African Journal of Science, Technology, Innovation and
Development
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rajs20
Technology adoption intensity and technical
efficiency of maize farmers in the Techiman
municipality of Ghana
Joana Deladem Kwawu, Daniel Bruce Sarpong & Frank Agyire-Tettey
To cite this article: Joana Deladem Kwawu, Daniel Bruce Sarpong & Frank Agyire-Tettey
(2021): Technology adoption intensity and technical efficiency of maize farmers in the Techiman
municipality of Ghana, African Journal of Science, Technology, Innovation and Development, DOI:
10.1080/20421338.2020.1866145
To link to this article:  https://doi.org/10.1080/20421338.2020.1866145
Published online: 25 Mar 2021.
Submit your article to this journal 
Article views: 101
View related articles 
View Crossmark data
Citing articles: 1 View citing articles 
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=rajs20
African Journal of Science, Technology, Innovation and Development, 2021
https://doi.org/10.1080/20421338.2020.1866145
© 2021 African Journal of Science, Technology, Innovation and Development
Technology adoption intensity and technical efficiency of maize farmers in the Techiman
municipality of Ghana
Joana Deladem Kwawu *, Daniel Bruce Sarpong and Frank Agyire-Tettey
Department of Economics, University of Ghana, Accra, Ghana
*Corresponding author email: jdkwawu001@st.ug.edu.gh
This paper analyses the determinants of the intensity of adoption of improved maize technology, technical efficiency and
constraints farmers faced in the Techiman Municipality of Ghana. To achieve the objectives, cross-sectional data were
collected from 407 maize farmers. The data collected were analyzed using descriptive statistics and econometric
models such as the Poisson model and the stochastic frontier model. The study found a positive and significant
influence of extension contact, formal training, land ownership, hired labour, farm size and mobile phone ownership
on the intensity of adoption of improved technology. The stochastic frontier model estimates also found maize farmers
to be on, average, 70% technically efficient with increasing returns to scale of 1.26. The intensity of adoption, age,
land ownership, livestock ownership and perception of soil fertility by the farmers with household size were found to
statistically contribute to the technical efficiency of farmers. The study concludes that intensity of adoption of
improved maize technology package elements increases productivity, and, therefore, recommends that subsidy
packages and credit should be made available to farmers through government and other financial institutions to
increase adoption intensity. This study addresses the gap in the use of improved and multiple maize technology in Ghana.
Keywords: Kendall coefficient, Poisson model, technology adoption, technical efficiency
Introduction interventions geared towards improving growth in the
Agricultural production has become the source of liveli- agricultural sector. Notable among the policies were the
hood for most African countries, including Ghana, due Food and Agricultural Sector Development Policy
to the role it plays in food security, poverty reduction, (FASDEP I, 2007 and FASDEP II, 2010). In 2011, the
and employment generation, and as a source of raw Medium Term Agriculture Sector Investment Policy
material for growing industries. Despite these massive (METASIP) was adopted as the investment plan of
contributions from the agricultural sector, a report by FASDEP II, with the aim of intensifying the use of
the World Bank (2015) suggests that sub-Saharan improved technologies in order to increase the pro-
African (SSA) countries still lag behind other countries ductivity of selected crops (maize, cassava, rice, yam
on the continent in terms of agricultural productivity. and cowpea) (MoFA 2010).
This makes it impossible for African countries to Cereals have been identified by Kuwornu, Suleyman,
achieve food security and poverty reduction. This was and Amegashie (2011) as the most widely consumed food
confirmed by Mensah-Bonsu et al. (2017) who stated in Ghana. This is seen by the total production of cereals of
that the contribution of the agricultural sector in SSA about 3 million tonnes in 2016 (FAO 2017). Among the
countries is far behind expected potential levels. The cereals consumed in Ghana, maize is the largest food
study attributed the productivity gap to the inability of staple, accounting for at least 50% to 60% of overall
SSA countries to take advantage of the technological pro- cereal production, according to the 2015 report on com-
gress that is being experienced globally. This raises the plete curriculum and guide to maize production in
question of how farmers in SSA countries can increase Ghana (Voto 2015). It is one of the major food crops cul-
productivity through enhanced adoption of improved agri- tivated in large quantities in Ghana. As of 2014, it was the
culture technology and modernized farm inputs. first staple food crop in the country and the second-largest
The Ghanaian economy, like other African countries food crop after cocoa with a planted area of 1,018,936
depends largely on the agricultural sector for its role in hectares (SRID & MoFA 2015). Maize production is
the growth and development of the country. From 2006 mainly dominated by smallholder farming households
to 2009, the agricultural sector contributed over 30% to who produce on a small scale and mostly reside in rural
the country’s GDP and also employed over 50% of the areas. Its cultivation serves as a staple crop for sales and
country’s labour force during these periods. This was con- food for human consumption as well as a feed for the
firmed by World Bank (2010) country report, which indi- poultry industry. Mashingaidze (2004) described maize
cated about 33.7% contribution of the agriculture sector to as the major source of protein and calorie intake as well
GDP and over 56% provision of livelihood of Ghana’s as a primary source of weaning for babies. A study con-
total labour force. Also, 90% of Ghana’s food needs are ducted by Tweneboah (2000) also reported maize as a
supplied by the agricultural sector, according to the major determinant of household food security in Ghana.
Food and Agriculture Organization (FAO 2010). In recog- In spite of the government’s vision of agriculture
nition of the importance of agricultural production in modernization through the use of improved technology,
Ghana, the Government through the Ministry of Food agricultural productivity is still characterized by low
and Agriculture (MoFA) initiated long-term policy improved technology adoption and technical inefficiency
African Journal of Science, Technology, Innovation and Development is co-published by NISC Pty (Ltd) and Informa Limited (trading as Taylor & Francis Group)
2 Kwawu, Sarpong and Agyire-Tettey
in production, leading to low productivity. Salifu, Alhas- technology package. MoFA (2011) also attributed the
san, and Salifu (2015) however, reported that the prere- low production of maize in the country to the low technol-
quisite for achieving food security without food aids in ogy adoption.
Ghana is by increasing agricultural productivity through The huge yield gap between the actual and the poten-
the use of improved agricultural technologies. Due to tial yield of maize production in Ghana, calls for the need
this, attempts were being made to increase farmers’ adop- to address the low adoption of improved maize technol-
tion of improved technology, productivity and technical ogy in Ghana (Wongnaa et al. 2019). The purpose of
efficiency through cereal crops since they are the major this study is to analyze the factors that determine the
contributors to agricultural production in Ghana. Several intensity of adoption of improved maize technology
improved technology packages of seed, agronomic prac- package elements and the technical efficiency achieved
tices, timely harvesting, proper storage and applying by farmers in the Techiman Municipality. The study
chemicals on the stored grain in the storage crib have further examines the sources of the technical inefficiency
been recommended by CSIR and MoFA for use by and the constraints farmers experienced in the adoption of
farmers. the improved technology package. The Poisson regression
Thus, researchers have paid much attention to model was used to study the intensity of adoption of
improved technology adoption in recent times, since it improved maize packages while the stochastic frontier
is believed to be the basis for production, increased pro- was employed in the estimation of the technical effi-
ductivity and income growth (Minten and Barrett 2008). ciency. This study contends that the partial adoption of
To assess the impact of these technologies, several the recommended CSIR maize technology packages
studies have been conducted by various researchers on suggests that the available technology is underutilized.
how maize productivity can be increased through the Increasing intensity of adoption of the technology
use of improved maize technologies and efficient utiliz- package elements and efficiency of farmers is essential,
ation of resources. For instance, Mensah-Bonsu et al. and a better option to grow outputs than introducing
(2017) used the count data model to examine the intensity new maize technologies. This study adds to the extant lit-
of adoption of land and water management practices in erature on technology adoption and technical efficiency in
Ghana. The study found a positive influence of extension Ghana (Abukari, Hussein, and Katara 2015; Alhassan,
visits, credit, and the experience of severe food shortage Salifu, and Adebanji 2016; Oppong, Onumah, and
on adoption with education and land per capita having a Asuming-Brempong 2016). It also fills the gap in the
negative impact. Alhassan, Salifu, and Adebanji (2016) intensity of adoption of improved technology in Ghana.
