Socioeconomic Determinants of Food Insecurity among Zambian Households: Evidence from a National Household Survey

Abstract

Meal frequency is an important indicator of food security and nutritional status. Defining food insecurity as a household’s inability to consume at least three meals a day, this study uses a logit model to investigate the socioeconomic determinants of food insecurity among Zambian households. Primary data from the 2010 Living Conditions Monitoring Survey data set developed by the Central Statistical Office were used. The 2010 Living Conditions Monitoring Survey used a nationally representative sample of about 20,000 households. This study found that urban households, households with higher income, and households with younger, more educated and male heads were more likely to be food-secure. Therefore, there is need to accelerate investments in formal education, narrow the rural–urban socioeconomic divide, and reduce gender inequities through deliberate policies to increase women’s access to and control over economic resources such as land.

Keywords

Food insecurity logit meal frequency Zambia

Introduction

Poverty is a complex concept. Although, it has no universally accepted definition, many definitions encompass the lack of access to basic needs – primarily food, and other necessities such as housing, clean water, clothing, basic infrastructure, and transportation (Narayan et al., 2000). Other scholars explicitly define poverty as food insecurity (Ghana, 1995a, as cited in Narayan et al., 2000).

Food insecurity exists when people have inadequate physical, social or economic access to food. It manifests itself in the lack of access to sufficient quality food to all household members at all times (Clay, 2002). Poverty is the main reason for this insecurity. Poor households have difficulty in obtaining adequate, safe and nutritious food.

It is estimated that 735.9 million people or 10.0% of the global population lived in extreme poverty in 2015 (World Bank, 2018a). Table 1 summarizes the global poverty situation by region in 2015. Clearly, extreme poverty is concentrated in South Asia and Sub-Saharan Africa (SSA). The latter, where Zambia is located, has a poverty headcount ratio of 41.1% and accounts for 413.3 million of the 735.9 million or 56.2% of the world’s poor.

Table 1.

Global poverty at the US$ 1.90 poverty line in 2015.

Region	Headcount (millions)	Headcount ratio (%)
East Asia and Pacific	47.2	2.3
Europe and Central Asia	7.1	1.5
Latin America and the Caribbean	25.9	4.1
Middle East and North Africa	18.6	5.0
South Asia	216.4	12.4
Sub-Saharan Africa	413.3	41.1
World Total	735.9	10.0

Source: Constructed from World Development Indicators Database (World Bank, 2018a).

During the 1990s and 2000s, the most popular policy prescription for reducing extreme poverty was the need to accelerate economic growth (Ferreira et al., 2009). However, in spite of high economic growth since the early 2000s in SSA and other poor regions, high extreme poverty rates and income inequalities persist (Bicaba et al., 2015). The growing consensus is that while economic growth is essential for sustainable poverty reduction, it is not a magic bullet. What is needed is pro-poor growth that supports agricultural and rural development, supports market development and trade, and most importantly, expands social protection (Handley et al., 2009).

The Zambian economy is heavily dependent on copper mining; accounting for over 70% of its exports (United Nations, 2015: 13). However, the mining sector employs less than 2% of the labour force (United Nations, 2015: 13). The majority of Zambians live in rural areas and 80% of the population depends on agriculture for their livelihood (Chapoto et al., 2012; United Nations Development Programme (UNDP) et al., 2013: 4). It is often argued that the agricultural sector is characterized by low productivity (Battikhi, 2009: 4).

Limited employment opportunities in the copper mining sector, the lack of industrial growth and the low productivity in agriculture manifest themselves in high poverty levels. According to the Central Statistical Office (CSO) (2007, 2011, 2016), the national average poverty rate which stood at 73.2% in 1998 declined to 54.5% in 2015. Extreme poverty declined from 57.9% in 1998 to 41% in 2015. Despite the decline in poverty rates, the absolute number of people falling into poverty increased on account of a high population growth rate.

The high poverty levels in Zambia are associated with rising income inequality. For example, the Gini coefficient rose from 0.53 in 1998 to 0.69 in 2015. In 2010, the richest 10% of the population received 52% of the national wealth compared to 0.5% received by the poorest 10%. Figure 1 depicts the trends in poverty and inequality in Zambia over the period 1990–2015.

Figure 1.

Trends in poverty indicators 1990–2015.

Although Zambia’s extreme poverty rate for 2015 was approximately equal to the SSA average of 41%, it has one of the highest extreme poverty rates when compared to most of its neighbouring countries. Particularly, it has a significantly higher extreme poverty rate compared to Angola (36.9%), Botswana (13.0%), Namibia (26.9%), Mozambique (43%) and Tanzania (28.2%); although it ranks better than Malawi (53%) and Zimbabwe (73.0%) (World Bank, 2018b).

The poverty and income inequality situation manifests itself in reduced meal frequency and poor nutritional outcomes. Table 2 exhibits the distribution of daily meal frequency for Zambian households over time. In 2015, almost 45% of all households could not afford three meals a day; the loose benchmark for characterizing a Zambian household as food-secure (CSO, 2007, 2016).

Table 2.

Distribution of meal frequency among Zambian households (percentage).

Meal frequency(per day)	2006	2010	2015
One meal	5	4.1	3.7
Two meals	51	46	41
Three meals	42	47.3	52
More than three meals	2	2.4	3.3

Source: CSO (2007, 2011, 2016).

The adverse poverty and food security situation significantly impacts children. The proportion and absolute number of children (0–18 age group) living in poverty is higher than any other age group. It is estimated that about five million children and adolescents lived in poverty in 2016. This is about 65% of the total child population. And about 46% of children and adolescents live in extreme poverty (UNICEF), 2016). Overall, this is reflected in low nutrition standards among children; reflected in the high prevalence of child stunting and wasting (see Table 3).

Table 3.

The state of under nutrition among Zambian children aged under five.

Year	Prevalence of wasting (%)	Prevalence of stunting (%)
1990–1994	6.3	46.4
1992
1998–2002	5.7	57.9
2000
2006–2010	5.6	45.8
2008
2012–2016	6.3	40.0
2015	–	–

Source: International Food Policy Research Institute (IFPRI) et al. (2017: 35).

