (Model 6).
The effects of the social and demographic measures changed little across models.
Households whose heads were employed in sales and services or unemployed were significantly more persistent in their food insecurity episodes than agricultural households. The more endowed food insecure households were, the greater than chances of becoming food secure. Age, gender, marital status, and religion of household heads were not significant predictors of transitions out of food insecurity. Urban households were significantly more likely to be persistently food insecure than rural households. Northern households seemed to be better equipped to transition out of food insecurity than southern households. In the south, the South east and the South south regions had significantly lower odds of transitions to food security from
101 food insecurity before and after controlling for other variables in the analyses. Net of other covariates, the odds of transitioning out of food insecurity declined significantly over time.
Discussion
Food security research in the developed world (e.g. Alaimo et al. 1998; Anderson et al. 2014;
Coleman-Jensen et al. 2014) has made the distinctions in the experiences of food insecurity among households with children and those with elderly persons. But analyses of the effects of children and older adults on food security status of households in the developing countries (e.g.
Babatunde et al. 2007; Miller et al. 2011; Obayelu 2010; Omonona and Adetokunbo 2007) still rely on dependency ratio which is now a contested measure of household socioeconomic dependency (Sanderson and Scherbov 2015). Also, whereas household socioeconomic status is negatively related to fertility in the developed world (Lovenheim and Mumford 2013), large family size is often viewed as a status symbol or as a form of insurance against adverse socioeconomic situations in sub-Saharan Africa (Boserup 1985). The present study examined how compositional differences in households, based on number of children and presence of older adults with a disability, relate to food insecurity. I found significantly higher risks of food insecurity among households with one or more older adults with a disability and those with children, relative to households with no such populations.
The findings of this study call for a rethinking of large family size as a status symbol in
Sub-Saharan Africa. A large share of Nigeria’s population may have lacked adequate incentives to reduce fertility, but food insecurity is a widespread impediment to having large families. As shown in this study, whatever proceeds are realized through child labor are obviously insufficient in offsetting the heightened risks of severe food insecurity among households with large number of children. Contrary to the general belief in divine provision for parents, having minors to feed
102 increases the burdens of household consumption and, therefore, the risk of food insecurity. The postulation (based on the demographic transition theory) that a large family may be instrumental to farm labor and therefore household socioeconomic status does not hold in this study.
Regardless of the occupation of household head, having children in the household predicted higher risks of severe food insecurity.
It is unclear how presence of children relate to household socioeconomic wellbeing in
Nigeria. Households with children may be selected based on certain characteristics that are related to socioeconomic wellbeing. In the U.S., people who reside with children compared to those with no children in the home have more socioeconomic resources, are more religious, and may have other conditions that are favorable to wellbeing (Deaton and Stone, 2014). However, in Nigeria, polygyny (which is a risk factor for larger family size) is widely practiced across the population, among both the rich and the poor (Tertilt 2005). Therefore, while this study builds on the resource theory in conceptualizing the presence of large number of children in a household as a vulnerable status, it acknowledges possible selection effects where economically advantaged households are more likely to be polygynous and therefore larger in size than monogamous ones.
Also, it is possible that limited household resources are diverted from food to caregiving in households with older adults with a disability or food insecurity contributed to adult disability.
Although I could not ascertain the causal relationship between household composition and food insecurity, the longitudinal design of this study suggests that presence of older adults with a disability in the household do shape the experiences of household food insecurity. Households that included one or more persons with a functional limitation were at significantly greater risks of persistent food insecurity.
103
My analyses provided only partial supports for the hypothesis that inadequate access to social support would explain some of the gap in the experiences of food insecurity between households with vulnerable populations and those without vulnerable populations. Though remittance was negatively related to food insecurity, financial support from friends, relatives, and money lenders were associated with higher, rather than lower, risks of food insecurity. In fact, borrowing from friends, relatives, and money lenders is a strong indicator of persistent food insecurity. It is plausible that the kinds of households using the forms of informal supports examined in this study were embedded within social networks that had too limited resources to significantly alleviate the experiences of food insufficiency among their peers.
Education did not have the expected strong protective effects on household food insecurity. Education did predict transitions out of food insecurity, but it might take a secondary or higher level of education for a household to exhibit significantly higher odds of transitioning out of food insecurity. But as expected, the longer it took a household to collect cooking fuel, the higher the risks of food insecurity. Also, as hypothesized, time spent collecting cooking fuel significantly reduced the effects of having children on the risks of severe food insecurity. The protective effect of having fewer children on household food insecurity declined with increasing time spent fetching cooking fuel. Having a professional household head predicted lower risks of severe food insecurity. Compared to agriculture, unemployment and sales and services were associated with lower odds of exiting food insecurity episodes. Households with large valuables were significantly less susceptible to food insecurity. Also, the more endowed a household was in terms of assets, the more likely it would transition out of food insecurity. My results suggest a pattern of urbanization of food insecurity in Nigeria; urban households were significantly more likely to transition into and had significantly lower chances of exiting food insecurity than rural
104 households. Northern households were generally more food secure than households in the southern parts of the country. Among the three geopolitical zones in southern Nigeria, households in the southwestern region appeared to have better food conditions and better able to escape food insecurity than those in the South east and the South west, net of differences in household sociodemographic characteristics.
In spite of the contributions of this study to understanding household structure and food insecurity, there are few limitations. Many of the measures used (e.g. food insecurity, household assets, social capital, and time spent collecting cooking fuel), even though they relate to individual members of the household, may vary across household members depending on patterns of resource allocation in each household. Second, social capital and disability were assessed at only two survey rounds and due to data limitations, the measures of social capital are mostly based on financial assistance (or financial need), leaving out other forms of social capital that might be crucial to the experiences of household food insecurity. More so, future analyses of social support in developing contexts should further explore the attributes of the members and/or the beneficiaries of the different informal groups included in this study as well as the circumstances that maximize or limit their impacts on poverty alleviation. Third, I analyzed two household compositional factors – presence of children and older adults with a disability – measured at the time of survey. Future studies should expand the list of vulnerable populations to include pregnant women, lactating mothers, and unemployed adults as well as change in household composition over time. Lastly, I used education as a proxy for food management strategies and time spent collecting cooking fuel as a proxy for overall time use in the household.
Future research should assess the effects of differential food management skills and other aspects of time use on household food insecurity.
105
Despite the above limitations, this study highlights the struggle for survival among households with two vulnerable populations (children and elderly persons with disability) in
Nigeria. The need for food and nutrition assistance programs in Nigeria is apparent. Programs such as cash transfers to poor households have been proven to significantly improve the food conditions of households in Malawi (Miller et al. 2011). Research has also shown tremendous benefits of food stamps in reducing food insufficiency (see Rose 1999). Further, as the findings of this study reveal, poverty alleviation efforts will be better able to improve household food conditions by assisting poor households with easily accessible cooking fuel.
106
Table 3. 1. Weighted Descriptive Statistics by Number of Children and Presence of Older Adults with a Disability in the Household Older adult with a Number of children in household disability in the household Five or No adult Has adult All No 1-2 3-4 more with a with a Variables households children children children children disability disability Household food security Food secure 49.2 47.6 47.2 49.4 57.8 50.2 39.5 Moderately food insecure 23.9 27.5 24.1 21.3 19.5 23.9 23.8 Severely food insecure 27.0 24.9 28.8 29.3 22.7 25.9 36.7 Use of social capital Received money from relatives or friends No 76.5 65.4 75.9 85.8 88.0 78.5 58.7 Yes 23.5 34.6 24.1 14.2 12.0 21.6 41.3 Received remittance from abroad No 97.5 96.8 97.1 98.2 99.1 97.7 95.5 Yes 2.5 3.2 2.9 1.8 0.9 2.3 4.5 Used informal savings group No 68.3 73.1 66.0 66.3 66.2 67.0 79.6 Yes 31.7 26.9 34.0 33.7 33.8 33.0 20.4 Used cooperative, savings association or micro finance No 90.4 90.5 88.9 90.2 94.9 90.0 94.1 Yes 9.6 9.5 11.1 9.8 5.1 10.0 5.9 Used informal group to borrow money No 86.0 89.3 87.4 82.7 80.5 85.8 87.9 Yes 14.0 10.7 12.6 17.3 19.5 14.2 12.1 Borrowed money from friends, relatives, or money lenders No 76.5 83.7 76.2 72.3 67.6 75.9 81.6 Yes 23.5 16.4 23.8 27.7 32.4 24.1 18.4 Education of household head No or other education 36.2 37.8 31.9 34.4 47.6 34.1 56.0 Primary education 25.8 25.3 26.6 26.3 24.0 26.3 21.2 Secondary education 22.7 19.9 25.2 25.2 17.6 24.1 9.6 Higher education 15.4 17.0 16.3 14.2 10.8 15.6 13.2 Time taken to collect wood in minutes 90 (256) 51 (144) 90 (228) 111 (302) 147 (393) 90 (260) 87 (213) Logged time taken to collect wood 2.6 (2.3) 2.0 (2.1) 2.6 (2.2) 2.9 (2.3) 3.5 (2.2) 2.6 (2.3) 2.6 (2.2) Missing time spent collecting wood No 86.0 79.3 86.0 89.6 96.2 85.9 87.3 Yes 14.0 20.7 14.0 10.4 3.8 14.1 12.7 Occupation of household head 107
Unemployed 9.6 19.1 7.4 4.5 2.0 6.6 37.1 Agriculture 43.9 34.7 40.7 50.1 63.6 44.8 35.5 Sales and services 25.2 26.9 29.4 20.7 18.2 25.9 18.4 Professional jobs 13.3 12.2 13.1 16.3 11.3 14.2 5.7 others 8.0 7.2 9.4 8.4 5.0 8.5 3.3 Naira value of assets 170816 110998 212653 134219 275250 177772 106646 (1135149) (419932) (1691176) (333064) (1418893) (1189132) (355657) Logged value of assets 10.6 (1.6) 10.2 (1.7) 10.7 (1.6) 10.8 (1.5) 11.0 (1.5) 10.6 (1.6) 10.0 (1.8) Age of household head 49.8 56.8 48.3 44.9 45.9 48.0 67.1 (15.5) (18.1) (14.9) (11.1) (10.0) (14.4) (13.9) Gender of household head Male 84.2 70.4 85.6 93.4 97.0 85.7 70.5 Female 15.8 29.6 14.4 6.6 3.0 14.3 29.5 Marital status of household head Married polygynous 17.1 8.5 11.8 19.6 48.2 17.4 14.8 Married monogamous 61.5 44.7 72.4 74.0 49.5 62.5 52.2 Unmarried 21.4 46.9 15.8 6.4 2.3 20.2 33.0 Religion of household head Muslim 43.2 28.8 40.4 48.7 75.9 44.0 35.7 Christian 55.2 69.5 57.9 49.8 23.1 54.4 62.8 Others 1.6 1.7 1.7 1.6 1.0 1.6 1.5 Urban residence Rural 59.0 51.5 54.3 65.3 78.3 58.5 64.0 Urban 41.0 48.5 45.7 34.8 21.8 41.6 36.0 Region of residence North central 12.5 10.4 12.1 15.1 13.8 12.9 8.6 North east 10.0 4.8 7.7 13.6 22.5 10.1 9.8 North west 21.2 9.1 18.3 27.3 47.0 21.6 17.1 South east 14.7 22.0 14.3 11.0 4.9 13.3 28.1 South south 14.8 16.1 16.1 16.2 5.6 14.8 15.2 South west 26.8 37.6 31.5 16.8 6.2 27.3 21.4 Unweighted n 4,719 1,358 1,523 1,195 643 4,259 460 (% of all households) (100%) (29%) (32%) (25%) (14%) (90%) (10%) Source: Nigeria General Household Survey 2010, Standard deviations in parentheses where appropriate
108
Table 3. 2. Experiences of Food Insecurity by Household Composition in Nigeria, Multinomial Logistic Regression Relative Risk Ratios (n = 4,719) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Moderate Severe Moderate Severe Moderate Severe Moderate Severe Moderate Severe Moderate Severe Predictors FI FI FI FI FI FI FI FI FI FI FI FI Household composition Children in the household
(0 = No children) 1-2 children 0.90 1.26* 1.04 1.66*** 1.03 1.64*** 1.05 1.66*** 1.02 1.59*** 1.01 1.56*** 3-4 children 0.76* 1.25† 1.00 2.04*** 0.99 2.06*** 1.00 2.03*** 0.98 1.93*** 0.96 1.94*** Five or more children 0.59*** 0.82 0.99 2.08*** 0.98 2.08*** 0.98 2.05*** 0.95 1.91*** 0.94 1.89*** Older adult with a disability in the household (0 = No adult with a disability) Household has adult with a disability 1.17 1.88*** 1.22 1.66*** 1.25 1.63** 1.23 1.70*** 1.21 1.65** 1.25 1.66** Social Capital Received money from 1.12 1.52*** 1.13 1.54*** relatives or friends Received remittance from 0.89 0.73 0.89 0.73 abroad Used informal 1.16 0.97 1.14 0.94 savings group Used cooperative, savings association or micro 1.00 1.11 0.98 1.08 finance Used informal group to 1.00 1.18 0.98 1.17 borrow money Borrowed money from friends, relatives, or money 1.69*** 2.99*** 1.68*** 2.95*** lenders Education of household head
(0 = Primary education) No or other forms of 0.73* 0.60*** 0.75* 0.63*** education Secondary education 0.98 0.99 0.98 0.95 Higher education 0.84 0.76 0.89 0.78 Time use Logged time taken to 1.06* 1.10*** 1.06** 1.11*** collect wood Missing time spent 1.02 0.96 1.07 1.06 collecting wood
109
Other household sociodemographic characteristics Occupation of household
head (0 = Agriculture) Sales and services 1.04 1.05 1.06 1.11 1.00 0.99 1.08 1.12 1.06 1.13 Professional job 1.06 0.71* 1.08 0.73† 1.05 0.70† 1.10 0.74† 1.10 0.77 Unemployed 0.62* 1.00 0.63* 1.01 0.61* 0.97 0.64* 1.07 0.64* 1.05 Others 0.82 1.07 0.83 1.14 0.79 1.01 0.85 1.14 0.84 1.16 Logged value of assets 0.91** 0.75*** 0.89*** 0.73*** 0.90** 0.74*** 0.91** 0.76*** 0.89*** 0.72*** Age of household head 1.00 1.00 1.00 1.00 1.00 1.01† 1.00 1.00 1.00 1.01 Female household head 1.06 0.74 1.05 0.72 1.11 0.81 1.05 0.73 1.09 0.77 Marital status of household head (0 = Married monogamous) Married polygynous 1.31* 1.06 1.27† 1.00 1.32* 1.07 1.29* 1.03 1.26† 0.99 Unmarried 0.97 1.01 0.97 0.99 0.98 1.02 1.00 1.05 0.99 1.02 Religion of household head
(0 = Muslim) Christian 0.87 0.99 0.85 0.96 0.84 0.93 0.87 0.99 0.83 0.92 Others 2.17* 2.84*** 2.13* 2.83*** 2.24* 2.96*** 2.25* 3.00*** 2.25* 3.10*** Urban residence 1.36** 2.02*** 1.38** 2.09*** 1.35* 1.97*** 1.41** 2.17*** 1.41** 2.20*** Region of residence
(0 = South west) North central 0.70* 0.83 0.67* 0.75† 0.72* 0.88 0.69* 0.80 0.68* 0.77 North east 0.94 0.81 0.92 0.77 1.01 0.92 0.95 0.80 0.99 0.85 North west 0.42*** 0.50*** 0.41*** 0.44*** 0.46*** 0.57** 0.41*** 0.47*** 0.44*** 0.48*** South east 2.13*** 2.46*** 2.18*** 2.58*** 2.08*** 2.38*** 2.09*** 2.33*** 2.13*** 2.42*** South south 2.97*** 3.68*** 3.06*** 3.90*** 2.94*** 3.64*** 3.03*** 3.76*** 3.13*** 4.01*** Intercept 0.56*** 0.46*** 1.17 4.03*** 1.10 3.85** 1.31 4.82*** 0.98 3.11** 1.04 3.41**
Source: Nigeria General Household Survey, 2010; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001; FI = Food Insecurity; reference category = food security
110
Figure 3. 1. Predicted Probabilities of Severe Household Food Insecurity by Number of children in the Household and Time Spent Collecting Wood (Fuel)
No children 1-2 children 3-4 children Five or more children
40%
35%
30%
25%
20%
15%
10% 1 2 3 4 5 6 7 8 Logged time taken to collect wood Note: Predicted probabilities are based on Model 6 of Table 3. 2 with interactions
Figure 3. 2. Predicted Probabilities of Severe Household food Insecurity by Number of Children in the Household and Occupation of Household Head No children 1-2 children 3-4 children 5+ children
