See discussions, stats, and author profiles for this publication at: [Link]

net/publication/338361554

Household Level Food Insecurity Assessment: Evidence from Panel Data,

Ethiopia.

Article in Scientific African · March 2020

DOI: 10.1016/[Link].2019.e00262

CITATIONS READS

63 914

1 author:

Teshager Sisha
University of Kentucky
6 PUBLICATIONS 64 CITATIONS

SEE PROFILE

All content following this page was uploaded by Teshager Sisha on 02 March 2020.

The user has requested enhancement of the downloaded file.

 Scientific African 7 (2020) e00262

Contents lists available at ScienceDirect

Scientific African
journal homepage: [Link]/locate/sciaf

Household level food insecurity assessment: Evidence from

panel data, Ethiopia
Teshager Assefa Sisha
Bahir Dar University, Ethiopia

a r t i c l e i n f o a b s t r a c t

Article history: Ethiopia is one of the fastest growing economies and with the top 10 highest population
Received 1 August 2019 growth, which is among the major resources that determine the dynamics of the wel-
Revised 9 October 2019
fare of the society. This study assesses the prevalence of food insecurity and the major
Accepted 27 December 2019
household level determinants of food insecurity at the national level and ranks the cop-
ing strategies households use. Longitudinal data of over 50 0 0 households is analyzed using
Editor: Dr B. Gyampoh logit model. Anthropometric indicators generally show an improvement over time, though
malnutrition is still one of the highest in the world. Child nutritional status in rural areas
Keywords:
is worse compared to urban areas. Also, it appears that a higher percentage of boys suffer
Food insecurity
from malnutrition than girls. The result of logistic regression revealed that average years of
Ethiopian socioeconomic survey
Nutrition schooling of members of the household, proximity to service centers, assets and availabil-
World bank ity of credit services positively affect household food security, whereas dependency ratio
and shocks increase the odds of a household to be food insecured. Households choose to
reduce the variety of food eaten as a primary strategy to reduce the impacts of food inse-
curity. Continuous investment in education and rural development could help improve the
household food insecurity.
© 2019 The Author(s). Published by Elsevier B.V. on behalf of African Institute of
Mathematical Sciences / Next Einstein Initiative.
This is an open access article under the CC BY license.
([Link]

1. Introduction

The issue of food security has continued to be an important agenda worldwide. According to UNICEF [22], new evidences
continued to point out a raise in world hunger after a prolonged decline, leaving a considerable work to be done towards
a food-secured world. Food security is likely to be a problem that mankind will have to address in the near and long-term
future [3].
The definition of food security has undergone several developments since its first notion during the Hot Springs Con-
ference of Food and Agriculture in 1943. The evolution of the concept, finally has ended up capturing four concepts called
“dimensions of food security”; Availability, Access, Utilization and Stability. The World Bank and the United Nations defined
food security (with minor differences) as “People having at all times, physical, social and economic access to sufficient,
safe and nutritious food which meets their dietary needs and food preferences for an active and healthy life”. Because of
its multidimensional character, however, identifying appropriate food insecurity indicators at household level is still not an

E-mail address: teshagerassefa@[Link]

[Link]
2468-2276/© 2019 The Author(s). Published by Elsevier B.V. on behalf of African Institute of Mathematical Sciences / Next Einstein Initiative. This is an
open access article under the CC BY license. ([Link]
2 T.A. Sisha / Scientific African 7 (2020) e00262

