6th Tigray Proceedings

Ethiopian Economics Association
(EEA)
P R O C EE D I N G S O F T H E SI X TH R E GI O NA L
C O N F ER E NC E O N T H E TI G R A Y R E GI O NA L
S T A T E EC O N OM I C D E V EL O PM E N T
Edited by
S a m u e l G e b rre
eselassie
Demirew Ge tachew
i
Published: April 2019
© Ethiopian Economics Association (EEA)
All rights reserved
ISBN – 978-999-44-54-68-6
ii
TABLE OF CONTENTS
F O R E W O R D ........................................................................................iv
Technical, Allocative and Economic Efficiencies and Sources of

Inefficiencies among Large-scale Sesame Producers in Kafta
Humera District, Western Zone of Tigray, Ethiopia: Non-
parametric approach .............................................................................1
Desale Gebretsadik, Jema Haji, and Bosena Tegegne
Comparative Analysis of Technical Efficiency of Row Plantation

Technology and Traditional Sowing Technology in Barley
Production in Eastern Zone of Tigray .................................................26
Gabriel Temesgen, Yohannes Hailu and Kibrom G/Kirstos
Implication of Animal Feed and Water Scarcity on Labor

Allocation, Food Production and Per Capita Food Consumption in
Tigrai Region, Ethiopia ..........................................................................47
Muuz Hadush
Explaining Inter-regional Differentials in Child Mortality in
Rural Ethiopia: A Count Data Decomposition Analysis ....................93
Yibrah Hagos Gebresilassiea and Phocenah Nyatanga
iii
FOREWORD
The Ethiopian Economic Association (EEA) and its Mekelle Chapter are
happy to issue the proceeding of the Sixth Annual Conference on the Tigray
Regional State Economic Development which was organized on June 30, 2018
at Axum International Hotel Conference Hall, Mekelle. EEA organized this
important regional conference as one of its objectives of broadening its
activities and coverage at regional level so as to contribute to the economic
advancement of regional state through enhancing economic policy formulation
capability; the dissemination of economic research findings; promotion of
dialogue on critical socio-economic issues; promotion of education in
economics in higher learning institutions; enhancing national, continental and
global networks of professionals and institutions; and advancement of the
professional interests of its members.
The Annual Regional Conferences that the Association has organized in

collaboration with its Mekelle Chapter has created important forums for
presenting and discussing development issues that are highly relevant to the
Regional Socio-economy. These forums have also provided incentives for
researchers to conduct research and present their findings on regular basis.
Indeed, the Annual Regional conferences were organized in an
interdisciplinary fashion, thereby widening the interactive coverage involving
both economists living here in the region and those living outside the region
and non- economists who are working and experiences on the region. The
Sixth Annual Regional Conference on Tigray Regional State Economic
Development has contributed towards a deeper understanding of the regional
economy and the complex challenges it faces. It attracted about 80 participants
including higher officials from Tigray Regional State council office and
expertise from Tigray Regional State Bureau of Plan and Finance, Mekelle
City Administration, Tigray Agriculture Research Institute, different Banks,
Universities of Mekelle, Adigrat, Raya and Axum, NGOs, private sector
representative and EEA members in the Region. The participants of the
conference expressed their satisfaction on the organization of the conference
and the content of the papers presented. They reflected that the papers largely
focused on local issue that can contribute to the development of the region.
iv
They also recommended that the issues raised in the discussion are critical that
need due attention by policy makers and implementing organs of the region.
In this publication, all papers which were presented at the Sixth Annual
Conference reviewed by external reviewers and comments and suggestions
including editorial comments were communicated to authors for improvement.
Finally, those papers which passed all the review and editorial process
published in the Proceeding of the Tigray Regional State Economic
Development.
At this juncture, I would like to take this opportunity to express my heartfelt

gratitude, on my own behalf and on behalf of the Ethiopian Economic
Association, to the many people and organizations that made the conference
resounding success. First and foremost, I thank the authors of the papers and
the audience whose active participations made the Conference meaningful.
The staffs of the Economics Department of the Mekelle University which runs
the EEA Mekelle Chapter, participants and the staff of EEA Secretariat
deserve a special recognition for their passion and perseverance in managing
the conference from inception to completion. Mekelle University also deserves
appreciation for hosting EEA Chapter by providing office.
Our special thanks go to our partners who have shared our vision and provided
us with generous financial support to materialize the activities of EEA. These
include; The Friedrich Ebert Stiftung of Germany and the Think Tank
Initiative of International Development Research Center (IDRC) of Canada.
Finally, I would like to extend my sincere gratitude to officials of the regions

and Mekelle University who spared their busy schedule and participated in the
conference.
Tadele Ferede (PhD)

President of the Ethiopian Economics Association
v
Technical, Allocative and Economic Efficiencies and
Sources of Inefficiencies among Large-scale Sesame
Producers in Kafta Humera District, Western Zone of
Tigray, Ethiopia: Non-parametric approach
Desale Gebretsadik1, Jema Haji2, and Bosena Tegegne3
Abstract
This study analyzed production efficiency and sources of inefficiency differentials of

sesame in Kafta Humera district of Tigray region, Ethiopia. It was specifically aimed
to address level of efficiency and lower productivity gap on which its determinant
factors were not identified; by measuring technical, allocative and economic
efficiencies and their sources of inefficiency differentials of sesame in Kafta Humera
District. For addressing these objectives this study used primary and secondary data
obtained from field survey and desk review. Multistage random sampling technique
was used to draw 126 large-scale sesame producers. Applying the Cobb-Douglas
functional form the average technical, allocative and economic efficiencies found
were 71%, 90% and 64% for large-scale producers. Regarding these producers;
education level, frequency of farm visit(number of farms follow up), experience in
sesame production, type of road and credited amount obtained were significant
sources of technical, allocative and economic inefficiencies. Distance of farm from
residence, ownership of living home and livestock and cooperative membership were
also significant sources of technical and economic inefficiencies. Depending on the
results found, this study recommend strengthening the introduction of improved seed,
mechanized labor substituting technologies and fertility enhancing inputs for
improving production level. For improving sesame production efficiency this study
recommends; capacitating large-scale producers through strengthening education,
strengthening the credit access at affordable interest rate, nearby sesame farm follow
up and frequently visiting of their farm with effective farm management will be better.
For improving the farm level efficiency of large-scale producers, it is also important
to initiate producers to hire certified experts, strengthening the productive utilization
of their livestock and their house to earn cash.
Key words: Cobb-Douglas function, Efficiency, Inefficiency sources, Kafta Humera, Large-
Scale, Production and Sesame
1
Tigray Agricultural Research Institute (TARI), Mekelle, Ethiopia; Email,
[email protected]
2
Haramaya University, Dire Dawa, Ethiopia; Email, [email protected]
3
Haramaya University, Dire Dawa, Ethiopia; Email, [email protected]
1
Desale, Jema and Bosena: Technical, Allocative and Economic Efficiencies and Sources…
1. Introduction
Agriculture is the most important sector of Ethiopia’s economy where about

95% of total arable area is cultivated by small-scale that produce more than
95% of total output obtained from the sector [22], [44], [58], [48]. It has
contributed to livelihoods of about 85%, employing about 85% labor force,
accounts about 45% of GDP and for foreign exchange currency of about 86%
[26]. Accordingly, the government of Ethiopia has taken initiatives that can
assure by improving efficiency through reducing losses and improving market
performance.
Sesame is one of the important oilseed crops adapted to semi-arid tropical

regions that best performs on well drained, moderately fertile soils with
temperature between 20-35°C [65]. Sesame is one of the six priority crops in
the AGP of Ethiopia [56]. In Ethiopia, sesame is being produced as cash crop
by small-scale who cultivate 0.42 million ha and produce 0.29 million tone
and by large-scale who cultivate 0.28 million ha and produced more than 0.22
million tones [22]. Nationally, sesame accounts for 3.35% of total area and
1.1% of total grain production [22]. It is produced in North Gondar and
Western Tigray lowlands, Wellega, Benishangul Gumuz and South-Omo;
which North Gondar and Western Tigray contributed more than 68% to the
national production [22].
Over the past years, sesame production has shown greater increase in area and
production but decreasing in yield. Looking at its trend, nationally sesame
covered 0.14 million ha area where 0.12 million tons was produced in 2004/5
[42] increased to 0.29 million tone production on 0.42 million ha in 2014/15
[42]. But, its productivity declined from 8.47 Qt/ha in 2004/5 [42] to 7.35
Qt/ha in 2013/14 [21] and to 6.87 Qt/ha in 2014/15 [22].
In Tigray region, about 176,030 small-scale [22] and about 1130 large-scale
[41] were engaged in sesame production that supplied 88.7% of their
production [21]. According to [21], Tigray region ranked second in area and
production. Western zone took the lion share in the region’s sesame area
2
Proceedings of the Sixth Annual Conference on the Tigray Regional State...
(76.33%) and total production (76%) for the average productivity of 7 Qt/ha
[22].
Given agriculture as backbone of the nation’s food security and as sesame is

the second agricultural product that earns foreign exchange; it is imperative to
conduct study on measuring production efficiency and inefficiency; so, large-
scale producers in the study area were fail to earn profit. Similarly, sesame
suffers lower productivity than the FAO estimated potential [65]. Also,
through farmer’s practice productivity ranges from 2 to 13.75 Qt/ha [67], [31]
which shows wider gap. So, this study was aimed to measure production
efficiency and inefficiency sources. Achievements of these objectives have
significance contributions on making an informed decision for optimum input
allocation and providing scientific information for decision makers, planners,
policy makers, input suppliers, supporting institutions, and other actors. This
study would also help as reference for other studies.
2. Methodology
Description of the Study Area: The study was conducted in Kafta-Humera

district, Ethiopia; bordered by Eritrea, Sudan, Tsegedie district, Welkayt
district and North western zone in the north, west, South, East and north east,
respectively. The study area has 24 kebeles with total population of 103,692
having 26,352 households covering 4,542.33Km2with 396,852ha cultivable
land [39]. There are also 1,130 large-scale producers who cultivate sesame
[41]. The study area is known for cultivation of sesame and sorghum
[39],[31] that obtains annual rainfall ranging from 400-650 mm in the months
from June to September [29].
Data Types, Sources and Methods of Data Collection: Both primary and
secondary data types were collected for this study where primary data sources
were collected using semi-structured questionnaires of formal survey
procedures from large-scale producers in four kebeles. Secondary data sources
are also collected from office of agriculture and rural development, HuARC,
different books and published and unpublished reports.
3
Sampling Procedure and Sample Size: This study used multi-stage sampling
technique for selecting sample producers. First, large-scale producers in the
district were selected purposively. Secondly, four kebeles (Mai Cadra, Baeker,
Adebay and Rawyian) were selected randomly. Then depending on probability
proportional to size of large-scale producers from each sample kebeles,
specified numbers of respondents were obtained based on the formula
developed by [69] considering confidence level of 90% and accepting the
error (e) of 9%,
N (3. 1)
n =
1 + N (e 2
)
Where n = sample size, N= total large-scale household equal to 1,130. Based

on the calculation, 126 large-scale sesame producers were sampled using
random sampling technique (Table 1).
Table 1: Number of sampled producers from each kebele

Kebele Total pop. Sample
Adebay 98 16
Baeker 149 25
Mai Cadra 409 68
Rawyian 107 17
Total 126
Source: [40]
3. Methods of Data Analysis
Production Efficiency and Sources of Inefficiency Differentials

In estimating technical, allocative and economic efficiencies and
inefficiencies, SFA of Cobb-Douglas function was applied; because, it allows
segregating of external effects from inefficiency. From [6], [47], SPF in
Cobb-Douglas form is defined as:
k
ln Y i = β 0
+ ln ∑ β
j =1
j x +εij i
(3. 2)
4
εν u
i= i− i (3. 3)
Where j= 1… k inputs; i= ith producer; Yi= sesame yield, Xij= jth input used,
βi= vector of unknown parameters, εi= disturbance term composed of vi
(error) and ui (inefficiency).
Production function could be either Cobb-Douglas or translog that requires

specification by likelihood ratio test. As it was developed by [14] Cobb-
Douglass production function of dual cost used to specify cost function with
its inefficiency where cost function represents dual approach [18]. The
stochastic nature of cost frontier would still imply the theoretically minimum
cost frontier; stochastic in nature, given as:
C = C P, Y * , α( ) (3. 4)
Or,
 k 
ln ci = α 0 +  ∑ p ,α i ,+α j Yi*
ij
 j =1  (3. 5)
Where i = ith household; Ci = minimum cost; j= 1…k, inputs used; Pij= input
price; Yi* = farm revenue adjusted for noise vi, and α's= parameters to be
estimated.
Variables of sesame production efficiency: These variables are inputs in

sesame production efficiency which could be production or cost inputs that
combined to determine the overall production efficiency.
Production function: This uses the Cobb-Douglas form that shows the
relation of dependent variable with its inputs. The dependent variable is given
by the following equation.
ln Y i
= β + β ln ld
0 i i
+ β ln sd
i i
+ β ln lb
i i
+ β ln pw
i i
+ β ln fq
i i
+ (ν u
i− i
)
(3. 6)
Where ld= land, sd = seed, lb= Labor, pw = Oxen/tractor power, fq = fertilizer

Estimation of cost functions for sesame production
5
This refers to production cost incurred by producer’s calculated taking price of

inputs give as follows:
ln c = ∂ +∂ ln cld + ∂ ln csd + ∂ ln clb + ∂ ln cft +∂ ln cpw + ∂ ln cmt + ∂ ln cop + (ν − u )

j 0j 1j j1 2j j2 3j 3j 4j 4j 5j 5j 6j 6j 7j 7j j j
(3. 7)
Where j= jth producer; cj = actual cost; i=1…7, ith input cost; βj= coefficients
to be determined; v-u = error; cld= land cost; csd= seed cost; clb= wage; cft=
fertilizer cost; cpw= plough cost; cmt= material cost and cop= operation cost.
The minimum cost input equation can be expressed as:
∆ci / ∆pi = x (p ie i , Y *i , α ) (3. 8)
So, optimization profit principle is to minimize cost subject to optimum

output. Minimum cost is derived using the methodology used in [10], [52],
[70] and [24]. Given input oriented function, the efficient cost function is
written as:
k
Min ∑ C = ∑ (x , p
j=i
j j
)
(3. 9)
Subject to
^
βj
Y * = Α ∏ Xj
i (3. 10)
^
Α = Exp β 0 (3. 11)
By substituting the expenditure function and the adjusted yield for stochastic
error in to the above minimization function to derive the following:
C  Y * , Y  = *u αj
H Y i ∏i Pi
 i i  (3. 12)
According to [60], the explained cost measures enable to estimate AE and
further EE.
6
Generally from the above explained concepts TE can be defined in the ratio of
observed outputs (Yi) to the corresponding frontier output (Yi*).
TE i = Y i / Y i = ∑i xit, Pi / ∑i xi, Pip

*
(3. 13)
Also economic efficiency (EE) is the ratio of the minimum costs adjusted or
expenditure (C*) to the actual total production cost or expenditure (C).
EE = c / c = ∑ x P / ∑ x P
*
i ie i i i
(3. 14)
From these two equations the AE can be derived as the ratio of EE to TE.
AE = EE / TE =∑ x P /∑ x P
ie i it i
(3. 15)
Sources of sesame production inefficiency

After measuring TE, AE, and EE, it is important to identify the major sources
of inefficiency derived from different variables. Following the adoption of
[14] for analysis inefficiency using Cobb Douglas functional form, estimation
of inefficiency is specified as:
Ui = σ0 +σ1i w1i + σ 2i w2i + σ3i w3i + ...+σ15i w15i

(3. 16)
Where Ui= inefficiency of ith producer; w1-w15= inefficiency variables.
Estimation of best production function

Selecting the best function relative to other functions is based on tests of
fitness to the data generated. In specifying the best production function this
study conducted hypothesis tests for the parameters of SFA using likelihood
ratio statistic defined by Equation 3.17, that H0 is for Cobb-Douglas and H1 is
for translog.
Likelihood ratio test: This is used to compare the goodness of fit of two
hypotheses given as in eq. 3.17.
LR = λ = − 2 ln [L (H 0 ) / L (H 1 )] = − 2 [ln L (H 0 ) − ln L (H 1 )] (3. 17)
7
Where, L[Ho] is value of H0, L[H1] is value of H1. This also enables to detect
either there is error or not; through comparison of χ2 by obtaining λ, γ and δ2.
λ = δu2 / δν2...or....δu / δν , (3.18) γ = δu2 / δ 2

(3. 19)
δ2 =δu2 +δν2 (3. 20)
Given the specification of SFA, inefficiency is present is defined by Ha: ≠ 0.

In selecting the best fitting model; so further the level of TE, AE and EE and
inefficiencies, the studies conducted by [20], [3], [30], [11], [28], [15], [17]
and [24], similarly used likelihood ratio test.
4. Results and Discussion
Demographic Features and Availability of Production Resources

On average there were six persons in each family with the composition of
three by three for male and female members (Table 2).
Table 2: Household characteristics of large-scale sesame producers

Variables Mean
Age (years) 48.4
Experience (years) 19.8
Family size (No) 6.50
Male members (No) 3.30
Female members (No) 3.20
Active family (No) 3.90
Non-active family (No) 2.50
Dependency ratio 0.38
Education level (years of school) 4.88
Extension contact (No) 1.34
Training obtained (No) 0.91
TLU 12.73
Off-sesame income (Birr) 61,361.27
Borrowed money (Birr) 347,960.30
Own income (Birr) 192,245.10
Labor hired /ha 23.17
Total land 159.86
Sesame land 128.60
Source: Survey result, 2016
8
The sampled sesame producers used hired labor at different production

activities that were 23 man-days/year per-ha owning the average land holding
size of 159.86 ha (Table 2). 98.67% of the sesame produced in 2015/16
production year was sold (Table 3).
Table 3: Amount of sesame allocated for different purposes

Purpose Mean (Qt.) %
Sold 295.91 98.67
Seed 3.6 1.2
Consumption 0.38 0.13
Summary Statistics of Sesame Production Inputs and Costs

The average sesame produced by the sampled producers was 299.43
Qt/household with cost of birr 854469.84 (Table 4).
Table 4.Summary of total production inputs and costs

Variable Unit Mean
Sesame produced Qt 299.43
Production cost Birr 854469.84
Labor used No 2979.44
Labor cost Birr 445387.67
Land size Ha 128.59
Land cost Birr 178981.01
Plough power hour Hr 64.30
Plough cost Birr 63530.04
Operating cost Birr 53458.69
Material cost Birr 7657.32
Seed amount Kg 436.98
Seed cost Birr 12681.64
Fertilizer cost Birr 92773.63
Fertilizer used Qt 72.24
Average sesame yield Qt/ha 2.46
Average production cost/ha Birr/ha 6644.43
9
Estimation of Production Function parameters

Specification tests: Different types of tests were applied for model validity
checking such as multi-collinearity, heteroskedasticity, and adjusted R-square.
Multicollinearity test using VIF for all variables was less than ten (i.e., 5.21),
indicating no severe problem (Table 5). Heteroskedasticity test using the
Breusch-Pagan test also show that there is no heteroskedasticity problem
(Table 5). Adjusted R-squared also was 0.92 indicating the variables explain
92% of the variability in sesame output (Table 5).
Table 5: SFA parameter coefficient for sesame production by large-scale

producers
Ln sesame production unit Coefficients P>|t|
Ln land Ha -1.15*** 0.007
Ln fertilizer Qt 0.024 0.276
Ln labor Man day 0.021 0.352
Ln plow power Tractor hr 2.31*** 0.00
Ln seed Kg -0.279** 0.025
Total (elasticity) 0.926
Constant 3.60*** 0.00
Wald chi2 (5 ) 2746.57 P= 0.00
Sigma_U 0.482 P= 0.00
Sigma_V 0.117 P= 0.00
gamma(γ) 0.944
sigma2 0.246
MLR -24.32
Adj. R-squared 0.9156
hettest. Prob. > chi2 0.217
VIF 5.21
*, **, ***, significant at 10%, 5% and 1%, significance level respectively
The study indicated that variables such as land size and plough power were
significant at 1%; while improved seed at 5% for determining large-scale
sesame production; however, variables such as fertilizer and labor become
insignificant (Table 5). The studies conducted by [13], [35], [64],[17] and
[24] found farm size was significant in determining production. Moreover, ,
[55], [70], [4], 0, [27], [35], [64], [15] and [24] found seed as significant
variable. So, it is observable that the result found by this study is similar with
10
the results obtained by the listed former studies. The inverse relationship
between farm size and yield was similar with the results of [19], [45], [68]
and [15]. The coefficients in Table 5 could be interpreted that, one percent
increment in sesame land size leads to 1.15% decline in yield. Similarly one
percent increment in seed results to 0.28% reduction of total production.
However, one percent increase in plough power hour leads to 2.31%
increment of production.
Elasticity of sesame production: The summation of production inputs’

coefficients was 0.93 (Table 5), indicating the one percent increase in inputs
simultaneously leads to 0.93% increment of production. This has consistency
with the result of [51], [35]and [24] found the scale ranging from 0.84 to
1.2105%.
Cost efficiency: This study found that both error terms (u and v) for sesame
producers were statistically significant at 1% (Table 7). Further, value of
gamma (γ=δu2 / (δ u2 + δ v2)); is γ = 0.9257 that implies 92.57% variability is
contributed by differences in decision maker’s inefficiencies (Table 6).
Regarding the cost function inputs, all variables have statistically significant
with positive sign; except operation and material costs that were insignificant
(Table 7).
Table 6:Tests of cost function model validity

Null hypothesis LR value calculated Critical value (5%) decision
H0: γ=0 92.57 11.07 Reject H0
H0: δ1 = ... δ10=0 77.27 9.39 Reject H0
Source: STATA.13, output
Material cost includes cost of agricultural materials, sack, harvesting and

threshing materials and tractor material and tools. Whereas operating cost
includes cost of; fuel and lubricants, tractor repairing, medical service and
feed expense for draft animals, transportation and loan. This study shares
similarities on cost parameters with the formerly conducted studies by [50],
[15] and [24], in which cost efficiency inputs were wage, seed cost, agro-
11
chemical costs, and amount produced. But also, cost of farm tools by [50] and
land rental cost by [15] in addition to the above explained once.
Table 7: Sesame production cost parameters

Total sesame Production cost Coefficient P>|z|
Ln Operation cost 0.006 0.44
Ln seed cost 0.033** 0.023
Ln fertilizer cost 0.007*** 0.001
Ln material cost -0.009 0.471
Ln plough cost 0.039*** 0.006
Ln labor cost 0.242*** 0.00
Ln land cost 0.083*** 0.004
Ln production 0.62*** 0.00
Elasticity 1.0155
Constant 5.22***
MLR 127.98
Sigma_v 0.0395***
Sigma_u 0.1393***
Sigma2 0.021***
gamma (γ) 0.9257
*, **, *** significance at 10%, 5%, and 1%, respectively
Estimation of technical, allocative and economic efficiencies of large-scale

sesame producers
Technical efficiency: The mean TE level found in this study was 71.46%
(25.6 - 96.03) (Table 8). This implies that if the average producer wants to
achieve the TE of his/her most efficient counterpart, he/she could realize
25.59% input saving [i.e., 1-(71.46/96.03) x100]. Similar the most inefficient
farmer reveals cost saving of about 73.34% [i.e., 1-(25.6/96.03) x100]. The
mean level of TE shows that there is an opportunity to increase efficiency on
average by 28.64% if inputs allocated properly.
12
Table 8: Category of sampled sesame producers based on their TE

Category Number of respondents Percent
TE<20 0 0.00
20<TE<30 2 1.59
30<TE<40 5 3.97
40<TE<50 9 7.143
50<TE<60 15 11.9
60<TE<70 20 15.87
70<TE<80 24 19.05
80<TE<90 37 29.37
TE>90 14 11.11
Mean TE 71.46
Source: Survey results, 2016
The average and range of TE in this study is consistent with the result of [34],
[8], [20], 0, [23], [28], [25], [27], [33], [64], [17] and [24]; ranging in 34-77%.
To give a better picture about TE distributions, a frequency distribution is

categorized by 10% interval; here, 40.48% of the producers were operating
blow mean (Table 8). This imply that in the long run there is a room for
improving the existing TE level of sesame producers providing a special
attention to introduce best alternative farming practices and improved
technologies.
Allocative efficiency: The average AE of large-scale sesame producers was

89.88% (56.94 - 98.16) (Table 9). With this deviation, if the average producer
wants to operate his/her AE to the most efficient, he/she could obtain cost
saving of 8.44% [i.e., 1-(89.88/98.16) x100], however the most allocatively
inefficient could save 42% [i.e., 1-(56.94/98.16) x100]. About 38% of the
sampled producers were operating below mean AE (Table 9).The result
obtained in this study is complementary with the results of [51],[50],[7], [46]
and [17] who found AE from 57 to 96%. Generally, AE of large-scale sesame
producers in Kafta Humera district show that most of the producers have
relatively similar allocation of resources with the unit prices attached to each
input, so leads higher AE.
13
Table 9: Distribution of AE of sesame producer categories

Category Number of respondents Percent
AE< 20 0 0.00
20< AE<30 0 0.00
30<AE<40 0 0.00
40<AE<50 0 0.00
50<AE<60 1 0.79
60<AE<70 0 0.00
70<AE<80 8 6.35
80<AE<90 43 34.13
AE>90 74 58.73
Mean 89.88
Economic efficiency (EE): Following the relative ratio of actual cost to the
hypothetical minimum cost, EE could be obtained which is the multiplication
of TE and AE. Applying this procedure this study found mean EE of 64.58
percent (22.37 - 92.76) (Table 10). Taking this range, if the average producer
wants to reach his/her EE to the most efficient counterpart, he/she could
experience the cost saving of 30.38% ([i.e., 1-(64.58/92.76) x100]. Similarly,
the most inefficient producer could save his/her cost by 75.88% [i.e., 1-
(22.37/92.76) x100]. The mean EE found in this study is similar with the
results of Abu et al. (2012), [49], [30], Abba (2012), [23], [11], [25], [27],
[15], [17] and [24].
As presented in Table 10, about 45% of the sampled producers’ EE was below
mean which is an indication that producers were unfairly efficient; meaning
there was greater variability in their achievement.
14
Table 10: Distribution of EE by large-scale sesame producers

