Greener Journal of Agricultural Sciences

Vol. 9(1), pp. 76-85, 2019

ISSN: 2276-7770
Copyright ©2019, the copyright of this article is retained by the author(s)
DOI Link: http://doi.org/10.15580/GJAS.2019.1.122118180
http://gjournals.org/GJAS

Impact of Improved Wheat Technology

Package Adoption on Productivity in
Ethiopia
Baye Belay, Fitsum Daniel* and Eyob Bezabeh
Ethiopian Institute of Agricultural Research, Addis Ababa, Ethiopia.

ARTICLE INFO ABSTRACT

Article No.: 122118180 This study examines the impact of adoption of improved wheat technology
Type: Research package (including improved wheat varieties, information regarding
DOI: 10.15580/GJAS.2019.1.122118180 improved wheat management practices as well as artificial/chemical
fertilizer) on productivity using 1,611 sample farm households in four major
administrative regions of Ethiopia. Propensity score matching (PSM)
Submitted: 21/12/2018 technique was employed since it is an increasingly utilized standard
Accepted: 27/12/2018 approach for evaluating impacts using observational data of a single
Published: 06/03/2019 period. It is found that full adoption of improved wheat technology package
appears to significantly increase productivity growth on the average by 51
*Corresponding Author to 55% for farm households in the study area. Thus, the study recommends
Fitsum Daniel that full adoption of improved wheat technology package could be an
E-mail: fitsum.daniel219 effective strategy to enhance productivity and, thereby, production that
@gmail.com contributes a lot to the structural transformation of the Ethiopian economy.
Phone: 251-0913 38 45 38

Keywords: impact; wheat;

improved technology package;
Ethiopia

Greener Journal of Agricultural Sciences, vol. 9, no. 1, pp. 76-85, 2019

 Baye et al / Greener Journal of Agricultural Sciences 77

INTRODUCTION On the other hand, most smallholder farmers

(i.e. 59% of total cultivated area) reside in the moisture
Like in many other sub-Saharan Africa countries, reliable cereal-based highlands among the five agro-
agriculture in Ethiopia is a basis for the entire ecological regions of Ethiopia distinguished by
socioeconomic structure of the country and has a major agricultural researchers (Taffesse et al., 2012).
influence on all other economic sectors and Accordingly, with farmers using virtually no irrigation,
development processes and hence it plays a crucial role reliable rainfall is an important condition to achieve good
in poverty reduction (Elias et al., 2013; GebreEyesus, agricultural productivity. In relation to this, as to the
2015). Despite the marginal decline in its share of GDP same source document, the Meher rainfall season is
in recent years, it is still the single largest sector in terms overwhelmingly important as it contributes about 96.9%
of its contribution to GDP as agricultural GDP constitutes of total crop production and 95.5% of total cereal
41% of total country's GDP (CSA, 2014/15). production in 2007/08.
As to Gebru 2006 citing CSA 2003, out of the With respect to all these facts, it is not
total production of agriculture, about 70% comes from questionable that accelerated and sustained growth in
crop production. According to Abegaz 2011, cereal the country‟s agriculture in general and in the crop sub-
crops constitute the largest share of farming household‟s sector in particular with special emphasis to the small-
production and consumption activities. Accordingly citing scale farmers will greatly help to achieve the various
Alemayehu et al., 2009, only five major cereals (barley, goals of the country (Gebru, 2006; MoFED, 2003;
maize, sorghum, teff and wheat) account for about 70% Gebre-Selassie & Bekele).
of area cultivated and 65% of output produced. Fertilizer Moreover, food needs as well as the industrial
use is also concentrated on cereals followed by pulses demand for agricultural products increase due to
and oilseeds respectively according to Endale 2011 population growth (Bor and Bayaner, 2009). All these
citing CSA 1995/96-2007/08. On the other hand, needs, according to them, require an increase in the
according to Endale 2011, data from the Ethiopian Seed agricultural production. The growth in agricultural
Enterprise show that improved seeds are mostly used in production in sub-Saharan Africa in the past was
wheat and maize cultivation with an average of 89 and achieved by expanding the amount of land cultivated
42 thousand quintal in the period 1994/95 to 2005/06, (Gebru, 2006). In relation with this, it is well known that
respectively. Moreover, Abegaz 2011 citing the in our country there are regions where there are large
Household Income, Consumption and Expenditure populations but limited land and vice versa (MoFED,
Survey of CSA indicated that the five major cereal crops 2003). Accordingly, most of the lands available for
account for 46% of household‟s total consumption. settlement are found in the lowlands that lack basic
Therefore, a closer look at what is happening in cereal infrastructural facilities and pose serious health hazards.
production has an important welfare and policy With little suitable land available for expansion of crop
implication in Ethiopia (Abegaz, 2011). cultivation, especially in the highlands, future cereal
According to Ketema and Kassa 2016 citing production growth will need to come from increasing
Shiferaw et al. 2013, wheat contributes about 20% of the land productivity mainly through the supply, duplication
total dietary calories and proteins worldwide. Ethiopia is and diffusion of continuously improving technology and
the second largest wheat producer in sub-Saharan information (Ayele et al. 2006 citing Reardon et al 1996;
Africa next to South Africa (Nigussie et al., 2015). Mann Taffesse et al. 2012; Elias et al. 2013; Matsumoto and
and Warner 2017 citing Minot et al. 2015 indicated that Yamano, 2010).
there are approximately 4.7 million farmers growing Cognizant of these as well as the fact that
wheat on approximately 1.6 million hectares productivity is the major component of growth and a
representing between 15 and 18% of total crop area and fundamental requisite in many form of planning
less than 1% of all wheat production takes place outside irrespective of the stage of development and economic
the four main regions of Ethiopia according to recent and social system as to Gebru 2006 citing Cheema
estimates. Wheat is one of the major staple crops in the 1978, the national wheat research program has released
country in terms of both production and consumption and disseminated a number of bread and durum wheat
(Kelemu, 2017). According to Kelemu 2017 citing FAO varieties since the 1950s and 1960s as to Ketema and
2014, it is the second most important food in the country Kassa 2016 citing Tesfaye et al. 2001. According to the
behind maize in terms of caloric intake. same source citing CSA 2015b, a closer look at the
The Ethiopian agricultural sector, as to Gebru proportion of the area covered by improved varieties of
2006 citing EEA 2004, is dominated by small-scale different crops showed that wheat took the second rank
farmers cultivating about 96% of the total area under (7.4%) next to maize (46.4%) among cereals. Given the
crop, producing more than 90% of total agricultural emphasis of increasing crop production through higher
output and 97% of food crops. With these statistics, one fertilizer use, import of chemical fertilizer augmented
can easily infer to what extent the small-scale farmers from 246,722 MT in 1995 to 375,717 MT in 2006 despite
are the key element in strengthening the effort towards the removal of fertilizer subsidies since 1997/98
agricultural growth and consequently to the overall according to Endale 2011 citing MOARD 2007/08. In this
economic growth (Gebre-Selassie & Bekele). regard, according to Ketema and Kassa 2016 citing CSA
78 Baye et al / Greener Journal of Agricultural Sciences