easy task [2,18,19]. In an effort to devise good indicators, Maxwell [9] discussed four differences in the approaches to define
food security, one of which is worth mentioning here, that is, the unit of analysis. The unit of analysis ranges all the way
from individuals and households to country level.
The variations in the approaches that have been used in defining food security made it difficult to label its existence
across countries and societies. For some, it encompasses nutritional/health aspect while, for others it is about having a food
for existence. Thus, it is apparent that food security has not yet been achieved by almost all countries around the world.
However, the problem is most pronounced in sub-Saharan and Asian countries.
Improper management of its enormous natural resources left African countries in uncertain social, economic and political
conditions. For example, in 2017, conflicts and social insecurity resulted in acute food insecurity in 18 African countries and
climate shocks affected 23 other countries [25]. Undernourishment and severe food insecurity is still increasing in almost
all regions of Africa [22].
Ethiopia, among the most famine-prone countries in Africa, has a long history of famines and food shortages, mostly
due to climate shocks [15]. A significant proportion of the country’s population is highly vulnerable to food insecurity for
any natural or man-made shocks. For example, the 2016 El Niño left about 10.2 million Ethiopians in need of emergency
food and nutrition assistance. Recent statistics show that, in the Somali National Region (one of the 9 National Regions
of Ethiopia), more than half (58% or 3.7 million people) of the region’s population were aided by World Food Program’s
emergency relief assistance due to a reduction in the amount of rainfall during one production season [24].
Ethiopia had adopted a food security strategy in 1996 for the first time and initiated a food security program in 1998.
But since then, food insecurity and under-nutrition have remained to be crucial problems to the society. Massive food aid
became ineffective instrument in responding to the problem. The performance of the agriculture sector proved itself to be
unable to feed the country’s population that grows at a rate of 2.85%. As under developed agrarian economy, the majority
of the food insecurity problems are reflections of the disturbances in the agriculture sector. In the past, most of the food
aids provided were following climate shocks that affected the sector. However, socioeconomic set up and policy issues of
the country need to be studied not only to help improve the agricultural sector but also to be able to see the nature of
the problem across the different socioeconomic segments of the population. This paper explores some of the socioeconomic
factors that could contribute to food insecurity and analyzes a panel data of over 50 0 0 Ethiopian households.
Food insecurity has long been viewed as a problem affecting rural population only [11]. Consequently, most studies
related to food insecurity tend to forgo urban dwellers. Recent studies, however, (eg. [21]) have started to notice the need
to consider the issue of food insecurity as the problem of urban population too. One of the reasons for the food-insecurity
in urban areas could be the population burden due to the migration from the rural areas. Increased migration from rural
to urban areas takes employment opportunities and puts a pressure on food prices in the urban populations. It also creates
a workforce loss in the rural population which in turn leads to a decline in food production. This is very similar to what
researchers used to call ‘poverty urbanization’, a common phenomenon in sub-Saharan Africa.
There are substantial variations, however, between urban and rural areas in economic opportunities, population density,
access to financial markets and service facilities all of which are likely to affect the lives of people in many dimensions,
including food security. Rural dwellers usually are able to produce their own food, while urban population are entirely
reliant on food purchased from markets.
Most socioeconomic studies consider rural and urban settings differently. The same justification is extended to the study
of food insecurity. For rural populations, the emphasis should be placed on the socioeconomic and agro-ecological situa-
tions and performance of small-holder farmers, while in urban economies the focus additionally considers issues such as
price volatility and market stability [17]. Urban dwellers are highly dependent on markets for their food demands and are
vulnerable to adverse food price shocks.
Financial services helps farmers and urban dwellers to make profitable investments that increase yields and smooth their
consumption patterns. Households with better access to credit and improved financial services are better able to raise their
living standards by engaging in more lucrative farming and non-farm income activities. These services are critical for rural
people since income comes all at once during harvest times, but they need funds at other times of the year, to purchase agri-
cultural inputs during the planting seasons and to smooth their consumption between harvests. Many small-holder farmers
in Ethiopia and other developing countries, however, reside in areas lacking access to basic financial services that it sup-
presses production and income and thus they cannot be confident to have enough food for the family all year round. Holden
and Shiferaw [5] provided an empirical evidence on the importance of credit services for increased productivity and hence
for better food security. The fact that Ethiopian small-holder farmers are practicing only subsistence farming makes it less
attractive for financial institutions to operate.
Education is an important component in human resources development. It is the first most important determinant of
poverty [20]. Education impacts on the food security of households through different channels depending on whether the
household resides in urban or rural areas. In urban settlements, the effect is on employment opportunities. That is, higher
educational attainment increases income and better decision-making capacity. In rural areas, education affects food security
through access to information on best agricultural practices and increased efficiency. Kidane et al. [8] analyzed primary data,
collected in Oromia region of Ethiopia, and found that educational attainment has a positive and significant impact on food
security status. Mutisya et al. [11], using longitudinal data, factored out the effects of education on household food security
and they concluded that the households, specially the urban poor, with lower educational attainment are more likely to
be food insecure than those with some basic education. Ngema et al. [13] using cross section of data from South Africa,
 T.A. Sisha / Scientific African 7 (2020) e00262 3