Category Frequency Percent
EE<20 -- --
20<EE<30 6 7.14
30<EE<40 6 7.14
40<EE<50 13 15.88
50<EE<60 21 16.67
60<EE<70 27 28.57
70<EE<80 26 19.84
80<EE<90 25 3.17
EE>90 2 1.59
Mean 64.58
Source: Survey data, 2016
Sources of technical, allocative and economic inefficiency of large-scale

sesame producers
Having information about TE, AE and EE, identifying the major sources of
inefficiency is the next important part of this study. Before using all the
proposed socio-economic and institutional variables into the model a test for
multi-collinearity using VIF is important. Accordingly, the VIF result of each
variable is below ten (i.e., 2.73) (Table 11), indicating no severe
multicollinearity problem. Based on the Breusch-Pagan test result of
heteroskedasticity also, the null hypothesis could not be rejected (Table 11).
The test for cost inefficiency model validity also indicated the result of VIF
for each variable in the model and the mean value of all variables is below 10
(i.e., VIF= 3.59) (Table 11). Based on the Breusch-Pagan test result of
heteroskedasticity also, the null hypothesis could not be rejected (Table 11).
The adjusted R-squared in both production and cost inefficiency also show the
variables explained 69%and 91.46%,respectively (Table 11). As a result all
the variables hypothesized are entered in to their respective models. The
significant sources of technical, allocative and economic inefficiencies
(Table11) are discussed as follows.
Education level of household head (eduhhd): The result of this study shows
that education level of household head significantly and negatively affect to
technical, allocative and economic inefficiency at 1%. If education level in
years of schooling becomes one year higher relative to others, one’s technical,
15
allocative and economic inefficiency decreases by 1.4%, 0.6% and 1.6%,

respectively. This may be, education enables producers to have greater ability
to understand, adopt and correlate inputs with lower cost and misuse.
According to [32]and [43], the relationship between education level and

efficiency is theoretically justified as education increases performing capacity
and so best match of resources; because education is proxy for managerial
ability. The result of this study is similar with the results found [62], [54], 0,
[27], [33], and [59]; but, in conducted to [4]. The result of allocative and
economic inefficiencies obtained in this study is in line with results of [10],
[12], [53], [61], [62], [49], [17] and [63].
Experience in sesame production (exp): It is found that experience of sesame

producers is significantly and negatively affected to technical, allocative and
economic inefficiency of sesame production at 5%, 10% and 5%, respectively.
This could be; because experience is a proxy for managerial aspects and
improves the skill and technical capacity that enables to best match inputs and
in cost saving aspect so attain higher productivity at minimum cost. The
relationship implied that, there is a reduction in technical, allocative and
economic inefficiencies by 0.6%, 0.37% and 0.5%, respectively as one’s
experience increases by one year. The technical inefficiency result is
consistent with the results of [2], [49], [30], [15]and [24]; but contradicts with
result found by [5]. Taking allocative and economic inefficiency the result
found is similar with results of [70], [4], [3], [49], [30] and [16]. However, it
is in contrast to the result of [23] and [24].
Membership in cooperatives (memb): The technical and economic

inefficiency of large-scale producers were significantly and positively
determined by being a membership in cooperative at 1%. Theoretically,
membership in social organizations helps producers in achieving efficiency; but,
this unexpected result could be that members might not discuss related to
sesame production while meeting and they may spend more time while
discussing other issues which compute time of sesame farm operation. Besides,
while producers want to take loan from their cooperative it takes more time; so,
they did not get their credit on its time, spending of time until getting loan which
computed sesame farm operating time. Depending on the result of this study, as
16
sesame producer’s become members of cooperative one’s technical and

economic inefficiencies raises by ten and 10.3%, respectively.
Distance of sesame farm from residence (distfh): It is found that farm

distance is significant and negatively related to technical and economic
inefficiencies at 1% and 5%. Accordingly, as farm distance increases by
01Km, their technical and economic inefficiencies decreases by 0.3%and
0.2%, respectively. This relation may be because there is high probability of
family members or manager to live in production site; so, whole day follow up
is observed that enables to better manage farms which lead to better efficiency
achievement.
Frequency of sesame farm visit (freqgo): It is found that this variable

significantly and negatively determines technical, allocative and economic
inefficiencies at 1%, 10% and 1%, respectively. So, according to the study
result as large-scale sesame producer increases his/her farm visit by one time,
his/her technical, allocative and economic inefficiencies decrease by 0.4%,
0.6% and 0.4%, respectively. Theoretically, the relation could be interlinked,
as there is nearby farm follow up that enables understanding of real happening
so solutions could be emanated.
Ownership of standardized home (ownhom): It is found that this variable

significantly and negatively determined technical and economic inefficiencies
of large-scale sesame producers at 1%. According to the result found, as large-
scale producer owns standard home that could hold as collateral their technical
and economic inefficiencies decreases by 23% and 22% relative to the one
who did not own. This may be as producers had their own house, they do not
pay house rent rather they may allocate the money for sesame production.
Also it can be used as collateral for obtaining loan. Thus, improves their
efficiency. This result matches with the result of [37].
Ownership of livestock (ownliv): This study found that TLU significantly

and negatively affected to technical and economic inefficiencies of the
sampled producers at 10%. This relationship implies that as large-scale
producer’s TLU increases by one, one’s technical and economic inefficiencies
decreases by 5.1% and 0.56%, respectively. This could be as livestock enables
to obtain off-sesame farm income. Regarding, the relationship of TLU and
17
TE, the result in this study is similar with the result of [64], but in
contradiction with the result revealed by [66]. In relation of TLU and EE the
result found is similar with the reports of Amos et al. (2007), [36], [53], [61],
[17] and [63].
Table 11: Sources of technical, allocative and economic inefficiencies

VIF
Technical Allocative Economic VIF cost
Variables production
inefficiency inefficiency inefficiency function
function
Age household head (years) 0.001 -0.001 0.001 1.99 2.40
Education level (schooling years) -0.014*** -0.006*** -0.016*** 1.69 1.92
Experience (years) -0.006** -0.0037* -0.005** 1.67 2.08
Dependency ratio (No) 0.004 0.004 0.006 1.9 2.06
Distance of farm (Km) -0.003*** 0.0002 -0.002** 1.88 2.03
No of extension contact (No) -0.01 0.005 -0.007 2.09 2.84
Frequency farm visit (No) -0.004*** -0.006* -0.004*** 1.93 2.21
Home ownership (Dummy) -0.23*** -0.022 -0.22*** 1.68 2.44
Livestock ownership (TLU) -0.051* -0.017 -0.056* 1.90 2.29
Number of training obtained (No) -0.003 0.001 -0.001 1.76 1.75
cooperative member (Dummy) 0.10*** 0.0204 0.103*** 1.92 2.07
Road type (Dummy) -0.033*** -0.008** -0.033*** 1.48 1.95
Ln Off sesame income (Birr) 0.003 0.0027 0.004 2.48 2.48
Ln loan obtained (Birr) 0.011* 0.003* 0.01** 1.53 2.25
Constant -0.6*** -0.87*** -0.54***
Mean 2.73 3.59
Adjusted R2 0.91 0.69
Chi2 P(X2 = 0.19)=0.12 P(X2 = 0.78)=0.38
*, **, *** significance at 10%, 5%, and 1% respectively
Amount of credit obtained for sesame production (loan): It is significantly

and positively related to technical, allocative and economic inefficiencies of
large-scale producers at 10%, 10% and 5%, respectively. According to this
result, as one obtains one percent of the amount he/she expected and invested
at sesame production one’s technical, allocative and economic inefficiencies
increases by 0.011, 0.003 and 0.01%, respectively. This may be that they spent
the loan obtained in payment of laborers and fertilizer purchase which did not
have significance contribution in TE achievement. It may also due to as loan is
obtained from informal money lenders, who were most familiar in the area
that requires higher interest rate.
18
Availability of road facility from farm to home (road): It is also found that
road facility is significantly and negatively related to technical, allocative and
economic inefficiencies of large-scale producers at 1%, 5% and 1%,
respectively. This is implying that as large-scale producer obtained access to
normal road, one’s technical, allocative and economic inefficiencies decreases
by 0.033%, 0.008% and 0.033%, respectively. It is because accessible road
enables to timely reach so manage farm activities timely and reduced amount
of grain loss while transporting.
19
References
[1] Abba, M.W. 2012. Technical efficiency of sorghum production in Hong local
government area of Adamawa State, Nigeria. Russian Journal of Agricultural
and Socio Economic Sciences, 6(6): 10-15.
[2] Abdukadir, S. 2010. Analysis of technical efficiency of groundnut production:
the case of small-scale farmers in Harari region. MSc Thesis, Haramaya
University, Haramaya, Ethiopia.
[3] Abu, G.A., Abah, D. and Okpachu, S.A. 2011. Analysis of cost and return for
sesame production in Nasarawa State: Implication for sustainable development
in Nigeria. Sustainable Development in Africa Journal, 13(3): 238-249.
[4] Abu, G.A., Ater, P.I. and Abah, D. 2012. Profit efficiency among sesame
farmers in Nasarawa State, Nigeria. Current Research Journal of Social Sciences,
4(4): 261-268.
[5] Adeyemo, R., Oke, J.T.O. and Akinola, A.A. 2010. Economic efficiency of
small-scale farmers in Ogun State, Nigeria. Tropicultura, 28(2): 84-88.
[6] Aigner, D.J., Chu, S.F. 1968. On estimating the industry production function.
American Economic Review, 58: 826-39.
[7] Alboghdady, A.M. 2014. Non parametric model for measuring impact of inputs
density on Egyptian tomato production efficiency. International Journal of Food
and Agricultural Economics, 2(4): 81-90.
[8] Amaza, P.S., Udoh, E.J., Abdoulage, T. and Kamara, A.Y. 2010. Analysis of
technical efficiency among community based seed producers in the Savannas of
Borno State, Nigeria. Journal of Food, Agriculture and Environment, 8(2): 1073-
1079.
[9] Amos, T., Chikwendu, D. and Nmadu, J. 2007. Productivity, technical efficiency
and cropping patterns in the Savanna Zone of Nigeria. Journal of Food,
Agriculture and Environment, 2(2): 173-176.
[10] Arega, D. Alene and Rashid, M. Hassan. 2006. The efficiency of traditional and
hybrid maize production in Eastern Ethiopia: An extended efficiency
decomposition approach. Journal of African Economics, 15(1): 91-116.
[11] Asad, A.S., Baig, I.A., Ayaz, S. and Mahmood, N. 2014. Impact of resource
management on the development of arid area productivity and livelihood: A case
study of Rawalpindi district. Journal of Scholarly Agricultural Science, 4(2): 90-
96.
[12] Aye, G. and Mungatana, D. 2010. Technological innovation and efficiency in the
Nigerian maize sector: Parametric stochastic and non-parametric distance
function approaches. Paper presented at the Joint 3rd African Association of
Agricultural Economists (AAAE) and 48th Agricultural Economists Association
20
of South Africa (AEASA) Conference. September 19-23, Cape Town, South

Africa.
[13] Baten, M. A., Anton, A. K. and Mohammad, A. H. 2009. Modeling technical
inefficiency effects in a stochastic frontier production function for panel data.
African Journal of Agricultural Research, 4(12): 1374-1382.
[14] Battese, G.E. and Coelli, T. 1995. A model of technical efficiency effects in a
stochastic frontier production functions for panel data. Empirical Economics,
20(2): 325-332.
[15] Berhan Tegegne Haile. 2015. Determinants of technical, allocative and
economic efficiencies among onion producing farmers in irrigated agriculture:
Empirical evidence from Kobo district, Amhara region, Ethiopia. African
Journal of Agricultural research, 10(20): 2180-2189.
[16] Biam, C. K., Okorie, A. and Nwibo, S.U. 2016. Economic efficiency of small-
scale soya bean farmers in Central Agricultural Zone, Nigeria: A Cobb-Douglas
stochastic frontier cost function approach. Journal of Development and
Agricultural Economics, 8(3): 52-58.
[17] Chakwera, S. J. 2015. A survey on technical, allocative, and economic efficiency
of maize production using the parametric stochastic frontier production function.
African Journal of Agricultural Economics and Rural Development, 3(6): 237-
245.
[18] Chambers, R. G. 1988. Applied Production Analysis: A Dual Approach.
Cambridge Syndicate University Press, Cambridge, Malaysia.
[19] Chand, R., Prasanna, P. A. and Shigh, A. 2011. Farm size and productivity:
Understanding the strengths of small-scales and improving their livelihoods.
Economic and Political Weekly Supplement, Review of Agriculture, XLVI: (26
&27): 5-11.
[20] Chimai, B. C. 2011. Determinants of technical efficiency in small-scale sorghum
farming in Zambia. MSc Thesis, Ohio State University, Ohio, USA.
[21] CSA (Central Statistical Authority) Federal Democratic Republic of Ethiopia.
2014 (2006 E.C.). Agricultural sample surveys: Area and production of major
crops (Private Peasant Holdings, Meher Season) Volume I, Addis Ababa,
Ethiopia.
[22] CSA (Central Statistical Authority) Federal Democratic Republic of Ethiopia.
2015 (2007 E.C.). Agricultural sample surveys: Area and production of major
crops (Private Peasant Holdings, Meher Season) Volume III, Addis Ababa,
Ethiopia.
[23] Dawit Kelemework, Jerey, H. D. and Esendugue, G. F. 2013. Productivity and
Efficiency of Small-scale Agriculture in Ethiopia. Selected paper prepared for
presentation at the Southern Agricultural Economics Association (SAEA)
Annual Meeting: February 3-5, 2013, Orlando, Florida.
21
[24] Ermiyas Mekonen, Endrias Geta and Belaineh Legesse. 2015. Production
efficiency of sesame in Selamago district of South Omo Zone, Southern
Ethiopia. Current Research in Agricultural Sciences, 2(1): 8-21.
[25] Evaline, C., Eric, K. B., Sospeter, O. N. and Kwena, K. 2014. Analysis of
technical efficiency of sorghum production in lower Eastern Kenya: A Data
Envelopment Analysis (DEA) approach. Journal of Economics and Sustainable
Development, 5(4): 58-65.
[26] FDRE (Federal Democratic Republic of Ethiopia). 2016. Ethiopian government
Economy Portal. (Accessed on 09 November, 2016; at:
http://www.ethiopia.gov.et/web/Pages/Economy).
[27] Getachew Magnar Kitila and Bamlak Alamirew Alemu. 2014. Analysis of
technical efficiency of small-scale maize growing farmers of
HoroGuduruWollega Zone, Ethiopia: A Stochastic Frontier Approach. Science
technology and arts research journal, 3(3): 204-212.
[28] Getahun Gemechu Abebe. 2014. Off-Farm Income and Technical Efficiency of
small-scale Farmers in Ethiopia: A Stochastic Frontier Analysis. MSc Thesis,
Swedish University of Agricultural Sciences (SLU), Uppsala, Sweden.
[29] Hagos Tadesse, Fetien Abay. 2011. Additive main effects and multiplicative
interactions analysis of yield performance of sesame genotypes across
environments in Northern Ethiopia. Journal of the Dry lands, 4(1): 259-266.
[30] Hidayah, I., Nuhfil, H., Ratya, A. and Budi, S. 2013. Production and cost
efficiency analysis using frontier stochastic approach: A case on paddy farming
system with integrated plant and resource management (IPRM) approach in
Buru district Maluku Province Indonesia. Economics and Sustainable
Development Journal, 4(1): 78-85.
[31] HuARC (Humera Agriculture Research Center). Report 2014. Productivity
improvement through the application of full package inputs on farmer’s field on
main crops (sesame, Sorghum): The case of lowland areas of Western Zone,
Tigray regional state. Humera, Ethiopia.
[32] Huffman, W.E. 1980. Farm and oﬀ-farm work decisions: The role of human
capital. Review of Economics and Statistics, 62(1): 14-23.
[33] Hussain, N., Ali, S., Miraj, N. and Sajjad, M. 2014. An estimation of technical
efficiency of garlic production in Khyber Pakhtunkhwa, Pakistan. International
Journal of Food and Agricultural Economics, 2(2): 169-178.
[34] Ibrahim, E.M. 2007. Technical efficiency of production in the mechanized rain-
fed sector: A case of Habila Scheme, Sudan. MSc Thesis, Khartoum University,
Khartoum, Sudan.
22
[35] Ibrahim, U.W., Umar, A.S.S. and Ahmed, B. 2014. Technical efficiency and its
determinants in water melon production in Borno State, Nigeria. Journal of
Economics and Sustainable Development, 5(27): 205-2011.
[36] Idiong, I.C., Agom, D.I, Effiong, E.O. and Ohen, S.B. 2009. Analysis of
technical and economic efficiencies in rice production systems in the Niger Delta
region of Nigeria. pp. 243-249. Proceeding of the 23nd Annual Conference of
the Farm Management Association of Nigeria (Faman), December, 14-17, 2009.
Farm Management Association of Nigeria, Mombasa, Nigeria.
[37] Jema Haji. 2008. Economic efficiency and marketing performance of vegetable
production in Eastern and central parts of Ethiopia, Doctoral Dissertation, Acta
Universitatis of Agriculture Sueciae (SLU) Uppsala, Sweden.
[38] Kahsu Kelali, Melisew Misker and Annemarie, G.K. 2014. Sesame yields and
post-harvest loss in Ethiopia: Evidence from the field. Summited to Sesame
Business Network (SBN), Gondar, Ethiopia.
[39] KHARDO (Kafta Humera Agricultural and Rural Development Office). 2013.
Annual report on crop production and productivity: In Kafta Humera woreda.
Kafta Humera, Ethiopia.
[40] KHARDO (Kafta Humera Agricultural and Rural Development Office). 2015.
Annual report on production and productivity of agriculture in Kafta Humera
woreda. Kafta Humera, Ethiopia.
[41] KHLAdO (KaftaHumera district of Land Administration Office). 2015.
Documented file on list of large-scale producers in Kafta Humera woreda with
their land size allocated. Kafta Humera, Ethiopia.
[42] Kindie Aysheshim. 2007. Sesame market chain analysis: The case of Metema
district Amhara regional state, Ethiopia. MSc Thesis, Haramaya University,
Haramaya, Ethiopia.
[43] Lockheed, M., Jamison, D. and Lau, L. 1980. Farmer education and farm
efficiency: A survey. Journal of Economic Development and Cultural Change,
29(1): 37-76.
[44] Mahelet, G. F. 2007. Factors affecting commercialization of small-scale farmers
in Ethiopia: The case of North Omo Zone, SNNP region. Paper presented at the
5th International Conference on the Ethiopian Economy, June 7-9. Ethiopian
Economic Association. Addis Ababa, Ethiopia.
[45] Maqbool, H. S., Shahid I. and Sheikh, A. D. 2012. Farm size–productivity
relationship: Recent evidence from central Punjab Pakistan. Economic and
Social Review, 50(2): 139-162.
[46] Mburu, S., Ackello-Ogutu, C. and Mulwa, R. 2014. Analysis of economic
efficiency and farm size: A case study of wheat farmers in Nakuru district,
Kenya. Economics Research International, Article ID 802706.
23
[47] Meeusen, W. and Vanden Broeck, D. J. 1977. Efficiency estimation from Cobb-
Douglas production functions with composed error. International Economic
Review, 18(2): 435-444.
[48] MoARD (Ministry of Agriculture and Rural Development). 2010. Ethiopia’s
agricultural and sector policy and investment framework (ASPIF): 2010-2020,
Addis Ababa, Ethiopia.
[49] Myo, N. A. 2012. Production and economic efficiency of farmers and millers in
Myanmar rice industry. Institute of Developing Economies, Japan External
Trade Organization, 471: 261-8545.
[50] Ogundari, K. and Ojo, S. O. 2007. Economic efficiency of small-scale food crop
production in Nigeria: A stochastic frontier approach. Journal of Social Science,
14(2): 123-130.
[51] Ogundari, K. and Ojo, S. O. 2005. The determinant of technical efficiency in
mixed crop food production in Nigeria. A stochastic approach. pp. 159-164.
Proceeding of the 1st annual conference on development in agriculture and
biological science of Nigeria, 21st April, 2005. School of Agriculture and
Agricultural Technology, Federal University of Technology, Akure, Nigeria.
[52] Ogundari, K., Ojo, S. O. and Ajibefun, I. A. 2006. Economics of scale and cost
efficiency in small-scale maize production: Empirical evidence from Nigeria.
Central European Agriculture Journal, (1): 15-26.
[53] Otitoju, M. and Arene, J. 2010. Constraints and determinants of technical
efficiency in medium-scale soybean production in Benue State, Nigeria. African
Journal of Agricultural Resource 5(17): 2276–2280.
[54] Rahman, K., Mia, M., and Bhuiyan, M. 2012. A stochastic frontier approach to
model technical efficiency of rice farmers in Bangladesh: An empirical analysis.
Scientific Journal Krishi Found. 10(2): 9-19.
[55] Rahman, S.A. and Umar, H.S. 2009. Measurement of technical efficiency and its
determinants in crop production in Lafia local government area of Nasarawa
State, Nigeria. Journal of Tropical Agriculture, Food, Environment and
Extension, 8(2): 90-96.
[56] SBN (Sesame Business Network) Support Programme. 2013b. Launching
Workshop Report: In North Western Ethiopia. May 15-16, 2013. North Western
Ethiopia, Gondar, Ethiopia.
[57] SBN (Sesame Business Network) Support Programme. 2014. Annual report
2013 and orientations for 2014. Gondar, Ethiopia.
[58] Seyoum Taffese Alemayohu, Bingxin, Y., Kassu Wamisho and Xinshen, D.
2007. Agricultural productivity and modern inputs in the Ethiopian economy.
ESSP Working Paper, International Food Policy Research Institute, Washington
24
D.C, USA and the Ethiopian Development Research Institute, Addis Ababa,
Ethiopia.
[59] Shalma, H.J. 2014. Economic analysis of soya bean production under Sasakawa
global 2000 project in Kaduna state, Nigeria. MSc Thesis, Ahmadu Bello
University, Zaria, Nigeria.
[60] Sharma, K. R., Leung, P. and Zaleski, H. M. 1999. Technical, allocative and
economic efficiencies in swine production in Hawaii: A comparison of
parametric and non-parametric approaches. Journal of Agricultural Economics,
20: 23-35.
[61] Shehu, J., Iyortyer, S., Mshelia, I. and Jongur, U. 2010. Determinants of yam
productivity and technical efficiency among yam farmers in Benue state Nigeria.
Journal of Social Science, 24(2): 143-148.
[62] Shumet Asefa. 2011. Analysis of technical efficiency of crop producing small-
scale farmers in Tigray, Ethiopia. MPRA Paper No., 40461.
[63] SisayDebebe, Jema Haji, Degye Goshu and Abdi-Khalil Edriss. 2015. Technical,
Allocative and economic efficiencies among smallholder maize farmers in
Southwestern Ethiopia: Parametric approach. Journal of Development and
Agricultural Economics, 7(8): 282-291
[64] Wassie Solomon Bizayehu. 2012. Application of stochastic frontier model on
agriculture: Empirical evidence in wheat producing areas of Amhara Region,
Ethiopia. BoD-Books on demand publishing, Gutenbergring, Germany.
[65] Wijnands, H. M., Biersteker, J. and Heil. R. 2007. Oilseeds business
opportunities in Ethiopia. The Hague.
[66] Wondimu Tefaye and Hassen Beshir. 2014. Determinants of technical efficiency
in maize production: The case of small-scale farmers in Dhidhessa district of
Illuababora Zone, Ethiopia. Journal of Economics and Sustainable Development,
5(12): 274-284.
[67] WTZAO (Western Tigray Zone office of Agriculture). 2014. Annual Report on
production and productivity of crop production: Lowland areas of the western
zone. Western zone Tigray, Humera, Ethiopia.
[68] Wutyi, N.S, Ismail, A. and Zainal, A.M. 2013. Farm efficiency and
socioeconomic determinants of rain-fed rice production in Myanmar: non-
parametric approach. Empirical Research Asian Journal, 3(11): 1401-1413.
[69] Yamane, T.I. 1967. Statistics: An introductory analysis: 2nd edition. New York:
Harper and Row.
[70] Zalkuwi, J.W, Dia, Y.Z. and Dia, R.Z. 2010. Analysis of economic efficiency of
maize production in Ganye local government area Adamawa state, Nigeria.
Report and Opinion, 2(7): 1-9.
25
Comparative Analysis of Technical Efficiency of Row
Plantation Technology and Traditional Sowing
Technology in Barley Production in Eastern Zone of
Tigray
Gabriel Temesgen1, Yohannes Hailu2 and Kibrom G/Kirstos3
Abstract
Ethiopian GDP highly depends on the production and productivity of

agriculture. Agricultural productivity is seen as one of the major contributors
to the development process. It is, therefore, essential to study the performance
of existing plantation technology systems in order to become informed about
this development process. This study examined the factors influencing
technical efficiency in barley farming in eastern zone of Tigray using a
stochastic frontier production function in which technical inefficiency effects
were assumed to be functions of both socioeconomic characteristics of the
farmer and farm-specific characteristics of the two seed plantation
technologies; namely row plantation system and traditional plantation system.
In this research work paper the researchers select randomly 300 farmers
comprising 155 row planters and 145 traditional planter farmers. The result
from the descriptive statistics indicates that the farmers who use traditional
plantation technology are producing on a lower production frontier than the
farmers who practice row plantation technology and the result is also
statistically significant at one percent in a two tail sample t test. The results
also revealed existence of high levels of technical inefficiencies in barley
production, especially among the traditional sowing farmers. The study found
that the magnitude of technical efficiency varied from one farmer to another
and ranged from 42.4% to 75.4%, with a mean of 68.2%. Consequently, due
to technical inefficiency farmers have lost close to 32% of the potential
output. The main factors that influenced the degree of inefficiency were age of
the household head, family size, livestock quantity, row plantation technology,
access to irrigation and cooperative membership. Based on the findings from
this study, the researchers recommend that farmers should have to get
1
Department of Economics, Adigrat University
2
3
26
trainings on how to plant seeds, on the use of better techniques and

application of fertilizer and other capital equipment. Moreover, the regional
government should have to develop small scale irrigation schemes to enhance
the productivity of row plantation technology. Last but not the least, though
the result from the Tobit model and t test statistics are significant, row
plantation technology user farmers are on higher frontier than their
counterparts. Thus integrating those who use traditional cultivation method
can lead to more viable production and productivity in using this technology.
Key words: Productivity, technical efficiency, stochastic frontier function, two limit
tobit model, and barley farming in Eastern zone of Tigray
1. Background and Justification
Agriculture is the backbone of the Ethiopian economy. This particular sector

determines the growth of all other sectors and consequently, the whole
national economy. On average, crop production makes up 60 percent of the
sector’s outputs, whereas livestock accounts for 27 percent, and other areas
contribute 13 percent of the total agricultural value added. The sector is
dominated by small-scale farmers who practice rain-fed mixed farming by
employing traditional technology, adopting a low input and low output
production system. The land tilled by the Ethiopian small-scale farmer
accounts for 95 percent of the total area under agricultural use and these
farmers are responsible for more than 90 percent of the total agricultural
output (GTP, 2010).
Agriculture is the mainstay of the Ethiopian economy and underpins its

development process. It is a sector with great potential for stimulating growth
and employment and eradicating poverty. Because of its importance to
national food security and poverty reduction, the government has, within the
Growth and Transformation Plan (GTP), articulated a clear vision for the
sector, placing it at the center of the country’s transformation agenda. The
initiatives that underlie the agriculture policy and plan aim to stimulate
investment and productivity of the sector to promote household and national
food security and to rally development partners to deliver effective
development aid to the sector. Transformation of the Ethiopia’s agricultural
sector requires scaling up efforts to increase agricultural production and
27
Gabriel, Yohannes and Kibrom: Comparative Analysis of Technical Efficiency of Row Plantation…
productivity by among others promoting domestic and foreign investment

through agricultural commercialization, increasing public investment in
agricultural infrastructure, promoting technology transfer and adoption,
ensuring efficient use of land, labor, technology and other inputs, and
specifically raising the productivity of smallholder farmers (GTP, 2010).
Among thus, row planning is one agricultural technology where high emphasis
is given for improving the productivity of small holder farmers in the Growth
and Transformation Plan of Ethiopia.
Beyond these investments and interventions, the introduction of new

technologies through a strengthened extension system has been a major area of
effort for the Transformation Agenda. The efforts in extending the Tef,
Improved Seed Variety, Row Planting, Reduced Seed Rate (TIRR) technology
package is an illustration of the significant yield increases that can be realized
from seemingly simple technologies.
The core “TIRR” technology package (Tef, Improved seed, Reduced seed rate,
and Row planting) prioritized for tef farmers by the agricultural extension
system 2013, led to significant increases in crop yields across the country.
Detailed analysis of the 2013 TIRR package, with a sample of 1,300 farmers,
showed average yield increases of 44% the control group and 72% the
experiment group (MoA, 2014/15).
During the GTP period, government aims to double the production of

smallholder farmers by implementing measures to raise and sustain high
agricultural productivity. The scope to increase production through area
expansion is continuously diminishing as land for agriculture gets exhausted,
making this approach less sustainable in the long term. In Ethiopia,
agricultural productivity among smallholder farmers is as low as 1.25 tons per
hectare for tef, there is also great variability in productivity across farmers
with the most productive farmer producing 3.66 tons per hectare compared to
the average yield of 1.83 per hectare for cereals (MoA, 2014/15).
One crucial element in the process of crop production is land preparation, as it

is decisive in obtaining a good harvest. Establishing a good crop, increasing
yield per hectare, reducing weed pressure, and improving soil moisture
28
retention all depend on good land preparation (tillage). Farmland is prepared

using the traditional ploughing instruments. Row planting is not easy for the
farmers as it needs more labor than broadcasting; therefore, many apply the
latter alternative. These and other activities, like weeding and soil fertility
management, are highly labor-intensive (Atsbaha G. and Tessema B. 2010).
There are many constraints to agricultural inefficiency including the small and
diminishing size of farm lands; inadequate extension services and follow-ups
by the respective office of agriculture; soil infertility; outdated modes of
production; and a lack of correct agricultural information.
This shows that there is great potential to increase production by raising yields
per hectare for all smallholder farmers to that of the most productive (model)
farmer. Significant productivity differences also exist across agro-ecological
zones. These differences provide additional prospects for increasing
production and productivity by providing incentives that induce farmers to
optimally exploit zonal specific advantages to enhance returns from
agricultural investment. Doing so will not only increase agricultural
production through specialization and commercialization of agricultural
production but will help to raise agricultural household income and
employment, and ultimately contribute to poverty reduction in the rural sector.
Table 1: Tradition and row plantation technologies