2015b, wheat is the most fertilized crop (82%) among all far would have low probability of influencing national and
crops and pesticide application is also most common on regional policies. Moreover, the focus of most studies
wheat as compared to that on other cereal crops. was measuring the impact of a single improved
Even though crop productivity and production agricultural technology or information rather than of a
remained low and variable in the 90s for the most part, package of agricultural technologies and information.
there have been clear signs of change over the past Thus, the objective of this study is to identify the impact
decade (Abate et al., 2015). According to Kelemu 2017, of adoption of improved wheat technology package
the average level of wheat productivity for the period of (including improved wheat varieties, information
2000-2014 is about 1.73 ton/ha while the average regarding improved wheat management practices as
growth rate in productivity is about 5.93%. During the well as artificial/chemical fertilizer) on wheat productivity
same period, total wheat production has been increasing per unit of land cropped in the four major administrative
at 10.14% growth rate per annum (Kelemu, 2017). regions of Ethiopia which are also known to be the major
As to Tsusaka and Otsuka 2013 citing FAO wheat producing regions in the country.
2011, although the production of staple food has been
increasing in sub-Saharan Africa, the rate of increase
has not been high enough to outstrip its high population MATERIALS AND METHODS
growth rate as a result of which per-capita agricultural
production in the region has declined by about 10% Analytical Framework for Evaluation of Adoption of
since 1960. These all obviously calls for a further and a Wheat Technology Package Impact on Productivity
better growth in agricultural productivity as well as
quality with minimum adverse impact on the The correct evaluation of the impact of a treatment like
environment. Kelemu 2017 citing Shiferaw and Okelo adoption of a technology package will require identifying
2011 indicated that of the cereals whose production is the “average treatment effect on the treated” defined as
soon likely to exceed domestic demand requirements, the difference in the outcome variables between the
wheat is the commodity that will most easily find an treated objects like farmers and their counterfactual. A
export market to supply. In view of this prospect, counterfactual is defined as “knowledge of what would
according to him, the need for increasing productivity of have happened to those same people if they
wheat is very crucial. simultaneously had not received treatment” (Olmos A.,
Holistic and appropriate evaluation of the efforts 2015 citing Shadish et al., 2002). In this context, as to
and corresponding results as well as reasons/strengths González et al. 2009, if Y represents the outcome
and weaknesses/ of the past few decades in general and variable and if D is a dummy variable that takes the
of the past recent years in particular is necessary in value of 1 if the individual was treated and 0 otherwise,
order to create a more fertile ground for the fast the “average treatment effect on the treated” will be
achievement of the aforementioned goal. In this regard, given by:
the role of historical data collected by different agencies
like CSA as well as of different socio-economic studies (1) TATT= E[Y (1) / D =1]− E[Y (0) / D =1]
carried out to provide vital policy and related
recommendations is indispensable. Studies that assess However, accordingly, given that the counterfactual (E[Y
the contribution of improved crop management practices (0) / D = 1]) is not observed, a proper substitute has to
information and technologies like improved crop varieties be chosen to estimate TATT. Using the mean outcome of
for the productivity growth of such important and widely non-beneficiaries-which is more likely observed in most
cultivated cereals like wheat in Ethiopia in the past of the cases-do not solve the problem given that there is
recent years are among studies that can be cited in a possibility that the variables that determine the
relation to this. However, studies carried out in the treatment decision also affect the outcome variables. In
country on this issue are not only few but also restricted this case, the outcome of treated and non-treated
to piece meal or location specific approach. Besides, individuals might differ leading to selection bias
most studies were biased towards those locations that (González et al., 2009). To clarify this idea, the mean
had high/better suitability and/or preference for the outcome of untreated individuals has to be added to (1)
production of the specific crop considered. Thus, a from which the following expression can be easily
nationally or regionally representative data could not be derived:
collected for the studies and the conclusions drawn so