concluded that education positively influenced the food security status of households. Kiani [7] and Abafita and Kim [1] also
observed the same relationship between educational attainment of the head of the household and the food security status.
Household assets determine the economic activities and livelihood status of a household in a given context. In the rural
context, for example, land has been considered to be the most important asset. This is so because land ownership and
access are closely linked to agricultural production and, as a result, to food security. Whereas in the urban settings, many
other assets are used for income generation. Haddad et al. [4] indicated that Ethiopian household food security, measured
by calorie intake or availability per capita, is positively affected by ownership of physical assets such as land and livestock
values. Using a panel data from the Ethiopian rural household survey, Hunnes [6] concluded that household assets contribute
positively to food security.
Size of a household is among the major factors contributing to the food insecurity of the rural small-holder farmers,
especially when the dependency ratio is higher. The larger the dependency ratio, the higher is the burden to the active
members of the family to feed the household, and hence there will be higher probability of being food-insecured. In many
developing economies, high family sizes are always associated with high dependency ratio, the dependency coming from
lower ages because of higher fertility rates. Ethiopian fertility rate is one of the highest in the world, specially in the rural
areas. Haddad and Co-authors [4,7,10,14] showed that large family sizes are always associated with higher odds of being
food-insecured.
Regassa [16] showed that households employ a range of coping strategies to respond to the high and sustained food
insecurity and hunger, ranging from minimizing the number of meals and amount of food consumption to out-migration of
household members during chronic food shortage. In fact, the coping strategies were dependent on some demographic and
socio-economic variables (such as age of the household head, educational status, access to main social services and others).
Most of the studies reviewed are cross-sectional in nature and fail to capture intra-household dynamics over time. In ad-
dition, many of them consider very specific segment of a population to represent an economy. Most importantly, very few
studies are done in Ethiopia in relation to food insecurity, notwithstanding its prevalence. This study, using recent longitudi-
nal national data, attempts to address these limitations and expands our understanding of the dynamics of food insecurity in
Ethiopia. It uses descriptive techniques to explore the prevalence of food insecurity and binary logistic regression to find out
the household level determinants of food insecurity. It also attempts to explore the coping mechanisms used by households
to mitigate the impacts.

2. Materials and methodology

2.1. Model specification

For this study, the unit of analysis is households, for reasons related to data availability. The author also found very
relevant to refer to food security as ‘removing the fear that there will not be enough to eat’ as used by Maxwell [9]. Self-
reported food insecurity status is used by recalling a situation of ‘food unavailability over the last 12 months’ and whether
the ‘household was worried that there would not be enough food in the past 7 days’, prior to the interview. These two
response variables can reasonably address the four components of food security. The first measures the incidence of food
unavailability and the second the fear that there would not be enough food for the family.
Moreover, to capture the status of food security in children, anthropometric measures are used. The commonly used such
indicators include height-for-age, weight-for-age, and weight-for-height. Measured by these indices, children with a score of
below minus two standard deviations (−2SD ) from the reference population are considered as moderately malnourished.
That is, children below −2SD height-for-age z-score are stunted (short for their age), children with below −2SD height-for-
age z-score are wasted (thin for their age) and children with below −2SD weight-for- height z-score are wasted (thin for
their height).
Household Asset Index: This is a composite measure, calculated using Principal Component Analysis (PCA). In the PCA
model, asset measures such as ownership of a car, motor bike, and bicycle or radio, gas cooker, sewing machine, bed and
mobile phone among others are included.
Household shocks were determined by whether the household had, in the past year, experienced any household shocks,
with the key ones being fire, loss of jobs, floods, illness, death, theft, drought, price fall or rise, violence and local unrest.
Households that reported experiencing a particular shock were coded as ‘yes’ and otherwise ‘no’.
Dependency ratio: We used the conventional definition of dependency to calculate this indicator. That is, the indicator
expresses the proportion of people aged either below 15 or above 64 years within a household as a fraction of the number
of family members aged between 15 and 64 years. International Labor Organization (ILO) allows workers older than the
age of 14 to be legally employed in earning economic activities. But the upper age limit varies across countries based on
the particulars of the countries’ demography, such as life expectancy. This study uses 64 as the older age to stay in the
active working group as used by CSA. Additional variables used in the logit model with their description are summarized
in Table 5. In addition to descriptive analysis, binary logistic regression model is fitted to a longitudinal data which has the
general form:
 p 
= μi + β xit + γ zi + αi
i
log (1)
1 − pi
4 T.A. Sisha / Scientific African 7 (2020) e00262

Table 1
Summary statistics for some of the variables by place of residence.

Rural Small Town Large Town All Area

2011 2013 2015 2011 2013 2015 2011 2013 2015 2011 2013 2015

Household level characteristics

Household size 5.1 5.3 5.2 3.9 4.2 4.3 – 3.4 3.7 5.1 4.7 4.8
Dependency ratio(%) 105 104 100 74 70 69 – 44 48 101 94 88
Average years of schooling in the HH 1.6 1.7 1.8 4.9 4.8 4.7 – 6.9 6.8 1.6 2.9 3.0
HH with access to credit (%) 96.2 95.5 96.2 90.8 95.8 94.9 – 96.1 96.1 94.6 95.7 96.0
Not enough food(%) (7 days) 16.6 15.7 20.1 20.2 16.8 18.5 – 15.4 17.1 17.0 15.7 19.2
Not enough food(%) (12 months) 32.9 34.4 30.6 20.9 19.9 18.6 – 13.8 13.6 28.5 27.4 26.3
Population characteristics
Literacy rate (%) 66.8 44.6 46.6 42.0 69.4 67.9 – 84.0 83.5 44.1 52.6 55.7
Stunted (%) 50.5 41.5 43.8 31.0 24.6 37.2 – 24.7 30.3 50.4 40.0 41.7
Underweight (%) 26.9 22.9 26.4 14.5 10.1 12.5 – 10.0 14.0 25.5 21.7 24.0
Wasted (%) 11.5 11.3 11.1 8.9 7.8 9.9 – 6.4 5.9 11.4 10.9 10.3