40%
35%
30%
25%
20%
15%
10%
5%
0% Agriculture Sales and services Professional jobs Note: Predicted probabilities are based on Model 6 of Table 3. 2 with interactions
111
Table 3. 3. Persistence and Transitory Household Food Insecurity among Households with Children and Older Adults with a Disability in Nigeria Older adult with a disability in the Number of children in the household household Five or No adult Has adult All No 1-2 3-4 more with a with a Food security status households children children children children disability disability Persistently food secure 24.4 22.9 22.4 24.6 32.4 25.2 16.9 Persistently food insecure 20.6 23.0 23.6 19.2 10.6 19.2 33.7 Transition into and out food insecurity 55.0 54.1 54.0 56.2 57.0 55.6 49.4 Unweighted n 4,821 1,273 1,546 1,263 739 4,351 470 Source: Nigeria General Household Survey, 2010-2013
112
Table 3. 4. Transitions into Food Insecurity Status among Vulnerable Populations in Nigeria: Weighted Descriptive Statistics Older adult with a disability in the Number of children in the household household All food Five or No adult Has adult secure No 1-2 3-4 more with a with a Variables households children children children children disability disability Change in food security status Persistently food secure 48.5 45.3 47.5 50.2 53.9 48.9 43.7 Transitioned into food insecurity 51.5 54.7 52.5 49.8 46.1 51.1 56.3 Social capital Received remittance from abroad No 97.4 95.9 96.9 98.5 99.1 97.5 95.6 Yes 2.7 4.1 3.2 1.5 0.9 2.5 4.4 Used informal savings group No 67.9 70.6 68.2 66.2 65.1 67.0 78.5 Yes 32.1 29.4 31.8 33.8 34.9 33.0 21.5 Used cooperative, savings association or micro finance No 89.3 87.7 88.1 89.6 94.4 89.0 93.8 Yes 10.7 12.3 11.9 10.4 5.6 11.1 6.2 Used informal group to borrow money No 85.9 88.0 88.4 84.7 78.9 85.7 87.9 Yes 14.1 12.0 11.6 15.3 21.1 14.3 12.1 Borrowed money from friends, relatives, or money lenders No 78.0 85.9 78.4 75.2 67.5 77.4 85.3 Yes 22.0 14.2 21.6 24.8 32.6 22.6 14.7 Education of household head No or other education 39.0 39.1 36.5 36.5 47.8 37.2 59.2 Primary education 23.1 24.9 20.1 24.7 23.4 23.7 16.2 Secondary education 21.0 17.9 24.2 23.1 16.8 22.0 10.1 Higher education 16.9 18.2 19.2 15.7 12.0 17.1 14.6 Time taken to collect wood 76 (210) 40 (108) 65 (156) 102 (303) 125 (247) 76 (210) 85 (208) Logged time taken to collect wood 2.4 (2.2) 1.7 (2.0) 2.4 (2.1) 2.7 (2.3) 3.3 (2.3) 2.4 (2.2) 2.3 (2.3) Missing time spent collecting wood No 86.6 79.8 87.2 88.6 94.8 86.7 85.8 Yes 13.4 20.2 12.8 11.4 5.2 13.3 14.2 Occupation of household head Unemployed 10.1 18.5 9.1 5.3 4.4 7.7 37.2 Agriculture 43.8 34.0 41.1 47.8 61.2 44.4 36.9 Sales and services 24.1 27.5 27.7 20.8 16.0 24.7 17.9
113
Professional jobs 14.2 12.7 14.5 17.1 11.6 14.9 5.9 others 7.8 7.4 7.6 9.1 6.9 8.3 2.2 Household wealth 200725 152271 235763 153519 295342 204097 161509 Naira value of assets (899780) (567146) (1064422) (408706) (1440128) (927403) (470091) Logged value of assets 10.8(1.6) 10.3 (1.8) 10.9 (1.6) 10.9 (1.4) 11.2 (1.5) 10.8 (1.6) 10.3 (1.9) 49.5 57.6 47.4 45.2 45.9 48.0 67.1 Age of household head (15.4) (17.8) (14.8) (11.8) (10.3) (14.4) (15.0) Gender of household head Male 86.4 71.9 86.9 94.6 98.5 87.6 71.8 Female 13.6 28.1 13.1 5.4 1.5 12.4 28.2 Marital status of household head Married polygynous 19.0 8.2 11.4 21.2 50.8 19.4 14.1 Married monogamous 61.7 45.7 73.4 73.8 47.9 62.4 53.7 Unmarried 19.3 46.1 15.2 5.0 1.3 18.2 32.2 Religion of household head Muslim 48.6 32.6 45.9 52.1 78.0 49.0 44.6 Christian 49.9 65.5 52.9 46.7 20.5 49.6 54.2 Others 1.5 1.9 1.3 1.3 1.5 1.5 1.3 Urban residence Rural 61.0 52.3 57.0 65.0 78.6 60.9 62.3 Urban 39.0 47.8 43.0 35.0 21.4 39.1 37.7 Region of residence North central 14.1 10.9 14.3 17.0 15.1 14.6 8.8 North east 11.9 5.4 9.3 15.4 23.5 11.7 14.1 North west 23.9 10.9 20.8 29.1 45.9 24.1 22.0 South east 9.5 15.9 10.4 5.8 2.1 8.8 18.4 South south 12.5 15.5 12.6 13.7 5.2 12.5 13.2 South west 27.9 41.3 32.6 19.0 8.3 28.3 23.4 Survey rounds since food security 1 round 38.2 46.8 37.4 35.5 28.3 38.4 36.2 2 rounds 32.9 27.6 34.0 36.0 35.2 32.8 33.7 3 rounds 29.0 25.5 28.7 28.5 36.5 28.9 30.2 Unweighted n 3,616 918 1,107 962 629 3,330 286 (% of food secure households) (100%) (25%) (31%) (27%) (17%) (92%) (8%) Source: Nigeria General Household Survey, 2010-2013
114
Table 3. 5. Odds Ratios, Logistic Regression Predicting Transitions Into Household Food Insecurity in Nigeria by Household Composition Predictors Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Household composition Number of children in the household
(0 = No children) 1-2 children 0.86† 1.00 0.99 0.99 0.97 0.96 3-4 children 0.79* 1.07 1.05 1.04 1.04 1.00 Five or more children 0.65*** 1.09 1.08 1.07 1.05 1.02 Older adult with a disability (0 = No adult with a disability) Household has adult with a disability 1.04 0.98 1.00 1.00 0.97 1.00 Social Capital Received remittance from abroad 0.71 0.76 Used informal savings group 1.27** 1.22* Used cooperative, savings association or micro finance 0.99 0.98 Used informal group to borrow money 0.84 0.84 Borrowed money from friends, relatives, or money lenders 1.00 0.98 Education of household head (0 = Primary education) No or other forms of education 0.87 0.87 Secondary education 0.99 0.99 Higher education 0.61*** 0.64** Time use Logged time taken to collect wood 1.04* 1.04* Missing time spent collecting wood 0.85 0.86 Occupation of household head (0 = Agriculture) Sales and services 1.11 1.10 1.09 1.14 1.12 Professional job 0.89 0.89 1.07 0.93 1.08 Unemployed 1.15 1.15 1.20 1.20 1.25† Others 1.31† 1.31† 1.30† 1.38* 1.36* Logged value of assets 0.87*** 0.87*** 0.89*** 0.87*** 0.89*** Age of household head 1.00† 1.00† 1.00 0.99* 1.00 Female household head 1.04 1.04 1.05 1.03 1.05 Marital status of household head (0 = Married monogamous) Married polygynous 1.06 1.05 1.04 1.05 1.03 Unmarried 0.85 0.86 0.85 0.88 0.88 Religion of household head (0 = Muslim) Christian 0.89 0.89 0.89 0.88 0.88 Others 0.89 0.89 0.86 0.89 0.87 Urban residence 1.28** 1.29** 1.29** 1.32** 1.34*** Region of residence (0 = Southwest)
115
North central 0.58*** 0.59*** 0.58*** 0.54*** 0.55*** North east 0.63*** 0.65*** 0.63*** 0.61*** 0.63*** North west 0.49*** 0.51*** 0.50*** 0.45*** 0.49*** South east 3.00*** 3.11*** 2.90*** 2.85*** 2.85*** South south 1.08 1.12 1.05 1.05 1.06 Number of survey rounds since food security (0 = 1 round) 2 rounds 1.16* 1.30*** 1.30*** 1.30*** 1.30*** 1.31*** 4 rounds 0.43*** 0.53*** 0.53*** 0.53*** 0.53*** 0.54*** Intercept 0.46*** 2.33** 2.12* 1.98* 2.16* 1.72† Source: Nigeria General Household Survey, 2010-2013; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001
116
Table 3. 6. Transitions Out of Food Insecurity Status among Vulnerable Populations in Nigeria: Weighted Descriptive Statistics Older adult with a disability in the Number of children in the household household All food Five or No adult Has adult insecure No 1-2 3-4 more with a with a Variables households children children children children disability disability Change in food security status Persistently food insecure 44.2 48.4 46.4 41.7 32.9 42.9 55.6 Transitioned out of food insecurity 55.8 51.6 53.6 58.3 67.1 57.2 44.5 Social capital Received remittance from abroad No 97.2 96.9 96.7 97.1 99.8 97.5 94.6 Yes 2.8 3.1 3.3 2.9 0.2 2.5 5.4 Used informal savings group No 65.4 71.4 62.4 62.6 65.0 64.2 75.0 Yes 34.6 28.6 37.6 37.5 35.0 35.8 25.0 Used cooperative, savings association or micro finance No 89.4 90.1 87.0 89.4 94.1 89.0 92.8 Yes 10.6 9.9 13.0 10.6 5.9 11.1 7.2 Used informal group to borrow money No 84.9 87.8 85.5 84.4 77.4 84.7 86.4 Yes 15.1 12.3 14.5 15.7 22.6 15.3 13.6 Borrowed money from friends, relatives, or money lenders No 74.1 82.2 74.5 71.1 59.3 73.4 79.5 Yes 25.9 17.8 25.5 28.9 40.7 26.6 20.5 Education of household head No or other education 32.5 35.7 27.0 31.5 42.2 30.3 51.4 Primary education 29.4 29.6 30.4 28.4 28.0 29.9 25.3 Secondary education 24.5 20.3 27.0 27.1 22.6 26.2 10.1 Higher education 13.6 14.4 15.6 12.9 7.2 13.6 13.3 Time taken to collect wood 84 (244) 59 (162) 74 (189) 99 (254) 143 (438) 85 (250) 78 (186) Logged time taken to collect wood 2.6 (2.2) 2.2 (2.1) 2.5 (2.2) 2.7 (2.3) 3.31 (2.2) 2.6 (2.2) 2.52 (2.2) Missing time spent collecting wood No 88.5 84.3 86.7 92.0 96.5 88.5 88.2 Yes 11.5 15.7 13.3 8.0 3.5 11.5 11.8 Occupation of household head Unemployed 10.6 20.0 9.0 5.2 3.7 7.2 40.1 Agriculture 40.6 32.5 36.8 44.6 62.6 41.4 33.9 Sales and services 27.0 28.0 30.4 24.1 20.7 28.1 17.0 Professional jobs 12.3 11.7 11.6 16.8 6.8 13.1 6.0 others 9.5 7.9 12.2 9.3 6.2 10.3 3.0
117
Household wealth 137000 84971 162129 122498 223311 142381 90866 Naira value of assets (1104894) (256607) (1727663) (267097) (1209194) (1161881) (333081) Logged value of assets 10.5 (1.6) 10.1 (1.6) 10.6 (1.6) 10.8 (1.5) 10.9 (1.4) 10.6 (1.5) 10.0 (1.8) 50.3 57.9 49.4 44.9 45.7 48.3 67.3 Age of household head (15.3) (17.2) (14.7) (11.6) (10.7) (14.2) (13.5) Gender of household head Male 81.9 67.8 82.4 90.7 96.2 83.2 70.4 Female 18.1 32.2 17.6 9.3 3.8 16.8 29.6 Marital status of household head Married polygynous 15.0 8.3 10.0 15.8 43.9 14.9 15.7 Married monogamous 62.0 43.9 70.3 76.1 52.7 63.2 51.2 Unmarried 23.0 47.9 19.7 8.1 3.4 21.9 33.1 Religion of household head Muslim 36.8 25.6 33.3 39.6 67.9 37.9 27.2 Christian 61.3 72.5 64.6 58.6 30.6 60.3 70.2 Others 1.9 1.9 2.1 1.9 1.6 1.9 2.6 Urban residence Rural 55.1 47.5 50.2 60.3 76.3 53.9 65.7 Urban 44.9 52.5 49.8 39.7 23.7 46.2 34.3 Region of residence North central 10.0 9.1 10.2 9.6 12.5 10.2 7.8 North east 8.6 3.0 5.9 11.8 22.8 8.6 8.0 North west 14.9 5.2 10.7 21.0 37.6 15.4 10.6 South east 18.1 26.8 17.9 13.1 8.0 16.4 32.4 South south 18.9 15.6 21.5 23.9 9.4 19.2 16.7 South west 29.6 40.3 34.0 20.7 9.7 30.2 24.5 Survey rounds since food insecurity 1 round 48.2 40.3 46.8 51.7 63.8 49.7 35.4 2 rounds 25.7 30.3 25.7 23.6 19.0 26.4 20.0 3 rounds 26.1 29.3 27.5 24.8 17.1 24.0 44.6 Unweighted n 3,390 900 1,108 897 485 3,018 372 (% of food insecure households) (100%) (27%) (33%) (26%) (14%) (89%) (11%) Source: Nigeria General Household Survey, 2010-2013
118
Table 3. 7. Odds Ratios, Logistic Regression Predicting Transitions Out of Household Food Insecurity in Nigeria by Household Composition Predictors Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Household composition Number of children in the
household (0 = No children) 1-2 children 1.04 0.84 0.86 0.85 0.85 0.87 3-4 children 1.23* 0.79† 0.80† 0.80† 0.81† 0.82 Five or more children 1.80*** 0.74* 0.75† 0.75† 0.76† 0.77† Older adult with a disability (0 = No adult with a disability) Household has adult with disability 0.63*** 0.73* 0.73* 0.72* 0.73* 0.71* Social Capital Received remittance from abroad 1.01 0.98 Used informal savings group 0.90 0.91 Used cooperative, savings association or micro finance 0.89 0.87 Used informal group to borrow money 0.98 0.98 Borrowed money from friends, relatives, or money lenders 0.84* 0.84* Education of household head (0 = Primary education) No or other forms of education 1.12 1.10 Secondary education 1.21† 1.21† Higher education 1.32† 1.30† Time use Logged time taken to collect wood 0.97† 0.97† Missing time spent collecting wood 1.00 1.00 Occupation of household head (0 = Agriculture) Sales and services 0.82† 0.82† 0.82† 0.80* 0.79* Professional job 1.16 1.16 1.08 1.15 1.08 Unemployed 0.73* 0.70* 0.70* 0.71* 0.67** Others 0.87 0.86 0.86 0.85 0.84 Logged value of assets 1.16*** 1.17*** 1.15*** 1.16*** 1.16*** Age of household head 1.00 1.00 1.00 1.00 1.00 Female household head 0.97 0.98 0.99 0.98 1.00 Marital status of household head (0 = Married monogamous) Married polygynous 1.01 1.03 1.02 1.02 1.05 Unmarried 0.98 0.97 0.97 0.97 0.96 Religion of household head (0 = Muslim) Christian 0.91 0.91 0.90 0.90 0.90 Others 1.19 1.20 1.19 1.17 1.19 Urban residence 0.77** 0.76** 0.76** 0.75** 0.73** Region of residence (0 = Southwest)
119
North central 1.24† 1.27† 1.24† 1.24† 1.27† North east 2.14*** 2.11*** 2.15*** 2.13*** 2.12*** North west 2.80*** 2.79*** 2.81*** 2.81*** 2.82*** South east 0.25*** 0.25*** 0.26*** 0.25*** 0.26*** South south 0.67** 0.65*** 0.68** 0.66** 0.65*** Number of survey rounds since food insecurity (0 = 1 round) 2 rounds 0.43*** 0.52*** 0.52*** 0.52*** 0.52*** 0.52*** 3 rounds 0.35*** 0.49*** 0.50*** 0.50*** 0.50*** 0.50*** Intercept 0.60*** 0.20*** 0.20*** 0.19*** 0.22*** 0.21*** Source: Nigeria General Household Survey, 2010-2013; † p<0.10, * p<0.05, ** p<0.01, *** p<0.001
120
CHAPTER 4: MULTILEVEL ANALYSIS OF HOUSEHOLD FOOD INSECURITY AND
CHILD MALNUTRITION IN NIGERIA
Introduction
The health and nutrition of mothers and children are central to human capital and economic development (Ozughalu and Ogwumike 2015; The World Bank 2006). Attaining nutritional security is also fundamental in any sustainable poverty alleviation effort (FAO, IFAD, and WFP
2013; The World Bank 2006). Although many countries achieved the Millennium Development
Goal (MDG) of halving the proportion of people living on less than one dollar a day between
1990 and 2015, less than one-fourth achieved the MDG goal of alleviating non-income poverty - hunger (The World Bank 2006). Malnutrition remains a formidable health problem confronting one out of every eight persons across the globe (FAO, IFAD, and WFP 2013). Nutritional insecurity or undernutrition is particularly harmful to the health of children. In 2013, nearly half of all the 6.3 million global deaths of children under the age of five were nutrition-related
(UNICEF 2014). Undernutrition is associated with poor health in adulthood, including heightening adults’ vulnerability to risks of contracting HIV/AIDS, elevating the risks of mother- child transmission of HIV, and rendering antiretroviral drugs less efficacious (Rose et al. 2014) but deprivation in early years of life may have long-reaching effects.
Nutritional status is tied to the physical, psychosocial, and cognitive development of children (Alaimo et al. 2001; Olson 1999). Stunting, wasting, and underweight, common manifestations of child malnutrition, can be multigenerational in their impacts. They often commence in-utero resulting in low birth weights, while malnourished girls continue the cycle by procreating underweight children in adulthood (Akinyele 2009; The World Bank 2006). Stunting is an irreversible impeded linear growth in children, resulting from prolonged nutritional
121 deficiency, inadequate care, and/or disease infection (The World Bank 2006). Whereas stunting reflects long term chronic malnutrition (Oldewage-Theron et al. 2006), wasting is an indicator of acute food insufficiency or illness (The World Bank 2006) while underweight is reflective of both acute and chronic malnutrition (National Population Commission 2014).
Food insecurity is a major driver of child malnutrition, including stunting, wasting, and underweight (Ajao et al. 2010; Adebayo and Ojo 2012). Food insecurity involves unreliable access to, and/or insufficient quality and quantity of, nutritionally adequate food (Hadley 2014;
Ivers and Cullen 2011; Sirotin et al. 2014). In Pakistan, Bangladesh, Ethiopia, Vietnam, and in other contexts, including some parts of Nigeria (Ajao et al. 2010; Ali et al. 2013; Baig-Ansari et al. 2006), stunting, wasting, and underweight are significantly more prevalent among children growing up in food insecure households than those in food secure households.
In spite of the high prevalence rates of household food insecurity and child undernutrition in Nigeria (Ajao et al. 2010; Akinyele 2009; Adebayo and Ojo 2012; Omonona and Adetokunbo
2007), only a few studies have examined the link between household food insecurity and child malnutrition in the country (Ajani et al. 2006; Ijarotimi and Oyeneyin 2005). The few existing analyses of household food insecurity and child malnutrition in the country are cross-sectional in nature, very limited in spatial scope, and mostly descriptive in nature.
A perplexing issue is the research findings showing relatively high prevalence rates of malnourishment among children residing in food secure households in Nigeria and other parts of sub-Saharan Africa. In Ondo state Nigeria, nearly one third (30%) of children aged 10-18 living in food secure households were stunted and 26% were underweight, even when as many as 44% of those in severely food insecure households had normal height-for-age Z-scores (which are indicators of stunting) and 21% had normal weight (Ijarotimi and Oyeneyin 2005). Similarly,
122 about one in four children (23%) living in food secure households in Akinyele local government area of Oyo state were stunted and 15 percent were found wasted (Atoloye et al. 2015). High rates of malnourishment among children in food secure households have also been reported in other parts of Africa. For instance, even though Kahsay, Mulugeta, and Seid (2015) found higher prevalence of stunting, wasting, and underweight in food insecure than in food secure households in rural parts of Ethiopia, nearly half (46%) of children living in food secure households were stunted, seven percent were wasted, and as many as 18% were underweight.