(2) TATT={E[Y (1) / D =1]− E[Y (0) / D =0]}−{E[Y (0) / D =1]− E[Y (0) / D =0]}

Here E[Y (0) / D= 1]−E[Y (0) / D= 0] represents the independent variable, that is, whether an individual
selection bias which will be equal to zero if treatment receives (or not) the intervention under scrutiny and (2)
was given randomly which can be achieved through the individuals are randomly assigned to the independent
use of experimental approach. variable. The first characteristic does not define the
The experimental approach, according to Olmos experimental approach: most of the so-called quasi-
A. 2015, has two characteristics: (1) it manipulates the experiments also manipulate the independent variable.
 Baye et al / Greener Journal of Agricultural Sciences 79

What defines the experimental method is the use of the treatment, one in the control group) sharing a similar
random assignment (Olmos A., 2015). However, due to propensity score are seen as equal, even though they
ethical or logistical reasons, random assignment is not may differ on the specific values of the covariates
possible as to Olmos A. 2015 citing Bonell et al. 2009. (Olmos A. 2015 citing Holmes 2014).
Moreover, accordingly, equivalent groups are not
achieved despite the use of random assignment which is Data and Variables
known as randomization failure. Usual reasons why
randomization can fail are associated with missing data The data utilized for this study is acquired from farm
which happened in a systematic way and sometimes can household survey undertaken during 2015/16 by
go undetected (Olmos A., 2015). Ethiopian Institute of Agricultural Research (EIAR) in
As a consequence of randomization failure, or collaboration with the International Maize and Wheat
because of ethical or logistical reasons, in a very large Improvement Center (CIMMYT). The sampling frame
number of real-world interventions, experimental covered seven major wheat growing agro-ecological
approaches are impossible or very difficult to implement. zones that accounted for over 85% of the national wheat
However, if we are still interested in demonstrating the area and production distributed in the four major
causal link between our intervention and the observed administrative regions of Ethiopia- Amhara, Oromia,
change, our options become limited. Some options Tigray as well as South Nations, Nationalities and
include regression discontinuity designs which can Peoples (SNNP). A multi-stage stratified sampling
strengthen our confidence about causality by selecting procedure was used to select villages from each agro-
individuals to either the control or treatment condition ecology, and households from each “kebele”/village.
based on a cutoff score. Another alternative is First, agro-ecological zones that account for at least 3%
propensity scores matching technique. Propensity of the national wheat area each were selected from all
scores matching is a statistical technique that has the major wheat growing regional states of the country
proven useful to evaluate treatment effects when using mentioned above. Second, based on proportionate
quasi-experimental or observational data (Olmos A., random sampling, up to 21 villages in each agro-
2015 citing Austin, 2011 and Rubin, 1983). Some of the ecology, and 15 to 18 farm households in each village
benefits associated with this technique, accordingly, are: were randomly selected. The data was collected using a
(a) Creating adequate counterfactuals when random pre-tested interview schedule by trained and
assignment is infeasible or unethical, or when we are experienced enumerators who speak the local language
interested in assessing treatment effects from survey, and have good knowledge of the farming systems.
census administrative, or other types of data, where we Moreover, the data collection process was supervised by
cannot assign individuals to treatment conditions. (b) experienced researchers to ensure the quality of the
The development and use of propensity scores reduces data.
the number of covariates needed to control for external
variables (thus reducing its dimensionality) and Productivity stands for the productivity of wheat per unit
increasing the chances of a match for every individual in of land cropped measured in kilogram per hectare.
the treatment group. (c) The development of a
propensity score is associated with the selection model, LnProductivity stands for the natural logarithmic
not with the outcomes model, therefore the adjustments transformation of Productivity.
are independent of the outcome. According to Olmos A.
2015, propensity scores are defined as the conditional HHAGE stands for the age of a household head.
probability of assigning a unit to a particular treatment
condition (i.e., likelihood of receiving treatment), given a HHSEX is a dummy variable indicating the sex of a
set of observed covariates: household head where HHSEX = 1 if the head is male
and 0 if otherwise.
(z = i |X)
FAMILY_SIZE stands for size of a household.
where z = treatment, i = treatment condition, and X =
covariates. In a two-group (treatment, control) HHEDU is a dummy variable indicating whether a
experiment with random assignment, the probability of household head is literate where HHEDU = 1 if the head
each individual in the sample to be assigned to the is literate/able to read and write/ and 0 if otherwise.
treatment condition is: (z = i│X)=0.5. In a quasi-
experiment, the probability (z = i│X) is unknown, but it CREDIT is a dummy variable indicating household's
can be estimated from the data using a logistic access to credit where CREDIT = 1 if the household has
regression model, where treatment assignment is got the credit it needed in 2013 and 0 if otherwise.
regressed on the set of observed covariates (the so-
called selection model). The propensity score then LANDHOLDING_SIZE stands for size of the land holding
allows matching of individuals in the control and of a household measured in hectare.
treatment conditions with the same likelihood of
receiving treatment. Thus, a pair of participants (one in
 80 Baye et al / Greener Journal of Agricultural Sciences