where pi is the probability that the household is categorized as food-insecured. zi is a column vector of variables that
describe the households but do not vary over time, and xit is a column vector of variables that vary both over households
and over time for each household. μt is an intercept that is allowed to vary with time, and β and γ are row vectors of
coefficients. α i represents all differences between households that are stable over time and not otherwise accounted for by
zi .
Fixed-effects models are used for looking at the determinants of within-subject variability. If there is no variability within
a subject, there is nothing to examine [12]. It will, therefor be costly for our data to use the fixed effects model. The reason
is a significant proportion of households in the survey remains food-insecured or food-secured in all the three periods.
Specially, variables such as gender, location of households, rural/urban settings, agro-ecology, etc. never changed over the
period under consideration. It is impossible for fixed-effects model to deal with variables that do not change over time. As
a result, most of the discussions are based on the random-effects model result reported in Tables 6 and 7.

2.2. Data description

This study analyzes data from the Ethiopia Socioeconomic Survey (ESS), collected by Central Statistics Agency (CSA) of
Ethiopia in collaboration with the World Bank. ESS is implemented in collaboration with the World Bank Living Standards
Measurement Study (LSMS) team as part of the Integrated Surveys on Agriculture Program (ISAP). The data can be accessed
online at [Link] ESS is a panel and nationally representative survey of over 50 0 0 households living
in rural and urban areas. The first wave was implemented in 2011–12, the second in 2013–14 and the third wave was carried
out in 2015–16.
The survey is implemented in a two-stage probability sampling procedure. The first stage involved selecting primary
sampling units, or CSA enumeration areas (EAs). A total of 433 EAs were selected based on probability proportional to size of
the total EAs in each region. In order to ensure sufficient sample size in the most populous regions (Amhara, Oromiya, SNNP,
and Tigray) and Addis Ababa, quotas were set for the number of EAs in each region. Finally, a sample weight with post-
stratification adjustments was calculated for the households to ensure all regions are represented. The household surveys
were made during the months of January, February and March. Brief summaries of the variables of interest are presented
in Tables 1–3.

3. Results and discussions

Educational outcomes of household members (aged 5 years and above) are captured in the survey by self-reported lit-
eracy, attainment, attendance/enrollment, and constraints such as proximity to primary and secondary schools and school
expenses. Table 2 shows gender-specific information on some of these educational outcomes. The discrepancy in literacy rate
between males and females is higher across places of residence and continued to exist over the three survey periods. At the
national level, the last survey indicated that more than half (64%) of males are literate, compared to females (48%). These
statistics, respectively, were 52.5% and 36% during the first survey, indicating an improvement over time in educational out-
comes. The number of girls and boys (age 7–18 years) who are not at school also declined over time. Gender differential,
with regard to enrollment in primary and secondary schools, seems to disappear recently. Generally, all educational outcome
measurements indicate the existence of slight improvements over the periods under consideration.
Anthropometric indicators generally show an improvement over time, though it is still one of the highest. Child nutri-
tional status in rural areas is worse compared to urban areas. Also, it appears that male children are highly affected. All the
nutrition indicators are much below the world average. In 2017, the prevalence of stunning worldwide, for example, was
approximately 1 in 5 children (22.2%), whereas the prevalence of wasting children was 1 in 14 (7.5%) [23]. These statistics
in Ethiopia, respectively, are 1 in 2 (45%) and 1 in 10 children (10%).
 T.A. Sisha / Scientific African 7 (2020) e00262 5

Table 2
Summary statistics for education and health outcomes .

Male Female All

2011 2013 2015 2011 2013 2015 2011 2013 2015

Education
Literacy rate (%) 52.5 59.3 64.1 35.9 42.9 47.5 44.1 52.6 55.7
Enrollment (Ages 7–18)
Not Enrolled 39.6 35.8 30.2 37.0 33.5 30.0 38.3 34.7 30.1
Primary 56.6 58.2 64.3 59.0 59.7 64.2 57.8 59.0 64.3
Secondary 2.5 4.4 5.5 2.7 5.1 5.8 2.6 4.8 5.6
Health
Children (6–59 months) who are :
Stunted (%) 50.7 40.6 42.2 50.0 39.4 41.2 50.4 40.0 41.7
Wasted (%) 13.6 11.7 12.2 9.1 10.0 8.3 11.4 10.9 10.3
Under weight(%) 29.5 22.7 24.0 24.0 20.8 24.1 26.9 21.7 24.0

Table 3
Reasons why the household did not have enough food for the last 12 months .