Similarly, Lwanga et al. (2015) reported that among children under the age of five living in food secure households in east-central Uganda, 33% were stunted, 30% wasted, and 35% were underweight. Even though there is currently limited empirical evidence, drawing on nutritional transition theory there are factors other than simply food access that may account for the numerous stunted, wasted, and underweight children who seem to be suffering in the midst of plenty.
The present study seeks to fill some of the gap in the existing literature on household food insecurity and child undernutrition in Nigeria by analyzing the patterns of stunting, wasting, and underweight among a nationally representative sample of children. Taking advantage of the panel nature of the Nigeria General Household Survey data, I analyzed the relationship between recent, distal, and chronic experiences of household food insecurity and under-five stunting, wasting, and underweight. Most importantly, I expanded the predictors of child nutritional status in previous studies to include community characteristics (community level infrastructural development and change in community socioeconomic development) that could help explain the high prevalence rates of stunting and wasting among food secure children in Nigeria.
123
Background
The United Nations stipulates that access to adequate food and freedom from hunger is a basic human right that every child is entitled to (Adebayo and Ojo 2012). However, the nutritional condition of most children in Sub-Saharan Africa is deplorable. Sub-Saharan Africa has the highest prevalence of undernourishment in the world and is the only region where malnutrition has been persistently high over the ages (FAO, IFAD, and WFP 2013; The World Bank 2006). In the past two decades, the prevalence rate of stunting among children under the age of five in
Sub-Saharan Africa remained consistent at about 40 percent (Remans et al. 2011), but, rapid population growth in the region precipitated an upsurge in the size of malnourished children (The
World Bank 2006).
The high prevalence rate of malnutrition in Sub-Saharan Africa is concerning but the region is far from being homogenous. The huge variation across many different countries within the region makes it imperative to understand the patterns of malnutrition within each of the Sub-
Saharan African nations. Nigeria has high rates of malnutrition. The HIV/AIDS pandemic compounded the problem of malnutrition in many East and South African countries over the past few decades (Adebayo and Ojo 2012; Rose et al. 2014). But there are more undernourished people in the giant of Africa - Nigeria - than in six3 out of the ten African countries with the highest HIV prevalence rates combined (FAO, IFAD, and WFP 2013).
Household food insecurity and malnutrition, especially child undernutrition, are highly prevalent in Nigeria (Ajani et al. 2006; Ajao et al. 2010; Akinyele 2009; FAO, IFAD, and WFP
2013; Adebayo and Ojo 2012; Omonona and Adetokunbo 2007; Ozughalu and Ogwumike
2015). More than two-thirds (68%) of Nigerian population survive on one dollar twenty five
3 Swaziland, Botswana, Lesotho, Namibia, Zambia, and Malawi
124 cents per day (The World Bank 2014); food accounts for the large share of household expenditure in Nigeria (Omonona and Adetokunbo 2007); and more often than not, poverty is closely linked to experiences of hunger and malnutrition in the country (Ozughalu and
Ogwumike 2015). At least half of Nigerians experienced food poverty in 2004 (Ozughalu and
Ogwumike 2015) and more than 12 million Nigerians were undernourished in 2011-2013 (FAO,
IFAD, and WFP 2013). According to the 2013 Demographic and Health Survey, about two out of every five Nigerian children under the age of five are stunted; 18 percent are wasted; and nearly one-third (29%) are underweight (National Population Commission 2014).
However, even though there is a relatively large body of work on household food insecurity in Nigeria, few studies have examined the link between household food insecurity and child malnutrition in the country (Ajani et al. 2006; Ijarotimi and Oyeneyin 2005). The few existing studies are very limited in their spatial scope and many are purely descriptive in nature.
For instance, Atoloye et al. (2015) analyzed food insecurity and childhood malnutrition in
Akinyele Local Government Area of Oyo State, Babatunde et al. (2007) examined the incidence of stunting and wasting among pre-school children in only 60 male- and female-headed households in Kwara state, and Ijarotimi and Oyeneyin (2005) investigated the link between household food insecurity and nutritional status of school-aged children in Ondo state. Ijarotimi and Odeyemi (2012) reported the prevalence of household food insecurity and nutritional status of children living in rural communities in Ondo State while Senbanjo and colleagues (2011) merely documented the prevalence and risk factors associated with stunting among school children in urban part of Abeokuta, a city in Ogun State. Others sampled households in parts of
Lagos and Ibadan (Ajani et al. 2006; Omonona and Adetokunbo 2007) or parts of Port Harcourt
(Ordinioha and Brisibe 2013). While these studies provide important insights into the problems
125 of food insecurity and malnutrition in Nigeria, they fail to account for the enormous spatial and sociodemographic heterogeneity across the country. Therefore, the present study examines the prevalence of stunting, wasting, and underweight among a nationally representative sample of
Nigerian children under the age of five, in relation to their experiences of household food insecurity. I expect to replicate the positive association between household food insecurity and stunting, wasting, and underweight at the national level. I also expect that distal and chronic experiences of food insecurity will be positively associated with stunting, more so than recent episodes of food insecurity which will be more likely related to wasting and underweight.
Food poverty is multidimensional in nature, encompassing not just food availability, access, and utilization, but also stability of access (Akinyele 2009; FAO, IFAD, and WFP 2013;
Omonona and Adetokunbo 2007). Existing studies of household food insecurity and child malnutrition in Nigeria (e.g. Ajao et al. 2010; Atoloye et al. 2015; Babatunde et al. 2007;
Ijarotimi and Oyeneyin 2005; Ijarotimi and Odeyemi 2012), rely almost exclusively on cross- sectional data and are, therefore, very limited in describing a critical aspect of household food insecurity in relation to child wellbeing - stability (Akinyele 2009). Food insecurity can be chronic or temporary in nature (Ojo and Adebayo 2012; Omonona and Adetokunbo 2007). The seasonal nature of agriculture, which is the dominant means of subsistence in countries like
Nigeria (Kurukulasuriya et al. 2006; Osei et al. 2010), highlights the inadequacy of cross- sectional analyses to adequately address the problem of household food insecurity and child malnutrition. More often than not, stunting is conceived of as a product of chronic food shortage or nutritional deficiency. However, factors other than feeding patterns, such as childcare practices, health services, sanitation, and environmental conditions, contribute to the prevalence of stunting reported in existing cross-sectional studies (Ajani et al. 2006; Baig-Ansari et al. 2006;
126
Osei et al. 2010). Thus, longitudinal data analyses are required to link patterns of experiences of household food insecurity or nutritional history to malnutrition among children.
Limited evidence from cross-sectional analyses suggests worsening food conditions and children’s nutritional status in Nigeria over time (Ajani et al. 2006; Ijarotimi and Oyeneyin 2005;
Ijarotimi and Odeyemi 2012). From being an exporter of cash crops, Nigeria metamorphosed to a major importer of food crops over the past few decades. The effect of the recently skyrocketing global food prices is however unequally felt across households in the country (Akinyele 2009;
Adebayo and Ojo 2012). Towering unemployment and underemployment rates may have combined with declining per capita income (Omonona and Adetokunbo 2007) to worsen the nutritional conditions of Nigerian households over time. In view of the deteriorating food conditions in Nigeria and the positive association between food security and children’s nutritional outcome, I expect that chronic food insecurity will be more detrimental to children’s nutritional status, particularly stunting, than transitory food insecurity.
The paradox of food security and poor child health
According to the UNICEF’s conceptual framework (UNICEF 2015), food insecurity or inadequate dietary intake is just one out of a host of factors influencing nutritional and health outcomes among children. According to this framework, factors contributing to children’s nutritional outcomes could be immediate (e.g. dietary intake as determined by household food insecurity and disease infection) or more distal factors influencing households and communities
(e.g. access to health services and other contextual socioeconomic characteristics). This means that, some children living in food secure households could be exposed to adverse environmental conditions that precipitate negative nutritional outcomes. It could also be that children living in food insecure households suffer from not just poor diets, but from a mixed dose of poor diets and
127 the impacts of poor environmental conditions. Therefore, it is important to examine stunting, wasting, and underweight in relation to food insecurity and community socioeconomic characteristics, especially access to potable water, health care, and sanitation facilities.
Additionally, the high prevalence rate of stunting among food secure children could result from the negative effects of changing diet and micronutrient intake with increasing urbanization and improved socioeconomic conditions in developing countries like Nigeria
(Oganah and Nwabah 2009; Popkin et al. 2012). According to the nutrition transition theory, the adoption of modern lifestyles as a population undergoes the processes of urbanization and socioeconomic development could lead to changes in dietary and micronutrient intake so drastically that nutritional status could be jeopardized (Popkin 1993). Rapidly increasing consumption of what is popularly known as a “Western diet” - refined carbohydrates, caloric sweeteners, cheap fats and oils - and the shift from indigenous foods, fruits, and vegetable to animal-source foods, which may not meet the nutritional requirements of early growth and development, among households in urbanizing areas could lead to the adverse health effects such as under-five stunting observed among children living in food secure households in Nigeria
(Popkin et al. 2012; Vorster et al. 2011).
In line with the expectations of the nutrition transition theory, I posit that Nigerian children will be more likely to be stunted, wasted, and underweight if they lived in communities with higher levels of infrastructural development (a measure of level of urbanization) and rapidly socioeconomically developing areas than in areas experiencing little or no socioeconomic change. I also expect that levels of infrastructural development and change in community socioeconomic development will partly account for the high prevalence of stunting, wasting, and underweight among children living in food secure households in the country. Further, I test for
128 the impacts of community-level characteristics on under-five wasting and underweight as well as on the association between household food insecurity and wasting and underweight.
Current Investigation
In the present study, I examine the following three research questions. First, I investigate the association between food insecurity and child health (stunting, wasting, and underweight).
Based on the traditional resource approach, I expect that household food insecurity will be positively related to stunting, wasting, and underweight. I examine the effects of recent, distal, and chronic experiences of food insecurity on child health. Given that stunting is an indicator of chronic malnutrition and wasting an indicator of acute nutritional deprivation (Reinhard and
Wijayarantne 2002), recent episodes of food insecurity are expected to have less pronounced impact on stunting while wasting and underweight may be responsive to both the immediate and past experiences of food insecurity. Second, I consider the food security paradox that many children in food secure households experience high levels of stunting. Drawing on nutrition transition theory, I evaluate how social context, measured in terms of community level infrastructural development and change in community socioeconomic development: 1) relate to child health, and 2) modify the experiences of stunting, wasting, and underweight among children living in food secure households. I employ multilevel logistic regression modeling technique to predict the odds of stunting, wasting, and underweight among Nigerian children under the age of five (level 1), living in households that are nested within about 400 communities
(level 2)4. Third, I conduct cross-level interactions between food insecurity status at the household level, and community level infrastructural development and change in community socioeconomic development (community level variables) in order to determine the effects of the
4 Community in this sense corresponds to the NGHS enumeration area from which the community focus group discussants were drawn.
129 community variables on stunting, wasting, and underweight among food secure children, relative to their counterparts in food insecure households. The NGHS data provide an opportunity to consider the timing and persistence of food insecurity and include critical variables to test key theoretical and conceptual frameworks related to food insecurity and child growth.
Drawing on research in Nigeria and in developing nations, I include a set of variables that have been traditionally associated with child malnutrition (Ajao et al. 2010; Babatunde et al.
2007; Baig-Ansari et al. 2006; Omonona and Adetokunbo 2007; Ordinioha and Brisibe 2013;
Osei et al. 2010; Ozughalu and Ogwumike 2015; Senbanjo et al. 2011). These include demographic characteristics of individual children and households (child’s age and gender, mother’s education, age and gender of household head, household size, dependency ratio, occupation of household head, and household wealth) and contextual factors (urban residence and region of residence).
Individual demographics: In Nigeria, the prevalence of stunting increases with age, peaks in the third year of life, and declines thereafter (National Population Commission 2014). Other studies document increasing prevalence of stunting with age up to the adolescent stage (Senbanjo et al. 2011). Similarly, Nigerian children seem to be more prone to wasting between the ninth and the eleventh month of life and less likely to be wasted after age four (National Population
Commission 2014). Malnutrition is more prevalent among boys than among girls in Nigeria
(National Population Commission 2014). Studies have widely shown the crucial role of maternal education in child’s health and nutrition (Ajao et al. 2010; Osei et al. 2010). Children of educated mothers are less vulnerable to household food insufficiency (Babatunde et al. 2007) and are less likely to be stunted or wasted (Baig-Ansari et al. 2006; National Population Commission 2014;
Senbanjo et al. 2011) than children born to uneducated mothers.
130
Household demographics: Households headed by persons of different age brackets are unequally vulnerable to food poverty (Ozughalu and Ogwumike 2015). Household food insecurity seems to worsen as the head of household ages (Babatunde et al. 2007; Omonona and
Adetokunbo 2007). Nigerian children living in female-headed households are more susceptible to the problem of household food insecurity and malnutrition than their counterparts residing in male-headed households (Babatunde et al. 2007; Omonona and Adetokunbo 2007). Household size indicates household food allocation and child’s home environment (Alaimo et al. 2001).
Incidence of food insecurity increases with household size (Babatunde et al. 2007; Omonona and
Adetokunbo 2007; Ozughalu and Ogwumike 2015) and large households have greater prevalence of impeded growth among children than smaller households (Baig-Ansari et al.
2006). The more dependents in a household relative to working age adults, the more vulnerable children are to the risks of food insecurity (Babatunde et al. 2007; Omonona and Adetokunbo
2007). Food poverty also varies across households depending on the means of subsistence of the household heads (Ozughalu and Ogwumike 2015). Professionals have lower incidences of food insecurity than traders and unemployed persons (Omonona and Adetokunbo 2007). Children experience more rapid growth and have reduced risks of stunting when they reside in households that are socioeconomically better off (Timaus 2012). Malnourished children are more concentrated in poor households than in wealthy households (National Population Commission
2014).
Contextual factors: Food poverty varies across the different geopolitical regions in
Nigeria (Ozughalu and Ogwumike 2015). Estimates based on the 2004 Living Standards Survey showed the highest incidence of food poverty in the south-south region and the lowest incidence in the south-east region (Ozughalu and Ogwumike 2015). Stunting and wasting are more
131 prevalent in the northwestern and the northeastern regions of the country (National Population
Commission 2014). Food insecurity may be more concentrated in the urban than the rural parts of Nigeria (Ozughalu and Ogwumike 2015). Food prices are higher in the urban than in the rural areas and as such, city residents spend more on food, consume less, and purchase most of their food (Garrett and Ruel 1999). As stated above, this study also draws on the nutrition transition theory to evaluate the effects of community level infrastructural development and change in community socioeconomic development on under-five stunting, wasting, and underweight.
Data and Methods
I utilized data from the panel component of the Nigeria General Household Survey (NGHS). The
NGHS is a nationally representative annual survey of 22,000 households conducted as part of the
Living Standards Measurement Study-Integrated Surveys on Agriculture (LSMS-ISA). In 2010, the NGHS was expanded to include a panel component that sampled 5000 households out of the
22,000 core sample of the NGHS. Unlike its cross-sectional counterpart, the panel survey is biennial by design. However, the panel households are visited twice per wave of data collection with the two visits corresponding to the post-planting and the post-harvest periods. Although the
NGHS primarily aims to collect household-level agricultural-related statistics, the survey also collects extensive information about household welfare and behavior which could aid in the analysis of household socio-demographic characteristics in relation to health and wellbeing.
The panel NGHS is the first panel survey implemented by the Nigeria National Bureau of
Statistics. The LSMS-ISA team in the World Bank’s Development Research Group also provides technical guidance in the design and implementation of the NGHS survey as well as assist with the analysis of the NGHS panel data. The survey was supported by various organizations including the Nigeria Federal Ministry of Agriculture and Rural Development, the National Food
132
Reserve Agency, the Bill and Melinda Gates Foundation, and the World Bank. The first and the second rounds of the panel survey used for the present analyses were carried out in August-
October 2010 (post-planting, round 1) and in February-April 2011 (post-harvest, round 2). The design, implementation, and coverage of the NGHS have been detailed elsewhere (National
Bureau of Statistics 20155).
At the post-harvest survey round (2011), the NGHS took the anthropometric measurements of most children under the age of five in each household. At the different rounds, the survey also documented retrospective accounts of household food insecurity during the week preceding the interview, by the senior female or person most knowledgeable about household food consumption. This wealth of data on both household food insecurity and nutritional status in the NGHS, coupled with the panel nature of the data, made it possible to analyze the relationship between household food insecurity status and under-five stunting, wasting, and underweight both in recent time (2011) and about six months before child growth was assessed
(2010). Further, the panel NGHS includes a community survey (completed in focus group settings by members of households in the different enumeration areas) that collects detailed information about the socioeconomic characteristics of the different communities from which the
NGHS sample was drawn. I was also able to control for a wide range of socio-demographic characteristics that have been shown to be associated with food insecurity and children’s nutrition.
Sample
At the second interview in 2011 when the first sets of anthropometric measurements were taken, there were a total of 3,423 children under the age of five in the households (the WHO
5 The data are also available for download through the World Bank’s Living Standard Measurement Study website (http://go.worldbank.org/BY4SLL0380).
133 macro used only provides estimates for children aged 0-60 months). Preliminary analyses indicated that 2,651 of the children had their measurements taken, 84 were absent, two were injured or sick, 26 refused measurement, 633 were not measured for other reasons, and twenty seven children were reportedly measured but had no records of weight or height in the data. I excluded two children with no reports of gender and 468 children with missing household food security status, my focal independent variable (i.e. missing on 3 or more of the 9 items used to construct the food security scale). The above restrictions left a sample of 2,181 children aged 0-
60 months. The sample sizes for my analyses of stunting, wasting, and underweight were slightly different because not all children had valid height and weight scores. The above sample is further limited for the analyses of stunting, wasting, and underweight (the three outcome variables) as described below.
Stunting
Stunting describes height/length-for-age Z-scores lower than two standard deviations below the WHO standard Z-scores among children. Out of the 2,181 children in my analytic sample, 57 children had missing heights and, therefore, no height-for-age z-scores; 21 had no estimated value of household wealth; and 25 were residents of communities with no known amenity. Thus, the final sample size for my analysis of household food insecurity and under-five stunting is 2,078.
Wasting
Children with weight-for-height/length Z-scores lower than two standard deviations below the WHO standard Z-scores were considered wasted. Given that both the height/length and the weight measures were required for the estimation of children’s weight/length-for-height scores, more children (342) out of the 2,181 children in my sample were missing on the indicator
134 of wasting. Also, 19 children were without values of household wealth and 25 had unknown community level of infrastructural development. Therefore, I analyzed wasting among 1,795 children 60 months or younger.