DSTMNMKT stands for distance to the nearest main between adopters and non-adopters of wheat
market from residence measured in kilometer. technology and information in terms of these
demographic factors as depicted by the descriptive
OXEN stands for the total number of oxen owned by a statistics. Male-headed households were found to be
household. more probable in adopting improved wheat technology
package fully and partially which is in line with the fact
TNOTRAREDS stands for the total number of traders that female-headed households are endowed with less
known by a household who could buy the produced resource and are less exposed to new information and
grain. ideas according to Admassie and Ayele 2004. Besides,
the descriptive statistics show that literate-headed
EXCONTACT is a dummy variable indicating whether a household are more probable in adopting improved
household had contact with government extension wheat technology package fully. This might be because
workers where EXCONTACT = 1 if the household had education may make farmers more receptive to advice
got contact with government extension workers and 0 if from an extension agent or more able to deal with
otherwise. technical recommendations that require a certain level of
numeracy or literacy (Admassie and Ayele, 2004). On
the other hand, households with relatively larger family
RESULTS AND DISCUSSIONS size are less likely to adopt improved wheat technology
package fully and partially which is a bit contradictory to
Descriptive Statistics the fact that such households, on one hand, do not face
labor shortage that may be needed to manage the
Various variables that were included in the propensity increased output which resulted from technology and
score matching model that describe the major observed information adoption and, on the other hand, higher
characteristics of the sample respondents are presented family size necessitates increased productivity to feed
in in table 1. While the average productivity of full and the family. Farmers with relatively smaller land holding
partial adopters of modern technologies & information is size tend to adopt improved wheat technology package
1.76 and 1.34 ton per hectare respectively, that of non- fully and partially which is in line with the fact that certain
adopters of modern technologies & information is only technologies may be appropriate for the intensive
0.93 ton per hectare. Thus, it tentatively shows that there management characteristic of smaller farms as to
is significant difference in productivity level between Admassie and Ayele 2004. Moreover, those farmers who
these two pairs of groups of households. Some of the had contact with government extension workers are
most important demographic determinants that influence more likely to adopt improved wheat technology package
adoption of a technology include family size, level of fully and partially than those that had not.
education and age. There exists a significant difference