Tradition sowing plantation
Row plantation technology
Woreda technology
Target Actual % Target Actual %
Atsbi Wenberta 6868 8810 128.28 5976 3620 60.576
K/Awlaelo 11556 13634 118 8020 4253.5 53
Hawzen 7007 10761 153.6 9305 3488 37.5
S/T/Emba 13262 14752 111.2 6538.5 4427 67.7
G/Afeshum 6910 5441 78.7 3613 4503 124.6
G/Mukada 7205.5 6551 90.9 3998.5 4653.3 116.4
Erob 634 909 143.4 566 200.96 35.5
Total 53442 60858 113.9 38017 25146 66.1
Source: Zonal BOARD 2015
29
In addition to the lack of agricultural technologies, the problem of low

productivity on smallholder farms is inadequate knowledge, skills and
resources (inputs such as fertilizer, labor, equipment, seeds and water) to
enable them adopt and efficiently utilize existing technologies to enhance
production and earning from farming. Thus, researches are necessary to
identify the low agricultural productivity. Thus, the existence of steady
economic inefficiency for decades in the nation and the prevalence of
production differences using different agricultural technologies in the region,
inadequate knowledge, and absence of scientific research carried out to assess
the socioeconomic determinants of economic inefficiency are the major
factors influencing this research to be realized.
However, unlike the remarkable achievements in the agricultural sector, there

are still gaps in the areas of household production and economic efficiency
that require further development, involvements of the government and other
development agents to share the fate of sustainable development. In addition,
most of the research works done on agricultural technologies focuses on the
impact of these technologies on livelihood of households. Thus, they never
relate with production and economic efficiency. This calls for the realization
of this research work to fill such gaps and provide scientific evidence on the
socio economic variables that determine economic inefficiencies.
Even though the interventions, the introduction of new technologies through a

strengthened extension system has been given a major area of effort for the
Transformation Agenda, however; studies on assessing the economic
efficiency of row plantation in comparison to traditional sow plantation is not
done yet. And to the best of the researchers understanding there were not
researches so far conducted systematically in this region where this study is
proposed to be conducted. For this reason, the study aims to compare
production and economic efficiency of row plantation system with the
traditional plantation system in cereal production using appropriate methods of
data analysis.
30
2. Objective of the Study:
General Objective:
The general objective of the study is to compare production and economic
efficiency of row plantation system with the traditional plantation system in
cereal production.
Specific Objectives:
• To compare the production efficiency (technical efficiency) of row
plantation production system with the conventional one
• To analyze the socioeconomic variables that may explain the differences
in the estimated levels of technical inefficiency
3. Methodology of the Study
Methodology
This study employed a mixed approach with an emphasis given to quantitative
household survey supplemented by the qualitative research method. The
quantitative research approach is to compare the production and technical
inefficiency of row plantation production system with the conventional one
and to analyze the socioeconomic variables that may explain the differences in
the estimated levels of technical inefficiency. In line with this, to capture some
variables which are non-quantifiable (either methodologically or due to other
reasons), qualitative methods of data analysis will also be used to describe the
cropping patterns of the two production systems.
Research Process
Based on the objectives, the research process with in this study was divided
into five stages. In the first stage, review of relevant secondary sources was
conducted which, in fact, served as the background for understanding the
research problem and hence set a research problem with in the ongoing
dialogue in the literature.
In the second stage, the random selection of the study woredas was done from
the seven woredas of eastern zone of Tigray based on the implementation of
the two production systems.
31
Thirdly, selection of tabias from different agro ecological zones was

undertaken to ensure diversity in the study and hence equal sample of
respondents was drawn using a systematic random sampling. In the fourth
stage, household survey using structured questionnaire and focused group
discussions was under taken. Lastly, since the purpose of the research is to
produce findings and the process of data collection is not an end by itself, data
analysis, interpretation and presentation of findings was conducted.
Therefore, a total of 300 households were selected from three agro ecological
zones from four tabias by using systematic random sampling method
individual household for questionnaire survey was selected.
Method of Data Analysis

As part of quantitative research methods, primary data was collected by means
of survey questionnaire. In the sample survey, in-depth information regarding
the social and demographic characteristic, different agricultural inputs,
livestock ownership and institutional variables were collected. All these data
were considered during the analysis to compare production and technical
efficiency of row plantation system with the traditional plantation system in
cereal production.
As part of quantitative data analysis, an econometric model was also used to

compare production and economic efficiency of row plantation system with
the traditional plantation system in cereal production.
Econometric Model specification

To compare production and economic efficiency of row plantation system
with the traditional plantation system in cereal production, a Stochastic
Frontier Analysis and two limit tobit model was employed in this study.
Frontier economic programming (version 4.1) software was used for
estimating the farm specific economic efficiency scores of cereals producers in
the study area. Following that the efficiency score is taken as a dependent
variable and is then regressed against farmer specific, demographic,
socioeconomic and institutional factors.
32
Boris et al. (1997) described that Cobb- Douglas functional form is used to
specify the stochastic production frontier, which is the basis for deriving the
cost frontier and the related efficiency measures. The specific Cobb- Douglas
production model estimated is given by:
= 0∗ ∗
By Transforming this in to double log linear model:
= 0∗ 1 +( − )
Where Yi represents cereal yield harvested and Xi represents cereal

production inputs by ith farmer. Whereas, 0, 1, 2, 3, 4, 5 $ % 6
are regression parameters to be estimated. From the error term
component ( − ), is a two sided (−∞ < < ∞) normally distributed
random error (v˜N [0, ( ]) that represents the stochastic effects outside the
)
farmer’s control (e.g whether, natural disaster,…), measurement error and

other statistical noise. While Ui is a one sided (Ui≥0) efficiency component
which is independent of Vi and is normally distributed with zero mean and
constant variance (( ) ) allowing the actual production to fall below the
frontier but without attributing all short falls in output from the frontier as
inefficiency.
Two limit tobit model with maximum likelihood estimation
Following Amemiya (1981), Waluse (2011) Essa et al (2011) and Endrias et

al. (2013) the two limit tobit model is defined as:
* ∗ +, = (0 + - (./ . +
0
Where * * is the latent variable representing the efficiency scores,
(0, (1, … . , (12 are parameters to be estimated, and TE (technical efficiency)
and of the ith farmer. Zi is demographic, socio economic and institutional
factors that affect efficiency level. And Ui is an error term that is
33
independently and normally distributed with mean zero and variance (2

(Ui˜IN (0, (2)). Farm specific efficiency scores for the smallholder cereal
producers range between zero and one. Therefore, two limit tobit model can be
presented as follows:
Yi= 1 if Yi*≥1
Yi= Yi* if 0 < Yi*<1
Yi= 0 if Yi*≤ 0
Two limit tobit model allows for censoring in both tails of the distribution
(Green, 2003). The log- likelihood that is based on the doubly censored data
and built up from sets of the two limit tobit model is given by:
56 − 1 5* −
3= - ∅( )+ - ∅( )
( ( (
7 89 7 7∗
51 −
+ - ln [1 − ∅ ( )]
(
7 8
Where Ioi= 0 (lower limit) and I1i= 1 (upper limit) where ∅ and ( are normal
and standard density functions.
In efficiency analysis, it is not only the level of inefficiency that is important,

but the identification of the socio economic and institutional factors that cause
it. Even though the approaches for the identification of these factors may vary
to some extent with the methodology employed, the most commonly followed
procedure in both approaches is what is usually referred to as the two step
procedure (Jema, 2008). First, the efficiency or an inefficiency index is
estimated. Second, the inefficiency or efficiency index is taken as a dependent
variable and is then regressed against a number of other explanatory variables
that are hypothesized to affect efficiency levels.
In a tobit model, each marginal effect includes both the influence of

explanatory variables on the probability of dependent variable to fall in the
uncensored part of the distribution and on the expected value of the dependent
variable conditional on it being larger than the lower bound. By following
34
McDonald and Moffitt (1980), Greene (2003) and Gould et al (1989) cited in
Endrias et al (2013) from the likelihood function decomposition of marginal
effects was proposed as follows two limit tobit model:
The unconditional expected value of the dependent variable:
>,(*) >,(* ∗) >[∅(/ ) − ∅(/ )] >[1 − >(/ )]

= [∅(/ ) − ∅(/ ). + +
> . > . > . > .
The expected value of the dependent variable conditional upon being the
limits
>,(* ∗) @/ ∅(/ ) − / ∅(/ )A @∅(/ ) − ∅(/ )A2

= ?. [1 + −B C
> . @∅(/ ) − ∅(/ )A @∅(/ ) − ∅(/ )A2
The probability of being between the limits
>[∅(/ ) − ∅(/ )] ?
= [∅/( ) − ∅/( )]
> . (
Where ∅( )= the cumulative normal distribution,

∅( )= the normal density function
D′ D′
/ = and/ = are standardized variables that come from the
E
likelihood function given the limits of Y* and
( = FG$ %$H% % $G 6 6I Gℎ ?6% .
To attain the major objective of this study, the data collected from the study
area were analyzed and interpreted. In the process of data analysis and
interpretation, major attention will be given to quantitative analysis although it
is going to be supported by qualitative technique.
35
Table 2: Variables and their expected signs

Variables Unit of measurement Expected sign
Seed Kg +
Labor Person equivalent days +
livestock TLU +
Dap Kg +
Urea Kg +
Farm size Tsimadi +
Access to irrigation 1= yes o= no +
Sex Male=1 female= 0 +
Age Person equivalent +
Education level of HHH Years of education +
Training Yes=1 no=0 +
Membership of
Yes=1 no=0 +
cooperatives
Credit Yes=1 no=0 +
Family size (adult
Persons +/-
equitant)
4. Results, Discussion and Analysis
Distribution of respondents by Woreda

We employed a stratified random sampling technique and the following
sample size was considered in the five woredas.
Table 4.1: List of woredas

Woreda Frequency Percent
Ganta Afeshum 75 25.00
Gulo Mikada 74 24.67
Kilte Awlaelo 76 25.33
Saesie Tsaeda Emba 75 25.00
Total 300 100.00
Source: Survey data (2016)
36
Table 4.2: List of Tabias

Tabia Frequency Percent
Dibla Siet 75 25.00
Aditesfa 74 24.67
A/tesfa 76 25.33
Sindeda 75 25.00
Total 300 100.00
As Table 4.1 above displayed that about 25% of the total respondents were
from Ganta Afeshum, 24.67% from Gulo Mekeda, 25.33% from Kilte Awlalo
and the remaining 25% were from Saesie Tsaeda Emba. And the tabia
representation of the households was also depicted in table 4.2.
Household size of the respondents

Respondents’ Household size in the four sample woredas may affect the
adoption of row planting. Because planting in rows requires high labor cost,
households with large family size may have a greater chance of adopting this
technology while households with small family size may find it challenging to
apply row planting technology.
As displayed in table 4.3 belowabout43/299 (14.38%) of the total respondents

had a small household size ranging between one to three family members,
among these 27.95%, 18.6%, 16.3% and 37.2% of respondents were in
Gulomekeda, K/Awlalo , Gant Afeshum and Saese Tsaeda Emba woredas
respectively. Due to this the proportion of respondents that had small family
size is relatively small that may not negatively affect the application of row
planting technology.About62.21%ofthetotalrespondentshadmediumhousehold
size composed of four up to seven members of family, of these27.4%, 23.1%
25.3% and 24.2% were from. Gulomekeda, K/Awlalo, Ganta Feshum and
Saese Tsaeda Emba woredas respectively. Among the total respondents 23.41
% had large family size ranging between eight and ten members.
37
Table 4.3: Family size of respondents by woreda

Household size
Total
Woredas 1-3 4-7 8-10
No % No % No % No %
Gulomekeda 12 27.9% 51 27.4 12 17.1 75 25.1
K/awlalo 8 18.6% 43 23.1 24 34.3 75 25.1
Gantafeshum 7 16.3% 47 25.3 20 28.6 74 24.7
Saesetsaeda 16 37.2% 45 24.2 14 20 75 25.1
Total 43 100% 186 100% 70 100% 299 100%
Table 4.4: Description of demographic variables

Standard
Variable Observation Mean Minimum maximum
deviation
HHH Age 300 48.49333 11.50876 25 80
HHH Gender 300 40= Female 260= Male
HHH experience in
300 23.533 13.62 1 60
farming
1= irrigated,
Plot type 300 0.572 0.321
0=otherwise
1=loan take,
Credit take 300 0.653 0.478
0=otherwise
1=participate,
Extension service 300 0.87 0.273
0=otherwise
TLU
The average age of the household head is 48.493. from the total 300
respondents 260 were male headed households while 40 were female headed
households. The experience in farming of the households ranges from 1 year
to 60 years with an average of 23.5 years. As the table above depicts the
average land holding of the farmers is 0.573 hectare.
38
Figure 4.1: Technology types employed

mployed by hhhs
52.00%
50.00%
%
48.00%
46.00%
Traditional Row plantation
plantation
Percent 48.33% 51.67%
From the total respondents 51.67 percent were row plantation users while the
remaining 48.33 percent were traditional sowing cultivators.
The average inputs allocated by the farmers are 2.19 Tsimad of land (Check
with reality), 15.055 man-days
days of family labor, 1.81 quintal of fertilizer, 11.9
oxen, and 13.37 compost. Using these inputs they got an average output of
266.5 KG with standard deviation of 142.45 kg of barley.
Table 4.5: Summary statistics

Plantation Land size Fertilizer Labor Compost
Oxen days
technology tsimad kg days quintal
Traditional 2.206897 1.501241 12.689655 8.758621 11.86552
technology (1.189634) (6.327388 ) (1.01554 ) (1.141896 ) (23.30022 )
Row 2.187097 2.105645 18.29032 14.90323 14.77742
technology (1.194071) (8.138873) (1.190109) (1.374006 ) (16.2549)
2.196667 1.813517 15.055 11.93333 13.37
Total
(1.189976) (7.313725) (1.121905) (1.301898) (19.99192)
39
The livestock ownership in the study area was on average of 4.996 TLU for
the farmers who use row plantation and 4.145 TLU for the farmers who use
traditional plantation while the average TLU of the sample respondents was
4.757.
Descriptive Analysis (Empirical Results)

The statistical summary in table 4.6 depicts that a typical household head who
cultivates his land using row plantation have, on average, 7.9% of inefficiency
while of the sampled households who use traditional cultivation system have
5.5% have technical inefficiency; the two sample t-test result shows that the
difference is statistically significant at 1% level. Thus, from this we can
deduce that row plantation is positively contributing to agricultural production
and productivity improvement.
Table 4.6: Two-sample t test with equal variances

Variable Mean (Std. Err.)
Traditional Plantation 0.55 (0.003259)
Row Plantation 0.79 (0.0025703)
Combined 0.682 (0.0021678)
Difference -0.24 (.0041227)
t = -5.7467*** Obs = 300
degrees of freedom = 298
The major benefits of planting crops using row plantation are listed by the
households. As stated by the farmers, row plantation have five major benefits
namely, easier weeding [1], easier harvesting [2], higher crop yield [3], use
less seed [4] and easier pest control [5]. On top of that, the benefits of row
plantation have multiple benefits. Thus, in the following table the multiple
responses are listed with their frequencies.
40
Figure 4.2: Benefits of row plantation

benefts of row plantation
All
26%
1, 2, 4, 5
33%
1, 2, 3, 4
14%
1, 3, 4, 5
1, 2, 3, 5
24%
3%
The average family size of the households was 5.56 with standard deviation of
2.07 and minimum of 1 and maximum of 11 members. The average adult
equivalent was 4.8 with standard variation of 1.78 members accompanied by
minimum of 0.74 and maximum of 10.34 adult equivalent members. The
findings of the study depicted that as the number of family members in the
household increases, the technical efficiency also increases too. 6.3 percent of
the households having family size of 11-2 had technical efficiency of 51.2
percent.
About 34.33 percent of the household with family size of 33-5 had technical
efficiency of 68.1 percent. As the number of household members increased to
6-8,
8, the level of technical efficiency also rose to 75.4 percent and decreased to
74.4 percent in 9-10
10 family size households and further decreased 69.9 percent
when the family size is 11 and above. Thus, we might suggest that level oof
technical efficiency and family size have directly related to each other.
41
Table 4.7: Family size of households and level of efficiency

Family size Freq. Percent Cum. Technical efficiency
[1-2] 19 6.33 6.53 0.512 (0.016)
[3-5] 103 34.33 40.67 0.681 (0.011)
[6-8] 124 41.33 82 0.754 (0.013)
[9-10] 24 8.00 90 0.744 (0.029)
5>=11 30 10.00 100 0.699(0.098)
Total 300 100.00 0.682 (0.038)
Econometrics Analysis
The study from its stochastic frontier model found that the magnitude of
technical efficiency varied from one farmer to another and ranged from 42.4%
to 75.4%, with a mean of 68.2%. The differences in the technical inefficiency
among the farmers is probably caused by farm management practices, the
socio economic characteristics of the households and other factors related to
natural factors.
The results from the Tobit regression model of the technical efficiency indexes
showed that scores of the technical efficiency varied from 42.4% to 75.4%,
with a mean of 68.2%.
The result from the Tobit model revealed that age, family size, TLU, row
plantation technology, membership of cooperative and access to irrigation are
among the major determinants factors of technical efficiency of smallholder
farmers producing barley. In this study, household age was found to be
negatively related to technical efficiency. This might be because of as age
increases households’ participation in labor related activities is decreased. On
the other hand, family size measured in adult equivalent is found positive and
significant. This might be because of efficient utilization of the available labor
force in the production efficiency of barley. Total livestock ownership was
measured using the standard tropical livestock unit (TLU). In this study, TLU
was found positive and significant at ten percent. One explanation for positive
association between cash technical efficiency and TLU might be livestock are
useful in cultivating land and useful in liquidity effect.
42
Table 4.8: Determinants of technical efficiency of barley production

(Tobit Regression)
Variable Robust
Coefficient P-Value
standard error
HHH age -.0639421 .0205987 0.002**
HHH gender (male) .0357281 .1106029 0.747
HHH years of education (primary) -.0019312 .0063909 0.763
HHH years of education (secondary) -.0014873 .0057688 0.797
HHH years of education (tertiary) .0005978 .0084691 0.944
Adult equivalent .1207017 .0332742 0.000***
Extension service (1=yes) .0861191 .539241 0.190
Credit take (1=yes) -.118265 .06031515 0.643
Training participation (1=yes) .0094242 .0467622 0.840
TLU .005914 .032773 0.074*
Plantation technology (1=row) .0204387 .0038395 0.000***
Membership of cooperative (1=yes) .138255 .24471135 0.023**
Access to irrigation (1=yes) .0333089 .0041482 0.000***
Constant -1.552087 .4970916 0.002
/sigma .0319756 .0013127
Number of obs = 300
LR chi2(12) = 98.51
Prob> chi2 = 0.0000
Log likelihood = 600.05196 Pseudo R2 = -0.0894
In addition, farmers who have cooperatives are found to be positively affecting

technical efficiency. This might be due to the demonstration effect that needs
to improve efficiency in production, disseminating agricultural information to
the farmers and helped them access to agricultural extension service easily.
The result from the Tobit regression model revealed that the production
frontier of the farmers who use row plantation as sowing technology is higher
than that of the farmers who use traditional sowing technology. This might be
due to row plantation technology has benefited like easier weeding, easier
harvesting, higher crop yield, use less seed and easier pest control. Last but not
the least, the study found that access to irrigation was found to be positive and
significant at one percent. This might be because of irrigation might decrease
the potential to crop failure and increases the opportunity to multiple cropping.
43
5. Conclusion and Recommendation
Conclusion
This paper has attempted to increase the understanding of the technical
efficiency and determinant factors of row plantation technology in comparison
with traditional sowing technology. More specifically, the aim of this study
was to capture the production efficiency (technical efficiency) of row
plantation production system with the conventional one using stochastic
frontier model and to analyze the socioeconomic variables that may explain
the differences in the estimated levels of technical inefficiency.
The results from the two tail test indicate that row plantation technology have
a better contribution to farmers production efficiency. Moreover, the result
from the Tobit model confirms the significance of row plantation in technical
efficiency.
The study found that the magnitude of technical efficiency varied from one
farmer to another and ranged from 42.4% to 75.4%, with a mean of 68.2%.
Consequently, due to technical inefficiency farmers have lost close to 32% of
the potential output. Moreover, the result from the stochastic frontier function
revealed that the production frontier of the farmers who use row plantation as
sowing technology is higher than that of the farmers who use traditional
sowing technology. In line with this, the data collected showed that, the inputs
used by the row plantation technology users is much higher than that of
traditional technology users.
In general, the result from the Tobit model revealed that age, family size,
TLU, row plantation technology, membership of cooperative and access to
irrigation are among the major determinants factors of technical efficiency of
smallholder farmers producing barley. Thus, the researchers recommend,
among others, the farmers who use traditional sowing technology has to
upgrade to the row plantation technology to gain the production efficiency. In
line with the integration access to irrigation, establishing cooperatives and
providing trainings to younger farmers to increase production efficiency.
44
Recommendation
Policy makers should pay due consideration to these factors that affect the
production efficiency.
Row plantation was found positive and significant in affecting technical

production. Thus, farmers should have to get trainings on how to plant seeds,
on the use of better techniques and application of fertilizer and other capital
equipment. In addition, the regional government should focus in integrating
those who use traditional cultivation method to this new technology to achieve
more viable production and productivity.
Encouraging the cooperativeness of youngsters with elders will improve the

technical efficiency of old farmers. Thus, the development groups and/or one- five
networks should have to consider different age groups to increase efficiency.
Having many livestock have found to be positively affecting technical

efficiency. Therefore, policies that encourage asset accumulation processes
through promoting investments in animal traction will create virtuous circle
between technical efficiency and assets creation.
Membership to cooperatives should have to be strengthening to gain extension

services and access market information.
The study found the impact of irrigation on production efficiency is direct and
immediate, therefore, there is still potential of integrating farm households’ for
those who don’t use row plantation technology in cropping to gain technical
efficiency.
Lastly, the study leaves for other researchers to study starting from the finding
that age of the household head, when gets older, affects production efficiency
negatively. That is, is this a lifecycle effect (meaning that the current
generation of young farmers may also leave from being efficiency when they
get older), or a generational shift? Investigating such questions could assist
policy makers in designing strategies to improve currently precarious farming
livelihoods, while facilitating a smooth exit from farming for those who wish
to take it.
45
References
Ameniya, T. (1981). Quantitative response models: a survey. Journal of economic

literature vol. xix, pp 1483-1536.
Atsbaha G. and Tessema B. (2010). A review of Ethiopian agriculture: roles, policy
and small scale farming system. Global growing caseteam. Country analyses:
Ethiopia and D. R. Congo
Boris, E., B. U. Ant and O. E. Pinheiro (1997). Technical, economic and allocative
efficiency in peasant farming: evidence from the dominical republic. The
developing economies, xxxv-1: 48-67.
Endrias Geta, Ayalneh Bogale, Belay Kassa and Eyasu Elias (2013). Productivity and
efficiency analysis of small holder maize producers in southern Ethiopia.
Journal of Hum Ecol, 41 (1): 67-75.
Essa, C. Mussa, Gideon A. Obare, Ayalneh Bogale, and Franklin P. Simtowe (2011).
Resource use efficiency of smallholder crop production in the central
highlands of Ethiopia. JEL classification: C21, C61, Q12.
Greene, W. H. (2003). Econometric analysis. 5th edition. Pearson education inc.,
upper saddle river, New Jersey 827p.
Israel, G. D. (1982). Determining sample size. PEOD6. IFAS extension, university of
Florida.
Jema Haji. (2008). Economic efficiency and marketing performance of vegetable
production in eastern and central parts of Ethiopia. PhD thesis. Swedish
University of Agricultural Sciences, Uppsala, Sweden, 64p.
Kopp R. J. and V. K. Smith (1980). Frontier production function estimates for steam
electric generation. A comparative analysis southern economic Journal.
McDonald, J. F., Moffitt, R. A. (1980). The use of tobit analysis. Review of
economics and statistics. Volume 62 issue (2), 318-321.
MoFED. (2010). Growth and Transformation Plan (2010/11 - 2014/15), Addis Ababa,
Ethiopia.
MoA. (214/15). Annual report. Transforming agriculture in Ethiopia. ATA, Addis
Ababa, Ethiopia.
Rebecca, B. T. (2011). Technical efficiency in Maize production by Small-scale
farmers in Ga-Mothiba, Limpopo province, South Africa. MSc thesis.
University of Limpopo private Bag X1106, Sovenga, South Africa.
Waluse, K. S. (2012). Determinants of common bean productivity and efficiency. A
case of small holder farmers in eastern Uganda. MSc Thesis. Egerton
University, Uganda.
46
Implication of Animal Feed and Water Scarcity on Labor
Allocation, Food Production and Per Capita Food
Consumption in Tigray Region, Ethiopia*
Muuz Hadush1
Inadequate quantity and quality of animal water and feed resources are major
factors limiting the productivity of livestock farming in Ethiopia. It is common
that households spend a considerable share of their daily time to search for
theses scarce resources by displacing available labor time away from more
productive farming activities and leisure consumption. This paper examines the
impact of time spent looking for animal water and grazing feeds on households’
agricultural food production and per capita food consumption expenditure using
NMBU-MU Tigray Rural Household Survey of 518 sample farmers. To address
our objectives, we employ IV 2SLS for estimating per capita food consumption
expenditure and double log for estimating food production drawing on non-
separable farm household model. Our results do support the hypothesis of a
negative relationship between total household labour input to crop farming and
resource scarcity. Likewise, the findings confirm that reducing time spent looking
for water leads to an increase in food production, per capita food consumption,
and food security. In addition for the median household, the total impact revealed
that decreasing searching time for water, grazing and collecting time for straw
leads to an increment in food security. The results from the quantile regression
further proved that the effect of these scarce resources is not uniform across the
food production and consumption distribution. In line with our suspicion, the
income variable was found to be endogenous and instrumental variables for it
were statistically significant and bear the expected signs.
Keywords: Animal Feed, Water Scarcity, Food Production, Food Consumption;