The above situation adversely affects children’s growth process and intergenerational equity. The situation is worse among rural households which are more vulnerable to shocks that adversely affect their access to food and have very little ability to absorb the impact of negative shocks (UNICEF, 2016).

It is difficult for many Zambian households to meet the required food quantity and quality standards. Their diets mainly comprise green vegetables and maize meal porridge, and occasionally sardines, chicken and goat meat (Chibuye, 2011: 16). The food often lacks dietary diversity and is nutritionally unbalanced (IFPRI et al., 2017).

The aim of this study is to establish the main socioeconomic factors that determine food insecurity among Zambian households. Using primary, cross-sectional data from the 2010 Living Conditions Monitoring Survey (LCMS), we estimate the likelihood that the household will suffer from food insecurity given its particular socioeconomic characteristics. The rest of the paper is structured as follows: Section 2 reviews the related literature while Section 3 presents the theoretical framework; Section 4 discusses the research methodology; Section 5 gives summary statistics on the relationship between socioeconomic factors and food security; Section 6 presents and discusses the results of the estimated empirical model; Section 7 discusses Zambian households’ poverty coping strategies; and Section 8 concludes and highlights policy implications.

Review of related literature

Several studies have investigated the effect of household’s socioeconomic dynamics on food insecurity. One such study is by Neumark-Sztainer et al. (2003), which showed that socioeconomic characteristics significantly influence weekly family meal frequency (a proxy for food security). Using cross tabulations, and log-linear modelling, the study found that socioeconomic status was positively related to meal frequency, similar to Fiese and Schwartz (2008), Fulkerson et al. (2014) and Haines et al. (2013).

Kalid et al. (2016) used a logit model to assess the factors that influence the household food security of internally displaced persons in Bosaso-Puntland in Somalia. They found that the gender of the household head, household size and food prices had a significant impact on household food security. Particularly, they found that male-headed households had better food security than female-headed households, similar to findings by Abdalla et al. (2013). They also found that larger households have poorer food security. Paradoxically, they found that higher food prices, overall, improved food security.

Zhou et al. (2016) also used a logit model to examine the factors that affect household food security in northern Pakistan. They found that age, gender, education attainment, remittances, unemployment, inflation, assets and disease were the most significant factors in determining food security. For example, they found that although male education was significant in improving food security, female education was not. Furthermore, it was noted that female-headed households were more vulnerable to food insecurity.

Silva et al. (2017) used a Poisson regression to determine the relationship between daily meal frequency and demographic and socioeconomic factors among children and adolescents in the Brazilian state of Minas Gerais. The study found that income and family size were significant factors in determining daily meal frequency. Specifically, the study found that low-income households and larger households, especially those with more males, had a lower meal frequency.

Zakari et al. (2014) used a logit model to determine factors that influence household food security in Niger. Their analysis revealed that gender, diseases and pests, labour supply, flooding, poverty, access to the market, the distance away from the main road and food aid were significant factors that influenced the probability that a household would have enough daily rations. The study revealed that female-headed households were less food-secure than male-headed households.

Abdalla et al. (2013) examined the relationship between socioeconomic factors and the household consumption gap and food security in Sudan’s North Kordofan state. They looked at socioeconomic factors such as the gender of the head of the household, household income and the number of dependents in the household. They found that female-headed households were more food-insecure than male-headed households. However, their study found that household income and the age of the household head had no significant impact on the food consumption gap and food security.

El-Gilany and Elkhawaga (2012) carried out a cross-section study to determine meal patterns among adolescent students in Mansoura, Egypt. They used a self-administered questionnaire to collect data on key sociodemographic characteristics of students and their families. Using a logit regression, they found that female students, students from urban areas, students with highly educated parents and those from higher socioeconomic classes were more likely to have at least three meals a day.

Sultana and Kiani (2011) used a logit model to examine the determinants of food security in Pakistan. Using data from a national survey carried out between 2007 and 2008, they found that urban residency and a higher dependency ratio had a negative effect on food security. Furthermore, the level of education had a positive impact on food security; while social capital was found to have no significant effect on food security.

The above literature review has identified some of the key socioeconomic determinants of food insecurity; including household income, educational status, residence, family size, gender, age, food prices and occupational status. However, a key limitation of the reviewed literature is the limited number of empirical studies on the socioeconomic determinants of food insecurity in SSA where Zambia is located. Most of the published empirical literature is drawn from Saharan and Arab Africa (Niger, Sudan, Somalia and Egypt) and Asia, particularly Pakistan. Therefore, this study seeks to fill the gap in the empirical literature on the socioeconomic determinants of food security in SSA, particularly Zambia. The next section outlines the theoretical framework used for the study.

Theoretical framework

This study outlines a theoretical framework based on Sassi’s (2015) framework for food and nutritional security, and Becker’s (1975) human capital theory. According to the former, there are various pathways through which household food and nutritional security is achieved. Sassi (2015) developed a framework to explain food and nutritional security centred around the concepts of household assets, food availability, food access and other confounding factors (see Figure 2).

Figure 2.

Framework for the analysis of household food and nutritional security.

Sassi points out that household assets include human capital, social capital, natural, financial, and physical assets. According to Becker’s (1975) human capital theory, human capital which is a set of skills and other attributes which can be used to produce economic value or earn income, can be augmented by among other things, investments in education and health. Healthier and educated people or households with more human capital investments are likely to produce more economic value because they are more productive. Additionally, aspects of social capital such as gendered societal power relations and hierarchy can be used to explain why women remain disadvantaged when it comes to accessing and controlling income-generating resources such as land (Claridge, 2004).

Put differently, household assets can be used to engage in productive activities through which the household can earn income to procure food. The household’s productive activities may include own-food production, cash crop production, and employment. Food produced can either be consumed by the household or sold. In this way, food availability is enhanced. Greater food availability may lead to reduced food prices; thereby increasing households’ food access – the ease with which a household or individual can acquire adequate food to meet their dietary requirements (Food and Agriculture Organization, 2006). In the short-run, households can employ various strategies to cope with insufficient food access. These include reducing meal frequency, limiting food portions, and consuming less favoured food. In addition, some assets can be sold to mitigate short-term adverse food security shocks.