Table 1(a): Descriptive statistics of important variables used in the probit model-Propensity score matching
Full Adopters of Non-Adopters of
Modern Technologies Modern Technologies Aggregate
Variables Unit t-stat.
& Information & Information Mean(se)
Mean(se) Mean(se)
Outcome variable
Productivity # 1756.79(30.70) 931.48(214.35) 1750.74(30.58) -2.305**
LnProductivity % 7.29(0.019) 6.60(0.248) 7.28(0.019) -3.16***
Variables that affect
probability of adoption
HHAGE # 45.50(0.35) 49.89(6.25) 45.53(0.35) 1.06
HHSEX (Male=1) 1=Yes 0.912(0.008) 0.78(0.15) 0.911(0.008) -1.41*
FAMILY_SIZE # 6.64(0.064) 8.33(0.93) 6.65(0.064) 2.25**
HHEDU (Read & write=1) 1=yes 0.66(0.014) 0.33(0.17) 0.65(0.014) -2.03**
CREDIT 1=yes 0.079(0.008) 0(0) 0.078(0.008) -0.88
LANDHOLDING_SIZE ha 1.56(0.037) 2.7(0.903) 1.56(0.037) 2.65***
DSTMNMKT km 9.09(0.172) 9.78(0.662) 9.09(0.171) 0.35
OXEN # 2.19(0.048) 1.67(0.24) 2.18(0.047) -0.94
TNOTRAREDS # 4.45(0.161) 3.33(1.01) 4.44(0.16) -0.59
EXCONTACT 1=yes 0.89(0.009) 0(0) 0.884(0.0092) -8.53***
***, **, * indicate significance at 1 percent, 5 percent and 10 percent level respectively.
Source: Own computation, 2018
 Baye et al / Greener Journal of Agricultural Sciences 81

Table 1(b): Descriptive statistics of important variables used in the probit model-Propensity score matching
Partial
Non-Adopters
Adopters of
of Modern
Modern Aggregate
Variables Unit Technologies t-stat.
Technologies Mean(se)
& Information
& Information
Mean(se)
Mean(se)
Outcome variable
Productivity # 1341.69(42.48) 931.48(214.35) 1332.25(41.88) -1.47*
LnProductivity % 7.01(0.034) 6.605(0.248) 6.997(0.034) -1.77**
Variables that affect
probability of adoption
HHAGE # 47.24(0.698) 49.89(6.25) 47.30(0.695) 0.57
HHSEX (Male=1) 1=Yes 0.929(0.0131) 0.78(0.147) 0.9258(0.0133) -1.72**
FAMILY_SIZE # 6.33(0.108) 8.33(0.93) 6.38(0.108) 2.79***
HHEDU (Read & write=1) 1=yes 0.542(0.0255) 0.33(0.167) 0.537(0.0252) -1.24
CREDIT 1=yes 0.039(0.00995) 0(0) 0.038(0.0097) -0.605
LANDHOLDING_SIZE ha 1.475(0.066) 2.7(0.903) 1.503(0.0685) 2.70***
DSTMNMKT km 8.94(0.283) 9.78(0.662) 8.96(0.277) 0.455
OXEN # 1.79(0.091) 1.67(0.236) 1.78(0.089) -0.2
TNOTRAREDS # 3.91(0.248) 3.33(1.014) 3.9(0.243) -0.36
EXCONTACT 1=yes 0.675(0.024) 0(0) 0.6598(0.024) -4.32***
***, **, * indicate significance at 1 percent, 5 percent and 10 percent level respectively.
Source: Own computation, 2018

Propensity Scores Estimation using Probit Model full adopters ranges between 0.591009 and 0.9997821
while it ranges between 0.58267 and 0.978982 for non-
The descriptive statistics of the key variables affecting adopters. And the region of common support for the
adoption of improved wheat technology package has distribution of estimated propensity scores of full
shown a tentative impact of improved wheat technology adopters and non-adopters ranges between 0.59100901
package adoption on increasing productivity. and 0.99978208. When matching techniques are
Nevertheless, a mere comparison of productivity has no employed, observations whose propensity score lies
causal meaning since improved wheat technology outside this range were discarded. The visual
package adoption is endogenous. And it is difficult to presentation of the distributions of the propensity scores
attribute the change to adoption of improved wheat is plotted in figure 1(a). The common support condition is
technology package since the difference in productivity satisfied as indicated by the density distributions of the
might be owing to other determinants. To this end, a estimated propensity scores for the treatment and
rigorous impact evaluation method; namely, Propensity control groups as there is substantial overlap in the
Score Matching has to be employed to control for distribution of the propensity scores of both full adopters
observed characteristics and determine the actual and non-adopters. On the other hand, the propensity
attributable impact of improved wheat technology score for partial adopters ranges between 0.5600793
package adoption on productivity in wheat producing and 0.9945421 while it ranges between 0.8195451 and
areas of Ethiopia. Propensity scores for full adopters and 0.9729395 for non-adopters. And the region of common
non-adopters as well as for partial adopters and non- support for the distribution of estimated propensity
adopters were estimated using a probit model to scores of partial adopters and non-adopters ranges
compare the treatment group with the control group. In between 0.56007927 and 0.99454209. When matching
this regard, only those variables that significantly affect techniques are employed, observations whose
probability of full and partial improved wheat technology propensity score lies outside this range were discarded.
package adoption were used in estimating the The visual presentation of the distributions of the
propensity scores. The test for „balancing condition‟ propensity scores is plotted in figure 1(b). The common
across the treatment and control groups was done and support condition is satisfied as indicated by the density
the result as indicated on figure 1 proved that the distributions of the estimated propensity scores for the
balancing condition is satisfied. treatment and control groups as there is substantial
Each observation‟s propensity scores are overlap in the distribution of the propensity scores of
calculated using a probit model. The propensity score for both partial adopters and non-adopters.
82 Baye et al / Greener Journal of Agricultural Sciences