also studied the influence of farmer’s socio-demographic
and varietal characteristics of maize on adoption of Literature review
improved maize varieties. Oppong, Onumah, and Technology adoption and determinants
Asuming-Brempong (2016) estimated technical effi- Doss (2003) defined adoption as the prolonged use of
ciency and production risk of maize production in suggested practices by farmers over an extended period
Ghana and found that seeds, land, the cost of intermediate of time. Dasgupta (1989), however, noted that adoption
input and herbicides have decreasing returns to scale is not a perpetual behaviour by farmers. This is because,
effect on maize output with a mean technical efficiency adopters can resolve to abolish their use of a given tech-
of 0.62. Kuwornu, Amoah, and Seini (2013) also investi- nology or innovation for various reasons such as personal
gated the technical efficiency of maize farmers in the characteristics, social, institutional characteristics and
Eastern Region. By employing a multi-stage sampling knowledge of a new technology that is more rewarding
procedure, the study found the mean technical efficiency than the initial one. Adoption of improved agricultural
of the farmers to be 0.51 with negative returns to scale technology has become a major concern in agricultural
based on farm inputs such as seed, fertilizer, and family production in both developed and developing countries.
labour. It is believed that invention of an improved technology
These studies, however, targeted the adoption of brings about a high adoption expectation by farmers or
improved maize seed varieties and fertilizer use, and neg- producers which play a vital role in agricultural pro-
lected other recommended agronomic practices in the ductivity. Several studies have been undertaken on tech-
improved maize technology package such as crop nology adoption and technical efficiency in developing
density and spacing, weed control, zero tillage and other countries. Most of these identified several factors that
soil fertility management practices that need to accom- affect the adoption process based on the contextual appli-
pany the improved maize variety and fertilizer application cability and underlying specific local condition in the
in order to achieve maximum yield. However, findings by study locales.
Aikins, Afuakwa, and Owusu-Akuoko (2012) have shown Wongnaa et al. (2019) examined the influencing
that adoption of an improved maize variety and chemical factors of adoption of improved maize production tech-
fertilizer application alone cannot solve the yield gaps that nologies in Ghana using a multinomial model. The
Ghana is currently experiencing. The study further pro- study found technology adoption to be influenced by edu-
posed the need to evaluate farmers’ adoption of rec- cational level, age, agricultural extension contact, access
ommended agronomic practices such as seeding rate and to credit, initial capital outlay, experience, land fragmen-
spacing, timely planting, proper crop density, weeds and tation, group membership, and previous year’s price of
pesticide control and other land use and management maize and availability of ready market. Mensah-Bonsu
practices recommended in the improved maize et al. (2017) used the count data model to examine the
African Journal of Science, Technology, Innovation and Development 3
intensity of adoption of land and water management prac- variations in the pattern of the estimated efficiency. Both
tices in Ghana. The study found a positive influence of studies, however, maintained the two approaches as the
extension visit, credit, the experience of severe food preferred choice for efficiency analysis. Coelli and
shortage on adoption, while education and land per Battese (1996) argued in favour of the parametric
capita had a negative influence on technology adoption. approach as the most suited for studies that are related
In a related study, Johnson (2013) used the count data to to agriculture. The study justified its conclusion on the
estimate the intensity of technology adoption of small- fact that exogenous factors and measurement errors are
scale oil palm producers in the Western Region. Extension intrinsic in agricultural production and since the non-para-
contact, access to credit, hired labour and type of share- metric approach does not account for them, it is most
holder were found to positively influence the intensity appropriate to prefer the parametric approach against the
of technology adoption, which confirms the findings of non-parametric approach.
Mensah-Bonsu et al. (2017). Ahmed et al. (2018) evaluated the technical, economic
Employing a panel dataset, Olwande, Sikei, andMath- and allocative efficiency of maize farmers in the Eastern
enge (2009) examined the determinants and intensive use Ethiopia with the help of a cross-section data from 480
of fertilizer adoption in Kenya. The study observed a maize plots. By employing stochastic production function
highly positive influence of age, gender, education of fitted on a Cobb–Douglas production function, the study
the farmer, dependency ratio, the presence of a cash found the amount of seed, land and DAP (Diammonium
crop, access to credit, distance to a fertilizer market and phosphate) as the factors that influence maize production
extension officer and agro-ecological potential on the in the study area. The study further found the potential of
adoption and intensive use of fertilizer. Similarly, a logit farmers to increase economic efficiency to depend more
model was extensively used by Mureithi and Ojiem on allocative efficiency than technical efficiency.
(2000) to assess the determinants of adoption of maize Oppong, Onumah, and Asuming-Brempong (2016) esti-
production technologies. The findings showed a positive mated technical efficiency and production risk of maize
influence of gender, hired labour, extension services and production in Ghana and used 232 farms from the
access to credit facilities on maize production technology Brong–Ahafo Region. The study found a decreasing
adoption in Kenya. Kaliba, Verkuijl, and Mwangi (2000) return to scale effect of seeds, land, the cost of intermedi-
also employed the Tobit and Probit models to examine the ate input and herbicides on maize output with a mean
determinants of adoption of improved technologies such technical efficiency of 0.62. While farm size contributes
as maize seeds and inorganic fertilizer in Tanzania. The to technical inefficiency negatively, ploughing, on the
study found extension services availability, rainfall, other hand, affects technical inefficiency positively.
variety characteristics and on-farm field trials and rainfall Kuwornu, Amoah, and Seini (2013) investigated the
to have a significant influence on the intensity of adoption technical efficiency of maize farmers in the Eastern
of improved technology. Region. By employing a multi-stage sampling procedure,
they selected 226 farmers from four geographical areas.
Technical efficiency and determinants The study found the mean technical efficiency of the
Uri (2002) defined technical efficiency as the commensur- farmers to be 0.51 with negative returns to scale based
able decrease in a number of inputs used by firms in order on farm inputs such as seed, fertilizer and family labour.
to achieve a given output level, measured as the efficient The study also found extension visit, farmer training in
use of the input. Technical efficiency simply is the phys- maize farming, membership of a farm-based organization,
ical proportion of output to input. The larger the pro- cash and in-kind credits and frequency of meetings of
portion obtained, the greater the achieved technical farm-based organizations as the main determinants of
efficiency. To obtain the technical efficiency of a farmer, technical efficiency in the Eastern Region of Ghana.
there is the need to equate the observed output of the
farmer to its corresponding frontier output (potential Theoretical framework and empirical model
output) based on a number of inputs used by the farmer Technology adoption decisions by farmers are contingent
(Ogundari and Ojo 2007). Farrell (1957) proposed a on several factors. One of these factors is the utility
number of frontier models for efficiency measurement farmers derive from making their adoption decisions.
broadly classified into two categories, namely the para- Farmers as rational consumers of improved agricultural
metric frontier technique and the non-parametric frontier technologies have been envisaged to select improved
technique. The parametric frontier technique is further technology package elements that maximize their profit-
divided into the deterministic frontier technique (Aigner ability. This study adopts the Von Neumann-Morgen-
and Chu 1968) and the stochastic frontier technique stern’s (VMN) Utility Theory propounded in 1947
(Aigner, Lovell, and Schmidt 1977). Data envelopment which is the basis for an expected utility theory. By apply-
analysis (DEA) has been widely used in relation to the ing the Von Neumann-Morgenstern’s Utility Theory, the
non-parametric frontier technique (Coelli et al. 2005). study assumes that smallholder farmers are rational and
Past empirical studies employed either the parametric thus decide to maximize their utility (Von Neumann and
or the non-parametric techniques to measure efficiency Morgenstern 1947).
(Endrias et al. 2010; Ephraim 2007). Aye and Mungatana According to Batz, Peters, and Janssen (1999), the
(2011) and Headey, Alauddinb, and Rao (2010) used both expected utility from the adoption of a new technology
efficiency techniques while paying keen attention to their is affected by the attributes of the technology (TC),
predictive capability. The findings revealed distributional characteristics of the farmer (FACH), the farming
4 Kwawu, Sarpong and Agyire-Tettey
system (FSCH) and the farming circumstances (FC). By A Count data analysis model which comprises of the
denoting the utility obtained from the new technology use of the Poisson Regression Model (equi-dispersion),
as Un, utility from the traditional technology as Ut and the Negative Binomial Regression Model (over-dis-
the adoption status of technology as Di, the expected persion) or Gamma Distribution (under-dispersion) has
utility from the adoption of a new technology is modelled been used intensively in previous studies in situations
as: where multiple technologies are to be adopted. According
to Sharma, Bailey, and Fraser (2011), the number of
Un = f (TC, FACH, FSCH, FC) (1) elements in a technology package adopted by a farmer
is interpreted as the intensity of adoption when count
Di = 1 if Un . Ut or (Un − Ut . 0) (2) data models are used. Extant literature that used the
count data model for technology adoption employs para-
Farmers will adopt a new technology only if the returns metric models such as the Poisson model and the Negative
from adoption are relatively high in terms of profit com- Binomial model. Mensah-Bonsu et al. (2017) and Sharma,
pared to the existing technology. Also, adoption will be Bailey, and Fraser (2011) employ the count data model to
possible in situations where the new technology has a estimate the speed and intensity of technology adoption.
relatively low risk compared to previous technology Mensah et al. (2017) used the Poisson and Negative Bino-
(Doss and Morris 2001). mial regression models to estimate the intensity of adop-
tion of land and water management practices in Ghana
Materials and method using maize farmers as a case study.