2011 2013 2015

Reasons Rural [Link] [Link] Rural [Link] [Link] Rural [Link] [Link]

Farm related 776 45 – 870 23 10 826 44 22

Market related 130 32 – 156 41 83 61 18 98
Other reasons 96 20 – 126 26 112 77 18 61
Chi square (P-value) 45.66 (0.000) 447.50 (0.000) 447.38 (0.000)

Fig. 1. Dynamics of food insecurity over the year.

Dependency ratio, one of the influential characteristics of a household, is higher in the rural population. Larger towns are
about two times lower than rural areas. On average, a household member has to support one other member in rural areas,
whereas in large towns one dependent is served by two members of the household. Most of the dependents in rural areas
come from the lower end of the age distribution, likely driven by higher fertility in rural areas. Family size, the number of
persons per household, is higher in the rural dwellers. In the rural community, about 5 persons live in a typical household.
All the three surveys indicate that urbanization comes with less family size. On average, small towns accommodate about
4 persons and large towns 3.5 persons per household. Feeding, providing shelter, paying school fees and health services
are more expensive in the urban areas. Since education attainment is higher, urban population are better aware of the
consequences of having larger families. Though many developments have been observed during the three survey periods,
the average household size did not show a decline in all the areas.
Despite several improvements in development indicators such as education and health outcomes, insufficient progress
has been observed towards reducing food insecurity over the period under consideration. A significant proportion (on aver-
age 1 in 4 households) of the population is uncertain to have food for the family all over the year. One important common-
ality is the seasonality of food shortage. Though it is more pronounced in rural areas, the months of June, July, August and
September are identified as slack periods for many households (Fig. 1).
In all the indicators of food scurity used in this study (anthropometic measurements, self-reported food availability and
average number of meals per day per household), the discrepancy between rural and urban areas remains apparent (Tables 1,
3 and 4). Both the dynamics and the severity of the problem behave differently for these settings since the driving factor
are different. The self-reported causes of this incidence are summarized in two categories in Table 3 to help identify factors
peculiar to places of residence. The impacts of farm-related causes (such as inadequate household stock due to drought/poor
rains, pest damage, small land size, lack of farm inputs, lack of farm tools) and market causes (such as high food prices,
market distances, unavailability of food in the market, high cost of transport to markets) depends on which segment of the
6 T.A. Sisha / Scientific African 7 (2020) e00262

Table 4
Average number of meals per day per household (5 years and above).

2011 2013 2015

Rural 2.75 2.72 2.83

Small town 2.89 2.86 2.99
Large town – 2.96 3.00

Table 5
Description of the variables used in the model.

Variables Description and measurement

Food insecurity Food insecurity = 1 if the household is food insecured,

Food insecurity = 2 otherwise
Female Female = 1 if head of the household is a female, Female = 2 otherwise
Rural Rural = 1 if the household resides in rural area,
Rural = 2 if the household resides in small towns and
Rural = 3 if the household resides in large towns
Average years of schooling Average years of schooling of all family members in the household
Shock Shock = 1 if the household had encountered any shocks, shock = 0 otherwise
Credit services Credit = 1 if the household has credit services credit = 0 otherwise
Asset A wealth indicator obtained from asset ownership using PCA
Distance from pop. center Distance (KM) of the household’s location from nearest population center

Table 6
Logistic regression (Food availability in the past 7 days) .

(1) (2) (3)

Pooled Fixed Effects Rondom Effects

Food-insecured (7 days)
Female 0.627∗∗∗ (8.32) 0.627∗∗∗ (8.32)
Rural 0.211∗∗ (3.26) 0.211∗∗ (3.26)
Annual Temperature −0.002 (−1.81) −0.002 (−1.81)
Average years of schooling −0.129∗∗∗ (−6.91) −0.129∗∗∗ (−6.91)
Dependency Ratio 0.497∗∗∗ (3.39) 0.497∗∗∗ (3.39)
Credit Services −0.810∗∗∗ (−6.41) −0.464∗∗ (−2.88) −0.810∗∗∗ (−6.41)
Shock 1.459∗∗∗ (21.36) 1.229∗∗∗ (14.23) 1.459∗∗∗ (21.36)
Distance from population center −0.004∗∗∗ (−3.57) −0.004∗∗∗ (−3.57)
Asset −0.083 (−1.94) −0.071 (−1.10) −0.083 (−1.94)
lnsig2u 0.328∗∗ (2.99) 0.328∗∗ (2.99)
N 11,283 3417 11,283

t statistics in parentheses.
Source: Model Result basd on LSMS data.
∗
p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.