Sassi also identifies other (confounding) factors outside the control of the households that can influence food and nutritional security. They can be physical, policy, or social in nature. Physical factors include weather patterns and soil conditions. Policy factors include government policies towards the agriculture sector, or indeed any policies that may affect households’ ability to access food. Social factors include any cultural attitudes, household features and other social institutions that influence access to food.

Methodology

Data and data sources

The dataset used for this study is derived from the 2010 Zambia LCMS. It was obtained upon formal request from the CSO. The LCMS was carried out by the CSO between January and April 2010. The aim of the survey was to get a perspective of the poverty situation in the country, and to compare it with previous surveys.

The survey covered a range of issues on the living conditions of households and individuals with respect to education, health, nutrition, income, housing, food production and sanitation. The 2010 LCMS covered about 20,000 households drawn from a sample of 1,000 standard enumeration areas (SEAs). The average response rate for the selected SEAs was 98% (CSO, 2011).

Even though the 2015 LCMS report is available to the public, we were unable to access the dataset from the relevant authorities. Nevertheless, analysis based on the 2010 LCMS still provides important insights into the key determinants of food insecurity. Furthermore, the results of this analysis serve as a good reference point for future analyses based on future surveys.

Data analysis

The data were analysed using the statistical package Stata 14. We began by computing the key descriptive statistics for the key socioeconomic variables: household income; education; region; family size; gender; age; and employment status. Each socioeconomic variable was then cross-tabulated with household daily meal frequency. Thereafter, the logit model presented below was estimated and discussed.

The empirical model

To investigate the socioeconomic determinants of food insecurity, a logit model was estimated. In this study, household daily meal frequency is used as an indicator of food insecurity. Meal frequency can also be an indicator of a household’s nutritional status; only to the extent that the food consumed, as assumed in this study, contains basic nutrients. However, due to data constraints, our study does not interrogate the nutritional contents of meals consumed. Nevertheless, Fulkerson et al. (2014) show the existence of a positive relationship between meal frequency and nutritional status.

We estimate the likelihood of a household having the standard daily meal frequency – assumed to be three or more. Following Chow (1983), let $P_{1}$ , the probability that the ith household having characteristics $z_{i}$ will have at least three meals a day, be given by:

P_{1} (z_{i}) = P_{1 i} = \frac{e^{β^{'} z_{i}}}{e^{β^{'} z_{i}} + 1}

(1)

Let $f_{1 i}$ be a dummy variable which assumes the value 1 if the ith household has at least three meals a day and 0 otherwise. A linear regression model can then be specified as follows:

f_{1 i} = β^{'} z_{i} + ϵ_{i}

(2)

where β is a row vector of coefficients and $ϵ_{i}$ is a random error term; Equation (2) estimates the probability of a household having at least three meals a day by a linear function of $z_{i}$ .

The marginal effects of the characteristics z, which measures the change in the expected probability of having a normal daily meal frequency (E [ $f_{1}$ / z ]) as a result of a unit change in z is given by:

\frac{\partial E [f_{1} / z]}{\partial z} = β f (β^{'} z)

(3)

f (β^{'} z) = \frac{e x p (β^{'} z)}{1 + e x p (β^{'} z)}

(4)

$f (β^{'} z)$ can be interpreted as a scale factor that transforms the parameters in equation (2) into marginal effects (Anderson and Newell, 2003; Griffiths et al., 1993).

The binary variable $f_{1 i}$ was denoted MFREQ. MFREQ assumes the dummy value 1 if a household can afford at least three meals per day and 0 otherwise. Let MFREQ* denote the unobservable probability of having at least three meals a day, then the binary logit model is given by:

\begin{array}{l} M F R E Q * = β_{0} + β_{1} H H I + β_{2} E D U + β_{3} R E G \\ + β_{4} F A M S I Z E + β_{5} A G E + β_{6} G E N D E R + β_{7} E M P + ε \end{array}

(5)

where HHI is household income, EDU is the educational level, REG is a dummy variable which assumes 1 for a rural household and 0 otherwise, FAMSIZE is the family size, GENDER is also a dummy variable which assumes the value 1 if the household head is male and 0 otherwise, AGE is the age of the household head, EMP is a dummy variable that assumes the value 1 if the household head is employed and 0 otherwise, $β$ coefficients are the true values of the parameters, and ε is the disturbance term.

It is assumed that a household has at least three meals a day if and only if $M F R E Q^{*} \geq 0$ . It is also assumed that the probability of a household affording at least three meals a day is a function of the social–economic factors in equation (5).

Summary statistics

Table 4 provides descriptive statistics of the socioeconomic variables from equation (5). These variables include monthly income (in Zambian Kwacha, ZMW), educational status, region, family size, gender, age and employment status. Furthermore, Table 4 shows that the average monthly household income was ZMW575.76 while the average age of the typical household head was 42.04 years. The average household size was five. Almost all household heads (99%) had some formal education while 44% of households were rural-based. In addition, 50% of the households were male-headed while 82.3% of all household heads were employed.

Table 4.

Descriptive statistics on key socioeconomic variables.

Variable	Mean(standard deviation/percentages)*
HHI	ZMW 575.76(ZMW 4, 875.8)
EDU (formal Schooling = 1)	0.999(99.88)
REG (rural = 1)	0.44(44.3)
FAMSIZE	5.30
GENDER (male = 1)	0.49(0.50)
AGE	42.04(13.89)
Employment Status (employed = 1)	0.33(82.3)

Note: *standard deviation for quantitative variables and percentages for indicator variables are in parentheses.

Source: 2010 LCMS.

The dependent variable is the likelihood that a household is food-secure. The LCMS questionnaire had a question on the daily average number of meals consumed by a household (excluding snacks). Table 5 summarizes the proportion of households who were classified as being food-secure. Out of 19,398 households, 57.1% reported having at least three meals per day while 42.9% had two or less meals per day.