Underweight
Underweight children had weight-for-age Z-scores lower than two standard deviations below the WHO standard Z-scores. Out of the above 2,181 children, thirty children lacked valid weight measures. I also excluded 21 children with missing information on household wealth, 25 with no reports of community socioeconomic characteristics, and one child with conflicting reports of age and weight measure. Excluding the above cases, my final underweight sample comprised 2,104 children.
Measures
I assessed all the predictors in my analyses at round one (first visit in 2010) and the outcome variables (stunting, wasting, and underweight) at the second survey round.
Outcome variables
I assessed the nutritional status of Nigerian children in relation to their experiences of household food insecurity using three widely researched anthropometric measures – stunting, wasting, and underweight. The NGHS collected information about the height and weight of all children under the age of five in each household. As in the Demographic and Health Survey, children under the age of two were measured while lying down (length) while older children had their heights taken while standing (height). I converted the weights and heights indices (lengths in the case of children under the age of two), into height-for-age Z-scores (HAZ), weight-for- height Z-scores (WHZ), and weight-for-age Z-scores (WAZ) using the WHO child growth standards. Children with height-for-age, weight-for-height, and weight-for-age Z-scores lower
135 than two standard deviations below the WHO standard Z-scores were categorized as stunted, wasted, and underweight respectively.
Focal predictor: Food security
I utilized the nine items in the refined Household Food Insecurity Access Scale (HFIAS) which assesses the access component of household food security (Coates et al. 2007). The
HFIAS questionnaire was adapted by the NGHS such that rather than asking about the frequency-of-occurrence of food insecurity separately as in the HFIAS questionnaire, the NGHS combines both incidence and frequency of household food insecurity in a series of questions about the number of days during which households recorded certain occurrences of food insecurity. Nevertheless, the food insecurity questions in the NGHS were very similar to the ones in the HFIAS questionnaire.
I compared the prevalence of stunting, wasting, and underweight among children residing in three major categories of households based on their food insecurity status reported at round 1.
These are: food secure households, moderately food insecure households, and severely food insecure households. The details of the reports of food insecurity among households in each of the three categories are presented in Appendix A. My classification of Nigerian households into the above three categories closely mirrors the official and well tested classification adopted by the Food and Nutrition Technical Assistance III Project (FANTA) team in the US (see Coates et al. 2007).
Focal variables: Community characteristics
I utilized two sets of questions asked in the NGHS community questionnaire to examine the level of development in each community and changes in community socioeconomic development over time. The first set includes indicators of level of community infrastructural
136 development as marked by the presence of the following: secondary school, public and private hospitals, bank, post office, cell phone distributor, internet café, and police station. I assessed the relationship between number of infrastructure in a community (ranging from 0-8) and child growth. The second set of questions pertains to changes in the community development over time. The participants of the focus groups who filled the community questionnaire were asked to compare the conditions in the community to what they were five years before the interview and report whether they agreed that certain features of their communities became much worse, worse, about the same, better, or much better. These include: access to transportation, availability of potable water, use of improved sanitation facilities, electricity, nutrition status of children, availability of vaccinations for children, availability of health care, care for pregnant women, adult literacy levels, quality of primary and secondary education, employment opportunities, and police services. I recoded responses of “better” and “much better” as 1 and 0 if otherwise. I used all the items to construct a composite ordinal scale ranging from 0-13 with higher values indicating greater socioeconomic development. To capture the context of development I also included control variables for urban residence (0 = rural, 1 = urban) and region of residence
(north-central, north-east, north-west, south-south, south-east, and south-west). The reference group is south-west.
Sociodemographic measures
Individual-level characteristics: I measured child’s age in three categories – 0-1 years
(reference), 2-3 years, and 4-5 years. The gender of the child is based on the head’s report with males coded as 1 and females 0. I compared nutritional status and household food insecurity among children born to mothers with four different levels educational attainment – no formal education, primary education (reference), secondary education, and higher levels of education.
137
Children with no reports of mother’s education were included in the last category – unknown education.
Household-level characteristics: I employed a continuous indicator of age of household head (in years). The gender of the head of household is also coded as 1 if the household is headed by a man and 0 if headed by a woman. Household size is a measure of the total number of individuals residing in each household. Household dependency ratio assesses the total number of dependents (children under the age of 15 and adults sixty five years and above) in each household as a fraction of the total number of working age adults (age 15-64) in the same household. Employment status of the household head is based on reports of employment activities within seven days preceding the survey. The questions asked whether or not the head of households: 1) worked for someone who was not a member of their households, 2) worked on a farm owned or rented by a member of their households, and 3) worked on their own account or in a business enterprise belonging to them or someone in their households. Household heads who reported engaging in any of the three work categories were considered employed and were compared to their unemployed counterparts in their experiences of household food insecurity. In a follow up question, respondents were asked to report the sector of their primary occupations. I combined the reports of employment status and the specific occupations to create the following employment categories: unemployed, agriculture (reference category), sales and services, professional jobs, and others. Lastly, I attempted to tease out the effect of food insecurity on stunting, wasting, and underweight from the effect of other dimension of household socioeconomic status by controlling for the log-transformed estimated values (in naira) of household’s durable assets.
138
Analytic strategy
First, I describe the sociodemographic characteristics and growth outcomes among children in my analyses. I then estimated the risks of being stunted, wasted, and underweight among Nigerian children five years or younger living in households clustered within different communities in a series of multi-level logistic regression models. The three levels in the multilevel logistic regression analyses are: individual (child age, child’s gender, and mother’s education), household (age of household head, gender of household head, household size, household dependency ratio, occupational status of household head, and household wealth), and community (community infrastructural development, change in community socioeconomic development, urban residence, and region of residence,). However, due to limited clustering at the household level (average of 1.2 children per household), I collapsed the individual- and the household-levels into one level – the individual/household level. Simulation studies show that small group sizes at each level of the multilevel hierarchy could produce biased estimates of the group level variance components in multilevel models (see Clarke 2008)
I predicted the odds of under-five stunting, wasting, and underweight among children living in moderately food insecure households and those in severely food insecure households, relative to their counterparts in food secure households in five models. The models include several correlates of household food insecurity and nutritional status at the individual level (age of child, child’s gender, and mother’s education), household level (age of household head, gender of household head, household size, household dependency ratio, employment status of household head, and household wealth), and at the community level (community infrastructural development, change in community socioeconomic development, urban residence, and region of residence). The first model is a variance components or intercept-only model which partitions the
139 total unexplained variance in child growth outcomes into the proportion due to community-level variations and the percentage attributable to differences among children living within the same communities. Model 2 includes indicators of household food insecurity and the third model controls for individual/household-level predictors of child growth. In Model 4, I examined the relationship between child growth, household food insecurity, and the community level predictors. The final model (Model 5) includes all the variables in the analyses. I further tested for significant interactions between the community socioeconomic variables and household food insecurity in predicting stunting, wasting, and underweight. All the individual- and household- level covariates were specified as fixed effects.
Results
Table 4. 1 presents the descriptive statistics for variables in my analysis of under-five stunting by household food insecurity. Only about half (51%) of Nigerian children under the age of five lived in food secure households in 2011 (survey round 2) when the anthropometric measures were taken. Food insecure children were fairly divided between moderately food insecure households (24%) and severely food insecure households (25%). Reflecting the food security paradox, a greater share of children were stunted (57%) than lived in food insecure households
(49%). Even more paradoxical was the greater share of stunted children in food secure households (61%) than in food insecure households (49%-58%). A little over a third of the children in this study were infants. Children living in food insecure households appeared to be slightly younger than those in food insecure households. Perhaps reflecting the imbalanced sex ratio in Nigeria, 53% of children in my analyses were of the male gender and this cuts across food secure and food insecure households. Nearly half (43%) of the children had mothers with no formal education and only 6% were born to highly educated mothers. More children in food
140 secure households (50%), than in food insecure households (32-39%), had uneducated mothers.
The average age of heads of households with under-five children was 43 and varied little across food security statuses. The vast majority of Nigerian children in this study lived in households headed by men. Severely food insecure households with children were more likely to have female heads than food insecure ones. On average, the children lived in households with eight people. The dependency ratio of households with children under the age of five was unsurprisingly higher (1.63) than the national average of less than one. Severely food insecure households may have more dependents than other households. Only a minority of children (3%) lived in households with unemployed heads; the vast majority, particularly food secure children, lived in agricultural households. There seems to be low correlation between logged value of household wealth and food security status among households with children.
Irrespective of its food security status, an average Nigerian household with under-five children had only two out of the eight amenities included in the infrastructural development scale. However, Nigerian communities seem to have undergone considerable socioeconomic transformations in the five years preceding the survey. On a scale ranging from 0-13, the communities scored six on the measure community socioeconomic change. Three-quarters of children in the study resided in the rural parts of Nigeria with slightly more urbanites among children living in food insecure households. My analyses had fair representation of children from all the six geopolitical zones in the country. Food secure children mostly lived in the northern
Nigeria while children in food insecure households were mostly based in the south.
The results of my multilevel logit estimation of under-five stunting by household food insecurity and community characteristics are presented in Table 4. 2. The first model partitions the total variability in stunting into two components – the total unexplained variance that is due
141 to differences among children living within the same communities (Level 1) and the portion that is attributable to variations across communities (Level 2). The intraclass correlation coefficient of 0.40 for the unconditional model (Model 1) indicates that two-fifths of the total variability in stunting among children under age five existed at the community level while 60% of variability was at the individual and household level. Model 2 shows no significant differences in the odds of stunting among children residing in food secure and those living in food insecure households.
Even after controlling for sociodemographic variations at the individual level, children’s odds of being stunted remained undifferentiated by household food security status (Model 3).
Child’s age and gender were not significantly related to under-five stunting. Also, education did not have the expected strong impact on children’s risks of stunting, save the higher odds of being stunted among children born to uneducated mothers than among children whose mothers had only primary education. The older the head of a house, the lower the odds of stunting. The odds of being stunted increased with increasing household size. The only other significant predictor of stunting at the individual/household level was household wealth; the more valuables in the household, the lower the odds of stunting among children (Model 3). Not only were community-level characteristics significantly related to stunting, but the effect of household food insecurity on the odds of stunting was also suppressed in models not accounting for spatial variations in Nigeria. As shown in Model 4, controlling for community-level characteristics, children living in food insecure households had between 24% and 39% higher odds of being stunted than those in food secure households. Level of infrastructural development did not significantly predict stunting but children living in communities that were perceived as rapidly developing and those in the urban areas had significantly better height-for-age scores.
Compared to those residing in the South west, children in the North west had elevated risks of
142 stunting while those in the South south and the South east were less prone to stunted growth. The reduction in the community-level variance from 2.21 in Model 1 to 1.29 in Model 4 shows that a substantial (42%) portion of the total variation in stunting was accounted for by the four community-level predictors in the model. In the final model (Model 5), moderate household food insecurity remained significantly related to higher odds of under-five stunting. The coefficients of mother’s education, age of household head, and household size were no longer statistically significant. Greater value of assets implied improved growth for children. Children residing in communities with improving socioeconomic features and those living in the urban areas had significantly better height-for-age scores. The regional variations in Model 5 resemble those presented in Model 4. The individual/household-level and community-level variables in Model 5 jointly explained about 44% of the between-community variation in under-five stunting.
Although the sample sizes for my analyses of wasting and underweight were slightly different from that of stunting, the distributions of the variables were strikingly similar to those presented in Table 4. 1. To conserve space, I discuss only the prevalence rates of wasting and underweight and the results of my multivariate analyses. Fourteen percent of Nigerian children under the age of five were wasted in 2011 (Table 4. 3) and more than one-third (36%) weighed insufficiently for their ages (Table 4. 5). Moderately food insecure households had the largest shares of wasted and underweight children, followed by food secure households, and then severely food insecure households. Also, whereas the children in the analyses of stunting and underweight were nested within 417 communities, children included in the analysis of wasting were sampled from 387 communities.
Table 4. 4 presents the results of multilevel logistic regression models of under-five wasting in Nigeria. Compared to stunting, a relatively smaller share of the total variation in
143 wasting (29%) was due to differences across communities in which children lived. Wasting was significantly more prevalent among children living in moderately food insecure households than food secure households. But children in severely food insecure households were not statistically different from those in food secure households in their weight-for-length scores. Region of residence was the only community-level predictor that significantly predicted the odds of being wasted among children. Children residing in the north-central and the north-east had significantly more anomalous weight-for-length scores than their peers in the south-west. At the individual level, the odds of wasting significantly declined with child’s age. Children born to mothers with no formal education were less likely to be wasted than those having mothers with primary education. Though only marginally significant, household wealth was protective of children against the risks of wasting. The community-level variables explained 16% of the community-level variation in wasting (Model 4) and together with the individual/household-level covariates, in Model 5, they accounted for 18% of the community-level variance in the intercept- only model – Model 1.
Lastly, I examined the growth of Nigerian children under age five based on their weight- for-age z-scores. As in the previous analyses, I predicted the odds of underweight with household food insecurity status and community-level characteristics in five multilevel logistic regression models while allowing the intercept to vary across communities. At least one-third of the total variability in underweight among children aged 0-5 was due to differences across communities
(Model 1). The results similarly showcase the important role of spatial disparities in food security and child health in Nigeria. Without accounting for the community-level variations
(Models 2 and 3), children living in severely food insecure households appear to be better off in terms of their weight-for-age scores. However, controlling for the level and change in
144 community socioeconomic status, urban residence, and region of residence, children residing in moderately food insecure households had 46% higher odds of being underweight than their counterparts living in food secure households (Model 4 and 5). At the community level, change in socioeconomic development and region of residence were significantly related to underweight among children. Children residing in rapidly developing communities were less likely to be underweight and children living in the northern regions had higher odds of being too thin for their ages than those in the south-west. Underweight was less prevalent in the south-east and south-south than in the south-west. Child’s age, child’s gender, and age of household head were three significant individual/household-level predictors of underweight status. Net of other covariates, the odds of being underweight declined as children aged; male children had 36% higher odds of being underweight than their female counterparts; and children living with younger household heads may be more likely to be deficient in weight. The community-level variables in the analysis explained half of the total community-level variation in children’s weight-for-age z-scores.
There were no significant interactions between the community socioeconomic variables and household food insecurity in predicting stunting, wasting, and underweight. Also, recent experiences of food insecurity (within seven days of the interview), and persistent food insecurity relative to transitory food insecurity, were unrelated to all the three outcomes (results available upon request).
Discussion
There is a growing body of research on household food insecurity in Nigeria. But the relationship between household food insecurity and child malnutrition at the national level is understudied.
The existing studies are not only limited in their spatial scope, but they are also cross sectional
145 and many are descriptive in nature. More so, scholars (e.g. Balistreri 2012) ponder what appear to be high rate of malnutrition among children residing in large share of food secure households in many contexts like Nigeria. In view of the above gaps in the literature, this study revisits the association between food insecurity and child malnutrition in Nigeria, focusing on the role of community-level infrastructural development and changes in community socioeconomic development on growth in childhood and on the relationship between food insecurity and child outcomes. The prevalence rates of under-five stunting, wasting, and underweight in this study are
57%, 14%, and 36% respectively. About half of Nigerian children under the age of five lived in food insecure households in 2011, but the majority lived in households that experienced food insecurity at some point between 2010 and 2013 (results not shown).
The relationship between household food insecurity and child malnutrition in Nigeria is more complicated than presented in previous studies. Previous estimates obscure the high levels of spatial heterogeneity in the experiences of food insecurity and, therefore, child growth in the country. Also, substantial portion (29-40%) of the total variability in under-five stunting, wasting, and underweight are attributable to community-level variations. Controlling for individual, household, and community characteristics, children residing in food insecure households were more prone to malnutrition than those living in food secure households.
However, moderate experiences of household food insecurity seem to be more detrimental to child health in Nigeria than severe food insecurity. In fact, children residing in severely food insecure households had similar odds of being underweight than their peers in food secure households.
Supplemental analyses (results not shown) showed no significant association between persistent food insecurity and child malnutrition, a finding that seems surprising at first glance.
146
However, it becomes less puzzling when considered in light of the alarming rate of food insecurity in Nigeria. Over the three-year period studied, only a minority of children (32%) lived in households that were persistently food secure and majority (89%) of the food-secured children lived in the northern region of the country (results not shown). This suggests that majority of
Nigerian children under the age of five would experience household food insecurity at some point in their lives making food insecurity measured over a relative short period of time (appear) less consequential for child growth.
Contrary to the predictions of the nutrition transition theory, regardless of household food security status, children living in rapidly developing communities in Nigeria have better health than those in less developing areas. Conversely, children residing in areas with slow rate of socioeconomic change, even if they reside in food secure households, had higher chances of being stunted and underweight, net of other covariates. As shown in Figure 1, it appears as if change in community socioeconomic development matter more for impoverished children, particularly those living in severely food insecure households than food secure children but the interactions between household food insecurity and community-level variables were not statistically significant. These findings help illuminate the paradoxical high rate of malnutrition among children in food secure households. Stunting, wasting, and underweight are shaped by factors other than household food insecurity. Children living in food secure households but in communities with low socioeconomic status and slowly developing communities are not immune to the risks of stunting and underweight. Similarly, a high level or rate of community development does not compensate for the negative impacts of household food insecurity on child development. Therefore, a multilevel approach to improving child nutrition and growth in
Nigeria promise to be more effective than efforts directed at individual children, households, or
147 communities alone. The expected negative impacts of urbanization and socioeconomic development on household nutrition and diets (if any), based on the nutrition transition theory, are yet to manifest in significantly higher risks of stunting, wasting, or overweight. Rather, children seem to be doing much better in areas that are rapidly developing. A possible explanation is that Nigeria is not far along enough in nutrition transition. It could also be rapid socioeconomic development is more related to child outcomes other than those examined in this study.
The study is not without limitations. First, due to limited clustering of children at the household level, I collapsed the household and the individual levels. This is consistent with prior work, but it is important for future studies with larger sample size to tease out the inter- household variations in under-five malnutrition. Due to limited clustering of children at the household level, I was unable to test for differences in the experiences of child malnutrition by birth order and by mother’s position in polygamous households. Second, the present analysis provides insights into the important role of community socioeconomic characteristics in understanding child growth. Future studies should examine a broader range of community-level characteristics. Third, even though I included a wide range of covariates of food insecurity and child outcomes, data limitations preclude the analyses of a number of mother’s and father’s characteristics such as age, employment, and income. As in previous research, I examined the effects of three household compositional factors - age of household head, household size, and household dependency ratio. Future analyses should test for differences in the association between household food insecurity and under-five malnutrition by number of working age adults and the ages of other children (above age five) in the household. I also analyzed only three indicators of child malnutrition – stunting, wasting, and underweight. Future work should
148 include under-five overweight and/or obesity and the possible co-existence of the different forms of malnutrition in the same social contexts. It is also important for researchers to consider the timing of anthropometric measurements relative to the timing of food insecurity. The present study improves on the previous cross-sectional designs by employing a lagged indicator of household food insecurity. Nonetheless, more advanced longitudinal data analyses are required to establish direct causal relationship between household food insecurity and child malnutrition.