Kernel density of PPS by treatment status

4
3
2
1
0

.4 .6 .8 1 1.2
Estimated propensity score

Full Adopters Non-adopters

Figure 1(a): Distribution of propensity scores of full adopters and non-adopters

Kernel density of PPS by treatment status

4
3
2
1
0

.4 .6 .8 1 1.2
Estimated propensity score

Partial Adopters Non-adopters

Figure 1(b): Distribution of propensity scores of partial adopters and non-adopters

Assessing Matching Quality proposed specification of the propensity score is fairly

successful in balancing the distribution of covariates
Ensuring good balance between treated and control between the two groups as indicated by decreasing
2
group is the most important step in using any propensity pseudo R for the partial adopters vs. non-adopters
score method. The before and after matching covariate case, decreasing mean standardized bias and the
balancing tests presented on table 2 suggested that the insignificant p-values of the likelihood ratio test.
 Baye et al / Greener Journal of Agricultural Sciences 83

Table 2(a): Propensity score matching quality test

Sample Ps R2 LR chi2 p>chi2 Meanbias Medbias R %Var
Unmatched 0.163 10.94 0.027 122.4 66.3 1.37 100
Matched 0.203 75.47 0.000 49.2 51.4 2.36* 100
* if B>25%, R outside [0.5; 2]

Table 2(b): Propensity score matching quality test

Sample Ps R2 LR chi2 p>chi2 Meanbias Medbias R %Var
Unmatched 0.097 6.41 0.093 93.3 63.6 1.19 100
Matched 0.080 27.50 0.000 19.5 21.1 2.51* 50
* if B>25%, R outside [0.5; 2]

Average treatment effects estimation a significantly higher rate of growth in productivity,

ranging from 51 to 55%, than their counterparts, the non-
Different impact estimators were employed to get adopters. Moreover, half of the matching techniques
estimated treatment effect that disclosed full as well as revealed that on the average partial adopters of
partial adoption of improved wheat technology package improved wheat technology package get a significantly
has a positive and significant impact on productivity higher rate of growth in productivity, ranging from 38 to
growth. Table 3 depicts the average impact of improved 51%, than their counterparts, the non-adopters.
wheat technology package adoption on productivity However, the growth rate of full package adopters that
growth following nearest neighbor matching (NNM), includes improved wheat varieties, information regarding
Stratification Matching, Radius (Caliper) Matching and improved wheat management practices as well as
Kernel Matching (KM) techniques. Accordingly, most of artificial/chemical fertilizer is clearly higher than that of
the matching techniques revealed that on the average the partial package adopters.
full adopters of improved wheat technology package get

Table 3(a): Average treatment effects estimation using different propensity score matching estimators
Mean of outcome
variable based on
Outcome Matching matched
ATT t-stat.
variable algorithm observations
Full Non-
Adopters Adopters
*
Nearest neighbor matching 7.17 6.62 0.548 1.439
Stratification matching 0.394 .
LnProductivity
Caliper matching 6.92 6.53 0.388 1.035
**
Kernel matching 7.17 6.66 0.508 2.181
**, * indicate significance at 5 percent and 10 percent level respectively.
Bootstrapped standard errors are based on 100 replications.
Source: Own computation, 2018

Table 3(b): Average treatment effects estimation using different propensity score matching estimators
Mean of outcome
variable based on
Outcome Matching matched
ATT t-stat.
variable algorithm observations
Partial Non-
Adopters Adopters
Nearest neighbor matching 7.01 6.59 0.428 1.209
Stratification matching 0.226 0.858
LnProductivity *
Caliper matching 7.36 6.84 0.514 1.605
**
Kernel matching 7.01 6.63 0.383 1.836
**, * indicate significance at 5 percent and 10 percent level respectively.
Bootstrapped standard errors are based on 100 replications.
Source: Own computation, 2018
84 Baye et al / Greener Journal of Agricultural Sciences