Table 5.

Distribution of household daily meal frequency.

Daily meal frequency	Number of households(percentage)*
Less than three meals	8,322(42.9)*
Three or more meals	11,076(57.1)*
Total	19,398

Source: 2010 LCMS.

Table 6 summarizes the distribution of household meal frequency by income groups, in terms of the poorest 20%, the poor 20%, the middle 20%, the rich 20% and the richest 20% of the sample. Clearly, the number of households accessing at least three meals a day increases with income levels. The top two income groups accounted for 1,903 out of 3,646 or 52.2% of those households that managed to have at least three meals per day. A simple Pearson Chi-square test for independence indicates a significant relationship between income levels and daily meal frequency.

Table 6.

Distribution of the meal frequency by income group.

	Poorest	Poor	Middle	Rich	Richest	Total
Less than three meals	760(12.2)*	771(12.4)*	483(7.7)*	328(5.3)*	227(3.7)*	2,569(41.3)*
Three or more meals	493(7.9)*	627(10.1)*	623(10.0)*	926(14.9)*	977(15.7)*	3,646(58.7)*
Total	1,253(20.2)*	1,398(22.5)*	1,106(17.8)*	1,254(20.2)*	1,204(19.4)*	6,215(100)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(4)) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Table 7 summarizes the relationship between meal frequency and the educational level of the household head. Table 4 shows that only 0.1% of the respondents had no formal education; 38.8% had primary education; 46.7% had secondary education; while 14.3% had tertiary education. It also shows that as much as 71.6% of households who had three or meals a day had a household head that had at least some secondary education; indicating that higher education is associated with higher food security. This preliminary observation is supported by the Pearson Chi-square test for independence which suggests the existence of a significant relationship between the educational level and daily meal frequency.

Table 7.

Distribution of daily meal frequency by educational status.

	No formal education	Primary education	Secondary education	Tertiary education	Total
Less than three meals	2(0.02)*	2,128(22.6)*	1,626(17.2)*	235(2.5)*	3,991(42.3)*
Three or more meals	9(0.1)*	1,535(16.3)*	2,784(29.5)*	1,117(11.8)*	5,445(57.7)*
Total	11(0.12)*	3,663(38.8)*	4,410(46.7)*	1,352(14.3)*	9,436(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(3) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Table 8 summarizes households’ daily meal frequency by region. It also indicates that only 3,753 out of 8,597 of 43.7% of the rural households managed to have at least three meals a day, compared to 7,318 out 10,793 or 67.8% of the urban households; indicating that that urban households were more food-secure, owing to better poverty coping strategies. The Pearson Chi-square test for independence also indicates the existence of a significant relationship between region and daily meal frequency.

Table 8.

Distribution of daily meal frequency by region.

	Rural	Urban	Total
Less than three meals	4,844(25.0)*	3,475(17.9)*	8,319(42.9)*
Three or more meals	3,753(19.4)*	7,318(37.7)*	11,071(57.1)*
Total	8,597(44.3)*	10,793(55.7)*	19,390(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(1) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Table 9 summarizes daily meal frequency by family size groups: 1–5; 6–10; 11–15; 16–20; 21–25; and 26–30. Furthermore, Table 9 shows that the proportion of households which had at least three meals a day increases as the household size increases: 1–5 (55.7%); 6–10 (58.8%); 11–15 (60.3%); 16–20 (63.2%); 21–25 (67%); and 26–30 (100%). This initial finding, supported by the Pearson Chi-square test for independence, is somewhat paradoxical. Given the high levels of poverty in Zambia, the expectation is that larger households’ coping strategies are likely to be limited.

Table 9.

Distribution of daily meal frequency by family size group.

	1–5	6–10	11–15	16–20	21–25	26–30	Total
Less than three meals	4,948(25.5)*	3,062(15.8)*	292(1.51)*	14(0.07)*	3(0.02)*	0(0.00)*	8,319(42.9)*
Three or more meals	6,219(32.1)*	4,377(22.6)*	444(2.29)*	24(0.12)*	6(0.03)*	2(0.01)*	11,072(57.1)*
Total	11,167(57.6)*	7,439(38.4)*	736(3.80)*	38(0.20)*	9(0.05)*	2(0.01)*	19,391(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(5) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Table 10 shows a cross tabulation of daily meal frequency and the gender of the household head. The number of male-headed and female-headed households included in the sample were almost evenly split, at 49% and 51%, respectively. The proportion of male-headed and female-headed households which reported to have had at least three meals a day was almost identical, at 56.8% and 57.4%, respectively. This seems to imply the lack of a significant relationship between the gender of the household head and daily meal frequency; a conclusion supported by the non-significance of the Pearson Chi-square test statistic.

Table 10.

Distribution of daily meal frequency by sex of the head of the household.

	Male-headed households	Female-headed households	Total
Less than three meals	4,107(21.2)*	4,212(21.7)*	8,319(42.9)*
Three or more meals	5,409(27.9)*	5,664(29.2)*	11,073(57.1)*
Total	9,516(49.1)*	9,876(50.9)*	19,392(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(1) = 0.473; *relative frequencies.

Source: 2010 LCMS.

Another factor that may influence a household’s meal frequency is the age of the household head. Table 11 gives a distribution of the daily meal frequency by the age of the household head, divided into four age groups: 16–25; 26–45; 46–65; and over 66. The majority of the household heads were aged between 26 and 45 years, and accounted for 58.2% of the households. The proportion of households in each age group which reported to have had at least three meals a day varies quite significantly across the four age groups: 51.0% for the 16–25 age group; 61.2% for the 26–45 age group; 54.7% for the 46–65 age group; and 38.6% for the over-66 age group. The oldest age group has the lowest proportion of households which could afford at least three meals a day; suggesting that households that are headed by the elderly are less likely to be food-secure. The Pearson Chi-square test for independence also suggests the existence of a significant relationship between daily meal frequency and the age of the household head.

Table 11.

Distribution of daily meal frequency by age of the head of the household.