Study area This study adopts the Poisson regression model to
This study was carried out in the Techiman Municipality, analyse the determinants of the intensity of adoption of
one of the 26 districts in the Brong-Ahafo region of Ghana improved maize technology package elements after the
which lies between latitude 8° 00′ N and 7° 35′ S and Cameron and Trivedi (1990) test for over-dispersion
longitudes 1° 49′E and 2° 30′ W. With a total land area was rejected. Elements in the package include improved
of 669. 7 km2, the Municipality has a total population of and certified seed, fertilizer application, crop spacing
147,788 as of the 2010 Population and Housing Census, and row planting, other soil fertilizer management prac-
representing 6.4% of the total population in the region tices, weed control and zero tillage.
with a sex distribution of 51.5% males and 48.5% Cameron and Trivedi (2009, 1998) and Greene (2003)
females. The Municipality has three vegetation zones, defined the Poisson regression model as the foundation
namely the Guinea Savannah Woodland, the Transitional for count data analysis. It is used to estimate the decision
zone and the Semi-deciduous zone located in the North- of a farmer on the number of improved maize technology
west, South-East and the North respectively. Both tropical package to adopt. The likelihood of adopting k number of
conventional and Semi-Equatorial climates are experi- improved maize technology elements given n independent
enced in the Municipality, which is characterized by mod- improved technology package element is represented by
erate to heavy rains. Annual mean rainfall in the the binomial distribution:
Municipality ranges from 1260 and 1600 mm with ( )
humidity of 75%–80% and 70%–72% during rainy n k n−k
seasons and dry seasons respectively and an average P(Y = k) = p (1− p) (3)k
Temperature of 28°C. The majority of the people in the
Municipality engaged in agriculture farming as their p is the probability of adopting the k maize technology, n
main source of livelihood. Apart from crops, the Munici- is the number of technologies in the package, Y is the
pality also reared livestock such as goat, sheet, guinea number of improved maize technology package elements
fowls and chicken. The Municipality has the largest that are adopted from the package. As n becomes large
market in the sub-region, which trades with traders from and p (probability of adoption) becomes small, repetition
Togo, Benin and Burkina Faso thus making it an inter- of a series of binomial choices supported by the random
national market. utility formulation converges to a Poisson distribution:
The study employs two different econometric tools to
analyze the determinants of intensity of adoption of ( )n e−lmk
improved maize technology package elements and techni- lim1 p
k(1− p)n−k = (4)
n k k!
cal efficiency due to the dual nature of the objectives of
the study. Most adoption studies have used the Logit
and Probit model to estimate the determinants of technol- The density function of the Poisson regression model is
ogy adoption based on the dichotomous nature of the specified as:
decision where ‘1’ represents adopters and ‘0’ for non- −mi y
adopters (Assefa and Gezahegn 2010; Moti et al. 2013). e mf (y/xi) = (5)
A Multinomial model used by Moti et al. (2013) is best y!
for evaluating more than two independent technologies
adoption decisions while the Multivariate model where the mean parameter expressed as a function of the
employed by Hailemariam, Menale, and Bekele (2013) xi and β is given as:
is useful for estimations involving more than two interde- ′
pendent technologies adoption decisions. E(y/xi) = m = exp (x b) and y = 0, 1, 2, . . . ..
African Journal of Science, Technology, Innovation and Development 5
f (y) is the likelihood that y will take a non-negative Empirically, the true frontier is not known; hence, the
integer value and m is log- linear value assumed to be best-practice farmer is used mostly as a proxy for the
related to Xi. true frontier. This study employed the parametric
This study, therefore, adopts the model used by approach (stochastic frontier approach) to estimate the
Nkegbe and Shankar (2014) to estimate the intensity of technical efficiency of maize farmers in the Techiman
adoption of improved maize technology package as: Municipality in the Brong–Ahafo Region of Ghana.
Selection of the stochastic frontier approach is based on
Prob(Yi ; yi|x p fc sii) ; f (x , x , x ) (6) its ability to account for stochastic noise and producer’si i i
inefficiency at the same time.
Y = Number of elements (count) adopted by farmer i; The stochastic frontier production model that Battesei
xp = Personal and household characteristics; x fc = and Coelli (1995) propounded in line with the originali i
Farm/plot and cropping characteristics; and xsi = Socio- model by Aigner, Lovell, and Schmidt (1977) is specifiedi
economic and institutional variables. as:
A farmer is said to be an adopter when he/she adopts
at least one of the six (6) elements in the improved maize Yi = f (Xi; b) exp (vi − ui)i = 1, 2, 3, . . . ., n (8)
technology package and a full adopter when he/she adopts
all the six (6) elements in the improved maize technology n = the farmers in the cross-sectional survey; Y = the
package. A farmer is, however, regarded as a partial ioutput level of the ith farmer; X = vector of input quan-
adopter when he/she adopts one (1) to five (5) elements itities from the ith farmer; vi − ui = disturbance errorfrom the technology package (Table 1). where vi is normally distributed with mean zero and con-
stant variance: N(0, s2v), (Coelli et al. 2005) and ui is a
Determinants of intensity of adoption of improved truncated normal distribution with mean mi and variance
maize technology package s2u. The parameters of the variance are estimated as: s
2
The empirical regression model used in this study to = s2 2u + sv
examine the determinants of intensity of adoption of
elements in the improved maize technology package is
specified as: = Yi = f (Xi; b) exp (vi − ui)TEi f = exp (−u ) (9)Y f (Xi; b)exp(v ii i)
ADOPTi =b0 + b1GEN+ b2AGEi + b3EDUFi
+ b EXVISITi + B5HHSi Yi is the observed output, Y∗i is the frontier output and4
+ b FTRAIN + b LANOWN the technical efficiency assumed the value between zero6 i 7 i
+ + and 1, (0 ≤ TEi ≤ 1). If ui = 0, it implies Y
f
i = Y ,
b8HLABOURi b9ACRED
i
i depicting that the farmers are technically efficient
+ b10FBO+ b11OMPi + b12DMCi (100% efficient).
+ + + s2b13PPERCEPTIONi b14TFSi ui Gamma (γ) = u and it specifies the error that is
(7) associated with thse2 technical inefficiency estimates. It
ranges from zero (0) to one (1). Where γ = 1, deviations
from the frontier are said to be due to technical ineffi-
Technical efficiency analysis ciency. Also, γ = 0 implies deviations are caused by
Measurement of technical efficiency compares the actual noise effect whilst 0 , g , 1 means both stochastic and
performance to the optimum performance (true frontier). non-stochastic errors are present in the data.
Table 1: Description of variables used in the Poisson model.
Dependent variable Description of variable
ADOPTION Number of improved maize technology package elements adopted by farmer i
Independent variables Description of variables Expected sign
GEN Gender of the farmer +/−
AGE Age of the farmer in years −/+
EDUF Years of schooling of the farmer −/+
EVISIT Extension contact +
HHS Household size +
FTRAIN Formal training +
LANOWN Land ownership by the farmer +
HLABOUR Availability of hired labour +
ACRED Access to credit +
FBO Member of farm-based organizations +/−
OMP Ownership of mobile phone +
DMC Distance to input market −
PPERCEPT Perception of price of improved maize technology package −
TFS Total farm size +
6 Kwawu, Sarpong and Agyire-Tettey
For the production function form, this study employed where Zi, i = 1, 2,… , 13 are the technical inefficiency
the translog production function as specified by Battese factors which are defined as adoption, gender, age, age
and Broca (1997). square, formal education, household size, livestock own-
∑ ership, land ownership, soil fertility perception, extension5 contact, distance to market centre, farm-based organiz-
lnQij = b0 + a1DDFi + a2DDAi + biInXi ations and access to credit.