Table 7
Logistic regression (Food availability in the last 12 months) .

(1) (2) (3)

Pooled Fixed Effects Rondom Effects

Food-insecured (12 months)

Female 0.462∗∗∗ (6.34) 0.462∗∗∗ (6.34)
Rural −0.336∗∗∗ (−5.08) −0.336∗∗∗ (−5.08)
Annual Temperature −0.00432∗∗∗ (−4.16) −0.0231 ∗
(−2.41) −0.00432∗∗∗ (−4.16)
Average years of schooling −0.181∗∗∗ (−9.84) −0.155∗∗∗ (−3.73) −0.181∗∗∗ (−9.84)
Dependency Ratio 0.853∗∗∗ (6.28) 0.759∗∗∗ (3.88) 0.853∗∗∗ (6.28)
Credit Services −1.016∗∗∗ (−8.14) −0.562∗∗∗ (−3.65) −1.016∗∗∗ (−8.14)
Shock 1.713∗∗∗ (27.89) 1.346∗∗∗ (17.82) 1.713∗∗∗ (27.89)
Distance from population center −0.00152 (−1.35) −0.00152 (−1.35)
Asset −0.207∗∗∗ (−4.56) −0.248∗∗ (−3.28) −0.207∗∗∗ (−4.56)
lnsig2u 0.473∗∗∗ (5.30) 0.473∗∗∗ (5.30)
N 11,251 4523 11,251

t statistics in parentheses.
Source: Model Result basd on LSMS data.
∗
p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.
 T.A. Sisha / Scientific African 7 (2020) e00262 7

population they are acting. In fact, as shown in Table 3, there happens to exist a statistically significant association between
the reasons listed and whether the household is from the rural area or not.
Only when considering food unavailability for the last 7 days, the percentages of food-insecured households do not
show a significant difference across rural and urban areas. This is so, probably because those surveys were done during
the harvest seasons when there is no agricultural product shortfall for the rural households (from February to May). A more
clear picture of food-insecurity for the rural population is captured when reference is made to food availability during the
past 12-months.
The proportion of households that are food-insecured never showed any decline over the periods under consideration,
the rural being at a higher rate, regardless of the various efforts being made by many stakeholders. In Table 1 we see that,
in both food insecurity measurement periods, households became more and more food-insecured over the survey periods
specially in the urban areas. This dynamics is the same for rural population even though the survey is conducted during the
harvest months (February-May) in which one expects the households to be in food-surplus.
In the logit model, the two slightly different response variables (food unavailability for the last 7 days and last 12 months)
are used to measure food insecurity status of the households. Since the data is a short panel, the fixed-effects model became
less appropriate as many of the variables in the study are time-invariant in the three years period. As a result, the model did
not provide estimates for time-invariant variables in the fixed-effects model. Most of the discussions, therefore, are based on
the outputs from the random-effects model. The variables and their measurements included in the model are summarized
in Table 5.
Though the dropout rate reported by the survey is fairly good, above 90%, the number of observations included in this
analysis slightly varies across the survey periods. This is so due to missing data for some of the variables under considera-
tion. Table 6 summarizes the coefficient estimates of the model using the self-reported response variable based on the last
7 days’ food unavailability for the family, whereas Table 7 is using the last 12 months self-reported information.
All the results of the model possess the expected signs. The overall significance of the models are good in both cases
(Tables 6 and 7), that is, when the response variable is measured by food unavailability in the past 7 days and 12 months
period, respectively.
Years of schooling is associated with lower probability of being food-insecured. The result indicates that increasing the
average years of schooling of the family by one year is associated with a decline in the odds of being food-insecure by about
12% when reference is made to the past 7 days and 17% when it is made to the past 12 months.
In developing economies where educational attainment is relatively low, there appears to exist a significant potential
gain through promoting the importance of investment in education. In Table 1, we see that the average years of schooling
is as low as 1.7 in the rural community. Even in larger towns, where the average years of schooling is about 7, a significant
gain could be achieved in reducing food insecurity status through investment in education.
The survey provides evidence for the positive contribution of credit services availability to the food security status of
households. Households with access to credit services have 64% lower odds of being food-insecured than those without the
services. In the survey a significant number of the households were found to have got credit from someone or an institution
whenever they needed. Access to credit services is an essential component of development in any economy. In addition, it
helps households to put themselves at a better food security status.
Members of a household who are unable (because they are not in the working age group) to contribute to the income
earned in the family increase the probability of the family to be food-insecure. These members of the family are only
consumers and burden to the household. The result of the logit model showed a statistically significant result, explaining a
common phenomenon of developing economies. The odds of being food-insecure increases by about 135% (Table 7) for every
additional family member who is not in the working age category. Higher dependency ratio is most commonly a problem
for rural households, the majority of the dependency coming from the lower age group. Promoting family planning helps
women limit the number of their children and hence reduces the risk of having no food for the family.
Wealth is accumulated over time from excess production during good periods and it can be used to feed the family
during bad periods. Households with excess wealth may never have fear of falling into the risk of having no food compared
to those who are poorer relatively. In the logit model asset is used as a wealth indicator, computed using Principal Compo-
nent Analysis (PCA) from several (36) assets ownership of the households. The result indicates that households with higher
wealth are at smaller risk of falling into food insecurity status (Table 7). Estimates from both the fixed- and random-effects
model provide statistically significant coefficients. Wealthier households have about 20% lower odds of being food-insecured
than poorer ones.
Shocks coming from various causes are expected to affect vulnerable population. The significant coefficient from the
regression result in Table 6 clearly indicated that shock is one of the causes for food insecurity. The fixed- and random-
effects model coefficients indicate that shocks are among the major causes for a household to be food-insecure. Households
who have encountered shocks (social or natural) in the last 12 months period have 330% higher odds of being food-insecured
than those who did not face any such shocks. This result further reveals the existence of a substantial proportion of the
population vulnerable to food insecurity.
Annual average temperature records and gender of head of the household that were included in the model turned out
to have a statistically significant coefficients. Female-headed households tend to have a higher probability of being food-
insecured than male counterparts, and specially those in the rural community are the most affected.
8 T.A. Sisha / Scientific African 7 (2020) e00262