CONCLUSION AND RECOMMENDATION Ethiopia. Ethiopian Development Research Institute

(EDRI), Research Report 5.
This study is undertaken to identify the impact of Bor O. and Bayaner A. (2009). How Responsive is the
adoption of improved wheat technology package that Crop Yield to producer prices? A panel data
includes improved wheat varieties, information regarding approach for the case of Turkey. NEW MEDIT N.
improved wheat management practices as well as 4/2009.
artificial/chemical fertilizer on wheat productivity in Cochran, W. G. and D. B. Rubin. (1973). „Controlling
Ethiopia. It used propensity score matching technique Bias in Observational Studies: A Review,‟ Sankhya,
which is a robust impact evaluation technique that ser. A, 35:4, 417–446.
identifies the impact which can be attributed to improved CSA. 2015. Major results of the 2007 GDP estimates.
wheat technology package adoption. The study also Central Statistical Agency (CSA), Addis Ababa,
employed and compared various matching algorithms to Ethiopia.
ensure robustness of the impact estimates. Finally, the D. Zeng, Alwang J., Norton G.W., Shiferaw B., Jaleta M.
study concludes that adoption of improved wheat and Yirga C. (2015). Ex post Impacts of Improved
technology package enabled farm households that Maize Varieties on Poverty in Rural Ethiopia.
adopted it fully to enjoy a relatively higher and Agricultural Economics 46 (2015) 1–12.
significantly positive productivity than their counterparts, Daniel F. (2018). Impact of Improved Wheat Varieties
the non-adopters as well as the partial adopters. This Adoption on Productivity: Ethiopia. LAP LAMBERT
indicates that full adoption of improved wheat technology Academic Publishing, Beau Bassin. ISBN: 978-613-
package has a huge potential in strengthening the 7-42438-4.
country‟s agricultural extension system that targets Daniel F. and Belay B. (2018). Impact of Improved
increasing production and productivity. Therefore, this Wheat Varieties & Information's Adoption on
study recommends to widely scale-up full package of Productivity in Ethiopia. GRIN Publishing, Munich.
improved wheat varieties and information as well as ISBN: 9783668808096.
other appropriate modern agricultural technologies and Elias A.,Nohmi M., Yasunobu K. & Ishida A. (2013).
information to all wheat producing farm households, and Effect of Agricultural Extension Program on
this should be accompanied by increasing availability of Smallholders‟ Farm Productivity: Evidence from
affordable improved wheat agricultural technologies and Three Peasant Associations in the Highlands of
information for the smallholder farmers to enhance their Ethiopia. Journal of Agricultural Science, Vol. 5, No.
livelihood which obviously calls for the well-coordinated, 8, 2013.
effective as well as efficient effort of all of the relevant Endale K. (2011). Fertilizer Consumption and
stakeholders of the agricultural sector of the country. Agricultural Productivity in Ethiopia. Ethiopian
Development Research Institute (EDRI), Working
Paper 003.
REFERENCES Endalkachew T. (2011). Impact of Soil and Water
Conservation on Crop Productivity in the Highlands of
Abate G.T., Bernard T., Brauw A. and Minot N. (2016). Ethiopia. A Thesis Submitted to the School of
The Impact of the Use of New Technologies on Graduate Studies of Addis Ababa University in Partial
Farmers‟ Wheat Yield in Ethiopia: Evidence from a Fulfillment of the Requirements for the Degree of
Randomized Control Trial. Selected Paper prepared Masters of Science in Economics, November, 2011.
for presentation at the 2016 Agricultural & Applied GebreEyesus A. (2015). A Study on Human Resource
Economics Association Annual Meeting, Boston, Management, Staff Turnover and Incentives in the
Massachusetts, July 31-August 2. National Agricultural Research System (NARS).
Abate T. , Shiferaw B., Menkir A. , Wegary D. , Kebede Ethiopian Development Research Institute (EDRI),
Y., Tesfaye K., Kassie M., Bogale G., Tadesse B. and Research Report 27.
Keno T. (2015). Factors That Transformed Maize Gebre-Selassie A. and Bekele T. A Review of Ethiopian
Productivity in Ethiopia. Food Sec. (2015) 7:965–981. Agriculture: Roles, Policy and Small-scale Farming
Abegaz G. (2011). Cereal Productivity in Ethiopia: An Systems. Global Growing Casebook.
Analysis Based on ERHS Data. Ethiopian Journal of Gebru A. (2006). The Determinants of Modern
Economics, Volume XX No. 2, October 2011. Agricultural Inputs Adoption and Their Productivity in
Admassie A. and Ayele G, (2004). Adoption of Improved Ethiopia (The Case of Amhara and Tigray Regions).
Technology in Ethiopia. Ethiopian Development A Thesis Submitted to the School of Graduate
Research Institute (EDRI), Research Report 3. Studies of Addis Ababa University in Partial
Asfaw S., Shiferaw B. and Simtowe F. (2010). Does Fulfillment of the Requirements for the Degree of
Technology Adoption Promote Commercialization? Master of Science in Economics (Economic Policy
Evidence from Chickpea Technologies in Ethiopia. Analysis), July, 2006.
Ayele G., Bekele M. and Zekeria S. (2006). Productivity González V., Ibarrarán P., Maffioli A. and Rozo S.
and Efficiency of Agricultural Extension Package in (2009). The Impact of Technology Adoption on
Agricultural Productivity: The Case of the Dominican
 Baye et al / Greener Journal of Agricultural Sciences 85