	16–25	26–45	46–65	Over 66	Total
Less than three meals	679(3.50)*	4,375(22.6)*	2,413(12.4)*	852(4.40)*	8,319(42.9)*
Three or more meals	708(3.65)*	6,915(35.7)*	2,914(15.0)*	536(2.76)*	11,073(XX)*
Total	1,387(7.15)*	11,290(58.2)*	5,327(27.5)*	1,388(7.16)*	19,392(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(3) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Table 12 summarizes daily meal frequency by the employment status of the household head. Out of the 7,684 household heads who reported their employment status, 82.3% were employed while 17.7% were unemployed. Out of all the households who reported to have had afforded at least three meals a day, 85.8% of them had household heads who were employed. Furthermore, 55.4% of the households with heads who were employed reported to have had at least three meals a day, compared to 42.5% for households whose heads were unemployed. This implies that households with employed heads are more likely to be food-secure; an observation supported by the Pearson Chi-square test for independence.

Table 12.

Daily meal frequency by employment status of the household head.

	Employed	Unemployed	Total
Less than three meals	2,817(36.7)*	782(10.2)*	3,599(46.8)*
Three or more meals	3,506(45.6)*	579(7.54)*	4,085(53.2)*
Total	6,323(82.3)*	1,361(17.7)*	7,684(100.0)*

Note: P-value of the Pearson Chi-square test statistic, Pr(Pearson chi2(1) = 0.000; *relative frequencies.

Source: 2010 LCMS.

Empirical model: results and discussion

In this section, we assess the socioeconomic determinants of meal frequency by estimating equation (5) using a logit regression. Table 13 presents the estimated model results.

Table 13.

Logit regression results.

Variable	Coefficient(standard error)
HHI	0.000027*
	(0.0000117)
EDU	0.7934161*
	(0.0440294)
REG	−0.4388775*
	(0.056773)
FAMSIZE	0.080761*
	(0.0106997)
AGE	−0.0084809*
	(0.0019297)
GENDER	0.1182025*
	(0.0546982)
EMP	−0.1129047
	(0.0748648)
Constant	−1.826883*
	(0.1946976)

LR chi2	797.48
Prob > chi2	0.0000
Pseudo R2	0.0872
n	6652

Note: *implies significance at the 5% level of significance.

Source: Authors’ computation.

The results show that household income has a significant positive impact on a household’s daily meal frequency. Thus, richer households are more likely to be food-secure. This finding is consistent with El-Gilany and Elkhawaga (2012) and Silva et al. (2017) who found that high-income households are more food-secure. The findings of this study are also consistent with Sassi’s (2015) theoretical prediction that households with more income have higher access to food and hence are likely to be more food-secure.

The level of education was also found to increase a household’s food security. This finding is in line with predictions of the human capital theory that an increase in an individual’s level of education increases their stock of human capital and hence their wage-earning capacity. At the household level, higher wage-earning capacity may translate into a higher ability to afford basic household food requirements. This finding is also consistent with Sultana and Kiani (2011) and El-Gilany and Elkhawaga (2012).

The location of a household can potentially influence a household’s food security. This study found that urban households were more food-secure. This expected finding may be explained by the fact that urban households have more household assets (human, social, physical, natural, and financial capital) and hence more sustainable coping strategies than rural households. This finding is also unsurprising given that poverty rates in Zambia are higher in rural than urban areas (CSO, 2016).

We also found that larger households are more food-secure than smaller households. This finding can be explained in terms of the higher social and productive capital that characterizes larger families; consistent with Sassi’s (2015) theoretical prediction. This high social and productive dividend allows these households to generate more income to procure food as well as to produce more food. However, this finding is contrary to empirical evidence that larger families are likely to have their coping strategies over-stretched, and hence likely to be less food-secure (Kalid et al., 2016; Silva et al., 2017).

The study also sought to determine the effect of the age and employment status of the household head on a household’s food security. The study found that households with older heads were less food-secure. This could be due to the fact that younger household heads were more likely to be energetic enough to mobilize resources to access basic food. Furthermore, old age drains one’s health and their ability to effectively exploit household assets by engaging in productive activities. Coupled with poor social security systems, households headed by older people have limited poverty coping strategies. However, the study found that employment status had no significant impact on food security.

Lastly, we examined the effect of the gender of the household head on food security. We found that male-headed households were more likely to be food-secure; consistent with findings by Abdalla et al. (2013), Zakari et al. (2014), Kalid et al. (2016) and Zhou et al. (2016). In the Zambian context, this expected finding may be explained by the argument that females are discriminated against in terms of access to and control of productive economic resources (Farnworth and Munachonga, 2010; Milimo et al., 2004). This finding is consistent with the theoretical assumption that gendered power relations skewed against women imply that they have lower access to key economic resources and hence are vulnerable to food insecurity.

In a broader perspective, the results of this study draw some similarities and differences when compared to other regional contexts. Most apparently, this study as with reviewed empirical studies from across the Saharan and Arab African regions (Abdalla et al., 2013 on Sudan; Zakari et al., 2014 on Niger; Kalid et al., 2016 on Somalia) and Asia (see Zhou et al., 2016 on Pakistan), identifies gender as a very important determinant of food insecurity. The finding that female-headed households are less food-secure simply reflects the view that women in many developing countries still suffer widespread discrimination in the political, social and economic arenas (United Nations Educational, Scientific and Cultural Organization, 2017).

The region of residence and the age and educational level of the household head, which are key determinants of food insecurity in Zambia, do not seem to be important determinants of food insecurity in other African contexts. These socioeconomic factors are much more important determinants of household food insecurity in Asian countries such as Pakistan (Sultana and Kiani, 2011; Zhou et al., 2016). However, in other Saharan and Arab African contexts such as Sudan (Abdalla et al., 2013), Somalia (Kalid et al., 2016) and Niger (Zakari et al., 2014), food prices are a leading determinant of food insecurity. Additionally, while household size is not a key determinant of food insecurity in SSA where Zambia is located, it is a key driver of food insecurity in Asia (see Sultana and Kiani, 2011 on Pakistan), Arab Africa (see Abdalla et al., 2013 on Sudan; Kalid et al., 2016 on Somalia), and Latin America (see Silva et al., 2017 on Brazil).