Table 8
Percentage of households with coping strategies applied at least for one day of the last 7 days.

Copping strategies 2011 2013 2015

Rely on less preferred food 26.09 18.47 30.93

Limit the variety of foods eaten 25.63 15.37 22.91
Limit portion size at meal times 22.31 13.58 16.47
Reduce number of meals eaten in a day 20.56 12.73 16.07
Restrict consumption by adults for small children to eat 11.71 6.85 10.26
Borrow food, or rely on help from a friend or relative 7.11 4.84 6.92
Have no food of any kind in your household 4.01 2.21 3.07
Go a whole day and night without eating anything 3.15 0.94 2.24

Households devise their own coping strategy when faced with different social and natural problems. Table 8 ranks the
strategies used by the households in the survey. More than a quarter of the households limit the quality and rely on less
preferred food to eat when they faced the risk of food insecurity. Some of the households with severe food insecurity
problem have to go out for help from someone or even could go the whole day and night without eating anything.

4. Conclusions

Based on the findings, the data from the national survey proved that food insecurity continued to be a problem for the
Ethiopian population. The percentage of food-insecured households never show a decline. It further revealed that urban
dwellers are relatively better with regard to food security as compared to the rural dwellers. The levels in both settings
stress the need for area-specific interventions to sustainably reduce the proportion of households suffering from hunger.
Interventions should also consider demographic features of the household to better address the problem. Educational
attainment of the household head, proximity to service centers and wealth positively affect to household food security,
whereas dependency ratio and shock increase the household food insecurity. Households choose to reduce the variety of
food as a primary strategy to reduce the impacts of food insecurity.
In developing countries, many young people are migrating to work in cities, believing that there is no future in farming
and that there are better prospects in urban areas, though quite the opposite is true. Providing advisory support to young
and women farmers in rural areas could help reduce the probability of households falling into food insecurity.

Declaration of Competing Interest

None.

Acknowledgement

None.

Ethical Approval and Consent to participate

None.

Availability of supporting data

The data sets analyzed during the current study are available and can be accessed on [Link]

Funding

None.

Authors’ contributions

None.

References

[1] J. Abafita, K.-R. Kim, Determinants of Household Food Security in Rural Ethiopia: an Empirical Analysis, Journal of Rural Development 37 (2) (2014)
129–157. ISNN 1019-0341.
[2] K. Castetbon, Measuring food insecurity, in: Sustainable Nutrition in a Changing World, Springer International Publishing, Cham, 2017, pp. 35–41,
doi:10.1007/978- 3- 319- 55942- 1_3.
[3] H.R. Dillard, Global food and nutrition security: from challenges to solutions, Food Secur. (2019) 1–4, doi:10.1007/s12571- 019- 00893- 3.
 T.A. Sisha / Scientific African 7 (2020) e00262 9