∑∑ n=11 5 5+ + Data sourceb
2 nm
InXnInXm 1i (10) The data used for the analysis were sourced from a cross-
n=1 m=1
section questionnaire that was administered to 408 maize
Q =Total output per hectare measured in kilogrammes farmers in the Techiman Municipality for the 2016 majori
per hectare (Kg/Ha); DDF=Dummy for fertilizer (1 if fer- maize cropping season. The study found one (1) question-
tilizer is used, 0 if not used); DDA =Dummy for agro- naire to be incomplete and it was rejected, thus 407 maize
chemical (Dummy, 1 if agrochemical is used, 0 if not farmers were used for the analysis. Four stage sampling
used); X = Fertilizer use, measured in Kg/Ha, X = Agro- technique was adopted to collect data. The first stage1 2
chemical (Pesticides and herbicides), measured in Litres/ involved a purposive sampling of the Techiman Munici-
Ha, X = Quantity of Seed, measured in Kg/Ha; X = pality as one of the leading maize growing areas in the3 4
Labour, measured in Man-days/Ha, X5 = Farm size,
country. The second stage involves a selection of 6
measured in Hectare; bj = Parameters to be estimated, major maize growing areas in the Municipality out of a
= v − u . total of 9 operations areas. At the third stage, 40 major1i i i
According to Onumah and Acquah (2010), lack of maize producing communities were selected from the
inclusion of dummies for fertilizer and agrochemicals in six (6) operational areas while the last stage saw the
the production function is likely to make the coefficient sampling of 408 farmers through stratified sampling.
of responsiveness of fertilizer and agrochemicals to The questionnaire which has both closed-ended and
maize output biased. open-ended questions was augmented with key informant
The productivity level of individual inputs (partial interviewed and focus group discussions to access the
elasticity of output) with respect to the inputs from the authenticity of the information collected from the farmers.
translog production function is computed as:
Results analyses and discussion
InE(Q ) Descriptive statistics= ∂ i1q { } Results from Table 2 shows the demographic statistics of∂InX ji ∑ the variables used in the regression models in the study.5
= + + = They include age, gender, educational level, labourbj b jjInX ji bnmInXnm bj (11) source, formal training in maize farming, extension
i=1
contact, access to credit, membership of farm-based
organization, ownership of the mobile phone, livestock
Since the coefficients of the translog production function
and land. Statistics from Table 2 show that men domi-
cannot be interpreted directly as elasticity, the variables
nated maize production in the Techiman Municipality
are rescaled into unit means. As a result of the rescaling,
with 81%. This is basically due to the fact that maize pro-
the square term of b jj and b jk equate to zero, whereas bj is duction is labour intensive as well as cost intensive
interpreted as elasticities.
leading to low participation of women. The results also
This study uses the firm’s output elasticity to deter-
show that about half of the farmers interviewed do not
mine whether the firm is exhibiting constant, decreasing
have formal education which is followed closely by
or increasing returns to scale. The summation of the
MSLC/JHS with about 31%. The majority of the
overall partial elasticity equals the returns to scale
farmers (96%) have farming as their main occupation
(RTS). It is specified mathematically as:
∑ while 61% of the farmers employed hired labour for
= their farming activities. In addition, about 24% of theRTS 1q (12) farmers received credit either from the bank, friends and
families or in kind during the 2016 major maize cropping
RTS = 1, implies constant returns to scale. RTS . 1, season while about 57% of the farmers were members of
increasing returns to scale; RTS , 1, decreasing returns farm-based organizations.
to scale. It can be observed from Table 3 that total maize pro-
duction in Techiman Municipality has a mean value of
Empirical model specification for technical about 1303 which ranges from about 111 to 6227. This
inefficiency model is lower than the 1800 kg/ha obtained by Wongnaa et al.
The technical inefficiency model for this study is specified (2019). Total fertilizer applied by the farmers also
as: ranges from 0 to 618 kg/ha with an average value of
∑ about 112 kg/ha. The mean value of fertility is also13 lower than the 140 kg/ha obtained in a study by
mi = d0 + diZi (13) Wongnaa et al. (2019) for maize farmers in the various
i=1 zones in Ghana. Agrochemical which comprised of the
African Journal of Science, Technology, Innovation and Development 7
Table 2: Socioeconomic characteristics of the farmer.
Socioeconomic variable Item Frequency Percentage (%)
Age 15–35 96 23.59
36–60 262 64.37
61–80 49 12.04
Gender Male 331 81.33
Female 76 18.67
Educational level No schooling 204 50.12
Primary school 52 12.78
MSCL/JHS 127 31.20
SHS/Voc/Tech 21 5.16
Tertiary 3 0.74
Labour source Hired labour 250 61.43
Family labour 129 31.70
Exchange 26 6.39
Others 2 0.49
Occupational status Main 392 96.31
Minor 15 3.69
Formal training in maize farming Yes 307 75.43
No 100 24.57
Extension contact Yes 333 81.82
No 74 18.18
Access to credit Yes 97 23.83
No 310 76.17
Membership of farm-based organizations Members 230 56.51
Non-members 177 43.49
Ownership of mobile phone Yes 45 11.06
No 362 88.94
Livestock ownership Yes 141 34.64
No 266 65.36
Land ownership Family 282 69.29
Hired 125 30.71
total amount of weedicides and herbicides also have a 0 to 6 where ‘0’ means no adoption and ‘6’ for full adop-
mean value of 7 litres/Ha with seed quantity (kg/ha), tion. Results from Table 4 show that only 14% of the
labour (man–days/ha) and farm size (hectares) averaging farmers adopted all the six (6) elements from the
at about 22, 305 and 2 respectively. The value of agro- package, whereas 2% of the farmers did not adopt any
chemical, seed and labour for this study was, however, of the elements from the package. The majority of the
higher than the 5.1, 18 and 64 obtained for agrochemical, farmers adopted at least four of the elements of the
seed and labour respectively by Wongnaa et al. (2019). improved maize technology package with improved
seed, other soil fertility management and weed control
Intensity of adoption of improved maize technology as the most adopted elements. This implies that farmers
package adopted improved seed while neglecting other agronomic
The study uses the responses derived from the number of practices that come with the improved maize technology
elements adopted by each farmer to construct the depen- package. The results concur the findings of Batz, Peters,
dent variable for the Poisson regression model. Analysis and Janssen (1999) who carried out a study in Embu Dis-
of the data showed that 97% of the farmers are aware of trict and found that farmers in the District adopted the
at least five (5) elements of the package. Only 0.25% of improved maize variety, but either partially or fully
the farmers were aware of one element from the ignored the agronomic practices that come with the
package. The number of elements adopted ranges from package.
Table 3: Summary statistics of variable in the stochastic frontier model and other socioeconomic variables.
Variable Unit Mean Std. D. Minimum Maximum
Output Kg/Ha 1302.952 727.6152 111.1973 6227.046
Fertilizer Kg/Ha 111.8147 145.1201 0 617.7625
Agrochemical Litres/Ha 6.8939 5.4126 0 23.06313
Seed quantity Kg/Ha 22.09646 6.744774 8.236834 88.9578
Labour Man- days/Ha 304.6112 221.3931 13.6402 963.7095
Farm size Hectares 1.781813 1.394072 0.4046863 10.11712
Household size Number 5.7961 3.4627 1 30
Distance to market centre Hours 1.4326 1.0382 0 5
Extension visit Number 6.8010 10.3718 0 48
Years of schooling Years 4.1793 4.6056 0 20
8 Kwawu, Sarpong and Agyire-Tettey
Table 4: Intensity of adoption of improved maize technology adoption of improved maize technology is labour inten-
package. sive, farmers who have access to hired labour are more
Number of elements Number of Percentage of likely to adopt improved maize technology. Farmers are
adopted adopters adopters able to supplement their family labour with hired labour
0 7 1.72 in order to adopt more improved maize technology. This
1 5 1.23 is consistent with Johnson’s (2013) study who found a
2 18 4.42 positive result among oil palm producers in the Western
3 104 24.82 Region of Ghana. Ownership of mobile phone also deter-
4 120 31.20
5 95 22.36 mined the intensity of adoption of improved technology
6 58 14.25 positively at the 5% level of significance. Farmers who
Total 407 100.0 own a mobile phone and use it to access production and
market information enriched themselves with better infor-
mation which increases their technology adoption likeli-
Empirical results hood. This study agrees with similar results obtained by
Determinants of adoption intensity of improved maize Kaba (2016) and Solomon et al. (2011). Farm size on
technology package the other hand also influenced the intensity of improved
The study failed to reject the post-estimation test for the technology adoption positively at the 1% level of signifi-
equality of the mean and the variance from the Stata diag- cance. Farmers with large farm size have a high prob-
nostics test. The study, therefore, employed the Poisson ability of adopting more elements from the improved
regression model to estimate the determinants of adoption maize technology package since large farm holdings
of improved maize technology package elements come with larger financial resources which enable
(Table 5). farmers to devote more lands to improved technology
Results as presented in Table 6 indicates extension adoption. This result conforms with similar results
contact, formal training, land ownership, hired labour, obtained by Abdul-hanan, Ayamga, and Donkoh (2013)
membership of farm-based organizations, mobile phone in Ghana.
ownership, distance to market centre, perception of Land ownership, farm-based organization, distance to
price of the improved maize package and farm size as market centre and price perception were also found to sig-
the main factors that significantly influence the intensity nificantly influence the intensity of adoption of improved
of adoption of improved maize technology package maize technology package elements negatively, at 1%,
adoption. 5%, 1% and 10% level of significance respectively.