Ethiopia
JEL Classification: Q01, Q16, Q57, Q13
1
Norwegian University of Life Sciences, Norway
Mailing Address: NMBU, School of Economics and Business Box 5003, 1432 Ås,
Norway,
Email: [email protected]
47
Muuz Hadush: Implication of Animal Feed and Water Scarcity on Labor Allocation, Food Production and…
1. Introduction
Land degradation significantly contribute directly to poverty, by reducing the

availability of important environmental goods and services to poor rural
households, leading to increasing the demands on labor needed to seek for
such goods in East Africa (Lal and Stewart, 2010; Kirui et al., 2014). Rural
households in developing countries heavily rely on environmental products
such as fuel wood, fodder, and water to meet their daily animal water and feed
requirements. One possible negative consequence is the reallocation of labor
time from farm, off farm and leisure activities to searching these scarce
resources. The scarcity of these resources may impact agriculture and food
security by influencing the allocation of factors of production, namely labor
since scarce resources require more time to spend on their collection.
Reduction in agricultural output stemming from less labor input is very likely
to have detrimental welfare consequence (Cooke, 1998; Cooke et al., 2008;
Mekonnen et al., 2015).
Rural households face considerable tradeoffs in the allocation of time between

crop production and collecting these scarce resource for animal feed and
energy sources (Cooke et al., 2008). Households that rely on agricultural
outputs as a source of food and those that spend considerable time for animal
feeding, watering may have then less time left to devote to food production.
This has a negative implication for future agricultural production and food
security in general (Mekonnen et al., 2014; Mekonnen et al., 2015; Yilma et
al., 2011). The scarcity affects household food production and consumption
either by affecting livestock production directly, affecting crop and off farm
income through labor reallocation or through its direct impact on time for
leisure consumption and food preparation. In poor households, searching and
collecting scarce resources are a significant cost of production where poor
farmers lack alternatives to these resources.
In Africa, livestock production depends mainly on natural resources such as

grazing land and water (Bezabih and Berhane, 2014) but feed shortage, water
scarcity and diseases are frequently ranked as the most binding constraints for
animal rearing (Bishu, 2014). The livestock sector is a key player in increasing
water use and water depletion (Steinfeld et al., 2006). A recent survey in rural
48
Ethiopia and South Africa found that feed and water shortage, labor scarcity
and lack of capital were major constraints limiting livestock production
(Descheemaeker, 2008; Tegegne, 2012). Ownership of livestock in Ethiopia
has steadily declined mainly due to low availability of feed and water (Abegaz
et al., 2007). Likewise, results from Hassen et al. (2010) revealed that shortage
of water and feed are common in dry season as compared to wet season in
Ethiopia. Thus, increasing scarcity of grazing land, water for animal and straw
can be a significant burden to poor households, as grazing and water are a key
factor of agricultural production in the country.
The research question that we want to answer is whether households reduce

labor input in agriculture as a result of increasing time allocation to searching
grazing, water for animal and collecting straw due to feed and water scarcity
and test whether the time allocation to search and collect these scarce reduces
crop production on the production side and household’s utility on the
consumption side by taking away time from leisure. In regard to this issue, we
add to a relatively small list of studies examining this relationship. One early
analysis is the article by Cooke (1998), which revealed that households that
have higher costs of collecting environmental products devote less time to
farming activities and thus reductions in agricultural output, thereby low
welfare in Nepal. The studies from Damte et al. (2012) and Mekonnen et al.
(2015) suggest that as a result of increasing water, grazing land and feed
scarcity, many households increase the time they spend on collecting these
resources. It is further suggested that increasing competition on household
members’ time allocation between searching and collecting scarce resource
and cropping, reduces agricultural output that further diminishes households’
food supply and incomes, and hence their capacity to achieve food security
and human welfare (Damte et al., 2012; Mekonnen et al., 2015; Tangka et al.,
2005).
The results of Mekonnen et al. (2015) in Ethiopia show that the shadow price
of fuel wood has a negative and significant impact on time spent on
agriculture; however, scarcity of water for humans has no effect on time spent
on agriculture. The only directly and slightly related to our study are of
Mekonnen et al. (2017), whose result indicated that farming productivity
decreases as time spent collecting dung increases in rural Ethiopia and
49
Bandyopadhyay et al.(2011), whose result indicates that amount of biomass

negatively affected rural per capita consumption expenditure in Malawi. To
the best of our knowledge, empirical studies examining the effect of grazing,
water and straw on food production and consumption are, unfortunately,
missing (Cooke et al., 2008; Khan, 2008; Tangka et al., 2005).
For this purpose, we draw on the agricultural farm household model (Singh et
al., 1986) as a framework for the analysis by incorporating the time spent for
searching these resources in to the model. Following Yotopoulos et al. (1976),
an econometric estimation was presented using the NMBU-MU Tigray Rural
Household Surveys dataset collected in 2015. In aggregate, the findings
confirm that reducing time spent looking for water by 1% leads to an increase
in food production by 0.155%, PCFE by 0.133% and food security by 0.142%
while a 1% decrease in time wastage for searching grazing land increase food
production, PCFE and food security by 0.279%, 0.086% and 0.102%
respectively. Besides, an increment of 0.328% in food production and
0.0731% of PCFE is achieved by 1% reduction in straw collecting time,
leading to an aggregate effect of 0.092% increment in food security.
The noble contribution of this paper is that it considers time allocation on

animal feeding and watering, and its effect on food production and food
consumption. This is important because livestock production in Ethiopia is an
important economic activity that promotes and sustains people’s livelihoods. It
is a major source of capital investment and employment: ensure food security
by providing milk and meat; improve soil fertility through manure (Herrero et
al., 2013). Few studies by Cooke (1998) and Kumar and Hotchkiss (1988) in
Nepal, and Mekonnen et al. (2015) in Ethiopia focused on the effects of scarce
environmental goods such as fuelwood, leaf fodder and grass on labor
allocation farming and farm activity, there is scarce evidence on how grazing,
water and straw scarcity affect household food production and food
consumption expenditure. This paper, unlike the previous studies, use unique
information on the entire set of food production and consumption, along with
the distance to grazing, water and crop residue of each household. The use of
IV estimation method also gives an extra information that treating income as
exogenous and hence estimating the consumption model using OLS would
give misleading result for both policy and inference. Finally, estimating the
50
effect of scarce resources on total food security provides extra information in

assessing farm management across ecological zone.
2. Review of Background and Empirical Studies
In Ethiopia, the agricultural sector is a cornerstone of the economic and social

life of the people. Livestock sector contributes about 12–16% of the total
GDP, and 40% of total agricultural GDP excluding the values of draught
power, transport and manure, and contributes to the livelihoods of about 60–
70% of the population (Asresie et al., 2015; Halderman, 2005). Ethiopia is a
home of 35 million tropical livestock unit (TLU), and on average, one TLU
requires about 25 liters of water per day and the total daily water requirement
for livestock is estimated at 875 million liters amounting to about 320 billion
liters per year. Despite its large population size, the contribution of livestock
production to agriculture is deteriorating (Ilyin, 2011). The major feed
resources are crop residues and natural pasture but their availability is
gradually declining as a result of crop expansion, settlement and land
degradation (Gebremedhin et al., 2009). Both human and livestock suffer from
the shortage of water and feed. Most of the year, animals have to walk long
distances in search of water and are usually watered once in two to three
days(Abegaz, 2005).
In many parts of the highlands, feed and water deficits start in December–
January, when the natural pastures are at their lowest quantity and the supply
of stored crop residues is beginning to diminish. There is usually a gap of four
to five months of the dry season before the start of the short rains. The gap
which lasts for about 150 days between October and March is, therefore, the
critical period in a feeding and watering system that is largely based on natural
grazing pasture (Sileshi et al., 2003). According to CSA (2010c), the total
agricultural land is reported to be about 16 million ha occupied by 12.9 million
households accounting for an average of 1.23 ha per household, out of the
total agricultural land, 75 % is used for temporary crops while grazing land
accounts for 9%. Total grazing land in the study region is estimated to be
47,431 km2 while tropical livestock unit (TLU) per km2 of grazing land was
increased from 44,000 TLU in 2001/02 to 55,000 TLU in 2007/08 (Tilahun
and Schmidt, 2012).
51
Based on Tesfaye (2010), the estimated crop residues from cultivated land in
the region is found to be about 1,229,651 tons dry matter/year. The region has
an estimated 878,322 ha of arable land available for crop production and
contributes about 45% of the animal feed demand. Belay et al. (2013) revealed
that the most important problems of livestock production perceived were feed
shortage (100%) and water shortage (27%) in Ethiopia. Livestock suffers from
a seasonal shortage of feed (grazing land) and water (Descheemaeker, 2008).
In the high altitude zone, livestock cover less than 1 km distance to reach
water compared to the low altitude zones (Hassen et al., 2010). As a result,
there is a shortage of labor for livestock management (Tegegne, 2012).
Nahusenay et al. (2015) found that adult males are much more responsible for
feeding animals (57%) and adult female accounts for 25% in feeding animals.
Cooke (1998) considered the effect of time spent on the collection of

fuelwood, leaf fodder and cut grass on labor time to agriculture and his result
revealed that a reallocation of time away from farm work and leisure may
occurred as environmental goods became scarce and costly in Nepal. The
work of Kumar and Hotchkiss (1988) linked time allocation behavior and
deforestation in Nepal. They found that time spent in farming declines with a
higher degree of deforestation (fuel scarcity). Mekonnen et al. (2015)
examined the effect of the scarcity of fuelwood and water on time spent in
agriculture using a panel data set collected from Ethiopia. The results of the
empirical analysis show that fuelwood scarcity, as reflected by the shadow
price of fuelwood, has a negative and significant impact on time spent on
agriculture; however, scarcity of water has no effect on time spent on
agriculture. Likewise, Cooke (2008) explained the effect of forest scarcity on
the livelihood of rural people in Nepal and found a negative effects on health,
labor burden and agriculture. Another related study by Damte et al. (2012) in
Ethiopia indicated that rural households respond positively to fuelwood
shortages by increasing their labor input for fuelwood collection even if they
fail to investigate whether the increase in labor comes from agriculture or
other activities.
According to Bandyopadhyay et al. (2011) study in Malawi, more time spends

on scarce fuelwood collection was associated with negative welfare even if the
effect on their overall welfare is small. Bhattacharya and Innes (2006)
52
highlighted that forest degradation spurs rural poverty in Sub-Saharan Africa.

In addition, Mekonnen et al. (2017) explored the effect of time spent on dung,
fuelwood and crop residue on agricultural productivity and the result indicated
that agricultural productivity decreases with increasing time spent on
collecting animal dung but increases with time spent on collecting crop
residue. None of the above studies examine the effect of grazing and water for
animal on time allocation, food production and food consumption (Cooke et
al., 2008; Khan, 2008).
3. Theoretical Model
In a mixed crop–livestock farming systems, Ethiopia owns a significantly

large livestock population (Tegegne, 2012). In the country, livestock
production mainly depends on natural resources such as grazing land, water
and own crop residue (Bezabih and Berhane, 2014). The contribution of
livestock to food and nutritional security is significant and serves as an
important source of livelihood (Swanepoel et al., 2010). However, crop -
livestock farming activity require huge inputs of labor either from own family
or labor market. In rural farm households, total time endowment is divided
into three main activities: farm activities, off-farm activities and leisure, where
rural farms take a significant share of total time endowment and a substantial
part of the production is retained at home for consumption. The scarcity of
grazing and water resources for animal may even takes the largest proportion
of family labor time in countries like Ethiopia, which is characterized by a
critical shortage of animal feed and water, having a negative implications for
agricultural production and food security (Tangka et al., 2005).
Considering the time spent on looking scarce resource, the total time
endowment is further divided into four main activities: farm activities, off-
farm activities, leisure and searching or collecting these resources activities.
Labor allocation for these scarce resources displaces household’s labor from
productive activities such as agricultural production and off-farm employment,
food preparation and leisure, resulting in low welfare (Cooke et al., 2008;
Mekonnen et al., 2015). The scarcity of grazing and water resources adversely
affects household food production and consumption either by affecting
livestock production directly, affecting crop and off farm income through
53
labor reallocation or through its direct impact on time for leisure consumption
and food preparation.
The theoretical framework for modeling the effect of resource scarcity on food
production and consumption is, in general, built within the framework of
household utility model. Modeling households’ decision of production and
consumption as a recursive method enables us to understand the households’
action as if it first maximizes profit (Straus, 1986). Following the work of
Singh et al. (1986), it makes sense first to maximize profit and then decide
consumption and leisure since income and utility are positively related. For
simplicity, the well-behaved quasi-concave household utility function have the
following form:
K = K( , L ; ɸ), (1)
where vector of home produced goods such as meals and purchased goods
consumed, and is consumption of leisure. The meal production is a function
of agricultural goods, off farm income , fuel sources such as straw or dung as
well as labor days the household spend on searching grazing land, water and
crop residue. The production of household goods is also influenced by the
vector of household characteristics,
= 0 OP0 , ,¸R, +S ; ɸT. (2)
An implicit production function which is assumed to be the quasi-convex

relating outputs and inputs, increasing in outputs and decreasing in inputs
(Strauss, 1986), and which allows for a separate production function for each
output or joint production function is therefore formally denoted by:
UOP0 ¸V, W¸+S T = 0, (3)
where is implicit production function, is vector of household productions

such as crops, and are vectors of variable inputs including labor, and fixed
inputs respectively. is labor time spent by the household on searching grazing
land, water and crop residue as a proxy of scarcity indicator. Expressing total
income of a farm household as the sum of its time endowment, value of
54
households’ production and other incomes such as transfer, minus the value of
variable inputs required for production, the budget constraint stating total
consumption equals total income can be presented as:
∑L0 Y0 0 = YL (+ + +S ) + ∑0 Y0 P0 − ∑Z Y V − YL 3 + , , (4)
where is commodities consumed, is price of output, is time endowment, is

wage, is household production, is price of variable inputs, is non-labor
variable inputs, is labor demand and is exogenous income. While the left hand
side of Equation (4) represents market value of commodity consumed with the
last term () being the value of leisure, the right hand side gives full income of
the household which consists of households time endowment, plus the value of
households total production, minus the value of variable inputs including
labor, and plus exogenous income which is generated outside the household
such as transfer from relatives or friends.
Generally, the household maximizes utility subject to production function,

budget constraint, and time constraint. Maximizing output by the households
depends only on the choice of variable inputs, and maximizing profit is the
same as maximizing full income given by the right-hand side of equation (4)
subject to the production function. Then, the household maximizes utility
subject to its full income upon achieving maximum income through profit
maximization. The Lagrangian function of the utility maximization subject to
full income and production function can be expressed as follows:
3 = KO 0 OP0 , ,¸R, +S ; ɸT, LT + [[(+ + +S ) + - Y0 P0 − - Y V − YL 3

0
+, − ∑L0 Y0 0 ] + [U(P, V, W¸+S )]. (5)
Assuming that interior solution exists, the first order conditions based on
Straus (1986) are:
\ℒ \_
= \D − [Y0 = 0, (5.1)
\D^ ^
\ℒ \a
\L
= γ \L − [YL = 0, (5.2)
55
\ℒ \a
=γ − [Y = 0, (5.3)
\bc \bc
\ℒ \_ \D^ \a
= + [YL + γ = 0, (5.4)
\de \D^ \de \de
\ℒ \D^ \a
\f^
= +[Y0 + γ \f = 0, (5.5)
\f^ ^
\ℒ
\g
=F(P, V, W¸+S )=0, (5.6)
\ℒ
\h
= YL (+ + +S ) + ∑0 Y0 P0 − ∑Z Y V − YL 3 + , − ∑L0 Y0 0 = 0.
(5.7)
Following Straus (1986), the solution to the first order conditions of the above
expressions yields standard demand function for inputs and outputs in terms of
all prices, the wage rate, time for searching and collecting scarce resource,
fixed land, and capital. Substituting optimal labor, and optimum output into
RHS of Equation (4) produces optimum income or full income under the
assumption of maximized profit. Likewise, the first order conditions of the
LHS of Equation (4) gives consumption demand function in terms of prices,
the wage rate, and income and household’s preferences represented by
household demographic characteristics. The effect of scarce resource on
\a
agricultural production i\d j is investigated through the production sector and
e
\D
its direct impact on household’s utility k\f^ l is explored through consumption
^
sector. Thus, the total effect which is sum of the two effects can be explained
using the budget constraint total income as
\m \D^ \7 \D
= + \d ^ ,
\de \7 \de
(6)
e
where* = ∑0 Y0 P0 − ∑Z Y V − YL 3 represents the net agricultural output or

profit from agricultural production and total income of the household is equal
to households time endowment, plus the value of households total agricultural
production, minus the value of variable inputs including labor, and plus
exogenous income in the RHS of Equation 4. The main question that interests
us is whether scarcity of these resources adversely affects crop production and
56
per capita food consumption expenditure. The hypothesis to be tested here is

that farmers that spend more time on searching these scarce resources are
likely to have less time for crop production and leisure consumption that is we
nL \a nD^
test weather, < 0, or < 0 and < 0 using walking distance2 to
nde \de nde
these resources sites as indicator of scarcity in the study area.
In this case, utility and production decision problems can indeed be solved
recursively, despite their simultaneity in time (Straus, 1986). Barnum and
Squire (1979) show that household characteristics can be introduced into the
model as linear functions and prove that introducing them as linear functions
will not change the analysis as long as household characteristics are treated as
fixed variables. Since solving the above system of equations becomes more
tiresome as the number of commodities consumed and outputs produced
increase, an alternative approach to estimating separated production function
for each output type is aggregate production. Aggregation gives a greater
chance to cancel out errors when some households report zero variable input
for some products but positive outputs and will not only reduce the number of
parameters to be estimated but also addresses the probable existence of
jointness (Strauss, 1986).
4. Description of Study Area and Dataset
Ethiopia is a federal country divided into 9 regions and 2 administrative cities.

Each region is subdivided into zones and zones into woredas. Woredas, in
turn, are divided into Peasant Associations (PA) or Tabias, an administrative
unit consisting of a number of smallest villages and individual households.
The study is conducted in Tigray region, the northern part of Ethiopia by
randomly selecting 632 sample households from 21 PAs. This study used a
cross-sectional data from NMBU-MU3, Tigray Rural Household Survey
(TRHS)4 dataset collected in 2015. TRHS includes a panel of five rounds
conducted in 1997/98, 2002/03, 2006/07, 2009/10 and 2014/2015 where the
2
See for a similar approach in the work of (Cooke, 1998; Cooke et al., 2008 and
Baland et al., 2010)
3
NMBU-MU refers to Norwegian University of Life Science-Mekelle University.
4
This dataset has been used by Gehbru Hosaena (2010); Holden et al. (2009, 2011).
Hagos and Holden (2011) and others.
57
author is involved only in collecting the data for the last round. The data has
been originally designed by a doctoral student from Ethiopia in Norway and
PhD students who joined the same university continued to use the same
design. The available panel dataset provides comprehensive household and
plot level data on household characteristics, agriculture and livestock
information, food consumption, rental market participation, land certificate
perception as well as community level data on GPS information including
rainfall, total cultivated, irrigated and grazing area, wages, and conservation
activities under safety net activities.
The primary data used in this paper is adapted from the last, 2014/2015,
household survey since some variables used in this estimation were only
added in the last round of the wave. Table 1 presents the basic socio-economic
characteristics of 518 farm households drawn from a total of 632 sample
farmers. For this study, the need for information regarding livestock activity
restricted us to use only 518 farmers, those who only owned cattle during the
study year (82 percent of the original data, 632). The dependent variable in the
production side is aggregate household agricultural production or monetary
value of all crops produced during the survey production season. In the
consumption side, the dependent variable is per capita food consumption
expenditure.
Crop production in Ethiopia is dominated by small-scale subsistence farm

households that on average cultivate less than a hectare of land. The main
agricultural products produced in the surveyed villages are Tef, barely, wheat,
maize, millet, sorghum, field pea, lentil, linseed etc. An average household
owns a production capital worth about 639 birr and has produced an average
agricultural output of worth 41,645 birr in the year. In addition, the average
livestock endowment of the sample households is 4 TLU which expected to
increase food security (Kassa et al., 2002), and average total income including
sales from agricultural outputs is worth 49,426 birr while the average per
capita food expenditure is 2,490 birr.
58
Table 1: Descriptive and Summary Statistics

N=518
Variables Description Mean SD Min max
Dependent Variables
Monetary value of per capita food
PCFE(ETB)e 2,490 3,722 22.22 34,962
expenditure
Output (ETB) Monetary value of crop production 41,645 87,517 152.40 892,500
b
Income(ETN) Monetary value of total income 49,521 92,642 300 892,730
Independent Variables
Market distance Distance to nearest market in minute 82.30 54.79 10 240
Water distance Distance to animal water source in minute 74.85 65.54 10 360
Feed distance Time to transport crop reside and grass 576.55 557.87 18 6,000
Family size Household family size in number 5.87 2.41 1 12
Age Household head age in years 56.83 15.20 18 99
1 if household head is orthodox and 0
Religion 0.82 0.38 0 1
Muslim
Gender 1 if household head is male 0.74 0.44 0 1
Education 1 if household head is literate 0.33 0.47 0 1
TLU Herd size in TLU 3.92 3.20 0.01 22
Time spent looking for grazing land in
Grazing distance 91.12 83.44 10 1,200
minute
Shocks (2012- Number of shocks due to theft, flood,
0.58 0.83 0 5
2014) death
Irrigation 1 if household head has access to irrigation 0.26 0.44 0 1
ashock13 1 if household face animal shock in 2013 0.04 0.20 0 1
1 if household face any shock in 2012-
Shock exposure 0.09 0.29 0 1
2014
Information 1 if hh had access to TV, radio and mobile 0.42 0.49 0 1
1 if hh get support from relatives and
Network 0.61 0.49 0 1
friends
Water harvest 1 if hh access water harvesting well, ponds 0.02 0.14 0 1
Location 1 if hh lives in highland(>2500masl) 0.06 0.24 0 1
Oxen Number of oxen the household head own 1.93 1.05 1 9
Area Total cultivated land in tsmdic 4.45 3.14 0.25 22
Family labor Total adult family labor in man day 85.52 69.33 1 778
Fertilizer Total fertilizer used in KG 68.55 49.24 0.5 425
Manure Total manure used in KG 775.60 1,585 1 20,000
Farm tool (ETB) Total monetary value farm toold 639.10 1,451 10 14,650
Notes: a: It includes crop, fruit and vegetable production
b: It includes income from Agriculture, off-farm, transfer and safety net
c: One Tsmdi is approximated to one-fourth hectare
d: Total monetary value of all farm implements such as plough parts, hoe, cart, sickle, spade
e: ETB refers to Ethiopian currency in which 1USD 23 ETB
59
Referring to figures of zonal distribution of production and per capita

consumption in the appendix, the average per capita food expenditure was
3200 Birr for Southern and around 2000 for the rest zones, showing that
average per capita food expenditure in Southern is much higher than the
overall average result; perhaps this is due to the densely populated livestock of
the zone compared to other zones. The same result in the appendix display that
average values of output and income of that household living in the southern
zone are 3.7 and 3.3 times larger than their respective values by an average
household in the other zone.
On average the households spend 75 minutes to reach a water source for

animal and 91 minutes to search for communal grazing land daily, maximum
time reaching up to 6 hours for water site and 8 hours for grazing land in the
data. Besides, the average time spent on transporting crop residue by the
households is 576.6 minutes, ranging from a minimum value 18 to maximum
value of 6000 minutes in the study area. Households that are situated far from
a water and grazing land source require longer time. The graphical display in
the appendix showed that farmers living in Raya Azebo district travel 110
minutes to reach grazing land followed by Easterners while those from central
zone spend minimal time. With regard to distance to animal water source,
North Westerners commute about 90 minutes followed by South Easterners.
Households from North West spend around 800 minutes to transport crop
residue while Easterners travel half of the distance of North West (400
minutes).
Farmers having a larger size of livestock holding (TLU) seem to be more

worried to supply enough feed to their animals and spend more time to search
for feed and water. In relation to this, Bishu (2014), whose study in Ethiopia
indicated that there was a shortage of water during the dry season for livestock
drinking in the study site (Abegaz, 2005; Tesfaye, 2010). It is therefore
hypothesized that any labor spent on searching scarce resources is inversely
related to the production and per capita consumption (Mekonnen et al., 2015).
The distance to the nearest market, on average, was 82 minutes. Thus, its
expected effect on consumption is negative, indicating that longer distance
leads to less frequency of visit and hence less likely to get market information
about selling and buying prices (Feleke et al., 2005; Shiferaw et al., 2003). As
60
of the survey, average land holding is 1 ha, which is less than the family
member size in the study area and holding large size is expected to play a
significant role in influencing households’ food production and food security
positively (Najafi, 2003).
Fertilizer and manure are used in most studies as a proxy for technology that
augments agricultural productivity and is expected to boost the overall
production, contributing towards attaining household food security. Each
household uses an average of 68.5 KG fertilizer and 775.6 KG manure during
the harvesting period, while the number of oxen by an average household is 2.
All inputs are expected to increase production and thus food consumption
(Brown, 2004; Di Falco et al., 2011). In many developing countries, oxen
serve as a source of traction, thereby significantly affecting households’ crop
production and consumption by enabling households to cultivate greater areas
of land (Govereh and Jayne, 1999). Hence, a positive relationship between ox
ownership and food expenditure and crop production is expected in this study.
On average, each household had 85.5 man day labor used for farm production.
The magnitude of this variable is smaller than the result from the previous
empirical finding of Sakketa and Gerber (2017) and Mekonnen et al. (2015),
who found the average household labor time is about 114 and 117 man day in
Ethiopia. I have also tried to look at the correlation between the time spent on
searching water, grazing land and crop residue and time spent on crop
farming. The result indicated that farm time and resource scarcity are
negatively associated in the study area. This is consistent with result of
Mekonnen et al. (2015), who investigated the impact of scarcity of fuelwood
and water for human on labor allocated to agriculture. Given adequate land,
adequate labor supply input is expected to foster production and is expected to
have a positive effect (Di Falco et al., 2011; Sarris et al., 2006).
Out of the total sample, 6.4% lives in highland parts of the region. Only 27%
of the households have access to irrigation and only 2% are involved in water
harvesting practices such as ponds and well. Nearly 39% of the households
report that they have been severely affected by eleven different level of shocks
including, drought, pests, flood, theft, illness and death, loss of job and home
damage in the last harvesting season, and 4.25 % of households report having
been affected by animal shocks one year before the harvesting season. Both
61
shocks are expected to affect production and consumption negatively (Abdulai

and Huffman, 2014; Dercon et al., 2005). Evidence showed that male headed
households have a better opportunity in terms of access to a resource such as
labor, land, modern input, education, credit and extension services compared
to female headed households (FAO, 2001). 74% of the households are male
heads with an average age of 57 years and family size of 5.87. Since resources
are very scarce, high family size may put much more pressure on consumption
than it contributes to production. The expected sign of consumption is then
negative because food requirements increase with the number of persons in a
household.
Nearly 32% of the household heads have at least a one or more years of
education. Thus, it is hypothesized that education is negatively related to
consumption value. Around 82% of the households are Orthodox followers
while 18% of the households are Muslim households in the study area. Out of
the 518 households in the sample, 61% got assistance either from their
relatives or friends and is expected to increase production and consumption
(Di Falco et al., 2011). More than 40 percent of household heads site attend
media via TV, radio and mobile phone about any development intervention.
Hence, it is expected that households with information are more likely to
produce more and be food secure. The expected effect on production and
consumption is positive (Di Falco et al., 2011).
4. Econometric Model Specification
This paper draws on the AHM which provides a holistic framework to analyze
the economic relations of production and consumption decision in the farm
household. We choose the recursive AHM since it has an advantage of
econometric estimation simplicity and fits best to the available data. Although
the separation property of the recursive model enables us to separate the
estimation of consumption and production sectors, it will result in inconsistent
estimators whenever one of the assumptions does not hold true. This problem
is even more significant for studies that deal on production side than
consumption (Delforce, 1994). But, as the focus of this study mainly inclines
to consumption side. The problem is less worrisome.
62
With regard to estimation, first, the production function was identified.