Republic. Inter-American Development Bank. Office Food Consumption in Southeastern Ethiopia.

of Evaluation and OversightWorking Paper: OVE/WP- Selected Poster prepared for presentation at the
05/09. September, 2009. International Association of Agricultural Economists
Hailu B.K., Abrha B.K. and Weldegiorgis K.A. (2014). (IAAE) Triennial Conference, Foz do Iguaçu, Brazil,
Adoption and Impact of Agricultural Technologies on 18-24 August, 2012.
Farm Income: Evidence from Southern Tigray, Nigussie A., Kedir A., Adisu A., Belay G., Gebrie D. and
Northern Ethiopia. International Journal of Food and Desalegn K. (2015). Bread Wheat Production in
Agricultural Economics Vol. 2 No. 4, (2014), pp. 91- Small Scale Irrigation Users Agro-Pastoral
106. Households in Ethiopia: Case of Afar and Oromia
Haregewoin T., Belay B., Bezabeh E., Kelemu K., Hailu Regional State. Journal of Development and
D. and Daniel F. (2018). Impact of Improved Wheat Agricultural Economics, Vol. 7(4), pp. 123-130.
Variety on Productivity in Oromia Regional State, Olmos A. (2015). Propensity Scores: A Practical
Ethiopia. Greener Journal of Agricultural Sciences, Introduction Using R. Journal of MultiDisciplinary
Vol. 8(4), pp. 074 – 081, April, 2018. Evaluation, Vol. 11, Issue 25, 2015.
Heckman, J., H. Ichimura, and P. Todd. (1998). Rosenbaum, P. and D. Rubin. (1983). The Central Role
Matching as an Econometric Evaluation Estimator. of the Propensity Score in Observational Studies for
The Review of Economic Studies 65(2): 261-294. Causal Effects. Biometrika 70:1, 41–55.
Kelemu K. (2017). Determinants of Farmers Access to Rosenbaum, P. and D. Rubin. (1985). Constructing a
Information about Improved Wheat Varieties: Case of Control Group Using Multivariate Matched Sampling
Farmers in Major Wheat Growing Regions of Methods that Incorporate the Propensity Score. The
Ethiopia. International Journal of Research in American Statistician Association, February1985, Vol.
Agricultural Sciences, Volume 4, Issue 1, ISSN 39, No. 1.
(Online): 2348 – 3997. Rubin, D. B. (1977). Assignment to Treatment Group on
Ketema M., and Kassa B. (2016). Impact of Technology the Basis of a Covariate. Journal of Educational
on Smallholder Wheat Production in Bale Highlands Statistics, 2, 1-26.
of Ethiopia: Application of Output Decomposition S. Tesfaye, B. Bedada and Y. Mesay. (2016). Impact of
Model. Turkish Journal of Agriculture- Food Science Improved Wheat Technology Adoption on
and Technology, 4(6): 446-454. Productivity and Income in Ethiopia. African Crop
Mann M. and Warner J. (2017). Ethiopian Wheat Yield Science Journal, Vol. 24, Issue Supplement s1, pp.
and Yield Gap Estimation: A Spatially Explicit Small 127-135.
Area Integrated Data Approach. Field Crops Taffesse A., Dorosh P. and Asrat S. (2012). Crop
Research 201 (2017) 60–74. Production in Ethiopia: Regional Patterns and
Matsumoto T. and Yamano T. (2010). The Impacts of Trends. Summary of ESSP II Working Paper 16,
Fertilizer Credit on Crop Production and Income in “Crop Production in Ethiopia: Regional Patterns and
Ethiopia. National Graduate Institute for Policy Trends”. ETHIOPIA STRATEGY SUPPORT
Studies (GRIPS) Policy Research Center Discussion PROGRAM (ESSP II) Research Note 11.
Paper 10-23, GRIPS, Tokyo. Tsusaka T. and Otsuka K. (2013). The Changes in the
MoFED. (2003). Rural Development Policy and Effects of Temperature and Rainfall on Cereal Crop
Strategies. Economic Policy and Planning, Ministry of Yields in Sub-Saharan Africa: A Country Level Panel
Finance and Economic Development, Addis Ababa. Data Study, 1989 to 2004. Environmental Economics,
Mulugeta T. and Hundie B. (2012). Impacts of Adoption Volume 4, Issue 2, 2013.
of Improved Wheat Technologies on Households‟

Cite this Article: Baye B; Fitsum D; Eyob B (2019). Impact of Improved Wheat Technology Package Adoption
on Productivity in Ethiopia. Greener Journal of Agricultural Sciences 9(1): 76-85,
http://doi.org/10.15580/GJAS.2019.1.122118180.