Extension contact was found to have a positive influ- Farmers who use family-owned lands usually do not
ence on intensity of adoption of improved maize technol- have the needed capital to purchase improved technology
ogy package elements at 1% level of significance. As since most of them operate on a small scale and in some
farmers have regular contact with extension officers, cases subsistence level. This study, however, contradicts
they receive reliable and timely information on improved the positive, but insignificant results obtained by
technologies which affects their likelihood of technology Mensah et al. (2017) who believed farmers with their
adoption positively and stimulates their adoption own land devoted more lands to improved technology
decisions. This finding is consistent with a study con- adoption. Interestingly, membership of farm-based organ-
ducted by Tura et al. (2010) in Central Ethiopia who izations has a negative effect on the intensity of technol-
found a positive influence of extension contact on ogy adoption which contradict the positive sign
improved technology adoption and explained that expected. Where farm-based organizations focus more
farmers obtain valuable information when they have attention on collecting dues and bargaining for good
regular contact with extension officers. The finding also prices with little time to educate group members about
revealed a positive and significant influence of formal the right technology to adopt, farmers are less likely to
training on intensity of technology adoption at the 5% receive information about improved technology leading
level of significance. This result concurs with the to low or no adoption. The result, however, contradicts
finding obtained by Kuwornu, Amoah, and Seini (2013) the positive finding obtained by Abdul-hanan, Ayamga,
in Ghana and Fikru (2009). According to Fikru (2009), and Donkoh (2013) in Ghana.
formal training of farmers is an awareness creation plat- The negative result obtained for distance to the market
form that encourages farmers to adopt improved maize centre met its expected sign. Farmers who are closer to the
technology. input market centre have low transportation costs in
The study also found a highly positive and 1% influ- accessing improved maize technology package elements
ence of hired labour on the intensity of adoption of which gives them a greater likelihood of adopting more
improved maize technology package elements. Since elements from the package. Mulugeta (2011) also found
Table 5: Test for equality of the mean and variance.
Variables Test statistics P-Value Decision
Goodness-of-fit 153.9654 Prob > chi2 (392) = 1.0000 Accept H0
Pearson good-ness-fit 127.1154 Prob > chi2 (392) = 1.0000 Accept H0
African Journal of Science, Technology, Innovation and Development 9
Table 6: Determinants of Intensity of improved maize technology package.
Variable Coefficient Robust S.E p-value
Constant 1.2630*** 0.1268 0.000
Gender 0.0230 0.0413 0.577
Age −0.0007 0.0013 0.587
Education 0.0030 0.0032 0.350
Household size −0.0024 0.0042 0.559
Credit 0.0363 0.0354 0.305
Extension contact 0.1813*** 0.0518 0.000
Formal training 0.0808** 0.0325 0.013
Land ownership −0.1350*** 0.0331 0.000
Hired labour 0.0785*** 0.0284 0.006
Farm-based org. −0.0724** 0.0367 0.049
Mobile phone 0.0837** 0.0328 0.011
Distance to market centre −0.0436*** 0.0160 0.006
Price perception −0.0602* 0.0320 0.060
Farm size 0.0312*** 0.0096 0.001
Number of observations 407
Pseudo R2 0.0270
Wald chi2 0.0000
Wald chi2 (14) 134.33
Log pseudo likelihood −728.4201
***p < 0.001; **p < 0.05; *p < 0.1
a negative influence of market distance on the intensity of respectively. Thus, farmers are more likely to achieve
adoption of improved technology in Dale Woreda. The greater output when higher amount of input such as
negative result of price perception on the intensity of tech- seed, labour and farm size is employed. Wongnaa et al.
nology adoption is an indication of how essential price is (2019) also found a positive influence of seed, labour
in the intensity of technology adoption decision-making. and farm size on maize output in the Guinea Savannah,
Farmers who perceive the price to be high are less Transition, Forest and Costal Savannah zones of Ghana.
likely to adopt more elements from the package than The study also found fertilizer square, agrochemical
their counterparts. This result is consistent with similar square, seed quantity square, fertilizer by labour, fertilizer
findings by Wandji et al. (2012) who noted that the per- by farm size, seed by labour and seed by farm size to be
ception about the characteristics of a new technology significant, thus justifying the use of the translog pro-
has a significant effect on its adoption. duction function. The study further found seed quantity
to have the highest coefficient of 0.4504, which implies
Technical efficiency analysis a 1% increase in the quantity of seed per hectare will
Test carried out to determine the appropriate functional increase output by 0.4504%. The positive result for seed
form for the stochastic frontier model rejects the Cobb– quantity and the highest coefficient value concurs with a
Douglas production function in favour of the translog similar result obtained by Essilfie, Asiamah, and Nimoh
functional form at 1% level of significance. The results (2011) who found a positive and high coefficient value
of the hypothesis tested are presented in Table 7. The between the quantity of seed used and total output of
hypothesis for the absence of inefficiency effects in the small-scale maize production in the Mfantseman Munici-
model was also rejected at the 1% level of significance. pality of Ghana. While the recommended seed per hectare
The study also rejects the hypothesis that the inefficiency of land is approximately 25 kg (USAID/IFDC 2015), the
model is zero while accepting the hypothesis which speci- study found the average rate of seeds planted per hectare
fies the absence of an intercept from the model and to be approximately 22 kg/ha which shows most farmers
analysis. do not follow the recommended rate of seed adoption.
Estimates of the maximum likelihood analysis found Mean technical efficiency was found to be approxi-
seed quantity, labour and farm size to be positive and sig- mately 70% on average, which ranges from 0.09 to
nificant 5%, 1% and 10% levels of significance 0.93. Thus, maize output of farmers could have likely
Table 7: Results of hypotheses tested.
Null hypothesis Test statistics Degree of freedom Critical value Decision
1. H0: βnm = 0 51.350 15 30.578*** Reject H0
2. H0: γ = δ0 = δ1 =… = δ13 = 0 73.965
a 15 37.005 Reject H0
3. H0: γ= 0 4.996
a 1 2.706 Reject H0
4. H0: α1 = α2 = 0 1.542 2 9.210*** Accept H0
5. H0: = δ1 =… = δ13 = 0 66.434 13 27.688*** Reject H0
Source: Author’s compilation from the Survey Data, 2017. aValues are a test of one-sided error. The critical values for all test involving γ are obtained
from Table 1 of Kodde and Palm (1986, 1246) whilst the critical values for the rest of the hypotheses are obtained from Chi-Square Table. *** represents
1% level of significance.
10 Kwawu, Sarpong and Agyire-Tettey
Table 8: Maximum likelihood estimates of the stochastic frontier model.