Multiple crop outputs are aggregated into a single output measure using the
medians of their reported village’s prices within each village following Jacoby
(1993) and Gutu (2016). Then, food demand equation (per capita food
expenditure) was specified using the utility maximization results of the AHM.
The parameters from production side were estimated using the Cobb-Douglas
production function since the output is a simple function of labor and capital.
However, it does not allow other variables than just the two which can
significantly affect production such as fertilizer and land. For this reason, the
General Cobb-Douglas (GCD) production function, developed by Diewert
(1973) was adopted in order to incorporate these variables into the production
function and denoted as:
qc^
* = ?∏ ∏ 0 i) p + ) p0 j , (7)
where is output, are quantities of the inputs, , and (This is the assumption of
constant return to scale). Assuming that for all , and taking natural log of
equation (7) produces a standard Cobb-Douglas equation with many inputs,
which is to be estimated in its natural log form:
* = rs + ∑ r p + , (8)
where, is the constant term in equation (8), and is the error term.
The GCD production function is often criticized for being restrictive due to its
assumptions of constant returns to scale (CRTS) and perfect competition in
both input market and output market even if it handles a large number of
inputs. Its assumptions make it difficult to measure technical efficiency levels
and growth effectively. But, the assumption about market does not
significantly affects the estimation power of Cobb-Douglas production
function as long as factors are paid according to their relative shares (Murthy,
2004). In addition, Miller (2008) argued that GCD can be estimated by
relaxing the CRTS assumption and then test whether the summation of the
coefficients is significantly different from one using the standard econometric
procedure.
63
In order to estimate consumption side, we are forced to approximate calorie

intake by per capita food expenditure due to limited data assuming that the
demand equation from the utility maximization of the recursive household
model has a functional form of log-linear. Thirumarpan (2013) and Asfaw et
al. (2012) used consumption expenditure to reflect the socio-economic welfare
of household and is a reliable indicator of food accessibility and degree of
vulnerability to food insecurity. Its capability of estimating respective
elasticities as its coefficient and modeling nonlinear effects makes it
applicable and preferable (Oum, 1989). Oum added that the log-linear demand
function resembles the demand function obtainable from a Cobb-Douglas
utility function with the drawback of invariant estimated elasticities across all
data points. Like in the production side, aggregate demand equation per
household is estimated for per capita food consumption expenditure rather
than estimating single demand equations for each product consumed or for
each individual member of the household.
t= s + u +∑ p + v, (9)
where t is households per capita food consumption expenditure; p for

= 1, ⋯ , , includes consumption side variables and household
characteristics; v is an error term which is assumed to be uncorrelated with the
production function error term . u and β are parameter coefficients of income
and the vectors of an exogenous variables, p . The effect on agricultural
production is investigated through the production sector and its direct impact
on household’s utility is explored through consumption sector.
Since farm and off-farm income is not randomly distributed among rural
households, this variable is likely to be endogenous, which could be caused by
omitted variables, measurement error, simultaneity or household unobservable
(Hidalgo et al., 2010), First, a reverse causality problem might exist, because
per capita food expenditure at the household level might also influence labor
productivity and thus farm productivity. Second, farm and off-farm income
might be influenced by household unobservable, which can lead to correlation
with the error term. In the presence of endogeneity, the use of the OLS
estimator biases the effect of income. In order to avoid an endogeneity bias,
we adopted Two-Stage Least Squares (2SLS) approach, using household
64
shock experience and a number of plots as instruments (Angrist and Evans,

1998) which are the most common instrumental variable estimator
(Wooldridge, 2009). This is similar to approaches that have been used by
Sarris et al. (2006) and Abdulai and Huffman (2014) in different contexts.
With this procedure, the structural equation is specified as
t = u+u x +∑ p + y, (10)
where t is percapita food expenditure, x is predicted values of the

endogenous income variables and is an error term, that is uncorrelated with ,
, u and are parameter coefficient of income and the vectors of an
exogenous variables, p . To obtain income ( ), first stage regression equations
is estimated by OLS based on the following specifications;
= r + /z + ∑ p + y, (11)
where is total farm and off farm income of the household, is parameter
coefficients of the vector of the instrumental variables, / which are assumed to
correlate with income but not with the error term, y in the structural
equation (10). The estimated per capita food consumption expenditure of the
household, in (10) is now assumed to be unbiased.
6. Economic Results
6.1 Estimation of Household Labour Allocation to Crop Farming
What is the consequences of increasing grazing, water and straw for

agricultural labour input? We answer this question by examining the link
between resource scarcity and labour input to crop farming in rural areas of
Ethiopia using similar estimation methods of Cooke (1998) using cross section
data in Nepal. In this paper, the variables of greatest interest are animal water
and feed scarcity measured by the time taken to collect them. A priori, animal
water and feed scarcity should reduce labour time on the crop farm because
they take away time from crop farms and leisure as people search for these
resources. The estimate of the effect of resource scarcity on time spent in crop
farming is presented in Table 6.
65
Our results do support the hypothesis of a negative relationship between total

household labour allocation to crop farming and resource scarcity at the
household level. With respect to the variables of interest, higher searching
times of water, grazing and collecting straw were shown to significantly
reduce labor time to crop farming. We found that that a 1% increase in
searching times of water, grazing and collecting straw results in a 0.0598%,
0.0929% and 0.0992% respectively decrease in time spent on crop farm. This
result finds favor among a number of researchers (Cooke, 1998; Cooke, 2008;
Bandyopadhyay et al., 2011; Mekonnen et al., 2015). We found significant
effects of other covariates as well. Land area in crops has a significant positive
effect on total household labour input to farming. Real off farm wage has a
significant positive effect on household farm labor input.
As expected, we also found that large family households spend more time on
crop farming. The households living in lowland areas spend more farm labour
input to farming than their counter part. Wealthier households who have more
livestock spend more time for farming. Higher on-farm income is associated
with household’s more time input to crop farming. Hiring labor from the local
market decrease labor family input to farming and higher altitude motivate
farmers to allocate more labor input to crop farming. These findings
correspond to the results of previous studies by Cooke (1998), Okwi and
Muhumuza (2010), Bandyopadhyay et al. 2011) and Mekonnen et al.(2015).
6.2. Estimation of Monetary Value of Aggregate Production
In order to estimate production sector of the farm households, we used

ordinary least square (OLS) on the log-transformed form of the GCD
production function specified in section 5. The dependent variable is aggregate
household agricultural production, which is the monetary sum of all crops
produced during the survey harvesting season. The estimates of the production
function and the effect of water, grazing the land and feed scarcity on
agricultural production are presented in Table 2 under 3 columns respectively.
In general, the estimation shows that all explanatory variables exhibit
significant and theoretically expected signs. Variables of interest in this paper
are time spent on looking water and feed resources included so as to capture
the effect of feed and water scarcity on agricultural production. The first
66
column presents the estimation of the food production function with water
scarcity taken into account as do the second and the third columns, putting
grazing land and feed transport into consideration. The result is in favor of our
hypothesis.
As expected Column (1) of Table 2 indicated that time spent on animal water
source is found to be negative significant, suggesting that a one percent
increase in time spent looking for water decreases agricultural production by
0.155 percent, and time spent on searching grazing land have stronger effect
than this variable as shown in Column (2) i.e., a one percent increase in time
spent searching for grazing decreases agricultural output by 0.279 percent.
Another feed scarcity related variable is time spent for transporting crop-
residue from threshing center to homestead. Increasing distance significantly
resulted in a negative sign as expected, implying that farmers that spend one
minute more for collecting crop residue produce about 0.328 percent less
output (Column 3). The output effect obtained here support the claim that time
spent for searching scarce resources displace labor time from production
activity and hence reduce crop production in line with the findings of (Damte
et al., 2012; Mekonnen et al., 2015; Tangka and Jabbar, 2005), who generally
concluded that collection of scarce resources such as water, firewood, and
grass negatively affect production activity by reducing labor time allocated to
crop farming.
The estimated coefficient for land (0.278, 0.304 and 0.201) shows that
increasing land size by one percent increases agricultural production, on
average, by almost 0.3 percent, implying that land is a vital input of
agriculture. The result is similar to what it was found by Nisrane et al. (2011),
whose study revealed that cultivated land had a positive effect on agricultural
production in Ethiopia. Moreover, Foster and Rosenzweig (2010) showed that
land size had a positive impact on net revenue in India while the empirical
results from Sarris et al. (2006) in Tanzania also appear to support the above
result.
67
Table 2: OLS Estimation of log Monetary Value of Aggregate Agricultural Production

(OLS) (OLS) (OLS)
Variables
Ln(output) Ln(output) Ln(output)
Ln(area) 0.278*** 0.304*** 0.201***
(0.0595) (0.0579) (0.0523)
Ln(manure) 0.0854** 0.0857** 0.0501
(0.0369) (0.0363) (0.0324)
Ln(oxen) 0.228** 0.248*** 0.186**
(0.0973) (0.0951) (0.0851)
Ln(fertilizer) 0.145** 0.174*** 0.150***
(0.0665) (0.0652) (0.0581)
Ln(family labor) 0.353*** 0.306*** 0.197***
(0.0650) (0.0641) (0.0581)
Hired labor(1/0) 0.472*** 0.481*** 0.307***
(0.0928) (0.0907) (0.0822)
Location(1/0) -0.493*** -0.453*** -0.544***
(0.174) (0.169) (0.150)
Ln(farm tool) 0.0566** 0.0561** 0.0162
(0.0254) (0.0249) (0.0224)
Ln(mktdistance) 0.0745 0.0808 -0.000798
(0.0551) (0.0538) (0.0485)
Info(1/0) 0.0959 0.0549 0.0264
(0.0851) (0.0836) (0.0746)
Well(1/0) -0.260 -0.218 -0.0514
(0.299) (0.292) (0.261)
Ln(shocks) -2.160*** -2.091*** -1.932***
(0.321) (0.311) (0.278)
Irrigation(1/0) 0.0627 0.0931 -0.0440
(0.0980) (0.0955) (0.0860)
Education(1/0) 0.284*** 0.246*** 0.243***
(0.0904) (0.0887) (0.0790)
Ln(water distance) -0.155***
(0.0475)
Ln(grazing distance) -0.279***
(0.0471)
Ln(feed distance) -0.328***
(0.0254)
Constant 6.873*** 7.383*** 9.496***
(0.500) (0.492) (0.476)
Observations 509 508 509
R-squared 0.394 0.423 0.538
Note: P-values are for slopes; ***P<0.01; **P<0.05 and *P<0.10 = Significant at 1%, 5% and
10% probability level respectively.
68
As expected fertilizer and manure use are found to be significant and positive
variables incongruent to the studies conducted by (Demeke et al., 2011;
Kidane et al., 2005; Nisrane et al., 2011; Di Falco et al., 2011) in Ethiopia. In
Ethiopia ox is the main capital input used for ploughing and threshing and can
be considered as an equivalent substitute of the uses of the tractor. In this
paper number of oxen is found to be significant, leading to a 0.23 percent
increase in the agricultural output. A similar result is found in the study of
Mekonnen et al. (2015) who found a positive effect of ox input food crop
productivity in Ethiopia.
In line with the predictions of economic theory, inputs such as farm capital
and labors are significantly associated with an increase in the quantity of
production value. A one percent increase in man day labor causes to increase
production by about 0.353 percent, a finding that is consistent with this notion
is of Di Falco et al. (2011) and Abdulai and Huffman (2014). But the
coefficient on seed input contrasts with the findings by Di Falco et al. (2011)
in Ethiopia and Bulte et al. (2014) in Tanzania, who both found a positive
significant on harvest. Farmers hiring one percent extra labor seems to
increase their production value by 0.481 percent, confronting with the result of
Sarris et al. (2006) whose result revealed a negative relation. Another capital
input included in the analysis is production capital which is the monetary
value of farm tools. It is found to be statistically significant. A one percent
increase in production capital has the ability to increase agricultural output by
0.056 percent. This finding is consistent with the earlier study by Sarris et al.
(2006).
Not surprisingly, we found that shock experience appears to be negatively

related to the household’s production. An increase in shock has a quite large
detrimental effect of food production (-2.16%) which is consistent with a
previous study (Abdulai and Huffman, 2014) who confirmed a negative effect
of drought or illness shock on production. The variable representing education
of the farmer is positive and significantly different from zero, suggesting that
more educated farmers are more likely to produce more in favor of Abdulai
and Huffman’s (2014) result.
69
6.3. Per Capita Food Expenditure Estimation
The objective of utility maximization by the household is analyzed using the

demand functions derived from maximized utility subject to budget constraint
and technology constraint of farm production and its estimated result is
presented in Table 3 using naïve OLS and IV method, where total income is
instrumented by shock occurrence and a number of plots of the household
head. Shock caused by crop theft and death of a household member is
expected to affect income and output negatively, thereby reducing food
expenditure (Abdulai and Huffman, 2014; Dercon et al., 2005). The exposure
of previous year’s shock (2012-2014) have a direct effect on the household
income and indirect effect on the consumption side through its effect on
income. The source of rural farm income is mainly from crop or animal
farming which is operated by family labor. Thus, farm income is expected to
decrease with increasing any shock on crop or animal farming caused by a
theft or illness of the household. Then, its effect on consumption reaches
through its effect on farm income.
70
Table 3: IV Estimation of log Per Capita Food Expenditure

(OLS) (IV) (OLS) (IV) (OLS) (IV)
Variables
lnPCFE lnPCFE lnPCFE lnPCFE lnPCFE lnPCFE
Ln(output) 0.0940*** 0.0562*** 0.0909*** 0.0563*** 0.0986*** 0.0623***
(0.0121) (0.0171) (0.0125) (0.0169) (0.0122) (0.0170)
Ln(livestock) 0.0336*** 0.0277** 0.0334** 0.0289** 0.0352*** 0.0295**
(0.0129) (0.0137) (0.0131) (0.0137) (0.0130) (0.0137)
Ln(Family size) -0.385*** -0.357*** -0.397*** -0.369*** -0.388*** -0.362***
(0.0529) (0.0562) (0.0535) (0.0566) (0.0534) (0.0564)
Gender(1/0) -0.119** -0.140** -0.0993* -0.117* -0.115* -0.136**
(0.0588) (0.0621) (0.0590) (0.0620) (0.0593) (0.0624)
Info(1/0) 0.0591 0.0370 0.0454 0.0247 0.0487 0.0260
(0.0539) (0.0570) (0.0545) (0.0573) (0.0544) (0.0573)
Location(1/0) -0.0411 -0.0543 -0.114 -0.133 -0.149 -0.173
(0.140) (0.147) (0.140) (0.146) (0.141) (0.147)
Ln(mktdistance) 0.00283 0.0196 0.00252 0.0200 0.00144 0.0174
(0.0337) (0.0358) (0.0340) (0.0360) (0.0340) (0.0359)
ashock13(1/0) -0.489** -0.379* -0.550*** -0.442** -0.540*** -0.440**
(0.191) (0.203) (0.192) (0.204) (0.193) (0.204)
Ln(shocks) 0.212 0.374* 0.307 0.465** 0.267 0.428**
(0.198) (0.214) (0.199) (0.214) (0.200) (0.215)
Religion(1/0) 0.121* 0.152** 0.101 0.130* 0.115 0.145*
(0.0700) (0.0741) (0.0705) (0.0743) (0.0706) (0.0744)
Network(1/0) -0.0833 -0.191*** -0.0761 -0.178*** -0.0729 -0.177***
(0.0554) (0.0666) (0.0559) (0.0665) (0.0558) (0.0667)
Age(years( -0.000477 -0.000808 -0.000535 -0.000848 -0.000554 -0.000886
(0.00174) (0.00183) (0.00175) (0.00184) (0.00175) (0.00184)
Ln(income) 0.0440*** 0.0593*** 0.0433*** 0.0581*** 0.0439*** 0.0587***
(0.00187) (0.00498) (0.00189) (0.00499) (0.00189) (0.00499)
Ln(wat distance) -0.122*** -0.133***
(0.0309) (0.0327)
Ln(graz distance) -0.100*** -0.0860**
(0.0336) (0.0354)
Ln(feed distance) -0.0642***
-0.0731***
(0.0240) (0.0253)
Constant 6.018*** 5.959*** 6.046*** 5.862*** 5.917*** 5.872***
(0.291) (0.306) (0.318) (0.337) (0.305) (0.319)
R-squared 0.710 0.670 0.705 0.668 0.705 0.667
First stage Shock -20.124*** -20.076*** -20.123***
( 2.184) (2.185) (2.183)
Landsize -0.476 -0.4976 -0.487
(0.308) (0.306) ( 0.306)
Observation 496 496 496 496 496 496
Note: P-values are for slopes; ***P<0.01; **P<0.05 and *P<0.10 = Significant at 1%, 5% and
10% probability level respectively.
71
Similarly, the land holding size of the household head can be considered as a
substitute for sources of wealth in rural areas and is expected to influence total
income positively (Sarris et al. (2006)). In the same fashion, cultivation of
more plots in rural areas of the country is a good indicator of wealth and
directly affects the farm income he/she harvests. Increasing number of plots is
expected to increase farm income directly but consumption indirectly thought
its effect on income. Table 3 compares results from naive OLS and 2SLS
estimates for all variables of interest, namely water, grazing land and crop
residue distance. The potential candidate instruments used in the estimation
were tested to check if they could pass the necessary requirements for an
instrument to be as an instrument.
Table 4 reports test results for all scenarios presented in Table 3. The Wu-
Hausman F-test with a p-value less than 0.05 rejected the null hypothesis that
OLS estimation is consistent or income is exogenous and motivates the use of
instruments. Besides, the Sargan chi2 –test fails to reject the null hypothesis
that all instruments are uncorrelated with the error term in the structural model
or all instruments are valid. This enables us to conclude that the instruments
pass the over-identification requirement for all estimates. Finally, instruments
were also tested if they could pass the second most important criteria that the
instrument should be correlated or relevant to the endogenous variable
income. To ensure the relevance of instruments, the Stock and Yogo (2005) F-
test was employed and F-values for three models are about 42 which is
extremely higher than the rule of thumb of at least greater than 10 (Table 4).
Table 4: Instrumental Variables Tests

Endogeneity validity Relevance
Estimates Criteria
Wu-Hausman Stock and Yogo,
Sargan (P-value)
(P-value) F-value
Water Scarcity Model (0.0008) (0.5562) 42.28
Feed Scarcity Model (0.0011) (0.5236) 42.27
Feed Collecting Model (0.0013) (0.5417) 42.56
The first stage regression results of two-stage least square (2SLS) which are
not reported here for the purpose of saving space show that both instruments
72
have a negative relationship with income but only shock variable is found to
be statistically significant in all scenarios (Table 3). Total income of the
household which have positive coefficient significantly affected per capita
food expenditure. Column (1, 3, and 5) of Table 3 shows the ordinary
estimates of the income effect by estimating the consumption model using
OLS estimator. The coefficient of income suggests that a 1% increase in
income increases per capita food expenditure by around 0.044 %, whereas the
2SLS result display that a one percent increase in total income leads to 0.59
percent increase in per capita food expenditure in all estimates. It turns out that
this naive ordinary estimate grossly underestimates the income effect than
effects from the IV-2SLS estimate. This implies that estimating the model
using OLS is not the correct approach and ignoring these differences would
bias the income effect. The findings of Njimanted et al. (2006) in rural
Cameroon, and Demeke et al. (2011) in rural Ethiopia also confirm that
household income is one of the key determinants of food expenditure and food
security in rural areas.
As hypothesized, time spent for searching animal feed and animal water
directly affected per capita food expenditure. Time spent looking for water and
grazing land has resulted in a negative sign as expected and they are found to
be an important factors of per capita food expenditure. A one percent increase
in minutes traveled to reach water source and grazing land leads to a 0.133 and
0.086 percent decrease in per capita food expenditure respectively (Table 3)
referring to the IV estimates. In addition, a one percent increase in minutes
traveled to collect crop residue from threshing fields to homestead leads to
0.073 percent decrease in per capita food expenditure. This supports the
argument by Tangka and Jabbar (2005), whose study conclude that feed
scarcity reduces livestock, crop, and non-farm productivity as well as access to
food, resulting in less food security and low welfare by traveling long distance
with an animal in search of feed and water in less developing countries.
We also report that agricultural output significantly affects households’ food

consumption. It is also the case that the OLS estimates significantly
overestimate the size of the coefficient of the output variable. The elasticity of
food consumption per capita with respect to the gross crop value is equal to
0.094 % for OLS and 0.056% for IV in the water scarcity estimates. Similar
effects are found in the feed and transport estimates presented in Table 3 of
Column 3 to 6.The larger elasticity originates from the fact that a larger share
73
of income is derived from agriculture in rural areas. This is in line with Sarris
et al. (2006) who found that that agricultural output significantly affects per
capita consumption expenditure in Ethiopia.
The variable livestock ownership is positively correlated with welfare,

suggesting that farmers with high herd size have a higher food consumption
expenditure. Studies by Sarris et al. (2006) in Tanzania and Dercon et al.
(2005) had similar findings in Ethiopia. Another significant variable is
household size, leading to 0.357 percent decrease in per capita food
expenditure for one percent increase in the number of family size, in line with
the findings of Dercon et al. (2005) in Ethiopia and Sarris et al. (2006) in
Tanzania but contradicts with the studies of Alene and Manyong (2006) in
Nigeria. The dummy variable for the gender of household head is also found
to be significant and has a negative sign against the findings of Dercon et al.
(2005) in Ethiopia.
Experiencing an animal shock at least once in the previous year lowers per
capita consumption by 0.379%, 0.442% and 0.440% for the three cases taking
the estimated value of IV in Table 3. Dercon (2004) found that a livestock
shock negatively affects per capita consumption expenditure in rural Ethiopia.
The coefficient of household’s religion is 0.152 % and is statistically
significant, implying that orthodox households have 0.152 percent per capita
consumption higher than Muslim group which is opposite to the result of
Oldiges (2012) and Sinha (2005), who together found a positive relation
between Muslim follower and per capita cereal consumption in India.
Although the location is insignificant, per capita food consumption for farmers
living in the highland is lower than for those living in the lowland area. This is
in favor of results from Asmamaw et al. (2015) whose study in Ethiopia
indicated that people from highlands are more chronically food insecure, and
consume less than 50% of total calorie requirements than in the lowlands.
The negative and significant sign of network shows that individuals who got
social supports have 0.191 % less per capita food expenditure, implying that
supports from relatives or friends are not adequate enough to cover food
expenditure for the recipient households (Sarris et al., 2006). Other
insignificant variables are proximity to market (positive), the age of the
74
household head (negative) in line with the study of Matchaya and Chilonda
(2012).
6.4. Total Effect of Feed and Water Scarcity on Food Security
This analysis finalizes its discussion by exploring the total effect of animal
water and feed scarcity on food security. In rural Ethiopia, households spend a
large portion of their daily productive time searching for water and grazing
land for the animal. Based on the descriptive statistics in Table 1, the median
household in this sample spends up to one 75 minutes to travel to a water
source, 91 minutes to search for grazing land and 577 minutes to transport
crop residue yearly. The labor hours allocated for these resources then reduces
the total time available for crop farming activities in addition to the reduction
in the households’ leisure consumption. Its effect on agricultural production is
investigated via the production sector and its direct impact on household’s
utility is analyzed through consumption sector. The aggregate of the two
shows the total welfare effect on the household’s livelihood.
Then, the total effect is simply calculated by taking the slope coefficient of
income in the consumption regression multiplied by the coefficient of time
allocation in the production estimation, plus the coefficient of time allocation
in the consumption regression. Based on Table 5, the total impact of time
spent searching for water, feed and transporting feed on per capita food
consumption expenditure is -0.142, -0.102 and -0.092 respectively. This
implies that for a one percent increase in minutes traveled to a water and feed
source, per capita food consumption decrease by 0.142%, 0.102%, and
0.092% respectively. If the median household in this data spends about 60
minutes to look for water and feed source and have per capita food
consumption expenditure of 2490 birr. For the median household, decreasing
traveling minutes to a water and feed source by 0.6 minutes will increase per
capita food consumption expenditure by 354 birr, 254 birr and 229 birr. The
results of this analysis based on per capita food expenditure can be good
indicators of a necessary condition for food security (FAO, 1996).
75
Table 5: Aggregate Effect of Resource Scarcity on Output, Food