Lastly, due to missing data, not all children under the age of five were included in the analyses.
Sensitivity tests showed no systematic patterns in the demographic characteristics of children excluded from the analyses. They were more likely to be wasted (18% versus 14% in the analytic sample) but less likely to be stunted (48% versus 57% in my sample) and underweight (31% versus 36%). Attrition rates were higher in the urban areas, in certain regions (North central,
South west, and South south), and among female children. I acknowledge that the results of the analysis could be impacted by the missing children.
In spite of the above limitations, this study expands the existing knowledge about the link between household food insecurity and child malnutrition in Nigeria by analyzing multilevel predictors of under-five stunting, wasting, and underweight. These results point to a complicated association between food insecurity and child malnutrition suggesting that children growing up in moderately food insecure households may be more prone to malnutrition than those residing in severely food insecure households. However, the study provides a possible explanation for the previous findings of high malnutrition rates among children residing in food secure households - the fact that household food security does not offset the effect of community socioeconomic development on child malnutrition. Regardless of their household food insecurity status,
Nigerian children living in rapidly developing social contexts (communities) were less likely to
149 be stunted or underweight. This finding calls for more multilevel analyses of household food insecurity and child malnutrition. Policies and programs aimed at improving child wellbeing in
Nigeria will also benefit from multilevel approaches. The designs and implementations of these programs should reflect the community characteristics.
150
Table 4. 1. Descriptive Statistics for Variables in Cross sectional Analyses of Stunting among Nigerian Children Aged 0-60 Months by Household Food Security Status Moderately Severely All children Food secure food secure food insecure Number of children 2,078 1,066 488 524 Child outcome (0-1) Stunted 0.57 (0.5) 0.61 (0.5) 0.58 (0.5) 0.49 (0.5) Individual-level characteristics Child's age (0-1) 0-1 0.36 (0.5) 0.32 (0.5) 0.41 (0.5) 0.40 (0.5) 2-3 0.45 (0.5) 0.47 (0.5) 0.41 (0.5) 0.44 (0.5) 4-5 0.19 (0.4) 0.21 (0.4) 0.17 (0.4) 0.16 (0.4) Male child 0.53 (0.5) 0.53 (0.5) 0.53 (0.5) 0.53 (0.5) Mother's education (0-1) No or other forms of education 0.43 (0.5) 0.50 (0.5) 0.39 (0.5) 0.32 (0.5) Primary education 0.23 (0.4) 0.20 (0.4) 0.26 (0.4) 0.26 (0.4) Secondary education 0.24 (0.4) 0.19 (0.4) 0.27 (0.4) 0.32 (0.5) Higher education 0.06 (0.2) 0.06 (0.2) 0.07 (0.3) 0.05 (0.2) Unknown education 0.04 (0.2) 0.05 (0.2) 0.02 (0.1) 0.05 (0.2) Household-level characteristics Age of household head (20-98) 42.74 (11.1) 42.27 (10.5) 43.56 (11.2) 42.93 (11.9) Male household head (0-1) 0.95 (0.2) 0.97 (0.2) 0.96 (0.2) 0.90 (0.3) Household size (2-26) 8.03 (3.3) 8.18 (3.3) 8.06 (3.9) 7.69 (2.8) Household dependency ratio (0-9) 1.63 (0.9) 1.63 (0.8) 1.55 (0.7) 1.70 (1) Occupation of household head (0-1) Unemployed 0.03 (0.2) 0.03 (0.2) 0.02 (0.1) 0.04 (0.2) Agriculture 0.57 (0.5) 0.61 (0.5) 0.55 (0.5) 0.52 (0.5) Sales and services 0.19 (0.4) 0.16 (0.4) 0.19 (0.4) 0.26 (0.4) Professional job 0.14 (0.4) 0.14 (0.4) 0.15 (0.4) 0.13 (0.3) Others 0.07 (0.3) 0.06 (0.2) 0.08 (0.3) 0.06 (0.2) Log of values of household wealth (6.62-16.15) 10.84 (1.5) 10.97 (1.4) 11.02 (1.4) 10.41 (1.6) Community-level characteristics Level of infrastructural development (0-8) 2.04 (2.4) 1.95 (2.4) 2.33 (2.5) 1.96 (2.4) Change in socioeconomic development (0-13) 5.75 (2.9) 5.6 (2.9) 5.69 (3.0) 6.09 (2.9) Urban residence (0-1) 0.26 (0.4) 0.22 (0.4) 0.28 (0.5) 0.30 (0.5) Region of residence (0-1) North-central 0.15 (0.4) 0.18 (0.4) 0.11 (0.3) 0.13 (0.3) North-east 0.2 (0.4) 0.20 (0.4) 0.29 (0.5) 0.14 (0.4) North-west 0.31 (0.5) 0.43 (0.5) 0.18 (0.4) 0.20 (0.4) South-south 0.14 (0.3) 0.08 (0.3) 0.20 (0.4) 0.20(0.4) South-east 0.12 (0.3) 0.05 (0.2) 0.16 (0.4) 0.24 (0.4) South-west 0.07 (0.3) 0.06 (0.2) 0.07 (0.3) 0.09 (0.3) Source: 2010 Nigeria General Household Survey; Children Aged 0-60 Months; Range in parentheses in column 1; Standard deviations in parentheses in
Columns 2, 3, 4, and 5; Number of communities = 417 and average number of children per community = 5
151
Table 4. 2. Multilevel Logistic Regression Models of Under-five Stunting in Nigeria (n = 2,078 children) Main Predictor Model 1 Model 2 Model 3 Model 4 Model 5 Fixed effects Intercept 1.28* 1.35* 3.94* 2.37** 9.38** Household food insecurity
(0=food secure) Moderately food insecure 0.98 1.03 1.39* 1.38* Severely food insecure 0.83 0.81 1.24 1.14 Individual-level predictors Child's age (0=0-1) 2-3 1.01 1.00 4-5 0.90 0.89 Male child 1.14 1.15 Mother's education
(0=primary education) No or other forms of education 2.07*** 1.24 Secondary education 0.87 1.04 Higher education 1.21 1.52 Unknown mother's education 1.48 1.12 Other household-level predictors Age of household head 0.98** 0.99 Male-headed households 1.11 0.86 Household size 1.04† 1.03 Household dependency ratio 1.01 0.99 Occupation of household head
(0=Agriculture) Unemployed 0.56 0.65 Sales and services 0.86 1.03 Professional jobs 0.96 0.95 Others 1.09 1.28 Log of values of household wealth 0.90* 0.88* Community-level characteristics Level of infrastructural development 1.05 1.05 Change in socioeconomic 0.94* 0.94* development Urban residence 0.55** 0.61* Region of residence (0=south-west) North-central 1.00 0.99 North-east 1.15 1.13 North-west 2.03* 1.85† South-south 0.16*** 0.16*** South-east 0.26*** 0.27*** Random effects Community-level variance (To) 2.21 2.13 1.53 1.29 1.23 Intra-class correlation coefficient (ICC) 0.40 0.39 0.32 0.28 0.27 Log likelihood -1289.4 -1288.7 -1262.7 -1228.7 -1220.6 Source: Nigeria General Household Survey (Panel), 2010-2013; *** p<0.001, ** p<0.01, * p<0.05, †p<0.1; To = variance of the random intercept at the community level; Number of communities = 417 and average number of children per community = 5
152
Table 4. 3. Descriptive Statistics for Variables in Cross sectional Analyses of Wasting among Nigerian Children Aged 0-60 Months by Household Food Security Status Moderately Severely All children Food secure food secure food insecure Number of children 1,795 892 429 474 Child outcome (0-1) Wasted 0.14 (0.3) 0.13 (0.3) 0.17 (0.4) 0.12 (0.3) Individual-level characteristics Child's age (0-1) 0-1 0.36 (0.5) 0.31 (0.5) 0.41 (0.5) 0.40 (0.5) 2-3 0.45 (0.5) 0.47 (0.5) 0.41 (0.5) 0.44 (0.5) 4-5 0.19 (0.4) 0.22 (0.4) 0.18 (0.4) 0.16 (0.4) Male child 0.53 (0.5) 0.54 (0.5) 0.52 (0.5) 0.53 (0.5) Mother's education (0-1) No or other forms of education 0.42 (0.5) 0.49 (0.5) 0.39 (0.5) 0.32 (0.5) Primary education 0.23 (0.4) 0.20 (0.4) 0.27 (0.4) 0.26 (0.4) Secondary education 0.24 (0.4) 0.20 (0.4) 0.25 (0.4) 0.31 (0.5) Higher education 0.06 (0.2) 0.06 (0.2) 0.07 (0.3) 0.06 (0.2) Unknown education 0.04 (0.2) 0.05 (0.2) 0.02 (0.1) 0.05 (0.2) Household-level characteristics Age of household head (20-98) 42.99 (11.3) 42.49 (10.8) 43.91 (11.4) 43.11 (12) Male household head (0-1) 0.95 (0.2) 0.97 (0.2) 0.96 (0.2) 0.89 (0.3) Household size (2-26) 8.06 (3.4) 8.23 (3.4) 8.17 (4.0) 7.66 (2.8) Household dependency ratio (0-9) 1.63 (0.9) 1.63 (0.9) 1.56 (0.8) 1.70 (1.0) Occupation of household head (0-1) Unemployed 0.03 (0.2) 0.03 (0.2) 0.02 (0.2) 0.04 (0.2) Agriculture 0.57 (0.5) 0.6 (0.5) 0.56 (0.5) 0.52 (0.5) Sales and services 0.2 (0.4) 0.16 (0.4) 0.20 (0.4) 0.26 (0.4) Professional job 0.14 (0.3) 0.14 (0.4) 0.14 (0.3) 0.12 (0.3) Others 0.07 (0.3) 0.06 (0.2) 0.08 (0.3) 0.06 (0.2) Log of values of household wealth (6.62-16.15) 10.85 (1.5) 11 (1.4) 11.03 (1.4) 10.42 (1.7) Community-level characteristics Level of infrastructural development (0-8) 2.07 (2.4) 1.97 (2.4) 2.32 (2.6) 2.03 (2.4) Change in socioeconomic development (0-13) 5.76 (2.9) 5.59 (2.9) 5.69 (3.0) 6.16 (2.9) Urban residence (0-1) 0.26 (0.4) 0.22 (0.4) 0.28 (0.5) 0.31 (0.5) Region of residence (0-1) North-central 0.12 (0.3) 0.15 (0.4) 0.08 (0.3) 0.11 (0.3) North-east 0.22 (0.4) 0.22 (0.4) 0.31 (0.5) 0.15 (0.4) North-west 0.31 (0.5) 0.43 (0.5) 0.17 (0.4) 0.20 (0.4) South-south 0.16 (0.4) 0.09 (0.3) 0.22 (0.4) 0.22 (0.4) South-east 0.13 (0.3) 0.06 (0.2) 0.17 (0.4) 0.23 (0.4) South-west 0.06 (0.2) 0.05 (0.2) 0.06 (0.2) 0.09 (0.3) Source: 2010 Nigeria General Household Survey; Children Aged 0-60 Months; Range in parentheses in column 1; Standard deviations in parentheses in
Columns 2, 3, 4, and 5; Number of communities = 387 and average number of children per community = 5
153
Table 4. 4. Multilevel Logistic Regression Models of Under-five Wasting in Nigeria (n = 1,795 children) Main Predictor Model 1 Model 2 Model 3 Model 4 Model 5 Fixed effects Intercept 0.09*** 0.09*** 0.22† 0.06*** 0.17† Household food insecurity
(0=food secure) Moderately food insecure 1.48† 1.54* 1.70* 1.77** Severely food insecure 1.01 0.99 1.27 1.21 Individual-level predictors Child's age (0=0-1) 2-3 0.66* 0.66* 4-5 0.53** 0.52** Male child 1.18 1.17 Mother's education
(0=primary education) No or other forms of education 0.84 0.62* Secondary education 0.87 1.02 Higher education 1.52 1.86 Unknown mother's education 2.46* 1.99 Other household-level predictors Age of household head 0.98* 0.99 Male-headed households 2.05 1.74 Household size 1.06* 1.04 Household dependency ratio 1.16 1.17 Occupation of household head
(0=Agriculture) Unemployed 0.80 0.79 Sales and services 0.89 1.11 Professional jobs 0.80 0.75 Others 1.10 1.24 Log of values of household wealth 0.90 0.89† Community-level characteristics Level of infrastructural development 0.99 1.00 Change in socioeconomic
development 0.96 0.95 Urban residence 1.01 0.91 Region of residence (0=south-west) North-central 2.81* 3.15* North-east 2.98* 3.16* North-west 2.10 2.46† South-south 0.75 0.69 South-east 1.42 1.28 Random effects Community-level variance (To) 1.35 1.34 1.31 1.14 1.10 Intra-class correlation coefficient (ICC) 0.29 0.29 0.28 0.26 0.25 Log likelihood -683.6 -681.4 -665.5 -670.5 -655.1 Source: Nigeria General Household Survey (Panel), 2010-2013; *** p<0.001, ** p<0.01, * p<0.05, †p<0.1; To = variance of the random intercept at the community level; Number of communities = 387 and average number of children per community = 5
154
Table 4. 5. Descriptive Statistics for Variables in Cross sectional Analyses of Underweight among Nigerian Children Aged 0-60 Months by Household Food Security Status Moderately Severely All children Food secure food secure food insecure Number of children 2,104 1,078 483 543 Child outcome (0-1) Underweight 0.36 (0.5) 0.39 (0.5) 0.41 (0.5) 0.27 (0.4) Individual-level characteristics Child's age (0-1) 0-1 0.36 (0.5) 0.32 (0.5) 0.41 (0.5) 0.39 (0.5) 2-3 0.45 (0.5) 0.47 (0.5) 0.41 (0.5) 0.44 (0.5) 4-5 0.19 (0.4) 0.21 (0.4) 0.18 (0.4) 0.17 (0.4) Male child 0.53 (0.5) 0.53 (0.5) 0.53 (0.5) 0.52 (0.5) Mother's education (0-1) No or other forms of education 0.43 (0.5) 0.49 (0.5) 0.39 (0.5) 0.33 (0.5) Primary education 0.23 (0.4) 0.21 (0.4) 0.26 (0.4) 0.26 (0.4) Secondary education 0.24 (0.4) 0.19 (0.4) 0.27 (0.4) 0.32 (0.5) Higher education 0.06 (0.2) 0.06 (0.2) 0.07 (0.3) 0.05 (0.2) Unknown education 0.04 (0.2) 0.04 (0.2) 0.02 (0.1) 0.05 (0.2) Household-level characteristics Age of household head (20-98) 42.85 (11.1) 42.45 (10.7) 43.63 (11.2) 42.96 (11.9) Male household head (0-1) 0.95 (0.2) 0.97 (0.2) 0.96 (0.2) 0.90 (0.3) Household size (2-26) 8.02 (3.3) 8.15 (3.3) 8.10 (3.9) 7.71 (2.8) Household dependency ratio (0-9) 1.63 (0.9) 1.63 (0.8) 1.55 (0.7) 1.71 (1) Occupation of household head (0-1) Unemployed 0.03 (0.2) 0.03 (0.2) 0.02 (0.1) 0.04 (0.2) Agriculture 0.57 (0.5) 0.60 (0.5) 0.56 (0.5) 0.52 (0.5) Sales and services 0.19 (0.4) 0.16 (0.4) 0.19 (0.4) 0.25 (0.4) Professional job 0.14 (0.4) 0.15 (0.4) 0.15 (0.4) 0.13 (0.3) Others 0.07 (0.3) 0.06 (0.2) 0.08 (0.3) 0.06 (0.2) Log of values of household wealth (6.62-16.15) 10.84 (1.5) 10.98 (1.4) 11.03 (1.4) 10.41 (1.6) Community-level characteristics Level of infrastructural development (0-8) 2.03 (2.4) 1.96 (2.4) 2.30 (2.5) 1.92 (2.4) Change in socioeconomic development (0-13) 5.73 (2.9) 5.58 (2.8) 5.70 (3) 6.05 (2.9) Urban residence (0-1) 0.25 (0.4) 0.22 (0.4) 0.28 (0.5) 0.30 (0.5) Region of residence (0-1) North-central 0.16 (0.4) 0.19 (0.4) 0.12 (0.3) 0.14 (0.4) North-east 0.2 (0.4) 0.2 (0.4) 0.29 (0.5) 0.14 (0.4) North-west 0.3 (0.5) 0.42 (0.5) 0.16 (0.4) 0.19 (0.4) South-south 0.14 (0.4) 0.08 (0.3) 0.20 (0.4) 0.21 (0.4) South-east 0.12 (0.3) 0.05 (0.2) 0.16 (0.4) 0.23 (0.4) South-west 0.07 (0.3) 0.06 (0.2) 0.07 (0.3) 0.09 (0.3) Source: 2010 Nigeria General Household Survey; Children Aged 0-60 Months; Range in parentheses in column 1; Standard deviations in parentheses in
Columns 2, 3, 4, and 5; Number of communities = 417 and average number of children per community = 5
155
Table 4. 6. Multilevel Logistic Regression Models of Under-five Underweight in Nigeria (n = 2,104 children) Main Predictor Model 1 Model 2 Model 3 Model 4 Model 5 Fixed effects Intercept 0.39*** 0.43*** 0.76 0.60 1.33 Household food insecurity
(0=food secure) Moderately food insecure 1.06 1.14 1.46* 1.46* Severely food insecure 0.63** 0.65** 0.90 0.86 Individual-level predictors Child's age (0=0-1) 2-3 0.77* 0.77* 4-5 0.64** 0.64** Male child 1.36** 1.36** Mother's education
(0=primary education) No or other forms of education 1.60** 0.98 Secondary education 0.73† 0.85 Higher education 0.95 1.23 Unknown mother's education 2.09* 1.56 Other household-level predictors Age of household head 0.98*** 0.99* Male-headed households 1.53 1.14 Household size 1.04† 1.02 Household dependency ratio 1.10 1.09 Occupation of household head
(0=Agriculture) Unemployed 0.81 0.90 Sales and services 0.64* 0.79 Professional jobs 0.85 0.86 Others 1.08 1.30 Log of values of household wealth 0.95 0.94 Community-level characteristics Level of infrastructural development 0.96 0.97 Change in socioeconomic
development 0.91*** 0.91*** Urban residence 0.82 0.89 Region of residence (0=south-west) North-central 1.40 1.34 North-east 2.39** 2.17* North-west 2.82*** 2.62** South-south 0.36** 0.37** South-east 0.56† 0.56 Random effects Community-level variance (To) 1.67 1.57 1.13 0.81 0.84 Intra-class correlation coefficient (ICC) 0.34 0.32 0.26 0.20 0.20 Log likelihood -1264.9 -1259.3 -1221.9 -1201.0 -1184.8 Source: Nigeria General Household Survey (Panel), 2010-2013; *** p<0.001, ** p<0.01, * p<0.05, †p<0.1; To = variance of the random intercept at the community level; Number of communities = 417 and average number of children per community = 5
156
CHAPTER 5: CONCLUSIONS
Adequate knowledge of household food security is essential because poverty influences nearly all demographic and health outcomes, including life expectancy, maternal and child mortality and morbidity, human capital development, fertility, contraceptive use, use of healthcare facilities, among others (Adeyemi et al. 2009; Bollen et al. 2002; Filmer and Pritchett 2001;
Gonzales et al. 2010; Gwatkin et al. 2007; Houweling et al. 2003; Montgomery et al. 2000).