Variables Parameters Coefficient Std Err. z-stats p-value
Constant β0 5.6154*** 0.1668 33.67 0.000
LnFertilizer/hectare β1 −0.2260 0.1511 −1.50 0.135
LnAgrochemicals β2 −0.1882 0.1327 −1.42 0.156
LnSeedquantity/hectare β3 0.4504** 0.2071 2.17 0.030
LnLabour/hectare β4 0.3886*** 0.1184 3.28 0.001
LnFarmsize β5 0.4174*** 0.1493 2.80 0.005
0.5 (LnFertilizer)2 β6 0.3553*** 0.1150 3.09 0.002
0.5 (LnAgrochemical)2 β7 0.1324* 0.0763 1.73 0.083
0.5 (LnSeedquantity)2 β8 −0.6584*** 0.2422 −2.72 0.007
0.5 (LnLabour)2 β9 0.0606 0.0850 0.71 0.476
0.5 (LnFarmsize)2 β10 0.0782 0.1116 0.70 0.483
LnFertilizer*LnAgrochem β11 0.0502 0.0397 1.26 0.206
LnFertilizer*LnSeedquan β12 0.0848 0.1027 0.83 0.409
LnFertilizer*LnLabour β13 0.0839* 0.0446 1.88 0.060
LnFertilizer*LnFarmsize β14 0.1577*** 0.0497 3.17 0.002
LnAgrochem*LnSeedquan β15 0.0787 0.1307 0.60 0.547
LnAgrochem*LnLabour β16 −0.0336 0.0519 −0.65 0.518
LnAgrochem*LnFarmsize β17 −0.0317 0.0552 −0.57 0.565
LnSeedquan*LnLabour β18 −0.4778*** 0. 1261 −3.79 0.000
LnSeedquan*LnFarmsize β19 −0.5807*** 0.1429 −4.06 0.000
LnLabour*LnFarmsize β20 −0.0289 0.0809 −0.36 0.721
Sigma square 0.8724
Gamma 0.9007
Mean technical efficiency 0.6973
Log likelihood −266.857
Wald chi2(20) 174.39
***p < 0.001; **p < 0.05; *p < 0.1
increased if farmers had operated at the optimal output study further shows productivity to be highly responsive
scale. This finding confirms similar results obtained by to the quantity of seed used which is an indication that
Abdulai, Nkegbe, and Donkoh (2018) and Wongnaa productivity is more responsive to seed than labour and
et al. (2019) in Ghana. The study also found a gamma land. This result is in conformity with the assumption of
(γ) value of 0.9007. This portrays that 90% of the vari- Kibaara (2005) which noted that there is a likelihood for
ation in output that the farmers obtained was due to inef- maize farmers to improve their maize productivity when
ficiency with random effects accounting for the remaining they make use of improved seed varieties.
10% (Table 8).
Productivity analysis Determinants of technical inefficiency of maize
The degree to which total maize output responds to a farmers
change in any of the inputs used in production is essential Table 10 shows the estimates of the determinants of tech-
in analyzing how productive the inputs are in increasing nical inefficiency of maize farmers in the Techiman Muni-
total maize output. Since the production inputs are esti- cipality. Adoption, which is a weighted sum of the number
mated at the mean, the coefficient obtained from the sto- of improved technology package elements adopted by the
chastic frontier model becomes the input elasticity. The farmers, has a negative coefficient as hypothesized and
study records return to scale of 1.26%, which indicates has a highly significant influence on technical efficiency
an increasing return to scale as presented in Table 9. A at the 1% level of significance. This implies maize
variation of all inputs in the same proportion of 1% is farmers who adopt more elements from their package
expected to lead to 1.26% increase in total output. This accompanied with other best practices of farming are
implies that the use of more inputs of seed, labour and technically more efficient. Abdul-hanan, Ayamga, and
farm size would increase maize output to an efficient Donkoh (2013) and Johnson (2013) found similar
level in the Techiman Municipality. The result of increas- results in Ghana and Western Region respectively. The
ing return to scale is consistent with the results of a study study also found age to influence technical inefficiency
by Wongnaa et al. (2019) who also found an increasing negatively, at the 5% level of significance, which
return to scale among maize farmers in Ghana. The implies that older farmers are more technically inefficient
compared with younger farmers. Kuwornu, Amoah, and
Seini (2013) also confined similar result among maize
Table 9: Elasticity of production and returns to scale. farmers in the Eastern Region of Ghana.
Variables Elasticity P-value The negative result of the soil fertility perception at
Seed quantity/hectare 0.45 0.030** 1% level of significance shows farmers who perceived
Labour/hectare 0.39 0.001*** their soil to be less fertile are more technically inefficient
Land 0.42 0.005*** compared with those who perceived their soil to be fertile.
Returns to scale 1.26 Farm ownership also presents a negative influence on
African Journal of Science, Technology, Innovation and Development 11
Table 10: Determinants of technical inefficiency.
Variables Coefficient Standard Err. z-value p-value
Constant 3.6839*** 1.0316 3.57 0.000
Adoption −1.9637*** 0.4784 −4.10 0.000
Gender −0.1073 0.1961 −0.55 0.584
Age −0.0944** 0.0405 −2.33 0.020
Age-square 0.0009** 0.0004 2.25 0.024
Formal education −0.0267 0.0184 −1.46 0.145
Household size 0.0402* 0.0238 1.69 0.092
Extension contact −0.1614 0.2192 −0.74 0.462
Credit access −0.1564 0.1873 −0.84 0.404
Farm-based organization 0.2774 0.2052 1.35 0.177
Distance to market centre 0.1762** 0.0840 2.10 0.036
Soil fertility perception −0.5006*** 0.1670 −3.00 0.003
Land ownership −0.5211*** 0.1893 −2.75 0.006
Livestock ownership −0.2967* 0.1612 −1.84 0.066
***p < 0.001; **p < 0.05; *p < 0.1
technical inefficiency at the 1% level of significance, in the Ashanti Region of Ghana. Though a large house-
suggesting farmers who grow maize on their own lands hold size is regarded as an increased source of family
are more efficient compared with those who used other labour for farm activities, it also has the disadvantage of
forms of land. Farmers who plant on their own lands increasing the financial burden on farmers in relation to
devote more time to their farming activities which trans- the consumption expenses of household members and
lates into increased productivity since they own every- other upkeeps. An increase in the financial obligation of
thing harvested from their crops. Johnson (2013) also farmers towards their household reduces the number of
found a negative effect of land ownership on technical resources available for farm activities, which increases
inefficiency in the Western Region of Ghana. In addition, the technical inefficiency level tremendously. The posi-
livestock ownership influenced technical inefficiency tive sign of the distance to market centres also met its
negatively, at the 10% level of significance. Farmers expected sign and is statistically significant at the 5% sig-
who reared livestock in addition to their farming activities nificance level. As market distance increases, farmers find
were more efficient technically. Ownership of livestock is it more difficult to easily purchase agriculture inputs and
used as a proxy for measuring the wealth status of raw materials they need for their farming activities. It also
farmers. It is also a source of fertilizer manure and increases the cost of transporting these goods and services
income for the purchase of farm inputs and improved to the farm site which makes it difficult for farmers to
technology. Thus, farmers who possess a large number effectively manage their operations efficiently. Thus,
of livestock tend to be technically more efficient. Kaba farmers closer to market centres are more efficient techni-
(2016) also found a negative effect of livestock ownership cally. This finding confirms the results obtained by Kaba
on technical inefficiency in South-Western Ethiopia. (2016). Abdul-hanan, Ayamga, and Donkoh (2013),
Household size, the distance to market centre and age however, found a negative influence of market distance
square on the other hand have a positive influence on tech- to technical inefficiency. According to the study, technical
nical inefficiency at the 5%, 10% and 5% levels of signifi- efficiency goes beyond proximity to an input market. It
cance, respectively. The result of household size demands undivided attention and whether farmers live
corroborates what Yiadom-Boakye et al. (2013) found in rural, peri-urban areas or urban areas is immaterial.
Table 11: Ranking of constraints faced by maize farmers.