[4] L. Haddad, S. Bhattarai, M. Immink, S. Kumar, Estimating the interactions between household food security and preschool diarrhea, Food Policy 23
(3–4) (1998) 241–261, doi:10.1016/S0306-9192(98)0 0 039-6.
[5] S. Holden, B. Shiferaw, Land degradation, drought and food security in a less-favoured area in the Ethiopian highlands: a bio-economic model with
market imperfections, Agric. Econ. (2004), doi:10.1016/[Link].20 02.09.0 01.
[6] D.E. Hunnes, The effects of weather, household assets, and safety-net programs on household food security in Ethiopia using rural household panel
data, Reg. Environ. Change 15 (6) (2015) 1095–1104, doi:10.1007/s10113-014-0680-2.
[7] A.S.A. Kiani, Determinants of food security at household level in Pakistan, Afr. J. Bus. Manag. 5 (34) (2011) 12972–12979, doi:10.5897/AJBM11.1441.
[8] H. Kidane, Z.G. Alemu, G. Kundhlande, Causes of household food insecurity in koredegaga peasant association, Oromiya zone, Ethiopia, Agrekon 44 (4)
(2005) 543–560, doi:10.1080/03031853.2005.9523727.
[9] S. Maxwell, Food security in developing countries: issues and options for the 1990s1, IDS Bull. 21 (3) (1990) 2–13, doi:10.1111/j.1759-5436.1990.
mp210 030 02.x.
[10] J. Mbukwa, A model for predicting food security status among households in developing countries, Int. [Link]. 2 (2) (2013) 544–555.
[11] M. Mutisya, M.W. Ngware, C.W. Kabiru, N. bakwin Kandala, The effect of education on household food security in two informal urban settlements in
Kenya: a longitudinal analysis, Food Secur. 8 (4) (2016) 743–756, doi:10.1007/s12571- 016- 0589- 3.
[12] J.M. Neuhaus, Fixed Effects Regression Methods for Longitudinal Data Using SAS, Taylor & Francis Group, 2006, doi:10.1198/jasa.2006.s118.
[13] P.Z. Ngema, M. Sibanda, L. Musemwa, Household food security status and its determinants in Maphumulo local municipality, South Africa, Sustainabil-
ity (Switzerland) 10 (9) (2018) 1–23, doi:10.3390/su10093307.
[14] A.O. Olayemi, Effects of Family Size on Household Food Security in Osun State, Nigeria, Asian Journal of Agriculture and Rural Development 2 (2)
(2012) 136–141. ISSN: 2224-4433.
[15] G. Ramakrishna, A. Demeke, An empirical analysis of food insecurity in Ethiopia: the case of North Wello, Afr. Dev. XXVII (1–2) (2002) 127–143.
[16] N. Regassa, Small holder farmers coping strategies to household food insecurity and hunger in southern ethiopia, Ethiop. J. Environ. Stud. Manag. 4
(1) (2011), doi:10.4314/ejesm.v4i1.5.
[17] S. Saravia-Matus, S.G.Y. Paloma, S. Mary, Economics of food security : selected issues, Bio-Based Appl. Econ. 1 (1) (2012) 65–80.
[18] M.D. Smith, W. Kassa, P. Winters, Assessing food insecurity in Latin America and the Caribbean using FAO’s food insecurity experience scale, Food
Policy 71 (2017) 48–61, doi:10.1016/[Link].2017.07.005.
[19] M.D. Smith, M.P. Rabbitt, A. Coleman- Jensen, Who are the World’s food insecure? New evidence from the food and agriculture organization’s food
insecurity experience scale, World Dev. 93 (2017) 402–412, doi:10.1016/[Link].2017.01.006.
[20] S.B. Thapa, Relationship between education and poverty in Nepal, Econ. J. Dev. Issues 15 & 16 (1–2) (2013) 148–161, doi:10.3126/ejdi.v15i1-2.11873.
[21] B. Tibesigwa, M. Visser, Small-Scale Subsistence Farming, Food Security, Climate Change and Adaptation in South Africa: Male-Female Headed House-
holds and Urban-Rural Nexus, Technical Report June, Economic Research Southern Africa (ERSA), 2015.
[22] UNICEF, The State of Food Security and Nutrition in the World 2018 - UNICEF DATA, 2018.
[23] WHO, Country Progress in Creating Enabling Policy Environments for Promoting Healthy Diets and Nutrition Global Nutrition, 2018, doi:10.1001/
archpedi.1979.02130080108026.
[24] World Food Program, WFP Ethiopia, Technical Report April, 2016.
[25] World Food Program, Global Report on Food Crises 2018, Executive Summary, 2018, pp. 1–16. ISSN 1728-9246.

View publication stats