Furthermore, we estimated the marginal effects of each socioeconomic variable (see Table 14). This is important for coefficient interpretational purposes, given that the coefficients of the estimated logit model are in log odds. Marginal effects measure the change in the expected probability of a household being food-secure, as a result of a marginal change in each of the socioeconomic variables.

Table 14.

Marginal effects.

Variables	Marginal effect(standard error)
HHI	0.000006*
	(0.000003)
GENDER	0.0259*
	(0.0120)
AGE	−0.00186*
	(0.00042)
EMP	−0.0247
	(0.0164)
REGION	−0.0961*
	(0.0122)
EDU	0.1738*
	(0.0087)
FAMSIZE	0.0177*
	(0.0023)

Note: *implies significance of the coefficient at 5%.

Source: Authors’ computation.

Table 14 shows that a ZMW 1000 increase in a household’s monthly income increases a household’s probability of affording at least three meals a day by 0.6%. This increase is statistically significant at the 5% level of significance. With respect to the gender of the household head, male-headed households were 2.59% more likely to be more food-secure than female-headed households. This increase was also statistically significant.

We also found that a one year increase in the age of the household head reduced the probability of a household affording at least three meals a day by 0.186%. Furthermore, we found that a one unit increase in the family size increased a household’s chance of being food-secure by 1.77%. However, as earlier noted, the impact of employment status on food security is not statistically significant.

It was also found that education and region had the largest marginal effects. Particularly, it was found that having an educated household head increased the probability of a household being food-secure by 17.38%. Furthermore, it was found that urban households were 9.61% more likely to be food-secure than their rural counterparts.

Household poverty coping mechanisms

Households in Zambia employ various poverty coping strategies – all the strategies put in place by households in a poor and difficult economic situation to restrict their expenses or earn some extra income to enable them to pay for basic necessities, such as food, clothing and shelter. This is to avoid falling too far below their normal level of welfare (Snel and Staring, 2001: 10).

The 2010 LCMS also collected data on households’ poverty coping strategies. Households are asked to mention their top three coping strategies associated with several shocks to household welfare such as droughts, floods, crop and livestock diseases. Table 15 summarizes the five most common coping strategies for Zambian households for each rank.

Table 15.

Poverty coping strategies among Zambian households.

	Strategy 1	Strategy 2	Strategy 3
1.	Did nothing	Did nothing	Did nothing
2.	Piece work on farms	Reduce the number of meals	Bought less non-food items
3.	Other pieces of work	Bought less food	Reduce the number of meals
4.	Reduce the number of meals	Bought cheaper food	Bought less food
5.	Seek refuge with other households	Substitute ordinary meals with mangoes, pumpkins and sweat potatoes	Seek refuge with other households

Source: 2010 LCMS.

Clearly, from Table 5, the most common response to shocks is to ‘do nothing’. This suggests that most Zambian households’ welfare deteriorates during economic shocks. The most widely used coping strategy is to reduce the number of meals that a household consumes. This finding is consistent with findings by Muuka and Kalyalya (1997), Simatele and Simatele (2009), Schrimpf and Feil (2012) and Carvalho and Nsemukila (2013). These findings are also consistent with Sassi’s (2015) theoretical arguments that households typically cope with dwindling food access by reducing meal frequency.

Another poverty coping strategy is to buy less and cheaper food and other non-food items. This observation is consistent with findings by Simatele and Simatele (2009). Furthermore, casual or piece works such as working for food or extra income are another popular coping strategy among Zambian households. Studies by Muuka and Kalyalya (1997) and Simatele and Simatele (2009) also emphasize the role that piece works play in smoothing Zambian households’ consumption during negative shocks. By engaging in casual or piece works, households are transforming their human capital (skills and knowledge) into productive activities through which they can earn income to procure food; consistent with Sassi’s (2015) approach to explaining food and nutritional security.

In addition, Zambian households also cope with adverse shocks by seeking refuge with their neighbours, friends or relatives. The 2010 LCMS also found that substituting ordinary meals with seasonal fruits or crops such as mangoes, pumpkins, and sweet potatoes is another coping strategy; similar to findings by Schrimpf and Feil (2012).

Other empirical studies on Zambia identify other poverty coping mechanisms adopted by Zambian households. These include: own-household crop farming and animal raring (Kalinda and Langyintuo, 2014); street vending (Simatele and Simatele, 2009); gathering wild fruits; and borrowing (Muuka and Kalyalya, 1997).

The above strategies are not unique to Zambia. For instance, households in peri-urban and rural Ghana cope by hunting, reliance on prepared food, sale of animals, charcoal, handcrafting, occasional employment, borrowing, gathering wild plants and crops, consuming of less preferred food, limiting food portion sizes, and reducing meal frequency (Chagomoka et al., 2016: 9).

In Kenya, households cope by reducing food consumption, borrowing, and removing children from school (Amendah et al., 2014; Hossain, 2005). In Asia, small households cope with poverty by reducing or changing their consumption patterns while large households tend to adjust their labour supply by sending a household member to work or increasing their own food production. More affluent households in urban areas use their assets to cope (Murakami, 2017: 16).

Conclusion and policy recommendations

A household’s daily meal frequency is an important indicator of its food insecurity and nutritional status. It is also an important indicator of the effectiveness and sustainability of its poverty coping strategies. Using meal frequency as a proxy for household food insecurity, this study sought to determine the key socioeconomic determinants of food insecurity in Zambia.

This study found that education was an important determinant of food insecurity. It increases people’s productivity and ability to acquire food. While education was found to be a key determinant of food insecurity in developing countries in SSA, this was not the case for Arab African contexts. Therefore, improving long-term food security and nutritional status in SSA requires accelerating investments in formal education. We also found that rural households are less food-secure than urban households. Region of residence was found to be a key determinant of food insecurity in many developing countries in SSA, Arab Africa and Asia. Therefore, there is need to narrow the rural–urban socioeconomic divide by accelerating income redistribution efforts and the implementation of social protection programmes targeting rural areas.