Constraints Mean rank Rank
Low selling price of maize 2.15 1
Low rainfall 2.55 2
High input price 2.82 3
Lack of credit 4.90 4
Low soil fertility 5.16 5
Insecure land tenure systems 7.35 6
Shortage of improved seed 7.43 7
Inadequate government incentives 7.45 8
Lack of extension contact 8.54 9
Pest and disease outbreaks 8.62 10
Lack of harvesting and drying equipment 9.02 11
Sample size 407
Kendall’s W 0.611
Chi-Square 2486.187
Degrees of freedom 10
Significance level 0.000***
***p < 0.01
12 Kwawu, Sarpong and Agyire-Tettey
Age square also has a positive influence on technical inef- all-inclusive approach in policy formulation to increase
ficiency at the 5% level of significance. For every the intensity of adoption of package elements by
additional year in age, farmers become less productive, farmers to nullify the remaining 30% deficit, with priority
which affects their output level and level of technical effi- given to technical efficiency. Furthermore, there is a need
ciency. The negative coefficient is in conformity with a to increase extension contact, provision of incentives to
similar study by Kuwornu, Amoah, and Seini (2013) encourage the youth to go into farming, farm-based
who also reported a negative influence of age on technical organizations and improved seeds to enable farmers
inefficiency. It, however, contradicts findings by Essilfie, operate at the optimal scale. Farmers are advised not
Asiamah, and Nimoh (2011) and Coelli and Battese only to increase their use of farm inputs, but to use
(1996) who reveal that younger farmers tend to be more them at the recommended rate in order to obtain
reluctant to adopt improved technology and new ways maximum output from the package elements. The study
of farming which makes them less efficient. recommends that farmers be encouraged through incen-
Estimates from Table 11, which presents the con- tive packages to adopt more elements from the improved
straint ranking of the maize farmers in the Techiman maize technology package since evidence has shown that
Municipality, indicates low selling price, low rainfall, adoption increases productivity as well as moving farmers
high input price and lack of credit as the four most closer to the frontier output. Since scale efficiency of
severe constraints out of the 11 constraints listed that maize farmers is higher than their technical efficiency,
maize farmers faced. Other constraints, in descending the study recommends that policies are targeted at
order, include low soil fertility, insecure land tenure encouraging farmers to use improved maize technology
systems, a shortage of improved seed, inadequate govern- and agronomic practices to increase their technical effi-
ment incentives, lack of extension contact, pest and ciency of maize production. Also, these policies should
disease outbreaks, and, lastly, lack of harvesting and be geared more towards enhancing technical efficiency
drying equipment. Results from the Kendall’s Coefficient than scale efficiency. Thus, where divergent effects exist
of Concordance also reject the null hypothesis that no between adoption intensity and technical efficiency of
agreement exists among the rankings of the farmer at output, technical efficiency should be accorded higher pri-
the 1% level of significance with approximately 61% ority since adoption is only a means to achieving technical
agreement in ranking. efficiency of output. Due to the unavailability of farm-
level panel data, this study was carried out with the help
Conclusion and recommendation of a cross-sectional data. Since a cross-sectional analysis
The objectives of this study were to analyze the factors is limited in tracing scale efficiency dynamics, this study
that determine the intensity of adoption of an improved recommends future researchers to explore the dynamics
maize technology package and the technical efficiency of scale efficiency through farm-level panel data in
achieved by farmers in the Techiman Municipality. The order to track efficiency of farmers over time.
study also aimed to examine the sources of the technical
inefficiency and the constraints maize farmers experi- Funding
enced in the adoption of the improved technology This work was supported by USAID/University of Ghana Insti-
package elements. The Poisson model was employed to tutional Capacity Building Agriculture Productivity Project.
achieve the first objective while the stochastic frontier
model was used to address the second and third objec- Disclosure statement
tives. The fourth objective was estimated using the Nopotential conflict of interestwas reportedby the author(s).
Kendall Coefficient of Concordance. The study found
extension contact, formal training, land ownership, hired ORCID
labour, farm-based organization, mobile phone owner- Joana Deladem Kwawu http://orcid.org/0000-0003-
ship, distance to market centre, price perception, and 2205-7035
farm size as significant policy variables that influence Frank Agyire-Tettey http://orcid.org/0000-0001-9414-
adoption of improved maize technology package elements 7342
in the Techiman Municipality of Ghana. Seed quantity,
labour and farm size also emerged as the significant vari-
References
ables that influence productivity, whereas age, age-square, Abdul-hanan, A., M. Ayamga, and S. A. Donkoh. 2013.
household size, distance to market centre, soil fertility “Smallholder Adoption of Soil and Water Conservation
perception, land and livestock ownership contribute sig- Techniques in Ghana.” African Journal of Agricultural
nificantly to technical inefficiency. Research 9 (5): 539–546.
The majority of the farmers were, however, producing Abdulai, S., P. K. Nkegbe, and S. A. Donkoh. 2018. “Assessing
the Technical Efficiency of Maize Production in Northern
on average at 70% of capacity, indicating that they are not Ghana: The Data Envelopment Analysis Approach.”
operating on the frontier. The implication is that farmers Cogent Food & Agriculture 4: 1.
could increase output by 30% if the full set of inputs Abukari, A., S. Hussein, and S. Katara. 2015. “Determinants of
were employed at their optimal scale. Hence, farmers in Farmers’ Adoption of Improved Maize Varieties in the Wa
the Techiman Municipality were found to be scale ineffi- Municipality.” American International Journal of
Contemporary Research 1 (4): 27–33.
cient. This resulted in a productivity estimate exhibiting Ahmed, M. H., A. Tazeze, A. Mezgebo, and E. Andualem. 2018.
increasing returns to scale of 1.26, indicating a fall in “Measuring Maize Production Efficiency in the Eastern
output below the optimal level. The study proposes an Ethiopia: Stochastic Frontier Approach.” African Journal
African Journal of Science, Technology, Innovation and Development 13
of Science, Technology, Innovation and Development 10 (7): Essilfie, F. L., M. T. Asiamah, and F. Nimoh. 2011. “Estimation of
779–786. Farm Level Technical Efficiency in Small Scale Production in
Aigner, D. J., and S. F. Chu. 1968. “On Estimating the Industry the Mfantseman Municipality in the Central Region of
Production Function.” American Economic Review 58: 826– Ghana: A Stochastic Frontier Approach.” Journal of
839. Development and Agricultural Economics 3: 645–654.
Aigner, D. J., C. A. K. Lovell, and P. Schmidt. 1977. FAO (Food and Agriculture Organization of the United
“Formulation and Estimation of Stochastic Frontier Nations). 2017. The future of Food and Agriculture,
Production Function Models.” Journal of Econometrics 6: Trends and challenges. Rome, Italy: Food and Agriculture
21–37. Organization of the United Nations.
Aikins, S. H. M., J. J. Afuakwa, and O. Owusu-Akuoko. 2012. Farrell, M. J. 1957. “The Measurement of Productive
“Effect of Four Different Tillage Practices on Maize Efficiency.” Journal of the Royal Statistical Society. Series
Performance Under Rain Fed Conditions.” Agriculture A (General) 120: 253–290.
and Biology Journal of North America 3 (1): 25–30. Fikru, Assefa Mengstie. 2009. “Assessment of Adoption
Alhassan, A., H. Salifu, and A. O. Adebanji. 2016. Behaviour of Soil and Water Conservation Practices in the
“Discriminant Analysis of Farmers’ Adoption of Improved Koga Watershed, Highlands of Ethiopia.” MPS Thesis.,
Maize Varieties in Wa Municipality, Upper West Region Cornell University.
of Ghana.” Springer Plus 5 (1): 1514. Food and Agriculture Organization of the United Nations, FAO.
Assefa, M., and J. Gezahegn. 2010. “Adoption of Improved 2010. Global Forest Resources Assessment. Report of the
Technology in Ethiopia.” Ethiopian Journal of Economics World Food Summit. Rome: Publishing Policy and
19 (1): 201–221. Support Branch, Food and Agriculture Organization of the
Aye, G. C., and E. D. Mungatana. 2011. “Technological United Nations.
Innovation and Efficiency in the Nigerian Maize Sector: Food and Agriculture Organization of the United Nations,
Parametric Stochastic and Non-Parametric Distance FAO. 2017. The future of Food and Agriculture, Trends
Function Approaches, Agrekon.” Agricultural Economics and Challenges. Rome, Italy. Food and Agriculture
Research, Policy and Practice in Southern Africa 50 (4): Organization of the United Nations.
1–24. Greene, W. H. 2003. Econometric Analysis. 5th ed. Upper
Battese, G. E., and S. S. Broca. 1997. “Functional Forms of Saddle River, NJ: Pearson Education Ltd.
Stochastic Frontier Production Functions and Models for Hailemariam, T., K. Menale, and S. Bekele. 2013. “Adoption of
Technical Inefficiency Effects: A Comparative Study for Multiple Sustainable Agricultural Practices in Rural
Wheat Farmers in Pakistan.” Journal of Productivity Ethiopia.” Journal of Agricultural Economics 64 (3): 597–
Analysis 8: 395–414. 623.
Battese, G. E., and T. J. Coelli. 1995. “A Model for Technical Headey, D., D. Alauddinb, and P. Rao. 2010. “Explaining
Inefficiency Effects in a Stochastic Frontier Production Agricultural Productivity Growth: An International
Function for Panel Data.” Empirical Economics 20: 325– Perspective.” Agricultural Economics 41: 1–14.