Expenditure, and Food Security
Estimates Effect On Output (Y) Effect On PCFE Total Effect
Effect Of Water Scarcity( -0.155 -0.133 -0.142
Effect Of Feed Scarcity () -0.279 -0.086 -0.102
Effect Of Feed Collection () -0. 328 -.0731 -0.092
Table 6: Estimation of Household Labour Allocation to Crop Farming

(OLS)
Variables
Ln (Family Labor)
Real wage(Wage/milk price) in ETB 0.0112***
(0.0035)
Ln(Wat distance) -0.0598*
(0.0360)
Ln(Graz distance) -0.0929**
(0.0402)
Ln(Feed distances) -0.0992***
(0.0287)
Ln (Family size) 0.3570***
(0.0659)
Ln (Mark distance) 0.0267
(0.0422)
Ln(land area) 0.3420***
(0.0462)
Ln(oxen number) 0.1420*
(0.0732)
Ln(livestock in TLU) 0.0312**
(0.0142)
Gender of household head(Male=1) 0.0677
(0.0722)
Age of Household head (Years) 0.0012
(0.0022)
Household head literacy(Literate=1) 0.0512
(0.0692)
Hired Labore(1/0) -0.1510**
(0.0698)
Household home altitude (GPS ) 0.0005***
(0.0001)
Ln(farm output value) 0.0601***
(0.0149)
Location(1/0) -0.4570***
(0.1640)
Constant 2.2990***
(0.5080)
Observations 502
R-squared 0.3400
Note: P-values are for slopes; ***P<0.01; **P<0.05and *P<0.10 = Significant at 1%, 5% and
10% probability level respectively
76
7. Conclusion and Suggestion
In the least developed countries, it is common that households spend a large

share of their daily hours available for production activities per day on
searching the animal water and feed as well as collecting crop residue. This
directly impacts farm production and utility consumption by displacing labor
from production and leisure activity. This study analyzes the economic
implication of animal water and feed scarcity on agricultural production and
consumption activities of rural farm households in North Ethiopia. For the
analysis, the agricultural farm household model has been adopted and time
spent for searching the animal water and feed resources, capturing water and
feed scarcity has been integrated into the model. The econometric model
derived from the recursive AHM and an empirical application has been
applied using a sample size of 518 extracted from Tigrai Rural Household
Surveys dataset in 2015 harvesting season.
The results in this paper provide an interesting picture of smallholders in

Ethiopia and hint at several areas that could be important for improving food
security. Our results do support the hypothesis of a negative relationship
between total household labour allocation to crop farming and resource
scarcity at the household level. As expected, it appears that time spent looking
for water and feed has a significant and negative effect on both production and
consumption sectors. In aggregate, reducing time spent looking for water by
one percent leads to an increase in food production by 0.155 percent, per
capita food consumption by 0.133 percent and food security by 0.142 percent.
Similarly, a one percent decrease in time wastage for searching grazing land
increase food production, per capita food consumption and aggregate food
security by 0.279 percent, 0.086 percent, and 0.102 percent respectively, and
an increment of 0.328 percent in food production and 0.0731 percent in per
capita food consumption is achieved by one percent reduction in feed
transporting time, leading to an aggregate effect of 0.092 increment in food
security. Thus, the total effect of water and scarcity on per capita food
consumption expenditure shows that reducing time spent on this resource can
bring a significant contribution to food security, and as a result improves the
welfare of the society.
77
Another major conclusion is that the use of inputs such as land, family and
hired labor, fertilizer, manure, oxen and farm physical tool appears to be
positively related to the household’s agricultural production, and are
significant determinants of farm productivity as predicted by the economic
theory. However, aggregate production seems to be impeded by the
occurrence of shock and agroecology, indicating that farmers experiencing
shock and living in the highland seem to suffer from less production. On the
consumption side variables such as agricultural output, income, livestock
ownership and religion affiliation are found to be major positive contributing
factors but shock occurrence, family size, male headship and social network
are found to reduce per capita food consumption. Results confirm the
theoretical prediction that having a higher number of family member and
shock exposure affect per capita food consumption expenditure adversely.
The empirical results presented in this paper lead to the following policy
conclusions. Two areas of policy intervention can be emerged as relevant. The
first involves policies and institutions that facilitate easier access to animal
water tap by advocating on emergency relief grounds. The second area of
policy intervention involves the introduction of more efficient animal feed
management strategy that can be implemented by helping households adopt
new technologies that improve cattle production and reduce land degradation.
Third, given the evidence in this paper, it appears that policies that seek to
promote information and reducing shock exposure would be useful in
enhancing household level food security.
In general, this study can be helpful for policy makers working to alleviate
animal water and feed problems in Ethiopia to justify their actions with an
empirical result. Besides, this study’s result can give a good lesson for policy
analysts that labor allocation for reaching water and feed source imposes a
negative impact on crop farm output and food consumption and hence on food
security. Helping farmers to have a nearby water and feed source do not only
alleviate labor constraints but also saves time that could be used for other
productive farming activities. Such strategy enables farmers to keep their
animals at the homestead in the form of stall feeding and tethering around the
backyard.
78
References
Abegaz, A. (2005), “Farm Management in Mixed Crop-livestock Systems in the

Northern Highlands of Ethiopia”, Ph.D. Thesis, Wageningen University.
Netherlands.
Abegaz A., H. Van Keulen, M. Haile, and S. Oosting (2007), Nutrient dynamics on
smallholder farms in Teghane, northern highlands of Ethiopia. In: Bationo
A, Waswa B, Kihara J, Kimetu J (eds) Advances in integrated soil fertility
management in sub-Saharan Africa: challenges and opportunities. Springer,
Germany, pp 365–368.
Abdulai, A., and W. Huffman (2014), “The Adoption and Impact of Soil and Water
Conservation Technology, An Endogenous Switching Regression
Application,” Land Economics, 90(1), 26-43.
Alene, A.D., and V.M. Manyong (2006), “Endogenous Technology Adoption and
Household Food Security, the Case of Improved Cowpea Varieties in
Northern Nigeria,” Quarterly Journal of International Agriculture, 45(3),
211-230.
Angrist, J.D., and W. Evans (1998), “Children and their Parents’ Labor Supply:
Evidence from Exogenous Variation in Family Size,” American Economic
Review, 88 (3), 450-477.
Asmamaw, T., M. Budusa, and M. Teshager (2015), “Analysis of Vulnerability to
Food Insecurity in the Case of Sayint District, Ethiopia,” Asian Journal of
Rural Development, 5(1), 1-11.
Asfaw, S., B. Shiferaw, F. Simtowe, and L. Lipper (2012), “Impact of Modern
Agricultural Technologies on Smallholder Welfare: Evidence from Tanzania
and Ethiopia,” Food Policy, 37(3), 283-295.
Asresie, A., and L. Zemedu (2015), “The Contribution of Livestock Sector in
Ethiopian Economy,” Advances in Life Science and Technology, 29, 79-90.
Baland, J.M., P. Bardhan, S. Das, D. Mookherjee, and R. Sarkar (2010), “The
Environmental Impact of Poverty: Evidence from Firewood Collection in
Rural Nepal,” Economic Development and Cultural Change, 59(1), 23-61.
Bandyopadhyay, S., P. Shyamsundar, and A. Baccini (2011), “Forests, Biomass use
and Poverty in Malawi,” Ecological Economics, 70(12), 2461-2471.
Bhattacharya, H., and R. Innes (2006), “Is There a Nexus between Poverty and
Environment in Rural India?” In AAEA Annual Meeting, Long Beach,
California.
Barnum, H.N., and L. Squire (1979), “An Econometric Application of the Theory of
the Farm Household,” Journal of Development Economics, 6(1), 79-102.
Belay, D., E. Getachew, T. Azage, and B.H. Hegde (2013), “Farmers’ Perceived
79
Livestock Production Constraints in Ginchi Watershed Area: Result of

Participatory Rural Appraisal,” International Journal of Livestock
Production, 4(8), 128-134.
Bezabih, M.Y., and G. Berhane (2014), “Livestock Production Systems Analysis,”
American International Journal of Contemporary Scientific Research, 1(2),
16-51.
Bishu, K.G. (2014), “Risk Management and the Potential of Cattle Insurance in
Tigray, Northern Ethiopia,” Ph.D. Thesis, University College Cork. Ireland.
Brown, L.R. (2004), World Food Security Deteriorating: Food Crunch in 2005 Now
Likely, Washington, DC: Earth Policy Institute. USA.
Bulte, E., G. Beekman, S. Di Falco, J. Hella, and P. Lei, (2014), “Behavioral
Responses and The Impact of New Agricultural Technologies: Evidence
from A Double-Blind Field Experiment in Tanzania,” American Journal of
Agricultural Economics, 96(3), 813-830.
Cooke, P.A. (1998), “The Effect of Environmental Good Scarcity on Own-farm Labor
Allocation: The Case of Agricultural Households in Rural Nepal,”
Environment and Development Economics, 3, 443-469.
Cooke, P., G. Köhlin, and W.F. Hyde, (2008), “Fuelwood, Forests, and Community
Management –Evidence from Household Studies,” Environment and
Development Economics, 13, 103-135.
Central Statistical Agency (2010c), Agricultural Sample Survey, Land Utilization,
Private Peasant Holdings, Meher Season, Central Statistical Agency (CSA),
Statistical Bulletin, 6(468), Addis Ababa, Ethiopia.
Damte, A., S.F. Koch, and A. Mekonnen (2012), “Coping with Fuelwood Scarcity.
Household Responses in Rural Ethiopia,” Environment for Development.
Discussion Paper Series, 12(01).
Delforce, J.C. (1994), “Separability in Farm-Household Economics: An Experiment
with Linear Programming,” Agricultural Economics, 10(2), 165-177.
Demeke, A.B., A. Keil, and M. Zeller (2011), “Using Panel Data to Estimate the
Effect of Rainfall Shocks on Smallholders’ Food Security and Vulnerability
in Rural Ethiopia,” Climatic change, 108(1), 185-206.
Dercon, S. (2004), “Growth, and Shocks: Evidence from Rural Ethiopia,” Journal of
Development Economics, 74 (2): 309–29.
Dercon, S., J. Hoddinott, and T. Woldehanna (2005), “Shocks and Consumption in 15
Ethiopian Villages, 1999-2004,” Journal of African Economies, 14(4), 559-
585.
Diewert, W.E. (1973), “Separability and a Generalization of the Cobb-Douglas Cost
Production, and Indirect Utility Functions,” Research Branch, Program
Development Service, Department of Manpower and Immigration.
Department of Economics, University of British Columbia. Canada.
80
Di Falco, S., M. Veronesi, and M. Yesuf (2011), “Does Adaptation to Climate Change
Provide Food Security? A Micro-perspective from Ethiopia,” American
Journal of Agricultural Economics, 93(3), 829-846.
Descheemaeker, K. (2008), “Baseline Data Report for the Ethiopian Study Sites:
Fogera: Kuhar Michael and Gubalafto: Lenche Dima, BMZ project in
Improving Water Productivity of Crop-Livestock Systems of Sub-Saharan
Africa,” IWMI (International Water Management Institute) and ILRI
(International Livestock Research Institute), Addis Ababa, Ethiopia.
Food and Agricultural Organization (1996), Rome Declaration on World Food
Security, World Food Summit. 13–17 Nov. 1996, Rome, Italy
Food and Agricultural Organization (2001), “Crop and Food Supply Assessment
Mission to Ethiopia,” FAO Global Information and Early Warning System on
Food and Agriculture, World Food Programme, Rome, Italy.
Feleke, S.T., R.L. Kilmer, and C.H. Gladwin (2005), “Determinants of Food Security
in Southern Ethiopia at the Household Level,” Agricultural Economics, 33,
351–363.
Foster, A., and M.R. Rosenzweig (2010), “Barriers to Farm Profitability in India:
Mechanization, Scale and Credit Markets,” In Conference Agriculture for
Development-Revisited, the University of California at Berkeley. USA.
Gebremedhin, B. (2009), “Feed Marketing in Ethiopia: Results of Rapid Market
Appraisal,” (No. 15). ILRI (aka ILCA and ILRAD).
Gehbru, H. (2010), “Land Policy Reforms and the Land Rental Market in Ethiopia.
Equity, Productivity, and Welfare Implications,” PhD-dissertation.
Department of Economics and Resource Management, Norwegian University
of Life Sciences. Norway.
Govereh, J., and T.S. Jayne (1999), “Policy Synthesis: Effects of Cash Crop
Production on Food Crop Productivity in Zimbabwe: Synergies or Trade-
offs,” USAID-Africa Bureau, Office of Sustainable Development.
Gutu, T. (2016), “Are Rural Youth Disengaging from Agriculture? Empirical evidence
from Ethiopia.” In 2016 AAAE Fifth International Conference, Addis Ababa,
Ethiopia (No. 246925). African Association of Agricultural Economists
(AAAE).
Hagos, F., and S.T. Holden (2011), Fertilizer Use by Smallholder Households in
Northern Ethiopia: Does Risk Aversion Matter? in R.A. Bluffstone and G.
Köhlin, eds., Agricultural Investment and Productivity. Building
Sustainability in East Africa, RFF Press: Washington, DC and London.
Halderman, M. (2005), “The Political Economy of Pro-poor Livestock Policy-making
in Ethiopia,” FAO working paper No. 19. Rome: FAO.
Hassen, A., A. Ebro, M. Kurtu and A. C. Treydte (2010), “Livestock Feed Resources
Utilization and Management as Influenced by Altitude in the Central
81
Highlands of Ethiopia,” Livestock Research for Rural Development. Volume

22, Article #229.
Herrero, M., P. Havlík, H. Valin, A. Notenbaert, M. C. Rufino, P. K. Thornton, M.
Blümmel, F. Weiss, D. Grace, and M. Obersteiner (2013), “Biomass use,
Production, Feed Efficiencies, and Greenhouse Gas Emissions from Global
Livestock Systems,” Proceedings of the National Academy of Sciences of the
United States of America. 110, 20888–20893.
Hidalgo, F.D., S. Naidu, S. Nichterm, and N. Richardson (2010), “Occupational
Choices: Economic Determinants of Land Invasions,” Review of Economics
and Statistics, 92(3), 505-523.
Holden, S.T., K. Deininger, and H. Ghebru (2011), “Tenure Insecurity, Gender, Low-
cost Land Certification and Land Rental Market Participation,” Journal of
Development Studies, 47(1), 31-47.
Holden, S.T., K. Deininger, and H. Ghebru (2009), “Impacts of Low-cost Land
Certification on Investment and Productivity,” American Journal of
Agricultural Economics, 91(2), 359-373.
Ilyin S. (2011), The Looming Threat of Overgrazing: Effects and Recommendations,
Congressional Hunger Center, available at:
http://Hungercenter.wpengine.netdna- cdn.com/wp-
content/uploads/2011/09/Looming-Threat-of-Overgrazing-Ilyin.pdf.
Jacoby, H.G. (1993), “Shadow Wages and Peasant Family Labour Supply: An
Econometric Application to the Peruvian Sierra,” Review of Economic
Studies, 60(4), 903-921.
Kassa, H., D. Gibbon, and B. P. Singh (2002), “Livestock improve household food
security and sustainability of Ethiopian small farms,” Journal of Sustainable
Agriculture, 21(2), 73-93.
Khan, H. (2008). “Poverty, Environment and Economic Growth: Exploring the Links
among Three Complex Issues with Specific Focus on the Pakistan’s Case,”
Environment Development Sustain, 10(6), 913-929.
Kidane, H., Z.G. Alemu, and G. Kundhlande (2005), “Causes of Household Food
Insecurity in Koredegaga Peasant,” Agrekon, 44(4), 543-560.
Kirui, O.K., and A. Mirzabaev (2014), “Economics of Land Degradation in Eastern
Africa,” ZEF Working Paper Series No. 128.
Kumar, S.K., and D. Hotchkiss (1988), “Consequences of Deforestation for Women’s
Time Allocation, Agricultural Production, and Nutrition in Hill Areas of
Nepal,” Research Report 69, International Food Policy Research Institute,
Washington, DC.
Lal, R., and B.A. Stewart (2010), Food Security and Soil Quality, CRC Press.
Matchaya, G., and P. Chilonda (2012), “Estimating Effects of Constraints on Food
Security in Malawi: Policy Lessons from Regressions Quantiles,” Applied
82
Econometrics and International Development, 12(2), 165-191.

Mekonnen, A., A. Damte, and R. Deribe (2015), “The Impact of Natural Resource
Scarcity on Agriculture in Ethiopia,” The Environment for Development
Discussion Paper-Resources for the Future (RFF), 15-13.
Mekonnen, D., E. Bryan, T. Alemu, and C. Ringler (2017), “Food Versus Fuel:
Examining Tradeoffs in the Allocation of Biomass Energy Sources to
Domestic and Productive uses in Ethiopia,” Agricultural Economics.
Miller, E. (2008), “An Assessment of CES and Cobb-Douglas Production Functions,”
Congressional Budget Office, USA.
Murthy, K.V. (2004), “Arguing a Case for Cobb-Douglas Production Function, Delhi,
India,” Review of Commerce Studies, 20-21, 75-91.
Nahusenay, A., and T. Tessfaye (2015), “Roles of Rural Women in Livelihood and
Sustainable Food Security in Ethiopia: A Case Study from Delanta Dawunt
District, North Wollo Zone,” International Journal of Nutrition and Food
Sciences, 4(3), 343-35.
Najafi, B. (2003), “An Overview of Current Land Utilization Systems and Their
Contribution to Agricultural Productivity,” Report of the APO Seminar on
the Impact of Land Utilization Systems on Agricultural Productivity,
Productivity Organization, Islamic Republic of Iran Asian.
Njimanted, G.F. (2006), “An Econometric Model of Poverty in Cameroon: A System
Estimation Approach,” International Review of Business Research Paper,
2(2):30-46.
Nisrane, F., G. Berhane, S. Asrat, G. Getachew, A.S. Taffesse, and J. Hoddinott
(2011), “Sources of Inefficiency and Growth in Agricultural Output in
Subsistence Agriculture: A Stochastic Frontier Analysis,” ESSP II Working
Paper 19.
Okwi, P.O., and T. Muhumuza (2010), “Effects of Deforestation on Household Time
Allocation among Rural Agricultural Activities: Evidence from Western
Uganda,” African Journal of Economic Policy, 17(1), 105-135.
Oldiges, C. (2012), “Cereal Consumption and per Capita Income in India,” Economic
and Political Weekly, 47(6), 63.
Oum, T.H. (1989), “Alternative Demand Models and their Elasticity Estimates,”
Journal of Transport Economics and Policy, 23(2), 163-187.
Sakketa, T., and N. Gerber (2017), “Rural Shadow Wages and Youth Agricultural
Labor Supply in Ethiopia: Evidence from Farm Panel Data,” ZEF-Discussion
Papers on Development Policy No. 236.
Sarris, A., S. Savastano, and L. Christiaensen (2006), “The Role of Agriculture in
Reducing Poverty in Tanzania: A Household Perspective from Rural
Kilimanjaro and Ruvuma,” FAO Commodity and Trade Policy Research
Working Paper, (19).
83
Shiferaw, F., R.L. Kilmer, and C. Gladwin (2003), “Determinants of Food Security in
Southern Ethiopia,” in American Agricultural Economic Association Meeting
at Montreal, Canada.
Singh, I. and S. Janakiram (1986), “Agricultural Household Modeling in a Multi-crop
Environment: Case Studies in Korea and Nigeria,” in Singh, I., L. Squire, and
J. Strauss, eds., Agricultural Household Models: Extensions, Applications,
and Policy, Johns Hopkins University Press, Baltimore, MD, pp. 95-l 15.
Sinha, K. (2005), “Household Characteristics and Calorie Intake in Rural India: a
Quantile Regression Approach, Working Paper.
Singh, I., L. Squire, and J. Strauss (1986), Agricultural Household Models:
Extensions, Applications, and Policy, Johns Hopkin University Press,
Baltimore, MD.
Sileshi, Z., A. Tegegne, and G.T. Tsadik (2003), “Water Resources for Livestock in
Ethiopia: Implications for Research and Development,” Integrated Water and
Land Management Research and Capacity Building Priorities for Ethiopia,
66.
Steinfeld, H., P. Gerber, T. D. Wassenaar, V. Castel, and C. de Haan (2006),
“Livestock’s Long Shadow: Environmental Issues and Options,” Food and
Agriculture Org.
Stock, J.H., and M. Yogo (2005), “Testing for Weak Instruments in Linear IV
Regression,” in Andrews D.W.K., H.S. James, and J.R. Thomas, eds.,
Identification and Inference for Econometric Models, New York: Cambridge
University Press
Strauss, J. (1986), “The Theory and Comparative Statics of Agricultural Household
Models: A General Approach,” in Singh, I., L. Squire, and J. Strauss, eds.,
Agricultural Household Models: Extensions, Applications, and Policy,
Baltimore: Johns Hopkins University Press.
Swanepoel, F., A. Stroebel, and S. Moyo (2010), The Role of Livestock in Developing
Communities: Enhancing Multifunctionality, Cape Town, South Africa:
University of the Free State and CTA.
Tangka, F.K., and M.A. Jabbar (2005), “Implications of Feed Scarcity for Gender
Roles in Ruminant Livestock Production,” Working Paper.
Tegegne, S.D. (2012), “Livestock Water Productivity (LWP) improvement in the
mixed crop-livestock system of Ethiopian Highlands, Amhara Region: a
gendered sustainable livelihood approach to target LWP interventions for
rural poverty reduction,” ZEF Ecology and Development Series No.85.
Tesfay, Y. (2010), “Feed Resources Availability in Tigray Region, Northern Ethiopia,
for the Production of Export Quality Meat and Livestock,” Ethiopia Sanitary
and Phytosanitary Standards and Livestock and Meat Marketing Program
(SPS-LMM) Report.
84
Thirumarpan, K. (2013), “Determinants of Household Food Expenditure among Rural

Households of Ampara District. Proceedings,” 04th International
Symposium, SEUSL,” Social Science and Humanities, 325-330.
Tilahun, H., and E. Schmidt (2012), “Spatial Analysis of Livestock Production
Patterns in Ethiopia, Development Strategy and Governance Division,” ESSP
II Working Paper 44.
Yotopoulos, P.A., L.J. Lau, and W.L. Lin (1976), “Microeconomic Output Supply and
Factor Demand Functions in the Agriculture of the Province of Taiwan,”
American Journal of Agricultural Economics, 58, 333-340.
Yilma, Z., E. Guernebleich, A. Sebsibe, and R. Fombad (2011), “A Review of the
Ethiopian Dairy Sector,” Ethiopia: FAO Sub-Regional Office for Eastern
Africa.
Wooldridge, J.M. (2009), Introductory Econometrics: A Modern Approach 4th eds,
Mason: South-Western SENGAGE Learning.
85
Appendix
Figure A1: Zonal Distance to Grazing Land
Figure A2: Zonal Distance to Water Source
86
Figure A3: Zonal Distance to Crop Residue Site
Figure A4: Zonal Per Capita Food Consumption Expenditure
87
Figure A5: Zonal Monetary Value of Farming Output
Figure A6: Zonal Total Income
88
Figure A7: PCFE Vs Market Distance
Figure A8: Output Vs Market Distance
Figure A9: Total Income Vs Market Distance
89
Figure A10: PCFE Vs Water Distance
Figure A11: Output Vs Water Distance
Figure A12: Income Vs Water Distance
90
Figure A13: PCFE Vs Feed Distance
Fig A14. Output Vs Feed Distance
Figure A15: Income Vs Feed Distance
91
Figure A16: PCFE Vs Shock Exposure
Fig A17 Output Vs Shock Exposure
92
Explaining Inter-regional Differentials in Child Mortality
in Rural Ethiopia: A Count Data Decomposition Analysis
Yibrah Hagos Gebresilassiea1* and Phocenah Nyatanga2
Abstract
Using data from the Ethiopian Demographic and Health Survey, 2016for a
total of 1,295 number of under-five child deaths, this study examined the
major determinants of inter-regional differentials in under-five child
mortality in rural settings of Ethiopia. An extended detailed Oaxaca-Blinder
decomposition technique to negative binomial regression model was applied
to examine the relative individual contribution of different covariates to
inter-regional differentials in under-five child mortality. Findings of
decomposition analysis indicated that large portion of the regional
differentials remained unexplained, being the lowest between Tigrai and
Benishangul-Gumuz (12 percent) and the highest in Tigrai-Gambella regions
(37 percent). The explained regional gap was due to differences in the
distributions of measured factors across regions mainly attributable to
differences in short birth-spacing, higher birth-order, antenatal healthcare
services visits, women without education, home delivery, large household
size, and poorest households’ economic status. Hence, understanding inter-
regional differentials in under-five child mortality and developing
appropriate policies and strategies could further reduce the rate of under-
five child mortality. Thus, on top of strengthening the health extension
programme in rural Ethiopia, this study suggests that substantial efforts must
also be made to improve the overall households’ economic status and
women’s education levels.
Keywords: Child, decomposition, determinant, mortality, region