Also, effective monitoring of stability and change in household socioeconomic conditions is crucial to policies and programs aimed at improving the living standards of the people (Howard
2011; Metallinos-Katsaras et al. 2012; Ryu et al. 2012).
Due to the dearth of panel data on food insecurity, research on the relationship between food insecurity and its correlates, as well as the underlying processes and mechanisms that modify those relationships in low income countries, is both “preliminary and partial” (Nord
2014: 3). There have been limited longitudinal analyses of the mechanisms underlying the experiences of food insecurity in relation to health and wellbeing in poor contexts like Nigeria.
The relationship between household food insecurity and other conventional measures of household socioeconomic wellbeing is also understudied. The existing studies of food insecurity and child health in Nigeria are mostly cross-sectional in design, descriptive in nature, and suffer from limited spatial and sociodemographic coverage.
The current study addresses these gaps in the literature on food insecurity and child health in developing contexts by employing longitudinal and multilevel analytic techniques to examine a wide range of individual-, household-, and community-level socioeconomic and demographic correlates of transitory and persistent food insecurity among nationally representative samples of Nigerian households and children. The study draws on three major
157 theoretical frameworks (resource perspectives, nutrition transition theory, and the family adaptive strategy framework) and the concept of social capital to provide new insights into processes underlying household food insecurity and their implications for child malnutrition.
Key Findings
The analyses in this dissertation are based on the recent and panel Nigeria General Household
Survey (NGHS) data collected between 2010 and 2013. The NGHS is particularly well-suited for the analyses because it collects extensive information about household socioeconomic conditions including food insecurity and household assets as well as sociodemographic characteristics that help expand existing knowledge of household food insecurity and child health. The community component of the survey also lends itself to multilevel approach to understanding the paradoxical high level of malnutrition among children residing in food secure households found in previous studies.
The second chapter (Chapter 2) of the dissertation focuses on food insecurity as an aspect of poverty using the multidimensional approach to poverty (Bollen et al. 2002; Howe et al.
2008). I analyzed the interrelationship between household food insecurity and a widely used measure of relative household socioeconomic status – household wealth. Compared to the
Demographic and Health Survey, the NGHS is unique in that it contains information about not just the ownership of an asset in the household, but also the numerical and actual naira
(currency) values of the different assets. The findings indicate that regardless of its measurement, household wealth constitutes an imperfect proxy for household socioeconomic conditions. I, however, showed an important avenue for improving the quality of data on household wealth. By taking into account the number and the quality (current estimated value) of household assets, values of household assets offer significant improvements over the conventional ownership of
158 assets in estimating household living standards. The value of assets was highly correlated with measures of both occasional (cross-sectional) and transitory experiences of food insecurity.
Nonetheless, the findings of the imperfect relationship between household food insecurity and household wealth reaffirm the multidimensional nature of socioeconomic wellbeing and the need to examine poverty in its different manifestations (Bevan and Joireman 1997; Falkingham and
Namazie 2002). The results of this chapter further shows the crucial role of panel data in understanding the experiences of food insecurity in developing countries like Nigeria. Whereas only half of Nigerian households were food insecure in 2010, the majority (76%) had experienced food insecurity by 2013. This is important because previous experiences of food insecurity may manifest in poor outcomes at a time when a household may have become food secure. The near-universal experiences of food insecurity in Nigeria have important implications for future research. Studies of behavioral and health outcomes in relation to household food insecurity should account for both the recent and distal experiences of food insecurity as well as the duration of each food insecurity episode.
An important correlate of food insecurity in previous studies is household composition
(Snyder et al. 2006). Having certain individuals (e.g. children and elderly persons living with disability) in the household seems to increase vulnerability to economic hardships and food insecurity. Along this line, food security scholars in developing countries have controlled for the dependency ratio in their analyses over the years. However, recent demographic shifts like increasing life expectancy and healthy living at older ages render treatment of all older adults as dependents inaccurate. Also, there are fewer older adults aged 65 and above in countries with low life expectancies. Moreover, the relationship between presence of children and household food insecurity in developing contexts like Nigeria may be more complicated than in the
159 developed world. On one hand, the practice of polygyny and its associated link to high fertility is perceived as a status symbol. On the other hand, a large number of children is normative in the lower socioeconomic strata partly because of the perceived economic returns to large family size through child labor. Further, many households with vulnerable populations manage to remain food secure while those with no vulnerable populations are food insecure, suggesting that certain factors buffer the effects of having vulnerable populations in the household on food insecurity.
In view of the above, I analyzed three mechanisms by which the presence of children and older adults with a disability in the household relates to persistent and transitory household food insecurity in Chapter 3 of this dissertation. The findings do not support the postulation of the demographic transition theory that children avert the risks of household food insecurity by contributing to household or farm labor. Rather, having children and elderly persons with a disability in the household was associated with significantly higher risks of food insecurity.
Unequal access to social capital and education did not explain the gaps in the experiences of food insecurity between households with vulnerable populations (children and older adults with a disability) and those without such populations. But, time spent collecting cooking fuel significantly reduced the effects of having children on the risks of severe food insecurity. Also, time spent fetching cooking fuel significantly modified the experiences of food insecurity among households with varying number of children. I found a declining protective effect of having fewer children on household food insecurity with increasing time spent fetching cooking fuel.
This means that without adequate access to cooking fuel, the benefit of having fewer children, in terms of household food security, may be minimal.
In Chapter 4, I revisited the association between food insecurity and child malnutrition.
The high rate of malnutrition among children growing up in food secure households suggests that
160 the link between household food insecurity and child malnutrition is not entirely direct.
Therefore, I drew on nutrition transition theory and used multilevel logistic regression modeling technique to evaluate how social context, measured in terms of community level infrastructural development and change in community socioeconomic development modifies the experiences of stunting, wasting, and underweight among children living in food secure households. The findings reveal a more complicated relationship between household food insecurity and child malnutrition than suggested in previous studies. Moderate experiences of household food insecurity seem to be more detrimental to child nutrition than severe food insecurity. Also, prior analyses masked the substantial community-level variations in under-five stunting, wasting, and underweight found in this study. Regardless of their food security statuses, Nigerian children living in rapidly developing communities were less likely to be stunted, wasted, and underweight than those in less developing areas.
Limitations
While this study provides new insights into food insecurity and child malnutrition, there are some limitations. First, I analyzed stability and change in household food insecurity over time using a standardized and well-tested measure. Even so, the duration of the experiences of food insecurity was limited to the seven days prior to the interview. Considering the magnitude of transitioning into and out of food insecurity reported in this study, retrospective reports of food insecurity episodes occurring between surveys and the duration of each episode would enhance the current understanding of the relationship between household food insecurity and wellbeing.
Also, I focused on first transitions into and out of food insecurity in the longitudinal analyses.
Future studies should further explore transitions across the different levels of food insecurity, particularly from moderate to severe food insecurity. Second, like other longitudinal data with
161 similar designs, the NGHS is affected by the problem of attrition. However, the attrition rates at the four NGHS rounds were 5% or lower. I also performed sensitivity tests to determine possible impacts of attrition on the findings. Third, the study is mostly based on households and not on individuals within each household. Even though household food insecurity and its correlates relate to individual members of the household, future studies should consider within-household variations in the experiences of food insecurity, particularly in relation to different patterns of household resource allocation. Fourth, the present study explored three mechanisms (social capital, food management skills, and time) by which food insecurity varies across households.
The analyses included the major sources of informal financial support found to be associated with poverty in previous studies. Future studies should explore the role of formal institutions in ameliorating food insecurity and assess the effects of differential food management skills and other aspects of time use on household food insecurity. Fifth, this study lays the foundation for the analyses of household structure and food insecurity in Nigeria by showing how presence of two vulnerable populations in the household relate to food insecurity. However, the list of vulnerable populations extends beyond children and elderly persons with disability to include pregnant women, lactating mothers, and unemployed adults. Lastly, to better understand the position of food deprivation relative to other aspects of household socioeconomic wellbeing, the meanings of the specific items in the Household Food Insecurity and Access Scale warrant more research attention. Households’ reports of their reliance on less preferred food, for instance, may be a function of not just their usual dietary patterns but also the feeding patterns within their socio-cultural climate. To the extent that norms about disclosure of household living conditions vary, reports of household food insecurity may also differ across sociocultural contexts.
162
Contributions
This dissertation contributes to existing knowledge of food insecurity and child health in three major ways. First, it provides a national portrait of household food insecurity in the largest and one of the most rapidly growing countries in sub-Saharan Africa – Nigeria. Previous studies of household food insecurity in this context largely rely on cross-sectional data with limited spatial scope. The profile of food insecure households presented in this study not only reflects the high level of spatial and sociodemographic heterogeneity across Nigeria, but also the dynamism of household struggle for survival in the recent years (2010-2013). Cross-sectional reports on household welfare point to high rate of poverty in Nigeria, but they did not come close to revealing the near-homogeneity of hardship in the country.
Second, this study broadens the existing discourse on household socioeconomic status in developing contexts by investigating how the experiences of household food insecurity or absolute deprivation compare to household socioeconomic status measured in relative terms as in household asset index. The analysis establishes that household food insecurity is not synonymous to material or economic hardship even though they are correlated. That a household contains electronic gadgets like television, radio, refrigerator, or even a car does not imply food sufficiency. Unsurprisingly, as this study shows, irregular financial supports through friends, family, and other informal groups do not contribute to household food insecurity as they did to other aspects of household economy (see Ezekiel 2014; Oleka and Eyisi 2014; Wanyama et al.
2008; Yusuf 2009).
Third, the present analysis of household food insecurity and child malnutrition illuminates the problematic findings of high rates of child malnutrition in food secure households in previous studies. Using multilevel modeling techniques, the study shows the relevance of both
163 household and contextual socioeconomic characteristics to child growth. While household food insecurity is important to child health, residence in an economically deteriorating community may retard gain in child development. Most importantly, that a child was found in a food secure household in a cross-sectional study does not necessarily mean that the child is food secure and vice versa. The transient nature of food insecurity in poor contexts like Nigeria mean that most children have a taste of destitution at some point in their lives. Perhaps the distinguishing factor in child outcome is the amount of time children spend in food insecure conditions which this study was unable to establish.
Summary
Research has established the multidimensional nature of poverty (Bollen et al. 2002; Howe et al.
2008) and the importance of the different socioeconomic facets in understanding health and wellbeing (Bevan and Joireman 1997; Falkingham and Namazie 2002). Yet, the Demographic and Health Survey (DHS), which serves as the major source of nationally representative data for studies of health and wellbeing in the developing world, contains limited information about household and individual socioeconomic characteristics. Most analyses of the DHS rely on respondents’ education, employment, and household wealth in assessing individuals’ and households’ socioeconomic standing relative to others in the study contexts. However, as this study has shown, household food insecurity, household wealth, employment, and even education present very different pictures of household socioeconomic wellbeing. In Nigeria, food insecure households are not predominantly uneducated or unemployed households. Neither were they
‘poor’, relative to their food secure counterparts, judging by the number of assets they had. In fact, the northern regions of the country that have been tagged the ‘poorest’ regions based on their limited accumulation of material wealth are more food secure than the southern regions that
164 are permeated with the materialistic western culture but mostly food insecure. This calls for a rethinking of the conceptualization and operationalization of household socioeconomic conditions in social studies. Policies and programs aimed at poverty alleviation also need to consider the multiple dimensions of household welfare in their interventions. Food insecurity is particularly important because food is a basic need for survival which every human is entitled to.
It is imperative that surveys like the DHS incorporate food insecurity into their measures. The conventional measure of household wealth could also be improved upon by including assessments of quantity and quality of each household asset.
Food insecurity is highly prevalent in Nigeria, but only a minority of households remain persistently food secure or food insecure. The food insecure households in cross-sectional studies
(about half of all households) includes households that were transitorily food insecure and the food secure households comprises households that have been and/or will be food insecure at other times. This underscores the importance of longitudinal data in helping to understand the dynamics of household food conditions. The transitory nature of food insecurity is particularly relevant to studies of health outcomes. Studies taking snapshots of household food insecurity and child growth at a single point in time revealed a concerning high rate of malnutrition among children residing in food secure households. But longitudinal approach presents a possible explanation for the above finding – the vast majority of children would experience food insecurity at some point before reaching adulthood.
In conclusion, this dissertation demonstrates the critical role of household food insecurity as a measure of absolute household socioeconomic condition and its distinction from other indicators of relative poverty. It also reveals the transitory nature of household food insecurity and its importance in studies of health and socioeconomic wellbeing. The analyses highlight two
165 vulnerable populations (children and elderly persons with a disability), whose presence in the household may be linked to heightened food insecurity. The findings further showcase the importance of reducing the amount of time it takes for households to access cooking fuel.
Finally, the study shows that experiences of household food insecurity, particularly moderate level of food insecurity, negatively impact child growth.
Future Research
This dissertation addresses some key questions, but there are many new directions for future research. The evolving panel data on household food insecurity in Nigeria and elsewhere opens up fertile research ground for food security scholars, particularly cross-national comparative analyses of food insecurity and health (physical, mental, and reproductive). The third wave of the
NGHS data collection (additional two rounds of data) is underway. The General Household
Survey is also being conducted in six other African countries (Uganda, Tanzania, Ethiopia,
Malawi, Mali, and Niger) as part of the Living Standards Measurement Study. With the increasing availability of high-quality panel data, the spatial scope of this dissertation could be broadened to include other developing and developed contexts.
Second, the present study of household socioeconomic wellbeing focuses on food insecurity and household wealth. Future research will greatly benefit from comprehensive assessments of deprivation, including water insecurity. The association between household income, household expenditure, and household food insecurity is yet to be established. Further, having established the interrelationships among the different measures of socioeconomic status, an avenue for future research is some combinations of multiple indicators of household socioeconomic wellbeing such as food and water deprivation.
166
Another key next step in this research is to employ multilevel modeling technique to further examine the role of social and economic contexts in shaping the experiences of household food insecurity and their impacts on health and wellbeing. Contextual factors such as community food prices, availability of, and accessibility to, food markets, community mode of subsistence, and community programs impacting food availability, should be included in future analyses because not only will they reveal the multiple levels of influence on household food insecurity, but they will also foster more effective interventions at the individual, household, and community levels. As pointed out earlier, within-family and within-household variations in the experiences of food insecurity would be captured in multilevel models. This study does not directly unravel the food dynamics in polygynous and multi-family households. A fruitful avenue for new work is to consider how a wide range of family contexts influence food insecurity and poverty. A detailed examination of the distribution of resources across complex families is warranted.
In this study, I analyzed three key indicators of child health – stunting, wasting, and underweight. There is, however, need for analyses of a broader range of adult and child health outcomes – physical, psychological, and reproductive – in relation to food insecurity among children and adults. The implications of food insecurity are potentially broad reaching, but poorly understood. This is important because effective policies to combat food insecurity and promote well-being require empirical analyses capturing these specific pathways of influence.
Lastly, an important correlate of household food insecurity highlighted throughout this dissertation is region of residence. Yet, there is limited theoretical underpinning for focused research on regional variations in the experiences of household food insecurity and child malnutrition in developing contexts. Understanding the regional differences in food insecurity
167 will help broaden the knowledge of household deprivation and child outcomes in Nigeria and in other developing nations.
Food insecurity is a serious issue and part of a larger set of social welfare concerns in many nations. Nigeria is undergoing rapid economic and development transformations, but the implications for the well-being of the population are not well established. Many countries across the globe have made tremendous progress in eradicating extreme poverty and hunger over the past decades. With much dedication and commitment, Nigerians and people everywhere can be food secure.
168
REFERENCES
Adebayo, P. F., & Ojo, E. O. (2012). Food security in Nigeria: An overview. European Journal
of Sustainable Development, 1(2), 199-221.
Adeyemi, S. L., Ijaiya, G. T., & Raheem, U. A. (2009). Determinants of poverty in sub-Saharan
Africa. African Research Review, 3(2), 162-177.
Afifi, M. (2009). Wealth Index association with gender issues and the reproductive health of
Egyptian women. Nursing & Health Sciences, 11(1), 29-36.
African Development Bank Group (2014). Nigeria Economic Outlook. Accessed March 20,
2015 at http://www.afdb.org/en/countries/west-africa/nigeria/nigeria-economic-outlook/
Ajani, S. R., Adebukola, B. C., & Oyindamola, Y. B. (2006). Measuring household food
insecurity in selected local government areas of Lagos and Ibadan, Nigeria. Pakistan
Journal of Nutrition, 5(1), 62-67.
Ajao, K. O., Ojofeitimi, E. O., Adebayo, A. A., Fatusi, A. O., & Afolabi, O. T. (2010). Influence
of family size, household food security status, and child care practices on the nutritional
status of under-five children in Ile-Ife, Nigeria. African Journal of Reproductive
Health, 14(4), 123-132.
Akinrinola, O. O., & Mafimisebi, T. E. (2010). Contributions of informal savings and credit
institutions to rural development: Evidence from Nigeria. Journal of Rural
Cooperation, 38(2), 187-202.
Akinyele, I. O. (2009). Ensuring food and nutrition security in rural Nigeria: An Assessment of
the challenges, information needs, and analytical capacity; 2010, International Food
Policy Research Institute, Nigeria Strategy Support Program (NSSP), 7, 1-90.
169
Alaimo, K., Briefel, R. R., Frongillo Jr, E. A., & Olson, C. M. (1998). Food insufficiency exists
in the United States: results from the third National Health and Nutrition Examination
Survey (NHANES III). American Journal of Public Health, 88(3), 419-426.