We also found that female-headed households were more food-insecure than male-headed households. This may reflect a socioeconomic arena that discriminates against women. Thus, in order to improve food security and the nutritional status of female-headed households, and broaden their poverty coping strategies, there is need to increase women’s access to and control over economic resources such as land. Most importantly, this study confirms already strong empirical evidence that gender inequality in many developing countries in Africa, Asia and Latin America often manifests itself in poorer food security and inadequate poverty coping strategies among female-headed households. More generally, this study also confirms the theoretical understanding that households’ endowments of assets such as social capital, human capital, financial capital, natural resources and physical capital are used by households to access food and broaden their poverty coping strategies.

A key limitation of this study is that it assumed that there exists a positive relationship between meal frequency and nutritional status. However, the dataset used did not tell us what food was eaten; whether it was balanced or not. Eating the recommended number of meals a day improves a household’s overall nutritional status only to the extent that these meals are of high nutritional quality. In spite of this limitation, this study highlights the importance of socioeconomic considerations for household food insecurity. Nevertheless, future studies could explore the relationship between meal frequency and nutritional status.

Footnotes

Acknowledgements

We thank Caesar Cheelo (Senior Research Fellow at the Zambia Institute for Policy and Research), Dale Mudenda (Senior Lecturer in Economics at the University of Zambia) and Venkatesh Seshamani (Professor of Economics at the University of Zambia) for their valuable comments on earlier drafts.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Maio Bulawayo

Justin Sichone

Author biographies

Maio Bulawayo is a Lecturer in the Department of Economics at the University of Zambia. He is also an Associate Research Fellow at the Southern African Institute for Policy and Research.

Manenga Ndulo is a Professor of Economics in the Department of Economics at the University of Zambia. He is also the Executive Director at the Southern African Institute for Policy and Research.

Justin Sichone is a Research Fellow at the Southern African Institite for Policy and Research.

References

Abdalla

Leonhäuser

I-U

Bauer

, et al. (2013) Household Coping Strategies and Food Consumption in Sudan. International Working Paper Series. Paper No. 13/06. Pavia: NAF-IRN. Available at: http://economia.unipv.it/naf/Working_paper/WorkingPaper/SAMAR/Samar_Abdalla.pdf (accessed 30 April 2018).

Amendah

Buigut

Mohamed

(2014) Coping strategies among the urban poor: Evidence from Nairobi, Kenya. PLoS One 9(1): e83428.

Anderson

Newell

(2003) Simplified Marginal Effects in Discrete Choice Models. Discussion Paper No. 3–38. Washington, D. C: Resources for the Future. Available at: https://ageconsearch.umn.edu/bitstream/10631/1/dp030038.pdf (accessed 30 April 2018).

Battikhi

(2009) Agriculture National Report Zambia. New York: United Nations.

Becker

(1975) Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. New York: National Bureau of Economic Research.

Bicaba

Brixiová

Ncube

(2015) Eliminating Extreme Poverty in Africa: Trends, Policies and the Role of International Organizations. Working Paper Series 223. Abidjan: African Development Bank. Available at: https://www.afdb.org/fileadmin/uploads/afdb/Documents/Publications/Working_Paper_223_-_Eliminating_Extreme_Poverty_in_Africa_Trends_Policies_and_the_Role_of_International_Organizations.pdf (accessed 30 April 2018).

Carvalho

Nsemukila

(2013) Update of the Situation Analysis of Children and Women in Zambia. Lusaka: UNICEF.

Chagomoka

Unger

Drescher

, et al. (2016) Food Coping Strategies in Northern Ghana: A Socio-Spatial Analysis along the Urban–Rural Continuum. Freiburg: University of Freiburg.

Chapoto

Haggblade

Hichaambwa

, et al. (2012) Agricultural Transformation in Zambia: Alternative Institutional Models for Accelerating Agricultural Productivity Growth, and Commercialization. Working Paper 64. Lusaka: Indaba Agricultural Policy Research Institute. Available at: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.4302&rep=rep1&type=pdf (accessed 30 April 2018).

10.

Chibuye

(2011) Interrogating Urban Poverty Lines: The Case of Zambia. Human Settlements Working Paper Series. Poverty Reduction in Urban Areas. Human Settlements Working Paper Series Poverty Reduction in Urban Areas – 30. London: International Institute for Environment and Development (IIED). Available at: http://pubs.iied.org/pdfs/10592IIED.pdf (accessed 30 April 2018).

11.

Chow

(1983) Econometrics. Tokyo: McGraw-Hill.

12.

Clay

(2002) FAO Expert Consultation on Trade and Food Security: Conceptualizing the Linkages. London: Overseas Development Institute.

13.

Claridge

(2004) Social Capital and Natural Resource Management: A Role for Social Capital? Brisbane: University of Queensland.

14.

CSO (2007) 2006 Living Conditions Monitoring Survey. Lusaka: CSO.

15.

CSO (2011) 2010 Living Conditions Monitoring Survey. Lusaka: CSO.

16.

CSO (2016) 2015 Living Conditions Monitoring Survey. Lusaka: CSO.

17.

El-Gilany

Elkhawaga

(2012) Socioeconomic determinants of eating pattern of adolescent students in Mansoura, Egypt. Pan African Medical Journal 13: 22.

18.

Farnworth

Munachonga

(2010) Gender Aware Approaches in Agricultural Programmes: Zambia Country Report. UTV Working Paper No 8. Lusaka: SIDA. Available at: https://www.oecd.org/countries/zambia/46150642.pdf (accessed 30 April 2018).

19.

Ferreira

FHG

Leite

Ravallion

(2009) Poverty reduction without economic growth? Explaining Brazil’s poverty dynamics, 1985–2004. Journal of Development Economics 93(1): 20–36.

20.

Fiese

Schwartz

(2008) Reclaiming the Family Table: Mealtimes and Child Health and Wellbeing. Social Policy Report. XXII. Washington, DC: Society for Research in Child Development.

21.