JEL Classification: D24, I14, J13, R23, R58
1
*Corresponding Email: [email protected]; Tel.: +251913139174, Adigrat
University, Adigrat-Tigrai, Ethiopia.
A College of Business and Economics, Adigrat University, Ethiopia
2
Lecturer in School of Accounting, Economics, and Finance, University of
KwaZulu-Natal
(UKZN), South Africa. E-mail: [email protected]; Tel: +2733 260 5522
93
Yibrah and Phocenah: Explaining Inter-regional Differentials in Child Mortality in Rural Ethiopia:…
1. Introduction
Child mortality is one among the key indicators of the well-being of

population and society, as measured by life expectancy and is considered as
one of the Human Development Index’s (HDI) dimensions used by the
“United Nations Development Programme” (UNDP) (Aigbe & Zannu, 2012;
NIMS et al., 2012; Patel & Sharma, 2013; UN, 2010). Reducing child
mortality can significantly increase the life expectancy and hence, human
capital, which is highly required for the overall development of one’s nation
(MOFED, 2004). The globe has made substantial improvement in overall
under-five child mortality reduction. Overall, under-five child mortality rate
(U5MR) has fallen dramatically from 12.7 million per year in 1990 to 5.9
million per year in 2015 (UNIGME, 2011, 2012, 2013, 2014, 2015). Despite
the progress that the globe has made in reducing the overall child mortality,
the rates of progress differ substantially across countries and regions (Shyama
et al., 2014). For example, East Asia and the Pacific have exceeded the
“Millennium Development Goal” (MDG-IV) target of a two-thirds reduction
in U5MR between 1990 and 2015, whereas sub-Saharan Africa has had only a
24 percent decline over the same period (UNIGME, 2015). Despite the
progress the Sub-Saharan Africa made, the region remains with the highest in
U5MR in the world (Demombynes & Trommlerová, 2012; UNIGME, 2011,
2012, 2013, 2014, 2015). Most of the global under-five child deaths still occur
in this region, where one child in every twelve dies before reaching five years
of age (UNIGME, 2015). Also, evidence indicates that there is a substantial
difference in the rate of progress within sub-Saharan Africa (UNIGME, 2015),
where one child in every nine dies before celebrating his or her fifth birthday
when compared to the death of one under-five child in every 152 in developed
countries (UNIGME, 2012).
Ethiopia is a sub-Saharan African country experienced sizeable progress in

under-five child mortality reduction at the national level, dropping from
211deaths in the 1990s to 88 deaths per thousand births in 2016 (CSA &
ICFInternational, 2012), however, the country remains among the highest
number of under-five child deaths in the world (UNICEF, 2015b). Although
Ethiopia has already achieved its U5MR by two-thirds (68 deaths per
thousand births) in 2012 (UNIGME, 2013, 2014) and dropped by 71 percent
94
with average annual rate of reduction of 5 percent between 1990 and 2015
(UNICEF, 2015a), previous studies indicated the existence of substantial
variations in the rate of progress across regions of the country (the regions are
Afar, Amhara, Benishangul-Gumuz, Gambella, Harari, Oromia, Somali,
Southern Nations Nationalities and People (SNNP) (Abebaw, 2013; CSA &
ICFInternational, 2012; CSA & ORCMacro., 2006; UNDP, 2012). Oftentimes,
the observed differences in the rate of progress across regions have been
masked by the overall rate of reduction in under-five child mortality at the
national average. Moreover, in Ethiopia, the inter-regional distribution of
under-five child mortality indicate the marked regional disparities (Abebaw,
2013; CSA & ICFInternational, 2012; UNDP, 2012). In 2000, for example, the
U5MR varied from as low as 169 deaths in Tigrai to as high as 233 death per
thousand births in Gambella (CSA & ORCMacro., 2000). Similarly, in 2016,
the U5MR also varied as low as 85 deaths in Tigrai to as high as 169 deaths
per thousand births in Benishangul-Gumuz. The rates of decline in under-five
mortality for all regions except Tigrai (85 deaths per thousand births) were
significantly lower than the national average rate (88 deaths per thousand
births) in 2016, indicating there was a disproportionate inter-regional gain in
under-five child mortality rates across times (CSA & ICFInternational, 2012).
Furthermore, despite the overall rate of reduction in under-five child mortality,

the magnitude of mortality rate inequalities has significantly varied between
regions and over time. For example, the under-five mortality rate of the
Ethiopian Somali and Benishangul-Gumuz regions have increased from 93
deaths in 2005 to 122 deaths per thousand births in 2016, and from 157 deaths
in 2005 to 169 deaths per thousand births in 2016, respectively. Similarly, the
U5MR for the Benishangul-Gumuz region has increased from 157 deaths in
2005 to 169 deaths per thousand births in 2016, a statistic (CSA &
ICFInternational, 2012) even higher than for Angola with an under-five child
mortality of 167 deaths per thousand, the highest in the world (UNICEF,
2014). This evidence shows that although most regions have reduced the
under-five child mortality with different levels of reduction, some of them
(Afar, Somali, and BG regions) have found to increase the U5MR instead
from 2005 to 2016 (CSA & ICFInternational, 2012; CSA & ORCMacro.,
2006). Moreover, compared to many other developing countries the
improvement that Ethiopia has made in overall child mortality reduction
95
remains very low. The country has been ranked 37th and is one among the ten
top countries with highest absolute umber under-five children deaths (184
deaths per thousand). Hence, Ethiopia accounts for three percent of the share
of global under-five child deaths in 2015 (UNICEF, 2015a; UNIGME, 2015).
More importantly, about 59 of every one thousand children in Ethiopia are still
dying before celebrating the age of five years (UNICEF, 2015a; UNIGME,
2015). Like in many developing countries, in Ethiopia mortality of under-five
children in rural areas are considerably higher than in urban areas (CSA &
ICFInternational, 2012; CSA & ORCMacro., 2006; Regassa, 2012). A child
born in rural areas has 38 percent higher probability of dying than a child of
urban counterparts (FMOH, 2014b). Previous studies have also noted that one
child in every 11 Ethiopian children under-five dying before reaching the fifth
births anniversary (CSA & ICFInternational, 2012; CSA & ORCMacro.,
2006). Furthermore, most of the Ethiopian population is still primarily rural.
Out of the total population (94 million), more than 15 percent (14.245 million)
of them are under-five children (UNICEF, 2014). Since the share of rural
population in Ethiopia is huge, combating under-five rural child mortality
could further speed up the overall U5MR reduction both at the national and
regional levels. The overall rate of progress that Ethiopia has made in U5MR
(59) is considerably lower than infant mortality (41 deaths per thousand births)
(UNIGME, 2015).
Furthermore, in Ethiopia, the regional disparities in under-five child mortality

rates were twice higher than in infant mortality rates (UNDP, 2012). This
suggests that the importance of addressing disparities in mortality of under-
five children to further reduce the overall child mortality of the country. More
importantly, much less is known about which factors explaining the regional
variations in under-five mortality rates, while majority of previous studies
have instead focused on factors influencing infant and under-five child
mortality rates in Ethiopia (Amouzou et al., 2014; Dejene & Girma, 2013;
Regassa, 2012; Tesfa & Jibat, 2014). These are the rationale as to why this
study is carried out and focused on U5MR in rural areas of Ethiopia. This
study, therefore, aims at identifying the major factors responsible for inter-
regional differentials in under-five child mortality levels in rural settings of
Ethiopia.
96
The remaining of the paper is systematised as follows: review of previous

studies is presented in section two. In section three, data source and
methodologies are described followed by analysis of results in section four.
Section five discusses the findings. The chapter concludes the study in section
six.
2. Review of previous literature
A substantial number of previous studies have evaluated the factors affecting

infant and under-five child mortality rates (Caldwell, 1979; Dejene & Girma,
2013; Kabir et al., 2001; Khadka et al., 2015; Shyama Kuruvilla et al., 2014;
Srinivasan, 2000). However, despite the overall improvement in under-five
child mortality rates across countries, the rate of progress was varied not only
across countries or between developing and developed countries, but within a
country. In Sub-Saharan Africa where Ethiopian is located, the marked
disparities in the rate of under-five child mortality remain very high across the
countries (UNIGME, 2012, 2013, 2014, 2015). Hence, within demographic
and development economics literature, currently, substantial interest has been
observed in identifying and quantifying the separate relative contribution of
specific determinants on how each explains the observed regional under-five
child mortality differentials across states or regions within a country and
across countries. In developing countries, there have been substantial regional,
provincial or cross- state differences in infant and under-five child mortality.
Hence, reducing the variations in child mortality within and between countries
could considerably contribute to the overall health of the
population(Houweling & Kunst, 2010). A study by Adedini et al. (2015)
examined the sources of regional differentials in infant and under-five child
mortality in Nigeria using 2008 demographic and health survey data. The
study has applied Cox-proportional hazard regression model to identify the
determinants of the regional differentials in child mortality (infant and under-
five child mortality) in Nigeria. The findings simply indicated that differences
in community infrastructure, households’ wealth index, households’ poverty
status, place of delivery and residence distributions across the regions were the
major factors of regional differentials in under-five child mortality while
difference in birth-order, birth-spacing, mother’s level of education, and
mother’s age at marriage distributions across regions were the most key
97
factors explaining the regional disparities in infant mortality rate in Nigeria.

The study concluded that to substantially reduce the overall child mortality of
the country, much efforts should be exerted in addressing the sources of
regional variations in these important health indicators by focusing on the
disadvantageous regions of the country, however, the authors could not
explain the percentage relative contribution of each covariate to the explained
regional gap. A study by Jhamba (1999) indicated that despite the dramatic
decline in child mortality among district of Zimbabwe, there have
considerable disparities across districts. Hence, mother’s education, the
percentage of households with access to improved water and toilet facility was
among the major determinants of regional variation in child mortality in
Zimbabwe. Other factors such as malaria epidemic, religious and cultural
determinants were also explained the district differentials in child mortality
rates in Zimbabwe. Similar regional differences in under-five child mortality
have been reported in many other developing countries (references). For
example, In Libya, a study by Ghaffar and Bhuyan (2000) examined the
factors explaining the regional differentials in child mortality in North-eastern
Libya. The study was based on the seven localities and then these localities
have developed into three regions namely, Benghazi, Darna, and Tobruk,
where five out of the seven localities are found in Benghazi.
In Nepal, the disparities in child mortality by ecological region was examined

by Goli et al. (2015). To examine the determinants of regional variation in
child mortality, they used an Oaxaca-blinder decomposition technique based
on Cox-Proportional hazard regression model using demographic and health
survey data. The results of Cox proportion regression indicated that children of
Mountainous areas had the highest probability of dying than children of the
same cohort living in the other two areas (Hill and Terai). The results of the
decomposition analysis revealed that differences due to the proportional
differences in children of four birth-order or higher, mother’s working status,
place of residence, households’ economic status, and father’s level of
education were reported to significantly explained the regional under-five
child mortality disparities. The decomposed covariates altogether explained 40
percent of the regional variations in under-five child mortality between the
mountain and the combined Hill and Terai regions while the larger 60 percent
of the components of the gap remained an unexplained part. Findings of the
98
decomposition analysis revealed that the differences in the proportional

distribution of parental educational levels (mother’s and father’s education)
contributed 34 percent of the regional variations in under-five child mortality.
However, 30 percent of the explained gap by parental education was attributed
to father’s level of education, the largest contributor to the ecological
differentials in under-five child mortality. The results further indicated that
households’ wealth status, households’ place of residence, higher birth-order
along with short spacing (less than 24 moths), and mother’s employment
status have contributed significantly to 25, 16, 11, and 5 percent of the
explained ecological regional differences in under-five child mortality,
respectively. In addition, mother’s religion and mother’s liberty on healthcare
decision have contributed 3 percent each to the explained regional gap in
under-five child mortality. Although its relative percentage contribution of the
explained gap is very small, mother’s exposure to mass media has also
contributed to under-five child mortality differences between the two
ecological regions. Furthermore, the study indicated that female under-five
children are in a les advantageous situation in terms of the survival rate in the
country compared to male cohort counterparts. The study has concluded that
though Nepal has made a remarkable progress and achieved the “Millennium
Development Goal” four (MDG-IV) in under-five child mortality reduction by
two-third, there has been variations in rate of progress in child mortality across
its ecological regions. Hence, the disparities in rate of progress of under-five
child mortality should be addressed from an ecological region outlook (Goli et
al., 2015).
In Mozambique, the geographic disparities in child mortality have been

examined using the Mozambican demographic and health survey data of 2003
(Macassa et al., 2012). The ten provinces have been geographically classified
into three regions; North, Central and South regions. The study has applied
Cox regression analysis to identify the factors explaining the regional
differences in under-five child mortality. An under-five child whose mother
was living in the North and the central regions had higher mortality risks than
a child of a mother who was living in the South regions. The study has also
indicated that there have been significant differences in levels of under-five
child mortality within the regions (among provinces of the same region).
However, although the authors have attempted to indicate why the regional
99
variations occurred in child mortality in Mozambique through discussing the

reviewed literature, empirically; the authors have nor explored which factors
and how much each did contribute to explain the geographic-specific
variations in under-five child mortality. Employing the Iranian demographic
and health survey data of 2000, a study by Hosseinpoor et al. (2006) examined
the contribution of determinants differentials in infant mortality. The analysis
was made using the concentration index based on logistic regression to
compute the contribution of specific socioeconomic determinants inequalities
in infant mortality. The magnitude of differences in households’ economics
status (36 percent), and mother’s education level (21 percent) were the largest
contributors to the regional infant mortality differences in Iran. The paper
further indicates that risky or short birth-spacing (13 percent), place of
residence (14 percent) and access to improved toilet facilities (12 percent)
contributed significantly to the regional disparities in infant mortality rates in
Iran. The findings have finally noted that provinces had different levels of
inequalities in infant mortality rates (Hosseinpoor et al., 2006).
Similarly, the study of Assi (2014) has attempted to assess the factors
explaining regional variations in under-five child mortality in Cote d'Ivoire
based on 2016-2012 Cote d'Ivoire demographic and health survey data using
logistic regression model. Findings indicated there were considerable
variations in child mortality across the region of Cote d'Ivoire. Mother’s
education at least who completed secondary education was associated with
under-five child mortality risk and was found to be statistically significant.
However, the study failed to identify the sources of the observed regional
variations in under-five child mortality in Cote d'Ivoire rather it has identified
the factors affecting under-five child mortality not the regional variations in
U5MR. More importantly, the study suggested further research be carried out
explaining the sources of regional differences in child mortality. Similarly, a
study by Akuma (2013) has evaluated regional differentials in infant mortality
using the 2009 Kenyan DHS. For analyses purpose, the author has examined
the regional differences in infant mortality by classifying provinces of the
country into two regions (groups) as low and high infant mortality regions
based upon the magnitudes or levels of infant mortality that the provinces had
and applied logistic regression model to analysis the data. Hence, the results of
the regression analysis revealed that there were regional disparities in infant
100
mortality across regions. The mother’s low level of educational attainment,

poor socioeconomic status, and short birth spacing were the major
determinants of infant mortality for the region of high mortality category that
causes the regional variations in infant mortality between the mortality regions
(high and low mortality regions). Finally, the author has concluded that the
sources of infant mortality differentials across provinces of Kenya are due to
differences in households’ economic status and social development. However,
the study did not consider other important demographic and other
socioeconomic factors while examining the regional differentials in infant
mortality that. More importantly, findings of the study might not really
indicate the sources the regional differentials in infant mortality in Kenya
Akuma (2013).
In Asia, a study by Khosravi et al. (2007) evaluated the mortality differentials

among the Iranian provinces. Child mortality rates varied among the provinces
from 25 to 47 per thousand births. The findings indicated that important
sources of variations in child mortality among the Iranian provinces. These are
the GDP per capita, life expectancy, and health care accessibility. Provinces
having a high GDP per capita and high life expectancy had the lowest rate of
child mortality. The Iranian study concluded that variations in child mortality
were worse in the rural areas than the urban areas of the country. However, in
Iran, the extent of variations in child mortality is lower than the child mortality
differentials for other developing countries. Another study evaluating “inter-
district variations in infant mortality in Sri Lanka” indicated that access to
health care services, (33 percent); safe drinking water, (16 percent); low
childbirth weight, (13 percent); and health care utilization (8 percent)
explained infant mortality differences across districts of the country. Findings
of the study noted that a unit increment in health care service accessibility and
utilization reduces infant mortality rate by 4.3 and 7.1 percent, respectively
(Chaudhury et al., 2006).
The reviewed literature revealed that there have been several factors affecting
regional differentials in infant and under-five child mortality; however, prior
studies available on this domain in Ethiopia have not given due emphasis on
examining determinants of regional disparities in infant and under-five child
mortality. Therefore, given the lack of empirical evidence on the relative
101
individual contribution of determinants to regional differentials in under-five

child mortality, there is a need to systematically examine the major drivers of
inter-regional differences in under-five child mortality in Ethiopia. The present
study, therefore, aims at quantifying and identifying the major factors
responsible for inter-regional differentials in under-five child mortality levels
in rural settings of Ethiopia.
3. Data and Methods

3.1 Data source
The study uses data from the Ethiopian “Demographic and Health Survey”
2016. The data are a cross-sectional and large-scale health survey carried out
in nationally representative sample households across all regions of the
country. The survey employed a multistage cluster sampling procedure to
select sample households that are nationally representative. Altogether, a total
8,881 households were selected. However, the present study was delimited to
a total of 5,481 households from nine administrative regions of rural Ethiopia.
There were a total of 5,437 under-five children ever born at the national level.
In this study, about 1,295 number of rural under-five deaths were considered
for further analysis after excluding those missing values for the variables
included in the regression analysis. Details of sampling procedure, data
collection tools, and sample design are available in the report of the CSA and
ICFInternational (2016).
3.2 Outcome variable
Analysis of this study was limited to rural children whose age is between 0-59
months as a primary health outcome variable (dependent variable), defined as
the probability of a child dying by age under-five years per thousand births
(CSA & ICFInternational, 2016). While examining the association between
under-five child mortality and explanatory variables, the unit of analysis was
number of under-five child deaths.
102
3.3 Explanatory variables (covariates)
Several previous studies have indicated that the importance of various

determinants that affects infant and under-five child mortality across various
countries (Akuma, 2013; Caldwell, 1979; CSA & ICFInternational, 2012;
CSA & Macro, 2006; Dev et al., 2016; Gupta, 1997; Hong et al., 2009;
Khadka et al., 2015; Mosley & Chen, 1984; Negera et al., 2013; Regassa,
2012). Hence, the inclusion of a set of explanatory variables in the analyses
was mainly guided by these previous studies and availability of data on these
potential explanatory variables. In the analytical framework employed in the
study analysis, these covariates are grouped into three distinct classifications:
I) proximate determinants such as the age of the child, gender of the child,
multiplicity of birth, birth-order, birth size, birth spacing, and mother’s age at
birth. II) socioeconomic determinants such as mother’s use of modern
contraceptives, antenatal visits, mother’s working status, mother’s and father’s
education level, sex and age of household head, household size, and
household’s wealth index as a proxy measure for household’s economic status.
III) Environmental determinants such as place of delivery, access to toilet
facilities, electricity facility, safe drinking water and household’s region of
residence.
3.4 An Oaxaca-Blinder decomposition model
Since the response variable is a count data variable, application of linear

regression models based O-B decomposition could not be an appropriate
technique of decomposition (Bauer et al., 2006). Thus, this warrant to use an
extended nonlinear decomposition technique to count data modeling approach
(Bauer & Sinning, 2008; Park & Lohr, 2010; Yun, 2004). The differences in
the average rate of under-five child mortality for any two groups (regions is in
the present context) can be explained by a set of independent variables
(O‘Donnell et al., 2008) and then are decomposed into two components.
Namely, i) the “explained component” is the part of the outcome measure
disparity due to differences in the magnitude of observable determinants
across the two regions (characteristics or covariates effect), labeled as EC). ii)
the “unexplained component” is the part of the outcome measure due to
differences in estimated effects of theses determinants across the two regions
103
(coefficients effect), labelled as UC) (Blinder, 1973; Fairlie, 2005; Oaxaca,

1973; Powers et al., 2011; Sen, 2014; Wagstaff et al., 2007).
Assume there are N number of under-five child deaths (U5MR•~ ) (indexed, €• =

1., ,., ‚• ) belonging to ℎ•household (h=1,. ,. ,. H)in R mutually exclusive and
collectively exhaustive regions, H• = 1,.,.,R, each region containing ‚•• , 0• is a
vector of j observable explanatory variables (as explained above), r0• represents
a vector of regression parameters to be estimated, and y0• denotes the error term.
Thus, following Bauer et al. (2006); Bauer and Sinning (2008); Park and Lohr
(2010); Yun (2004) and Sinning et al. (2008), the O-B decomposition of two
regions, continuing with Tigrai (TG) as a reference category and Harari (HR) as
a comparison regions for example is computed by:
∆x‡ˆ,‰†
„…† = Š‹(Œ•Ž†••
‘ ‡ˆ
) − Š‹(Œ•Ž†‘•• ‰† ) = ’“”•„…† (Œ•Ž†‘•• ‡ˆ |˜–‡ˆ −
–‡ˆ
“”–‡ˆ„…†(Œ•Ž†••†=‰†|˜–‰†+“”‡ˆ„…†(Œ•Ž†™‰‘=‰†|˜–‰†−
“”–‰†„…†(Œ•Ž†••‘=‰†|˜–‰† [01]
The first bracketed segment on the right-hand side of equations [01] represents
the “explained component”, the differences in U5MR due to differences in the
magnitude of observable characteristics across the two regions
(“characteristics effect or covariates effect”). The second bracketed segment
represents the “unexplained component”, the regional differences in under-
five child mortality rates due to effects of the estimated coefficient of the
observable attributes across the two regions (“coefficients effect”).
A separate decomposition analysis was performed for the nine regions

continuing with Tigrai region as a reference category to examine how much of
the overall regional disparity or the relative regional differentials specific to
one of the covariates ( 0• ) is attributable to differences in covariates
(covariates effect) and differences in returns of these covariates (coefficients
effect). the present discussion focused only on explained part of the
components gap (covariates effect) because influencing the behavioural
responses to the characteristics (captured by the coefficient effects) is more
complicated (Jann, 2008; O'Donnell et al., 2009; Oaxaca & Ransom, 1999).
104
The statistical analyses are computed using Stata version 14 by adopting the
“user-written mvdcmp Stata command” on nonlinear regression-based detailed
decomposition technique of average outcome differentials proposed by Powers
et al. (2011) and O‘Donnell et al. (2008).
4. Empirical results
The results of decomposition analysis show that of the regions being

compared with a benchmark of Tigrai region, Somali region seems
exceptional in that its aggregate, characteristics, and coefficients effects were
significantly smaller than in the case of the other regions. The results of
detailed decomposition analysis indicated that the relative contribution of
determinants to the regional differentials in under-five child mortality rates
differ significantly across groups of regional comparisons of Ethiopia (Table
1). The relative contribution of a determinant (factor) reflects the differences
between the groups of regional comparisons distributions of that covariate
(variable) and the differences in the magnitude of the association of the
variable with under-five child mortality (Van de Poel et al., 2009). Therefore,
among the socioeconomic determinants, the most important relative
contributions come from antenatal health care visits, maternal education,
households’ economic status, household size, and use of modern
contraceptive. The differences in the proportion of children born to mothers
have received antenatal healthcare services contributed a substantial 12, 9, 26,
55, 13, 37, and 32 percent to the explained Tigrai-BG, Tigrai-Harari, Tigrai-
Amhara, Tigrai-Oromia, Tigrai-Somali, Tigrai-Afar, and Tigrai-SNNP
regional gaps in under-five child mortality, respectively. On the contrary, the
antenatal visit has been found to reduce 8 percent of Tigrai-Gambella regional
under-five child mortality difference. Similarly, results of decomposition
analysis revealed that the differences in under-five child mortality for Tigrai-
BG, Tigrai-Somali, Tigrai-Afar, and Tigrai- Gambella regions were explained
by the proportional differences in children to mothers with no education,
which accounts for 3, 23, 6, and 21 percent of the total explained regional
differences. However, this covariate contributed significantly to a 79, and 3
percent reduction of the Tigrai-Oromia and Tigrai- SNNP regional
differentials in under-five child mortality, respectively. In the present study,
households’ economic status measured in households’ wealth index was the
105
most important socioeconomic determinants of the regional gap. The

differences in the proportion of children to households in the poorest third
index category contributed significantly to a 4, 12, 3, and 10 percent of the
explained Tigrai-Harari, Tigrai-Amhara, Tigrai-Oromia and Tigrai-SNNP
regional gaps in under-five child mortality, respectively. However, this factor
has also been found to significantly reduce 8 percent of the explained Tigrai-
Gambella regional gap.
Furthermore, results indicated that the proportion of children to larger

household size explained significantly the Tigrai-BG, Tigrai-SNNP, and
Tigrai-Amhara regional gaps in under-five child mortality by about 9, 4 and 9
percent, respectively. On the contrary, differences in the proportion of children
to households with relative larger household size were found to significantly
narrow down by about 10, 8 and one percent, respectively of the covariates
effect of Tigrai-Harari, Tigrai-Afar, and Tigrai-Gambella regional differences.
The regional gaps were also partly explained by differences in proximate
factors. The differences in the proportion of children of four or higher between
regions contributed a substantial 9, 4, 29 and 12 percent, respectively to the
explained Tigrai-Harari, Tigrai-Oromia, Tigrai-SNNP and Tigrai-Gambella
regional gap in under-five child mortality. However, its relative effect was the
reverse for the other groups of regional comparisons and was found to
significantly reduce 13, 10, 8 and one percent of the explained Tigrai-Amhara,
Tigrai-Somali, Tigrai-Afar and Tigrai-BG regional differences U5MR,
respectively. Similarly, the differences in the proportion of children whose
birth size are less than average constituted significantly 36 percent of the
Tigrai-Gambella regional gap while it has significantly reduced by less than
one percent of the explained Tigrai-Amhara and Tigrai-Somali regional gaps
in under-five child mortality rates. Interestingly, differences in children who
had an average birth size contributed significantly 15 percent of the explained
Tigrai-Gambella regional gaps in U5MR while it has also been found to
significantly reduce the Tigrai-BG regional gap by less than one percent. As
to the relative contribution of short birth spacing (birth spacing less than 24
moths), differences in proportional distributions of children of short birth
spacing contributed 17, 5, 53, 2, 39, and 3 percent, respectively of the
covariates effect in under-five child mortality for Tigrai-BG, Tigrai-Amhara,
Tigrai-Oromia, Tigrai-Somali, Tigrai-Afar and Tigrai-SNNP regions, and the
106
differences were statistically significant. However, unlike to the other groups

of regional comparisons, short birth spacing significantly narrow down by 34
percent of the Tigrai-Harari regional child mortality differences.
Most importantly, children born to mothers with less than 20 years age at first
birth contributed significantly 6 and .78 percent, respectively to the Tigrai-
Amhara and Tigrai-Somali regional gaps while it has also been found to
reduce 8 percent of the Tigrai-Gambella regional differences in under-five
child mortality. Also, the differences in the proportion of children who have
been delivered at home (out of health facilities) contributed significantly 3, 9,
10 and 3 percent of the Tigrai-Harari, Tigrai-Amhara, Tigrai-Somali, and
Tigrai-Gambella regional differences in under-five child mortality,
respectively.
More importantly, results of decomposition analysis revealed that the negative