Alaimo, K., Olson, C. M., & Frongillo, E. A. (2001). Food insufficiency and American school-
aged children's cognitive, academic, and psychosocial development. Pediatrics, 108(1),
44-53.
Ali, D., Saha, K. K., Nguyen, P. H., Diressie, M. T., Ruel, M. T., Menon, P., & Rawat, R.
(2013). Household food insecurity is associated with higher child undernutrition in
Bangladesh, Ethiopia, and Vietnam, but the effect is not mediated by child dietary
diversity. The Journal of Nutrition, 143(12), 2015-2021.
Aliber, M. (2009). Exploring Statistics South Africa's national household surveys as sources of
information about household-level food security. Agrekon, 48(4), 384-409.
Alufohai, G. O., Ejenavi, F., & Koyenikan, M. J. (2012). Effect of credit on small ruminant
production in Delta state, Nigeria: Implications for sustainable development. OIDA
International Journal of Sustainable Development, 5(10), 91-100.
Amaza, P. S., Umeh, J. C., Helsen, J., & Adejobi, A. O. (2006). Determinants and measurement
of food insecurity in Nigeria: some empirical policy guide. International association of
agricultural economists annual meeting, August, pp. 12-18.
Anderson, P. M., Butcher, K. F., Hoynes, H. W., & Schanzenbach, D. W. (2014). Beyond
income: What else predicts very low food security among children? University of
Kentucky Center for Poverty Research Discussion Paper Series, DP2014-06. Accessed
June 22, 2016 at http://www.ukcpr.org/sites/www.ukcpr.org/files/documents/DP2014-
06_0.pdf
170
Anderson, S., & Baland, J. M. (2002). The economics of ROSCAS and intrahousehold resource
allocation. Quarterly Journal of Economics, 963-995.
Anema, A., Vogenthaler, N., Frongillo, E. A., Kadiyala, S., & Weiser, S. D. (2009). Food
insecurity and HIV/AIDS: current knowledge, gaps, and research priorities. Current
HIV/AIDS Reports, 6(4), 224-231.
Atoloye, A. T., Ogunba, B. O., & Samuel, F. (2015). Spatial pattern of household food insecurity
and childhood malnutrition in Akinyele local government area, Nigeria. International
Journal of Health Sciences, 3(1), 235-250.
Ayantoye, K., Yusuf, S., Omonona, B., & Amao, J. (2011). Food insecurity dynamics and its
correlates among rural households in south-western Nigeria. International Journal of
Agricultural Economics and Rural Development, 4(1), 43-55.
Babatunde, R. O., & Martinetti, E. C. (2011). Impacts of migrant remittances on food security
and nutrition of Farming Households in Kwara State, Nigeria. Contributed paper for the
International Conference: Shocks in Developing Countries. June 30–July 1, 2011, Paris,
France. Accessed June 1, 2015 at
http://pagesperso.dial.prd.fr/dial_pagesperso/dial_eve_shocks/pdf/150_Babatunde.pdf
Babatunde, R. O., Omotesho, O. A., & Sholotan, O. S. (2007). Socio-economic characteristics
and food security status of farming households in Kwara State, North-Central Nigeria.
Pakistan Journal of Nutrition, 6(1), 49-58.
Babatunde, R., Owotoki, G., Heidhues, F., & Buchenrieder, G. (2007). Vulnerability and food
insecurity differentials among male and female–headed farming households in
Nigeria. Pakistan Journal of Social Sciences, 4(3), 414-418.
171
Baig-Ansari, N., Rahbar, M. H., Bhutta, Z. A., & Badruddin, S. H. (2006). Child's gender and
household food insecurity are associated with stunting among young Pakistani children
residing in urban squatter settlements. Food & Nutrition Bulletin, 27(2), 114-127.
Balistreri, K. (2012). Family structure, work patterns and time allocations: Potential mechanisms
of food insecurity among children. University of Kentucky Center for Poverty Research
Discussion Paper Series, DP2012-07. Accessed June 22, 2016 at
http://uknowledge.uky.edu/cgi/viewcontent.cgi?article=1031&context=ukcpr_papers
Balogun, O. L., & Yusuf, S. A. (2011). Effect of social capital on welfare of rural households in
South Western States, Nigeria. Journal of American Science, 7(3), 506-514.
Bartfeld, J. S., & Ahn, H. M. (2011). The School Breakfast Program strengthens household food
security among low-income households with elementary school children. The Journal of
Nutrition, 141(3), 470-475.
Belachew, T., Hadley, C., Lindstrom, D., Gebremariam, A., Michael, K. W., Getachew, Y., ... &
Kolsteren, P. (2011). Gender differences in food insecurity and morbidity among
adolescents in southwest Ethiopia. Pediatrics, 127(2), e398-e405.
Belachew, T., Lindstrom, D., Gebremariam, A., Jira, C., Hattori, M. K., Lachat, C., ... &
Kolsteren, P. (2012). Predictors of chronic food insecurity among adolescents in
Southwest Ethiopia: A longitudinal study. BMC Public Health, 12(1), 1-11.
Berhane, G., & Gardebroek, C. (2011). Does microfinance reduce rural poverty? Evidence based
on household panel data from northern Ethiopia. American Journal of Agricultural
Economics, 93(1), 43-55.
Bevan, P., & Joireman, S. F. (1997). The perils of measuring poverty: Identifying the `poor' in
rural Ethiopia. Oxford Development Studies, 25(3), 315-343.
172
Bhargava, A., Jolliffe, D., & Howard, L. L. (2008). Socio-economic, behavioural and
environmental factors predicted body weights and household food insecurity scores in the
Early Childhood Longitudinal Study-Kindergarten. British Journal of Nutrition, 100(02),
438-444.
Bigsten, A., & Shimeles, A. (2008). Poverty transition and persistence in Ethiopia: 1994–
2004. World Development, 36(9), 1559-1584.
Bigsten, A., Kebede, B., Shimeles, A., & Taddesse, M. (2003). Growth and poverty reduction in
Ethiopia: Evidence from household panel surveys. World Development, 31(1), 87-106.
Birchall, J. (2003). Rediscovering the cooperative advantage: Poverty reduction through self-
help. Cooperative Branch International Labour Office, Geneva. Accessed May 31, 2015
at http://dspace.stir.ac.uk/bitstream/1893/3242/1/Birchallpoverty%20book%5B1%5D.pdf
Bollen, K. A., Glanville, J. L., & Stecklov, G. (2002). Economic status proxies in studies of
fertility in developing countries: Does the measure matter? Population Studies, 56(1), 81-
96.
Booysen, F., Van Der Berg, S., Burger, R., Von Maltitz, M., & Du Rand, G. (2008). Using an
asset index to assess trends in poverty in seven Sub-Saharan African countries. World
Development, 36(6), 1113-1130.
Boserup, E. (1985). Economic and demographic interrelationships in sub-Saharan
Africa. Population and Development Review, 383-397.
Caldwell, J. C. (2005). On net intergenerational wealth flows: An update. Population and
Development Review, 31(4), 721-740.
Callier, P. (1990). Informal finance: the rotating saving and credit association: An
interpretation. Kyklos, 43(2), 273-276.
173
Canagarajah, S., & Nielsen, H. S. (2001). Child labor in Africa: A comparative study. The
ANNALS of the American Academy of Political and Social Science, 575(1), 71-91.
Carter, M. R., & May, J. (1999). Poverty, livelihood and class in rural South Africa. World
Development, 27(1), 1-20.
Chen, S., & Ravallion, M. (2004). How have the world's poorest fared since the early 1980s? The
World Bank Research Observer, 19(2), 141-169.
Chen, S., & Ravallion, M. (2007). Absolute poverty measures for the developing world, 1981–
2004. Proceedings of the National Academy of Sciences, 104(43), 16757-16762.
Clarke, P. (2008). When can group level clustering be ignored? Multilevel models versus single-
level models with sparse data. Journal of Epidemiology and Community Health, 62(8),
752-758.
Coates, J., Swindale, A., & Bilinsky, P. (2007). Household Food Insecurity Access Scale
(HFIAS) for measurement of food access: Indicator guide. Washington, DC: Food and
Nutrition Technical Assistance Project, Academy for Educational Development.
Cole, S. M., & Tembo, G. (2011). The effect of food insecurity on mental health: Panel evidence
from rural Zambia. Social Science & Medicine, 73(7), 1071-1079.
Coleman-Jensen, A., Gregory, C., & Singh, A. (2014). Household food security in the United
States in 2013, ERR-173, U.S. Department of Agriculture, Economic Research Service,
September.
Coleman-Jensen, A., Rabbitt, M. P., Gregory, C., & Singh, A. (2015). Household Food Security
in the United States in 2014, ERR-194, U.S. Department of Agriculture, Economic
Research Service, September.
174
Cook, J. T., Frank, D. A., Casey, P. H., Rose-Jacobs, R., Black, M. M., Chilton, M., ... & Cutts,
D. B. (2008). A brief indicator of household energy security: Associations with food
security, child health, and child development in US infants and toddlers. Pediatrics,
122(4), e867-e875.http://www.rti.org/pubs/full_hunger_report_final_07-24-14.pdf
Cook, J. T., Frank, D. A., Levenson, S. M., Neault, N. B., Heeren, T. C., Black, M. M., ... &
Chilton, M. (2006). Child food insecurity increases risks posed by household food
insecurity to young children's health. The Journal of Nutrition, 136(4), 1073-1076.
Dean, W. R., Sharkey, J. R., & Johnson, C. M. (2011). Food insecurity is associated with social
capital, perceived personal disparity, and partnership status among older and senior adults
in a largely rural area of central Texas. Journal of Nutrition in Gerontology and
Geriatrics, 30(2), 169-186.
Deaton, A., & Stone, A. A. (2014). Evaluative and hedonic wellbeing among those with and
without children at home. Proceedings of the National Academy of Sciences, 111(4),
1328-1333.
Devereux, S., & Maxwell, S. (2001). Food security in Sub-Saharan Africa. Pietermaritzburg,
South Africa: University of Natal Press.
Devine, C. M., Jastran, M., Jabs, J., Wethington, E., Farell, T. J., & Bisogni, C. A. (2006). “A lot
of sacrifices:” Work–family spillover and the food choice coping strategies of low-wage
employed parents. Social Science & Medicine, 63(10), 2591-2603.
Dewing, S., Tomlinson, M., le Roux, I. M., Chopra, M., & Tsai, A. C. (2013). Food insecurity
and its association with co-occurring postnatal depression, hazardous drinking, and
suicidality among women in peri-urban South Africa. Journal of Affective Disorders,
150(2), 460-465.
175
Diamond-Smith, N., Raj, A., Prata, N., & Weiser, S. (2015). Household food insecurity and
women’s health in Nepal. Paper presented at the 2015 Annual meeting of the Population
Association of America. San Diego: California.
Ezekiel, P. O. (2014). A study on co-operative societies, poverty reduction and sustainable
development in Nigeria. IOSR Journal of Business and Management, 16(6), 132-140.
Falkingham, J., & Namazie, C. (2002). Measuring health and poverty: A review of approaches to
identifying the poor. London: DFID Health Systems Resource Centre. Accessed June 13,
2015 at http://www.heart-resources.org/wp-content/uploads/2012/10/Measuring-health-
and-poverty.pdf
FAO, IFAD and WFP (2013). The state of food insecurity in the world 2013: The multiple
dimensions of food security. Rome: FAO. Accessed June 14, 2015 at
http://www.fao.org/docrep/018/i3434e/i3434e.pdf
Filmer, D., & Pritchett, L. H. (2001). Estimating wealth effects without expenditure data-or tears:
An application to educational enrollments in states of India. Demography, 38(1), 115-
132.
Food and Agriculture Organization (FAO) of the United Nations (2014). The state of food
insecurity in the world: Strengthening the enabling environment for food security and
nutrition. Accessed May 30, 2015 at http://www.fao.org/3/a-i4030e.pdf
Foster, J. E. (1998). Absolute versus relative poverty. American Economic Review, 88(2), 335-
341.
Garrett, J. L., & Ruel, M. T. (1999). Are determinants of rural and urban food security and
nutritional status different? Some insights from Mozambique. World
Development, 27(11), 1955-1975.
176
Godfray, H. C. J., Beddington, J. R., Crute, I. R., Haddad, L., Lawrence, D., Muir, J. F., ... &
Toulmin, C. (2010). Food security: the challenge of feeding 9 billion people. Science,
327(5967), 812-818.
González, C., Houweling, T. J., Marmot, M. G., & Brunner, E. J. (2010). Comparison of
physical, public and human assets as determinants of socioeconomic inequalities in
contraceptive use in Columbia - moving beyond the household wealth index.
International Journal for Equity in Health, 910-21.
Green-Lapierre, R. J., Williams, P. L., Glanville, N. T., Norris, D., Hunter, H. C., & Watt, C. G.
(2011). Learning from" knocks in life": Food insecurity among low-income lone senior
women. Journal of Aging Research, 2012, 450630-450630.
Gundersen, C. (2013). Food insecurity is an ongoing national concern. Advances in Nutrition: An
International Review Journal, 4(1), 36-41.
Gwatkin, D. R., Rutstein, S., Johnson, K., Suliman, E., Wagstaff, A., & Amouzou, A. (2007).
Socio-economic differences in health, nutrition, and population within developing
countries: An overview. Accessed June 13, 2015 at
http://siteresources.worldbank.org/INTPAH/Resources/IndicatorsOverview.pdf
Hadley, C. (2014). Food Insecurity. The Wiley Blackwell Encyclopedia of Health, Illness,
Behavior, and Society.
Hadley, C., Lindstrom, D., Tessema, F., & Belachew, T. (2008). Gender bias in the food
insecurity experience of Ethiopian adolescents. Social Science & Medicine, 66(2), 427-
438.
Hanmer, L. C., Pyatt, G., & White, H. (1999). What do the World Bank's Poverty Assessments
teach us about Poverty in Sub-Saharan Africa? Development & Change, 30(4), 795-823.
177
Hoddinott, J., & Yohannes, Y. (2002). Dietary diversity as a food security indicator. Food
consumption and nutrition division discussion paper, 136, 2002.
Hodler, R. (2006). The curse of natural resources in fractionalized countries. European
Economic Review, 50(6), 1367-1386.
Houweling, T. A., Kunst, A. E., & Mackenbach, J. P. (2003). Measuring health inequality among
children in developing countries: Does the choice of the indicator of economic status
matter? International Journal for Equity in Health, 2(1), 8.
Howard, L. L. (2011). Does food insecurity at home affect non-cognitive performance at school?
A longitudinal analysis of elementary student classroom behavior. Economics of
Education Review, 30(1), 157-176.
Howe, L. D., Hargreaves, J. R., & Huttly, S. A. (2008). Issues in the construction of wealth
indices for the measurement of socio-economic position in low-income
countries. Emerging Themes in Epidemiology, 51-14.
Ijarotimi, O. S., & Odeyemi, A. O. (2012). Prevalence of food insecurity among rural
communities and its effects on nutritional status of children (8-15 years) in Ondo State,
Nigeria. Journal of Medicine and Medical Sciences, 3(1), 5-15.
Ijarotimi, O. S., & Oyeneyin, O. O. (2005). Effect of economy restructuring on household food
security and nutritional status of Nigerian children. Journal of Food Agriculture and
Environment, 3(3/4), 27.
International Food Policy Research Institute (IFPRI) (2012). Reducing the risk of food and
nutrition insecurity among vulnerable populations. Accessed May 25, 2015 at
http://www.ifpri.org/sites/default/files/publications/ifrcpaper.pdf
178
Ivers, L. C., & Cullen, K. A. (2011). Food insecurity: Special considerations for women. The
American Journal of Clinical Nutrition, 94(6), 1740S-1744S.
Jensen, H. & Zhylyevskyy, O. (2013). The effect of household financial, time and environmental
constraints on very low food security among children. University of Kentucky Center for
Poverty Research Discussion Paper Series, DP2013-05. Accessed June 22, 2016 at
http://uknowledge.uky.edu/cgi/viewcontent.cgi?article=1015&context=ukcpr_paper
s
Joshi, N. P., Maharjan, K. L., & Piya, L. (2010). Poverty and food insecurity in Nepal: A review.
Munich Personal RePEc Archive (MPRA), Paper No. 35387 1-19.Accessed June 13,
2015 at http://mpra.ub.uni-muenchen.de/35387/1/MPRA_paper_35387.pdf
Joyce, K. M., Breen, A., Ettinger de Cuba, S., Cook, J. T., Barrett, K. W., Paik, G., ... & Frank,
D. A. (2012). Household hardships, public programs, and their associations with the
health and development of very young children: Insights from children’s
HealthWatch. Journal of Applied Research on Children: Informing Policy for Children at
Risk, 3(1), 4.
Kahsay, A., Mulugeta, A., Seid, O. (2015). Nutritional Status of Children (6-59 Months) from
Food Secure and Food Insecure Households in Rural Communities of Saesie Tsaeda-
Emba District, Tigray, North Ethiopia: Comparative Study. International Journal of
Nutrition and Food Sciences, 4(1), 51-65.
Kalichman, S. C., Watt, M., Sikkema, K., Skinner, D., & Pieterse, D. (2012). Food insufficiency,
substance use, and sexual risks for HIV/AIDS in informal drinking establishments, Cape
Town, South Africa. Journal of Urban Health, 89(6), 939-951.
179
Knueppel, D., Demment, M., & Kaiser, L. (2010). Validation of the household food insecurity
access scale in rural Tanzania. Public Health Nutrition, 13(03), 360-367.
Kurukulasuriya, P., Mendelsohn, R., Hassan, R., Benhin, J., Deressa, T., Diop, M., ... &
Mahamadou, A. (2006). Will African agriculture survive climate change? The World
Bank Economic Review, 20(3), 367-388.
Lamidi, E. O. (2015). State variations in women’s socioeconomic status and use of modern
contraceptives in Nigeria. PloS one, 10(8), e0135172.
Laraia, B. A., Siega-Riz, A. M., Gundersen, C., & Dole, N. (2006). Psychosocial factors and
socioeconomic indicators are associated with household food insecurity among pregnant
women. The Journal of Nutrition, 136(1), 177-182.
Lindelow, M. (2006). Sometimes more equal than others: how health inequalities depend on the
choice of welfare indicator. Health Economics, 15(3), 263-279.
Liverpool-Tasie, L. S., Kuku, O., & Ajibola, A. (2011). Review of literature on agricultural
productivity, social capital and food security in Nigeria (No. 21). International Food
Policy Research Institute (IFPRI).
Lovenheim, M. F., & Mumford, K. J. (2013). Do family wealth shocks affect fertility choices?
Evidence from the housing market. Review of Economics and Statistics, 95(2), 464-475.