Food and Agriculture Organization (2006) Food Security. Policy Brief Issue No. 2. Rome: Food and Agriculture Organization. Available at: http://www.fao.org/fileadmin/templates/faoitaly/documents/pdf/pdf_Food_Security_Cocept_Note.pdf (accessed 30 April 2018).

22.

Fulkerson

Larson

Horning

, et al. (2014) A review of associations between family or shared meal frequency and dietary and weight status outcomes across the lifespan. Journal of Nutrition Education and Behaviour 46(1): 2–19.

23.

Griffiths

Carter

Judge

(1993) Learning and Practicing Econometrics. New York: John Wiley and Sons, Inc.

24.

Haines

McDonald

O’Brien

, et al. (2013) Healthy habits, happy homes: Randomized trial to improve household routines for obesity prevention among pre-school aged children. JAMA Pediatrics 167(11): 1072–1079.

25.

Handley

Higgins

Sharma

, et al. (2009) Poverty and Poverty Reduction in Africa: An Overview of Key Issues. Working Paper 299. London: Overseas Development Institute. Available at: https://www.odi.org/sites/odi.org.uk/files/odi-assets/publications-opinion-files/860.pdf (accessed 30 April 2018).

26.

Hossain

(2005) Poverty, Household strategies and coping with urban life: Examining ‘Livelihood Framework’ in Dhaka city, Bangladesh. Bangladesh e-Journal of Sociology 2(1): 1–8.

27.

IFPRI, Concern Worldwide and Welthungerhilfe (2017) Global Hunger Index 2016: The Inequalities of Hunger. Washington. DC: IFFRI. Available at: https://www.globalhungerindex.org/pdf/en/2017.pdf (accessed 30 April 2018).

28.

Kalid

Abdirizak

Abdinasir

, et al. (2016) Assessment of factors influencing household food security of IDPs in Bosaso-Puntland, Somalia. International Journal of Medical and Health Research 2(3): 28–31.

29.

Kalinda

Langyintuo

(2014) Livelihood strategies, shocks and coping mechanisms among rural households in Southern Zambia. Research Journal of Social Sciences 6(4): 120–133.

30.

Milimo

Munachonga

Mushota

, et al. (2004) Zambia: Strategic Country Gender Assessment. Washing-ton, DC: World Bank. Available at: http://siteresources.worldbank.org/EXTAFRREGTOPGENDER/Resources/ZambiaSCGA.pdf (accessed 30 April 2018).

31.

Murakami

(2017) Household Consumption Risk and Coping Strategies in Tajikistan: Evidence from Household Surveys. Working Paper No. 738. Tokyo: Asian Development Bank. Available at: https://www.adb.org/sites/default/files/publication/316606/adbi-wp738.pdf (accessed 30 April 2018).

32.

Muuka

Kalyalya

(1997) Too rich to be poor? A glimpse of the poverty situation in Zambia in the 1990s. Scandinavian Journal of Development Alternatives and Area Studies 16(1): 139–156.

33.

Narayan

Patel

Schafft

, et al. (2000) Voices of the Poor: Can Anyone Hear Us? New York: Oxford University Press.

34.

Neumark-Sztainer

Hannan

Story

, et al. (2003) Family Meal Patterns: Associations with Sociodemographic Characteristics and Improved Dietary Intake Among Adolescents. Minneapolis: University of Minnesota.

35.

Sassi

(2015) The welfare cost of maize price volatility in Malawi. Bio-Based and Applied Economics 4(1): 77–100.

36.

Schrimpf

Feil

(2012) Traditional Food Crisis Coping Mechanisms: A Regional Perspective from Southern Africa. Stuttgart: Diakonisches Werk der EKD.

37.

Silva

Candia

Pequeno

, et al. (2017) Daily meal frequency and associated variables in children and adolescents. Journal de Pediatria 93(1): 79–86.

38.

Simatele

(2009) The Evolution and Dynamics of Urban Poverty in Zambia. Washington, DC: Noval Publications.

39.

Snel

Staring

(2001) Poverty, migration, and coping strategies: An introduction. European Journal of Anthropology 38: 7–22.

40.

Sultana

Kiani

(2011) Determinants of food security at household level in Pakistan. African Journal of Business Management 5(34): 12972–12979.

41.

UNICEF (2016) Malnutrition in Zambia, Harnessing Social Protection for the Most Vulnerable. London: UNICEF.

42.

United Nations (2015) Zambia Country Analysis: Summary. Lusaka: United Nations country office. Available at: http://zm.one.un.org/sites/default/files/un_country_analysis_report.pdf (accessed 30 April 2018).

43.

United Nations Development Programme (UNDP) Comprehensive African Agricultural Programme (CAADP) and New Partnership for Africa’s Development (NEPAD) (2013) Zambia Agricultural Investment Opportunities Brief: CAADP Investment Facilitation Programme. Available at: https://www.abghq.com/downloads/Zambia.pdf (accessed 30 April 2018).

44.

United Nations Educational Scientific and Cultural Organization (2017) UNESCO and Gender Equality in Sub-Saharan Africa. Paris: United Nations Educational, Scientific and Cultural Organization. Available at: http://www.unesco.org/new/fileadmin/MULTIMEDIA/HQ/AFR/images/3781_15_E_web.pdf (accessed 30 April 2018).

45.

World Bank (2018a) World Development Indicators Database. Available at: http://databank.worldbank.org/data/reports.aspx?source=world-development-indicators (accessed 30 April 2018).

46.

World Bank (2018b) Decline of Global Extreme Poverty Continues but Has Slowed. Available at: https://www.worldbank.org/en/news/press-release/2018/09/19/decline-of-global-extreme-poverty-continues-but-has-slowed-world-bank (accessed 30 April 2018).

47.

Zakari

Ying

Song

(2014) Factors influencing household food security in West Africa: The case of Southern Niger. Sustainability 6: 1191–1202.

48.

Zhou

Abdullah

Shah

, et al. (2016) Factors affecting household food security in rural Northern Hinterland of Pakistan. Journal of the Saudi Society of Agricultural Sciences. Available at: https://dx-doi-org.web.bisu.edu.cn/10.1016/j.jssas.2017.05.003