relative contribution of the male under-five child shows that female of the
same cohort was in a less advantageous situation in terms of survival rate in
the other comparison regions except for Harari and Somali regions. Likewise,
the negative contribution of age of the child at the time of death in explaining
regional gaps in under-five child mortality indicated that children of the
comparison regions were relatively younger than children of the same cohort
of the references category (Tigrai) except for Harari and Amhara comparison
regions. However, results of decomposition analysis further indicated that no
statistically significant regional differences were observed due to differences
in the proportional distribution of access to improved toilet facility; electricity
facility and safe drinking water across the regional comparisons (see Table 1).
107
Table 1: Detailed decomposition of inter-regional differentials in under-five child mortality (between Tigrai RC and other regions)
Tigrai-BG Tigrai-Afar Tigrai-Amhara Tigrai-Oromia Tigrai-Somali Tigrai-SNNP Tigrai-Gambella Tigrai-Harari
Covariates region region region region region region region region
Contribution (in percentage)
Child’s age -1.15 -1.89 2.72 -11.31 3.03 -3.50 -2.93 2.64
Child =Female -.05 -0.06 -.01 -.12 .03*** -.39 -.01*** .51
Birth order>4 -.68 -7.60*** -13.02*** 4.47*** -9.96*** 29.14** 12.45** 8.63***
Birth size < average -.08 3.90 -.53** -.76 -.28** -.60 36.17*** 14.73
Birth size = average -.62** -4.44 .57 .61 .36 -.89 14.71** -4.31
Multiple birth .11 -0.31 -1.07 -1.52 -.33** -.47 .13 2.85
Short birth interval 16.68*** 38.88*** 5.06*** 53.15** 2.40** 31.27** -1.63** -33.76**
Maternal age at birth <20 -.05 12.63 5.78** -5.11 .79** -11.74 11.49 3.52
Maternal age at birth >35 .09 -6.04 -17.09** -9.34 -2.26 -1.60 -8.31*** 25.92
Contraceptive use -.46 14.66 -8.57* -.24 -10.14* -5.19 .68 9.22
Antenatal visits 11.72* 36.83* 26.62** 55.48* 12.89* 32.21* -8.12* 8.53*
Mother’s education 2.85** 6.15** -.90 -79.54** 22.97** -3.52** 21.42*** 3.08
Mother’s work status -5.63 .78 -.40 -16.31 -.17 -9.98 -12.15 9.53
Female HH head .04 -0.66 -.21 -.55 .69 .01 -6.31 -2.96
Age of HH head -20.96 -6.85 1.46 -1.74 -.96 -40.00*** -6.27 -14.74
Father’s education -.54 -17.57 -.28 -1.54 6.31 .67 -.54** -8.57
Poorest third 4.15 7.37 12.51* 2.92* -7.60*** 9.98* 3.90 4.44**
Middle third .09 -21.49 .07 17.29 -1.04 .52 -18.21** -6.28*
Household size 8.84*** -8.45*** 9.26*** 6.69 5.20 4.33*** -.81*** -9.99***
Toilet facility -.33 -32.14 -.61 -3.05 -.50 -2.37 -.16 -3.14
Electricity facility -1.34 -0.37 -2.21 1.32 -1.69 .04 3.08 -3.20
Home delivery -1.12 24.86 9.17*** 24.10 9.6*** -5.79 3.41* 2.70***
Safe drinking water 1.74 -4.58 -1.12 -4.00 2.86 14.32 -2.09 3.84
Source: Own computation, 2016 EDHS; Notes: The contribution of each covariate (characteristics) has been expressed in percentage. RC indicates the reference category. The relative contributions of
individual covariates can be positive (>0 percent) or negative (<0 percent) and can exceed 100 percent. A positive value (sign) shows the component contributes to the greater differentials of U5MR
between Tigrai and the other regional comparisons whereas a negative contribution designates the opposite. Asterisks denote the level of significance: *** p<0.01, ** p<0.05, and * p<0.1. HH
represents household
108
5. Discussions
While Ethiopia has made a remarkable improvement in reducing the overall

child mortality at the national level, evidence indicated that there were
variations in rates of progress across its administrative regions (Abebaw,
2013; CSA & ICFInternational, 2012; CSA & ORCMacro., 2006; UNDP,
2012). To author’ best knowledge, this study is the first to decompose the
major determinants of inter-regional differentials in under-five child mortality
into components gap (explained and unexplained parts). The results of the
negative binomial regression analysis indicated that most determinants have
the expected associations with the under-five child mortality rates and
supported by previous studies (Dejene & Girma, 2013; Khadka et al., 2015;
Regassa, 2012). The results of regression analysis show that there have been
substantial differences in estimated coefficients of all regressed determinants
on under-five child mortality, indicating substantial variations in degree of
effects on under-five child mortality across regions.
Identifying the factors that explain most inter-regional differentials in under-

five child mortality rates could help in minimizing the regional gaps and to
speed up the rate of reduction in under-five child mortality both at regional
and national levels of Ethiopia. The results of O-B decomposition analysis
indicated that there have been substantial regional variations in under-five
child mortality across regional comparisons. Only small part of regional gaps
in under-five child mortality was explained (28 percent), being the lowest in
Tigrai and Benishangul-Gumuz regions (12 percent) and the highest in Tigrai-
Gambella regional comparisons (37 percent).
However, the substantial part of the regional differentials in under-five child

mortality remained unexplained (72 percent), range from 62 percent (for
Tigrai- Gambella regions) to 88 percent (for Tigrai-BG regions) which entails
due attention. The results of decomposition analysis also indicated the
substantial differences in socioeconomic, proximate and environmental
determinants in explaining the regional gaps with socioeconomic factors being
the major determinants of regional differentials in under-five child mortality
followed by proximate. factors. More specifically, results of the detailed
decomposition analysis reported the specific relative contribution of
109
determinants to the regional gaps in under-five child mortality. The

differences in the proportion of children born to mothers who have received
antenatal healthcare services contributed a substantial to the explained
regional gaps in under-five child mortality with different magnitude of effect
and significance levels across regions. Evidence indicated that though the
trends in antenatal health care coverage shows increasing rate, there has been
wide disparities observed across regions of Ethiopia, ranging from the lowest
41 percent in Somali to the highest 100 percent in SNNP, Harari, Oromia and
Tigrai regions (FMoH, 2014a). In low and middle-income countries, the
socioeconomic disparities in child mortality are the key public health problem
(Houweling & Kunst, 2010). Women education was considered as a major
determinant factor of reducing under-five child mortality (Caldwell, 1979).
Likewise, the contribution of the proportion of children to women with no
education constituted to regional gaps in under-five child mortality for most
regional comparisons. Comparable regional disparities in child mortality were
reported in other developing countries. For example, in Iran, mother’s level of
education contributed 21 percent of the regional differences in infant mortality
rates (Hosseinpoor et al., 2006), in Nepal, 4 percent of the explained regional
differentials in under-five child mortality was attributed to mother’s level of
education (Goli et al., 2015). Also, in Nigeria, there have been regional child
mortality differentials due to differences in women’s education level (Adedini
et al., 2015). Moreover, Jhamba (1999) and Akuma (2013)indicated that
maternal education was the major determinants of regional variation in child
mortality in Zimbabwe and Kenya, respectively.
The most striking regional differentials almost across the groups of regional
comparisons occurred due to differences in the proportion of children from the
poorest third index households. In line with present study, significant
difference in child mortality was observed due the major difference in
households’ wealth index in Nigeria (Adedini et al., 2015), Nepal (Goli et al.,
2015), Kenya (Akuma, 2013), and in Iran (Hosseinpoor et al., 2006). A
mother who gave a birth at less than 20 years old could face delivery and
pregnancy related problems due to the mother’s biological immaturity. Also,
the mother could not have basic knowledge on how to care babies (Pandey et
al., 1998) and as a result, a child born to this mother could have more likely to
significantly die than a child of a mother whose age is above 20 years (Babson
110
& Clarke, 1983). In the present study, the differences in distribution of

maternal age at first birth less than 20 years was among the major
determinants of regional differentials in under-five child mortality with
different magnitude of effects and level of significant. In line with the present
findings, it was also evident that the proportional differences in children of
mothers whose age at first birth less than 20 years across regions explained the
regional variations in child mortality significantly in Nepal (Goli et al., 2015),
and in Nigeria (Adedini et al., 2015). How a child birth order determines child
mortality and explains regional gaps in child mortality? A child of the first
order is most probably to born from a young woman who is not biologically
ready to accept and care for a baby. On top of this, the young woman has very
limited basic knowledge on how to care for a baby (NIMS et al., 2012; Pandey
et al., 1998). A child of higher birth-order, in contrast, is most probably to
born to an older woman and is likely to be influenced by competition from
older siblings in terms of resources (NIMS et al., 2012). Hence, in the present
study, higher birth-order was among the major proximate determinants of
under-five child mortality. The differences in the proportional distribution of
children of birth-order of four or higher across the regions explained the
regional gaps in under-five child mortality with different magnitude of effects
and levels of significant. The present finding was consistent with some
previous studies from Nepal (Goli et al., 2015), and in Nigeria (Adedini et al.,
2015). Prior studies have indicated that birth spacing and child mortality has a
direct relationship (Srinivasan, 2000; Sweemer, 1984). A woman who
experienced short birth spacing may not recover instantly her health and then
can deter baby’s growth. Therefore, a child born to less than 24 moths birth
spacing (short birth spacing) have more likely to die than a child born to a
birth spacing of more 24 months (Hobcraft et al., 1983; NIMS et al., 2012).
Likewise, a child born to less than 24 months birth spacing had more likely to
die. The differences in the proportional distribution of children of short birth
spacing across regions explained the regional differences in U5MR for most
regional comparisons with different degree of effects and levels of significant.
Findings on short birth spacing were in line with some of the existing
literature in Iran (Hosseinpoor et al., 2006), Nigeria (Adedini et al., 2015)and
in Nepal (Goli et al., 2015).Findings of this study further indicated that birth
size less than average (2500g) affects under-five child mortality across
regions. The differences in the distribution of birth size less than average
111
explained significantly to 39 percent of the regional variations in under-five

child mortality for Tigrai-Gambella regions. In Sri Lanka, low birth-weight
explained the inter-district disparity in infant mortality rate (Chaudhury et al.,
2006). However, the present findings revealed that for most regions child size
at birth less than average contributed to reducing the regional differentials in
under-five child mortality with small size effect. Furthermore, the unequal
distributions of children who have been delivered at home (out of health
facilities) attributed significantly to the explained regional gap in under-five
child mortality; however, the relative percentage contribution of this variable
was small. This result was in line with previous empirical studies from Nigeria
(Adedini et al., 2015).
6. Conclusions
The present study has identified the inter-regional differentials in under-five

child mortality in rural Ethiopia was due to different levels of determinants
that are often associated with under-five child mortality. The results of
decomposition analysis indicated that households’ economic status, mothers’
levels of education, birth-order, birth-spacing, antenatal visits, household size,
and place of delivery attributed were the key determinants of regional
disparities in under-five child mortality. The under-five child mortality
disparities were largely due to the reflection of the wide regional differentials
of these determinants. The findings of this study can help to draw a critical
attention in developing specific national and regional policy based on the
relative contribution of individual covariates to explained regional gaps that
help in reducing child mortality disparities among regions of the country.
Hence, on top of strengthening the Ethiopian health extension programme
across regions, this study suggests that addressing those identified potential
determinants focusing on improving households’ economic status and
women’s education could help to minimize regional disparities in under-five
child mortality and ensure universal health care coverage of the country. Also,
further sustained effort is needed to speed up the rate of reduction in under-
five child mortality both at the national and regional levels against a certain
target set, for example, 75 percent disparity reduction goal in under-five child
mortality among regions in 2025.
112
References
Abebaw, D. (2013). Infant and child health in Ethiopia: Reflections on regional

patterns and changes. Journal of International Development, 25(4), 536-548.
doi:10.1002/jid.1842.
Adedini, S. A. , Odimegwu, C. , Imasiku, E. N. , Ononokpono, D. N. , & Ibisomi, L.
(2015). Regional variations in infant and child mortality in Nigeria: a
multilevel analysis. Journal of biosocial science, 47(02), 165-187.
doi:10.1017/S0021932013000734
Aigbe, G. o. , & Zannu, A. E. (2012). Differentials in infant and child mortality rates
in Nigeria: Evidence from the six Geopolitical Zones. International Journal
of Humanities and Social Sciences, 2(16), 206-214.
Akuma, J. M. (2013). Regional Variations of Infant Mortality in Kenya: Evidence
from 2009 KDHS Data. Mediterranean Journal of Social Sciences, 4(9), 425 -
433 doi:10.5901/mjss.2013.v4n9p425
Amouzou, A. , Kozuki, N. , & Gwatkin, D. R. (2014). Where is the gap?: the
contribution of disparities within developing countries to global inequalities
in under-five mortality. BMC public health, 14(1), 216 - 220.
doi:10.1186/1471-2458-14-216
Babson, S. G. , & Clarke, N. G. (1983). Relationship between infant death and
maternal age: comparison of sudden infant death incidence with other causes
of infant mortality. The Journal of pediatrics, 103(3), 391-393.
doi:10.1016/S0022-3476(83)80409-0
Bedane, A. S. , A., T. F. , Shamenna, A. T. , & Abshoko, A. D. (2016). Variations in
Under-five Child Mortality among Regional States of Ethiopia: A Multi-level
Modelling Approach. British Journal of Applied Science & Technology,
15(2), 1-16. doi:10.9734/BJAST/2016/24448
Blinder, A. S. (1973). Wage Discrimination: Reduced Form and Structural Estimates.
Journal of Human Resources, 8(4), 436 - 455. doi:10.2307/144855
Caldwell, J. C. (1979). Education as a factor in mortality decline an examination of
Nigerian data. Population studies, 33(3), 395-413. doi:10.2307/2173888
Chaudhury, R. H. , Gunasekera, P. , & Gunasekera, D. (2006). Child Health. World
Health Day Message from the Regional Director, 10(1), 96 - 103.
CSA , & ICFInternational. (2016). Ethiopia Demographic and Health Survey 2016.
Addis Ababa, Ethiopia and Calverton, Maryland, USA: Central Statistical
Agency and ICF International.
_______. (2012). Ethiopia Demographic and Health Survey 2011. Addis Ababa,
Ethiopia and Calverton, Maryland, USA: Central Statistical Agency and ICF
International. 110 - 153.
113
CSA , & Macro, O. (2006). Ethiopia Demographic and Health Survey 2005. Addis
Ababa, Ethiopia and Calverton, Maryland, USA: Central Statistical Agency
and ORC Macro.
Ethiopia and Calverton, Maryland, USA: Central Statistical Authority and
ORC Macro., 97 - 139.
Ethiopia and Calverton, Maryland, USA: Central Statistical Agency and
ORC Macro. 103 - 206.
Dejene, T. , & Girma, E. (2013). Social determinants of under-five mortality in
Ethiopia: Event history analysis using evidence from Ethiopian Demographic
and Health Survey (EDHS). Health 5(5), 879 - 884.
doi:10.4236/health.2013.55115
Demombynes, G. , & Trommlerová, S. K. (2012). What has driven the decline of
infant mortality in Kenya?
Dev, R. , Williams, M. F. , Fitzpatrick, A. L. , & Connell, F. A. (2016).
Topographical Differences of Infant Mortality in Nepal. Kathmandu
University medical journal (KUMJ), 14(54), 96.
Fairlie, R. W. (2005). An extension of the Blinder-Oaxaca decomposition technique to
logit and probit models. Journal of economic and social measurement, 30(4),
305-316.
FMoH. (2010). Health Sector Development Plan IV 2010/11 – 2014/15. Addis Abba
2010. Federal Democratic Republic of Ethiopia Ministry of Health. .
_______. (2014a). Ethiopia's Fifth National Health Accounts 2010/2011. Addis
Ababa, Ethiopia.
_______.(2014b). Ethiopia Ministry of Health, Ethiopian Public Health Institute,
United Nations Children’s Fund, Countdown to 2015: Ethiopia's progress
towards reduction in under-fve mortality. UNICEF Ethiopia, Addis Ababa.
Ghaffar, A. , & Bhuyan, K. C. (2000). Regional variations in child mortality in north-
eastern Libya. Journal of Family Welfare, 46(1), 47-56.
Goli, S. , Bhandari, P. , Atla, U. M. R. , & Chattopadhayay, A. (2015). Childhood
Mortality Differentials by Ecological Region in Nepal. Population, Space and
Place, 23(e1977). doi:10.1002/psp.1977
Gupta, M. D. (1997). Socio-economic status and clustering of child deaths in rural
Punjab. Population studies, 51(2), 191-202.
doi:10.1080/0032472031000149906
Hobcraft, J. , McDonald, J. W. , & Rutstein, S. (1983). Child-spacing effects on infant
and early child mortality. Population Index, 49(4), 585-618.
Hong, R., Ayad, M., Rutstein, S., & Ren, R. L. (2009). Childhood mortality in
Rwanda: levels, trends, and differentials. DHS Further Analysis Report(66).
114
Hosseinpoor, A. R., Van Doorslaer, E., Speybroeck, N., Naghavi, M. , Mohammad,

K. , Majdzadeh, R., Vega, J. (2006). Decomposing socioeconomic inequality
in infant mortality in Iran. International Journal of Eepidemiology, 35(5),
1211-1219.doi:10.1093/ije/dyl164
Houweling, T. A. J. , & Kunst, A. E. (2010). Socio-economic inequalities in childhood
mortality in low-and middle-income countries: a review of the international
evidence. British medical bulletin, 93(1), 7-26. doi:10.1093/bmb/ldp048
Jann, B. (2008). The Blinder-Oaxaca decomposition for linear regression models. The
Stata Journal, 8(4), 453-479.
Jhamba, T. (1999). Regional variations in childhood mortality in Zimbabwe.
Geography, 84(4), 319-330.
Kabir, A., Islam, M. S., Ahmed, S. M. S., & Barbhuiya, K. (2001). Factors influencing
infant and child mortality in Bangladesh. J Med Sci, 5, 292-295.
Khadka, K. B., Lieberman, L. S., Giedraitis, V. , Bhatta, L. , & Pandey, G. (2015).
The socio-economic determinants of infant mortality in Nepal: analysis of
Nepal Demographic Health Survey, 2011. BMC pediatrics, 15(1), 152.
Khosravi, A. , Taylor, R. , Naghavi, M. , & Lopez, A. D. (2007). Differential
mortality in Iran. Population health metrics, 5(7), 478 - 493.
Macassa, G. , Ghilagaber, G. , Charsmar, H. , Walander, A. , Sundin, Ö. , & Soares,
J. (2012). Geographic differentials in mortality of children in Mozambique:
their implications for achievement of Millennium Development Goal 4.
Journal of Health, Population and Nutrition, 30(3), 331-345.
Mosley, W. H. , & Chen, L. C. (1984). An analytical framework for the study of child
survival in developing countries. Population and development review, 10, 25-
45.
Negera, A. , Abelti, G. , Bogale, T. , Gebreselassie, T. , & Pearson, R. (2013). An
analysis of the trends, differentials and key proximate determinants of infant
and under-five mortality in Ethiopia. ICF International: Calverton, Maryland
USA.
NIMS, ICMR , & UNICE. (2012). Infant and Child Mortality in India: Levels, Trends
and Determinants, National Institute of Medical Statistics (NIMS), India
Council of Medical Research (ICMR), and UNICEF India Country Office,
New Delhi, India.
O'Donnell, O. , Nicolás, Á. L. , & Van Doorslaer, E. (2009). Growing richer and
taller: Explaining change in the distribution of child nutritional status during
Vietnam's economic boom. Journal of Development Economics, 88(1), 45-
58. doi:10.1016/j.jdeveco.2008.01.004
O‘Donnell, O. , Doorslaer, E. V. , Wagstaff, A. , & Lindelow, M. (2008). Analysing
Health Equity Using Household Survey Data: A Guide to Techniques and
115
Their Implementation. Washington, DC: World Bank Institute Learning

Resource Series, the World Bank.
Oaxaca, R. (1973). Male-Female Wage Differentials in Urban Labor Markets.
International economic review, 14(3), 693 - 709. doi:10.2307/2525981
Oaxaca, R. L. , & Ransom, M. R. (1999). Identification in detailed wage
decompositions. Review of Economics and Statistics, 81(1), 154-157.
doi:10.1162/003465399767923908
Pandey, A. , Choe, M. K. , Luther, N. Y. , Sahu, D. , & Chand, J. (1998). Infant and
child mortality in India. National Family Health Survey Subject Reports No.
11. International Institute for Population, Mumbai.
Park, T. A. , & Lohr, L. (2010). A Oaxaca–Blinder decomposition for count data
models. Applied Economics Letters, 17(5), 451-455.
doi:10.1080/13504850801964307
Patel, T. A. , & Sharma, D. B. (2013). Inter-state variation in Neonatal Mortality Rate
among Indian states. Community Med, 4(1), 54-58.
Powers, D. A., Yoshioka, H. , & Yun, M. (2011). mvdcmp: Multivariate
decomposition for nonlinear response models. The Stata Journal, 11(4), 556-
576.
Regassa, N. (2012). Infant Mortality in the Rural Sidama Zone, Southern Ethiopia:
Examining the Contribution of Key Pregnancy and Postnatal Health Care
Services. Journal of Nursing, Social Studies, Public Health and
Rehabilitation 1(2), 51 – 61.
Shyama K., Julian S., David B., Sadia ., Daniele C.F. , Rafael C., Bustreoh, F.
(2014). Success factors for reducing maternal and child mortality. 533 - 544.
doi:10.2471/BLT.14.138131
Sinning, M. , Hahn, M. , & Bauer, T. K. (2008). The Blinder-Oaxaca decomposition
for nonlinear regression models. Stata Journal, 8(4), 480 - 492.
Srinivasan, S. (2000). Determinants of infant and child mortality in Tamil Nadu and
Uttar Pradesh. International Institute for Population, Mumbai, 18-23.
Sweemer, C. D. (1984). The influence of child spacing on child survival. Population
studies, 38(1), 47-72. doi:10.1080/00324728.1984.10412822
Tesfa, B. , & Jibat, N. (2014). Tesfa, B., & Jibat, N. (2014). Health extension program
as innovative health care service: The socio-cultural factors affecting its
implementation in Jimma Zone, South Western Oromia. International
Journal of Sociology and Anthropology, 6(8), 235-245.
doi:10.5897/IJSA2014.0556
UN. (2010). The Millennium Development Goals Report, 2010.
UNDP. (2012). Analyzing Regional Performance and Disparities in Health Outcomes
in Ethiopia: Development Brief, UNDP, Ethiopia.
116
UNICEF. (2014). The State of the World’s Children 2015: Reimagine the Future:
Innovation for Every Child.
_______. (2015a). Committing to Child Survival: A Promise Renewed. Progress
Report 2015, UNICEF New York.
_______. (2015b). Committing to Child Survival: A Promise Renewed. United
Nations Children’s Fund (UNICEF) Progreess Report.
UNIGME. (2011). Levels and Trends in Child Mortality, Report 2011, Estimates
Developed by the UN Inter-agency Group for Child Mortality Estimation
(UNIGME), UNICEF, World Health Organization, World Bank Group, and
United Nations, New York.
_______. (2012). Levels and Trends in Child Mortality, Report 2012, Estimates
United Nations, New York:
Wagstaff, A., O'Donnell, O. , Van Doorslaer, E. , & Lindelow, M. (2007). Analyzing
health equity using household survey data: a guide to techniques and their
implementation. World Bank Publications.
Yun, M. (2004). Decomposing differences in the first moment. Economics Letters,
82(2), 275-280.
117

6th Tigray Proceedings

Uploaded by

Copyright:

Available Formats

6th Tigray Proceedings

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

6th Tigray Proceedings

Uploaded by

Copyright:

Available Formats

Ethiopian Economics Association

© Ethiopian Economics Association (EEA)

All rights reserved

Technical, Allocative and Economic Efficiencies and Sources of

Desale Gebretsadik, Jema Haji, and Bosena Tegegne

Comparative Analysis of Technical Efficiency of Row Plantation

Gabriel Temesgen, Yohannes Hailu and Kibrom G/Kirstos

Implication of Animal Feed and Water Scarcity on Labor

Yibrah Hagos Gebresilassiea and Phocenah Nyatanga

The Annual Regional Conferences that the Association has organized in

At this juncture, I would like to take this opportunity to express my heartfelt

Finally, I would like to extend my sincere gratitude to officials of the regions

Tadele Ferede (PhD)

Desale Gebretsadik1, Jema Haji2, and Bosena Tegegne3

This study analyzed production efficiency and sources of inefficiency differentials of

Agriculture is the most important sector of Ethiopia’s economy where about

Sesame is one of the important oilseed crops adapted to semi-arid tropical

Given agriculture as backbone of the nation’s food security and as sesame is

Description of the Study Area: The study was conducted in Kafta-Humera

Where n = sample size, N= total large-scale household equal to 1,130. Based

Table 1: Number of sampled producers from each kebele

3. Methods of Data Analysis

Production Efficiency and Sources of Inefficiency Differentials

Production function could be either Cobb-Douglas or translog that requires

Variables of sesame production efficiency: These variables are inputs in

Where ld= land, sd = seed, lb= Labor, pw = Oxen/tractor power, fq = fertilizer

This refers to production cost incurred by producer’s calculated taking price of

ln c = ∂ +∂ ln cld + ∂ ln csd + ∂ ln clb + ∂ ln cft +∂ ln cpw + ∂ ln cmt + ∂ ln cop + (ν − u )

∆ci / ∆pi = x (p ie i , Y *i , α ) (3. 8)

So, optimization profit principle is to minimize cost subject to optimum

Α = Exp β 0 (3. 11)

TE i = Y i / Y i = ∑i xit, Pi / ∑i xi, Pip

Sources of sesame production inefficiency

Ui = σ0 +σ1i w1i + σ 2i w2i + σ3i w3i + ...+σ15i w15i

Estimation of best production function

LR = λ = − 2 ln [L (H 0 ) / L (H 1 )] = − 2 [ln L (H 0 ) − ln L (H 1 )] (3. 17)

λ = δu2 / δν2...or....δu / δν , (3.18) γ = δu2 / δ 2

δ2 =δu2 +δν2 (3. 20)

Given the specification of SFA, inefficiency is present is defined by Ha: ≠ 0.

4. Results and Discussion

Demographic Features and Availability of Production Resources

Table 2: Household characteristics of large-scale sesame producers

The sampled sesame producers used hired labor at different production

Table 3: Amount of sesame allocated for different purposes

Summary Statistics of Sesame Production Inputs and Costs

Table 4.Summary of total production inputs and costs

Estimation of Production Function parameters

Table 5: SFA parameter coefficient for sesame production by large-scale

Elasticity of sesame production: The summation of production inputs’

Table 6:Tests of cost function model validity

Material cost includes cost of agricultural materials, sack, harvesting and

Table 7: Sesame production cost parameters

Estimation of technical, allocative and economic efficiencies of large-scale

Table 8: Category of sampled sesame producers based on their TE

To give a better picture about TE distributions, a frequency distribution is

Allocative efficiency: The average AE of large-scale sesame producers was

Table 9: Distribution of AE of sesame producer categories

Table 10: Distribution of EE by large-scale sesame producers

Sources of technical, allocative and economic inefficiency of large-scale

>,() >,( ∗) >[∅(/ ) − ∅(/ )] >[1 − >(/ )]