Lwanga, F., Wanyenze, R. K., Matovu, J. K., & Orach, C. G. (2015). Food security and
nutritional status of children residing in sugarcane growing communities of east-central
Uganda: a cross-sectional study. Journal of Food Security, 3(2), 34-39.
Manning, W. D., & Brown, S. (2006). Children’s economic well‐being in married and cohabiting
parent families. Journal of Marriage and Family, 68(2), 345-362.
180
Martin, M. A., & Lippert, A. M. (2012). Feeding her children, but risking her health: The
intersection of gender, household food insecurity and obesity. Social Science &
Medicine, 74(11), 1754-1764.
Maxwell, D. (1999). The political economy of urban food security in Sub-Saharan Africa. World
Development, 27(11), 1939-1953.
Mberu, B. U., & Reed, H. E. (2014). Understanding subgroup fertility differentials in
Nigeria. Population Review, 53(2), 23-46.
McIntyre, L., Glanville, N. T., Raine, K. D., Dayle, J. B., Anderson, B., & Battaglia, N. (2003).
Do low-income lone mothers compromise their nutrition to feed their children? Canadian
Medical Association Journal, 168(6), 686-691.
McKenzie, D. J. (2005). Measuring inequality with asset indicators. Journal of Population
Economics, 18(2), 229-260.
Metallinos-Katsaras, E., Gorman, K. S., Wilde, P., & Kallio, J. (2011). A longitudinal study of
WIC participation on household food insecurity. Maternal and Child Health
Journal, 15(5), 627-633.
Metallinos-Katsaras, E., Must, A., & Gorman, K. (2012). A longitudinal study of food insecurity
on obesity in preschool children. Journal of the Academy of Nutrition and
Dietetics, 112(12), 1949-1958.
Miller, C. M., Tsoka, M., & Reichert, K. (2011). The impact of the Social Cash Transfer Scheme
on food security in Malawi. Food Policy, 36(2), 230-238.
Miracle, M. P., Miracle, D. S., & Cohen, L. (1980). Informal savings mobilization in
Africa. Economic Development and Cultural Change, 28(4), 701-724.
181
Mishra, V., Bignami-Van Assche, S., Greener, R., Vaessen, M., Hong, R., Ghys, P. D., ... &
Rutstein, S. (2007). HIV infection does not disproportionately affect the poorer in sub-
Saharan Africa. Aids, 21, S17-S28.
Moen, P., & Wethington, E. (1992). The concept of family adaptive strategies. Annual Review of
Sociology, 18233-251.
Montgomery, M. R., Gragnolati, M., Burke, K. A., & Paredes, E. (2000). Measuring living
standards with proxy variables. Demography, 37(2), 155-174.
Muga, G. O., & Onyango-Ouma, W. (2009). Changing household composition and food security
among the elderly caretakers in rural western Kenya. Journal of Cross-Cultural
Gerontology, 24(3), 259-272.
National Bureau of Statistics (2015). Basic information document, Nigeria General Household
Survey–Panel 2012/13. Accessed March 10, 2015, at
http://siteresources.worldbank.org/INTLSMS/Resources/3358986-
1233781970982/5800988-1282216357396/7337519-1388758418241/GHS-
Panel_BINFO_2012_13_Feb_2015_new.pdf
National Population Commission (NPC) [Nigeria] and ICF International. (2014). Nigeria
Demographic and Health Survey 2013. Abuja, Nigeria, and Rockville, Maryland, USA:
NPC and ICF International. Accessed April 9, 2015 at
https://dhsprogram.com/pubs/pdf/FR293/FR293.pdf
Nkpoyen, F., & Bassey, G. E. (2012). Micro-lending as an empowerment strategy for poverty
alleviation among women in Yala Local Government Area of Cross River State,
Nigeria. International Journal of Business and Social Science, 3(18), 233-241.
182
Nord, M. (2012). How much does the Supplemental Nutrition Assistance Program alleviate food
insecurity? Evidence from recent programme leavers. Public Health Nutrition, 15(05),
811-817.
Nord, M. (2014). What have we learned from two decades of research on household food
security? Public Health Nutrition, 17(01), 2-4.
Nord, M., & Kantor, L. S. (2006). Seasonal variation in food insecurity is associated with heating
and cooling costs among low-income elderly Americans. The Journal of
Nutrition, 136(11), 2939-2944.
Obayelu, E. A. (2006). Global food price increases and nutritional status of Nigerians: The
determinants, coping strategies, policy Responses and implications. ARPN Journal of
Agricultural and Biological Science, 5(2), 67-80.
Oganah, B., & Nwabah, N. I. (2009). Nutrition transition and its consequences: Implications for
the Nigerian child. International Journal of Home Economics, 2(2), 20-31.
Ogwumike F. O. (2001). An appraisal of poverty reduction strategies in Nigeria. Economic and
Financial Review, 39(4), 45-71.
Okunmadewa, F. Y., Yusuf, S. A., & Omonona, B. T. (2007). Effects of social capital on rural
poverty in Nigeria. Pakistan Journal of Social Sciences, 4(3), 331-339.
Oldewage-Theron, W. H., Dicks, E. G., & Napier, C. E. (2006). Poverty, household food
insecurity and nutrition: Coping strategies in an informal settlement in the Vaal Triangle,
South Africa. Public Health, 120(9), 795-804.
Oleka, C., & Eyisi, N. (2014). The effect of informal financial institutions on poverty alleviation
in Nigeria. Journal of Economics and Sustainable Development, 5(6), 100-107.
183
Olowe, O. O., Okunmadewa, F., & Yusuf, S. (2014). Effects of social capital on well-being of
rural households in southwest Nigeria. Journal of Agriculture and Veterinary Science,
7(12), 13-21.
Olson, C. M. (1999). Nutrition and health outcomes associated with food insecurity and
hunger. The Journal of Nutrition, 129(2), 521S-524S.
Oluwatayo, I. B. (2009). Towards assuring households-food security in rural Nigeria: Have
cooperatives got any place. Journal of Agricultural Economics and Rural Development
(JAERD), 2(1), 52-60.
Omonona, T. B., & Adetokunbo, G. (2007). An analysis of food security situation among
Nigerian urban households: evidence from Lagos State, Nigeria. Journal of Central
European Agriculture, 8(3), 397-406.
Ordinioha, B., & Brisibe, S. (2013). Urbanization, household food security and childhood
malnutrition: A comparison of two communities in Rivers State, South-South
Nigeria. Journal of Food Security, 1(1), 1-5.
Orubuloye, I. O., Caldwell, J. C., & Caldwell, P. (1993). The role of religious leaders in
changing sexual behaviour in Southwest Nigeria in an era of AIDS. Health Transition
Review, 93-104.
Osei, A., Pandey, P., Spiro, D., Nielson, J., Shrestha, R., Talukder, Z., ... & Haselow, N. (2010).
Household food insecurity and nutritional status of children aged 6 to 23 months in
Kailali District of Nepal. Food and Nutrition Bulletin, 31(4), 483-494.
Otaha, I. J. (2013). Food insecurity in Nigeria: Way forward. African Research Review, 7(4), 26-
35.
184
Oyefara, J. L. (2007). Food insecurity, HIV/AIDS pandemic and sexual behaviour of female
commercial sex workers in Lagos metropolis, Nigeria. SAHARA-J: Journal of Social
Aspects of HIV/AIDS, 4(2), 626-635.
Ozughalu, U. M., & Ogwumike, F. O. (2015). Food poverty profile for Nigeria. The Journal of
Developing Areas, 49(2), 183-201.
Pilkauskas, N. V., Garfinkel, I., & McLanahan, S. S. (2014). The prevalence and economic value
of doubling up. Demography, 51(5), 1667-1676.
Popkin, B. M. (1993). Nutritional patterns and transitions. Population and Development Review,
138-157.
Popkin, B. M., Adair, L. S., & Ng, S. W. (2012). Global nutrition transition and the pandemic of
obesity in developing countries. Nutrition Reviews, 70(1), 3-21.
Portes, A. (1998). Social capital: Its origins and application in modern society. Sociology, 24, 25-
56.
Quandt, S. A., & Rao, P. (1999). Hunger and food security among older adults in a rural
community. Human Organization, 58(1), 28-35.
Qvortrup, J. (1997). A voice for children in statistical and social accounting: A plea for
children’s right to be heard. Constructing and Reconstructing Childhood: Contemporary
Issues in the Sociological Study of Childhood, 85-106.
Radimer, K. L., Olson, C. M., & Campbell, C. C. (1990). Development of indicators to assess
hunger. The Journal of Nutrition, 120, 1544-1548.
Ramsey, R., Giskes, K., Turrell, G., & Gallegos, D. (2011). Food insecurity among Australian
children Potential determinants, health and developmental consequences. Journal of
Child Health Care, 15(4), 401-416.
185
Ravallion, M., & Chen, S. (1997). What can new survey data tell us about recent changes in
distribution and poverty? World Bank Economic Review, 11(2), 357-382.
Reinhard, I., & Wijeratne, K. B. S. (2000). The use of stunting and wasting as indicators for food
insecurity and poverty. Poverty Impact Monitoring Unit, Integrated Food Security
Programme. Trincomalee: Sri Lanka. Access June 15, 2015 at
http://www.sas.upenn.edu/~dludden/stunting-wasting.pdf
Remans, R., Pronyk, P. M., Fanzo, J. C., Chen, J., Palm, C. A., Nemser, B., ... & Sachs, J. D.
(2011). Multisector intervention to accelerate reductions in child stunting: an
observational study from 9 sub-Saharan African countries. The American Journal of
Clinical Nutrition, 94(6), 1632-1642.
Rose, A. M., Hall, C. S., & Martinez-Alier, N. (2014). Aetiology and management of
malnutrition in HIV-positive children. Archives of Disease in Childhood, archdischild-
2012.
Rose, D. (1999). Economic determinants and dietary consequences of food insecurity in the
United States. The Journal of Nutrition, 129(2), 517S-520S.
RTI International (2014). Current and prospective scope of hunger and food security in America:
A Review of Current Research. Accessed May 25, 2015 at
http://www.rti.org/pubs/full_hunger_report_final_07-24-14.pdf
Ryu, J. H., & Bartfeld, J. S. (2012). Household food insecurity during childhood and subsequent
health status: The early childhood longitudinal study—kindergarten cohort. American
Journal of Public Health, 102(11), e50-e55.
Sahn, D. E., & Stifel, D. (2003). Exploring alternative measures of welfare in the absence of
expenditure data. Review of Income & Wealth, 49(4), 463-489.
186
Sanderson, W. C., & Scherbov, S. (2015). Are We Overly Dependent on Conventional
Dependency Ratios? Population and Development Review, 41(4), 687-708.
Schmidhuber, J., & Tubiello, F. N. (2007). Global food security under climate change.
Proceedings of the National Academy of Sciences, 104(50), 19703-19708.
Senbanjo, I. O., Oshikoya, K. A., Odusanya, O. O., & Njokanma, O. F. (2011). Prevalence of
and risk factors for stunting among school children and adolescents in Abeokuta,
southwest Nigeria. Journal of Health, Population, and Nutrition, 29(4), 364.
Service Canada (2014). Income assistance for families and children and for seniors. Accessed
May 30, 2015 from http://www.servicecanada.gc.ca/eng/subjects/benefits/
Sirotin, N., Hoover, D. R., Shi, Q., Anastos, K., & Weiser, S. D. (2014). Food insecurity with
hunger is associated with obesity among HIV-Infected and at risk women in Bronx, NY.
PloS One, 9(8), e105957.
Snyder, A. R., McLaughlin, D. K., & Findeis, J. (2006). Household composition and poverty
among female‐headed households with children: Differences by race and
residence. Rural Sociology, 71(4), 597-624.
Stevens, A. H. (1994). The dynamics of poverty spells: Updating Bane and Ellwood. The
American Economic Review, 84(2), 34-37.
Stevens, C. A. (2010). Exploring food insecurity among young mothers (15–24 years). Journal
for Specialists in Pediatric Nursing, 15(2), 163-171.
Sumarto, S., Suryadarma, D., & Suryahadi, A. (2007). Predicting consumption poverty using
non-consumption indicators: Experiments using Indonesian data. Social Indicators
Research, 81(3), 543-578.
187
Szabo, S., Hossain, M. S., Adger, W. N., Matthews, Z., Ahmed, S., Lázár, A. N., & Ahmad, S.
(2015). Soil salinity, household wealth and food insecurity in tropical deltas: Evidence
from south-west coast of Bangladesh. Sustainability Science, 1-11.
Tertilt, M. (2005). Polygyny, fertility, and savings. Journal of Political Economy, 113(6), 1341-
1371.
The World Bank (2006). Repositioning nutrition as central to development: A strategy for large-
scale action. Washington, DC: The International Bank for Reconstruction and
Development/The World Bank.
The World Bank (2008). Measurement of Living Standards. Washington, D.C.: World Bank.
Accessed May 13, 2016 at
http://siteresources.worldbank.org/INTPAH/Resources/Publications/459843-
1195594469249/HealthEquityCh6.pdf
The World Bank (2010). Food price watch, December, 2010. Poverty Reduction and Equity
group of the World Bank. Accessed May 19, 2015 http://www-
wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2011/11/21/00038619
4_20111121005124/Rendered/PDF/656480NEWS00PU0oodPriceWatchDec2010.pdf
The World Bank (2011). Food price watch, Februrary, 2011. Poverty reduction and equity group
of the World Bank. Accessed May 19, 2015 http://www-
wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2011/09/05/00033303
7_20110905015942/Rendered/PDF/643610BRI0Food00Box0362535B0PUBLIC0.pdf
The World Bank (2011b). Food price watch, April, 2011. Poverty Reduction and equity group of
the World Bank. Accessed May 19, 2015 http://www-
188
wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2011/09/20/00038619
4_20110920023846/Rendered/PDF/644590NEWS0Foo00Box0361538B0PUBLIC0.pdf
The World Bank (2014a). 2014 World Development Indicators. Accessed March 20, 2015, at
http://data.worldbank.org/sites/default/files/wdi-2014-book.pdf
The World Bank (2014b). Nigeria Economic Report: Improved economic outlook in 2014, and
prospects for continued growth look good. Accessed June 14, 2015 at
https://www.worldbank.org/en/country/nigeria/publication/nigeria-economic-report-
improved-economic-outlook-in-2014-and-prospects-for-continued-growth-look-good
Timæus, I. M. (2012). Stunting and obesity in childhood: A reassessment using longitudinal data
from South Africa. International Journal of Epidemiology, 41(3), 764-772.
Togunde, D., & Carter, A. (2006). Socioeconomic causes of child labor in urban
Nigeria. Journal of Children & Poverty, 12(1), 73-89.
Tsai, A. C., & Weiser, S. D. (2014). Population-based study of food insecurity and HIV
transmission risk behaviors and symptoms of sexually transmitted infections among
linked couples in Nepal. AIDS and Behavior, 18(11), 2187-2197.
UNICEF (2011). Under-five mortality rankings -2010. Accessed October 18, 2014 at
http://www.unicef.org/sowc2012/pdfs/UNDER-FIVE-MORTALITY-RANKINGS.pdf
UNICEF (2014). Levels and trends in child mortality. New York: United Nations children’s
fund. Accessed May 11, 2015, at
http://www.unicef.org/media/files/Levels_and_Trends_in_Child_Mortality_2014.pdf
United Nations Press Release (2013). World population projected to reach 9.6 billion by 2050
with most growth in developing regions, especially Africa. Accessed June 13, 2015 at
http://esa.un.org/wpp/Documentation/pdf/WPP2012_Press_Release.pdf
189
United Nations, Department of Economic and Social Affairs, Population Division (2014). World
Urbanization Prospects: The 2014 Revision, Highlights (ST/ESA/SER.A/352). Accessed
March 20, 2015, at http://esa.un.org/unpd/wup/Highlights/WUP2014-Highlights.pdf
USDA Food and Nutrition Service (2015). Women, Infants, and Children (WIC).
http://www.fns.usda.gov/wic/women-infants-and-children-wic
Vorster, H. H., Kruger, A., & Margetts, B. M. (2011). The nutrition transition in Africa: can it be
steered into a more positive direction? Nutrients, 3(4), 429-441.
Vozoris, N. T., & Tarasuk, V. S. (2003). Household food insufficiency is associated with poorer
health. The Journal of Nutrition, 133(1), 120-126.
Wanyama, F. O., Develtere, P., & Pollet, I. (2008). Encountering the evidence: Co-operatives
and poverty reduction in Africa. Journal of Co-operative Studies, 41(3), 16-27.
Woltil, J. (2012). The impact of emotional social support on elders' food security. Sociation
Today, 10(2).
Yusuf, N., Ijaiya, G. T., & Ijaiya, M. A. (2009). Informal financial institutions and poverty
reduction in the informal sector of Offa Town, Kwara State: A case study of Rotating
Savings and Credit Associations (ROSCAs). J Soc Sci, 20(1), 71-81.
Zeller, M., Sharma, M., Henry, C., & Lapenu, C. (2001). An operational tool for evaluating
poverty outreach of development policies and projects. Washington, DC: International
Food Policy Research Institute.
190
APPENDIX A
Appendix A. Measures of Food Insecurity Frequency of occurrence Never Rarely Sometimes Often Occurrence of food insecurity (0 day) (1 day) (2-3 days) (4-7days) Relied on less preferred foods Limited the variety of food eaten Limited portion size at meal times Reduced number of meals eaten in a day Restricted consumption to allow small children to eat Borrowed food, or relied on help from a friend or relative Had no food of any kind in the household Went to sleep at night hungry because there was not enough food Went a whole day and night without eating anything Adapted form of Household Food Insecurity Access Scale (HFIAS) for measurement of food access Moderately Severely Food secure food insecure food insecure
191
APPENDIX B
Appendix B. Weights assigned to each item in the asset index (First component score coefficient) Item Weight Furniture (3/4 piece sofa set) 0.215 Furniture (chairs) 0.145 Furniture (table) 0.224 Mattress 0.125 Bed 0.051 Mat -0.133 Sewing machine 0.087 Gas cooker 0.072 Electric stove 0.062 Gas stove -0.013 Kerosene stove 0.306 Fridge 0.249 Freezer 0.131 Air conditioner 0.010 Washing Machine -0.069 Electric Clothes Dryer -0.089 Bicycle -0.094 Motorbike 0.034 Cars and other vehicles 0.145 Generator 0.251 Fan 0.352 Radio 0.095 Cassette recorder 0.051 Hi‐Fi (Sound System) 0.082 Microwave 0.004 Iron 0.274 TV Set 0.368 Computer 0.092 DVD Player 0.329 Satellite Dish 0.112 Musical Instrument 0.011 Mobile phone or other assets 0.174 Uncultivated land -0.193 Source: Nigeria General Household Survey 2010