332. Johnson, E. N. 2013. “Adoption of Improved Technology and
Batz, F. J., K. J. Peters, and M. Janssen. 1999. “The Influence of Farm Level Technical Efficiency of Small-Scale Oil Palm
Technology Characteristics on the Rate and Speed of Producers in the Western Region of Ghana.” Thesis.,
Adoption.” Agricultural Economics 21: 121–130. Department of Economics and Agribusiness, University of
Cameron, A. C., and P. K. Trivedi. 1990. “Regression-Based Ghana.
Test for Over Dispersion in the Poisson Model.” Journal Kaba, S. D. 2016. Agricultural Technology Adoption, Crop
of Econometrics 46: 347–364. Diversification and Efficiency of Maize-Dominated
Cameron, A. C., and P. K. Trivedi. 1998. Microeconometrics Smallholder Farming System in Jimma Zone, South-
Using Stata. College Station, TX: Stata Press. Western Ethiopia. Haramaya: Haramaya University.
Cameron, A. C., and P. K. Trivedi. 2009. Microeconometrics Kaliba, R. M. A., H. Verkuijl, and W. Mwangi. 2000. “Factors
Using Stata. College Station, TX: Stata Press. Affecting Adoption of Improved Maize Seeds and Use of
Coelli, T., and G. Battese. 1996. “Identification of Factors Inorganic Fertilizer for Maize Production in the
Which Influence the Technical Inefficiency of Indian Intermediate and Lowland Zones of Tanzania.” Journal of
Farmers.” Australian Journal of Agricultural Economics Agricultural and Applied Economics 32 (1): 35–47.
40 (2): 103–128. Kibaara, W. B. 2005. “Technical Efficiency in Kenyan’s Maize
Coelli, J., P. Rao, J. Christopher, O. Donnell, and E. Battese. Production: An Application of the Stochastic Frontier
2005. An Introduction to Efficiency and Productivity Approach.” Unpublished Masters Thesis., Department of
Analysis. 2nd ed. New York: Springer Science and Agricultural and Resource Economics, Colorado State
Business Media. University.
Dasgupta, S. 1989. Diffusion of Agricultural Innovations in Kodde, D., A. and F. C. Palm. 1986. “Wald Criteria for Jointly
Village India. Department of Sociology and Anthropology. Testing Equality and Inequality Restrictions.”
New Delhi: Wiley Eastern Limited. Econometrica 54 (5): 1243–1248.
Doss, C. R. 2003. Understanding Farm Level Technology Kuwornu, J. K. M., E. Amoah, and W. Seini. 2013. “Technical
Adoption: Lessons Learned from CIMMYT’s Micro- Efficiency Analysis of Maize Farmers in the Eastern
Surveys in Eastern Africa. CIMMYT Economics Working Region of Ghana.” Journal of Social and Development
Paper 03-07. Mexico, DF: CIMMYT. Sciences 4 (2): 84–99.
Doss, C. R., and M. Morris. 2001. “HowDoes Gender Affect the Kuwornu, J. K. M., D. M. Suleyman, and D. P. K. Amegashie.
Adoption of Agricultural Innovations?: The Case of 2011. “Analysis of Food Security Status of Farming
Improved Maize Technology in Ghana.” Agricultural Households in the Forest Belt of the Central Region of
Economics 25 (1): 27–39. Ghana.” Russian Journal of Agricultural and Socio-
Endrias, G., K. Belay, B. Ayalneh, and E. Eyasu. 2010. Economic Sciences 1 (13): 134–157.
“Productivity and Efficiency Analysis of Smallholder Mashingaidze, Arnold B. 2004. “Improving Weed Management
Maize Producers in Southern Ethiopia.” Journal of Human and Crop Productivity in Maize Systems in Zimbabwe.”
Ecology 41 (1): 67–75. Ph.D. Thesis., Wageningen University.
Ephraim, W. 2007. Sources of Technical Efficiency Among Mensah-Bonsu, A., D. B. Sarpong, R. Al-Hassan, A. Asuming-
Smallholder Maize Farmers in Southern Malawi. African Brempong, I. Egyir, E. John, and K. Osei-Asare. 2017.
Economic Research Consortium, Research Paper 172. “Intensity of and Factors Affecting Land and Water
Nairobi. Management Practices Among Smallholder Maize
14 Kwawu, Sarpong and Agyire-Tettey
Farmers in Ghana.” African Journal of Agricultural and Farmers: Evidence from Parametric and Non-Parametric
Resource Economics 12 (2): 142–157. Count Data Models.” Journal of Agricultural Economics
Minten, B., and C. B. Barrett. 2008. “Agricultural Technology, 62 (1): 73–92.
Productivity, and Poverty in Madagascar.” World Solomon, A., S. Bekele, S. Franklin, and G. H. Mekbib. 2011.
Development 36 (5): 797–822. “Agricultural Technology Adoption, Seed Access
MoFA (Ministry of Food and Agriculture). 2010. Agriculture in Constraints and Commercialization in Ethiopia.” Journal of
Ghana: Facts and Figures (2009). Accra: Statistics, Development and Agricultural Economics 3 (9): 436–447.
Research, and Information Directorate. SRID (Statistical Research and Information Department), &
MOFA (Ministry of Food and Agriculture). 2011. Agriculture in MoFA (Ministry of Food and Agriculture). 2015. National
Ghana: Facts and Figures (2010). Accra: Statistics, Crop Production Estimates 2014. Accra, Ghana: Statistics,
Research, and Information Directorate. Research, and Information Directorate.
Moti, J., Y. Chilot, K. Menale, H. D. Groote, and S. Bekele. Tura, M., A. Dejene, T. Wondesen, T. R. G. Tesfahun, M.
2013. “Knowledge, Adoption and Use Intensity of Mwangi, and G. Mwabu. 2010. “Adoption and Continued
Improved Maize Technologies in Ethiopia.” Invited Paper Use of Improved Maize Seed in Ethiopia.” African
Presented at the 4th International Conference of the Journal of Agriculture Research 17: 2350–2358.
African Association of Agricultural Economists, Tweneboah, C. K. 2000. Modern Agriculture in the Tropics,
Hammamet, September 22–25. Food Crops. Accra, Ghana: Co-wood Publishers.
Mulugeta, A. 2011. “Factors Affecting Adoption of Improved Uri, N. 2002. “The Adoption of Incentive Regulation and Its
Haricot Bean Varieties and Associeted Agronomic Effect on Technical Efficiency in Telecommunications in
Practice in Dale District, SNNPRS.” MSc Thesis., the United States.” International Journal of Production
Hawassa University, Hawassa. Economics 2: 1–4.
Mureithi, J. G., and J. Ojiem. 2000. “The Role of Green Manure USAID/IFDC. 2015. “Seed Guide: Recommended Commercial
Legumes in Smallholder Farming Systems in Kenya: An Maize, Rice and Soybean Varieties Available for Northern
Overview of Legume Research Network Project.” Paper Ghana.” file:///C:/Users/delad/Downloads/Seed%20Guide
Presented in a 9 Symposium on Fallow Management in %20-%20Recommended%20Commercial%20Maize,%
the Tropics, Minneapolis, MN, November 5–9. 20Rice%20and%20Soybean%20Varieties%20Available%
Nkegbe, P., and B. Shankar. 2014. “Adoption Intensity of Soil 20for%20Northern%20Ghana%20(1).pdf.
and Water Conservation Practices by Smallholders: Von Neumann, J., and O. Morgenstern. 1947. Theory of Games
Evidence from Northern Ghana.” Bio-Based and Applied and Economic Behavior, First Edition, 1944; Third Edition,
Economics 3 (2): 159. 1953. Princeton, NJ: Princeton University Press.
Ogundari, K., and S. O.Ojo. 2007. “Economic Efficiency of Small Voto. 2015. “A Complete Curriculum and Guide to Maize
Scale Food Crop Production in Nigeria: A Stochastic Frontier Production in Ghana.” http://aamiafrica.com/wp-content/
Approach.” Journal of Social Science 14 (2): 123–130. uploads/2017/10/maize-production-guide-ghana-2016.pdf.
Olwande, J., G. Sikei, and M. Mathenge. 2009. Agricultural Wandji, D. N., V. Pouomogne, J. N. Binam, and R. Y. Nouaga.
Technology Adoption: A Panel Analysis of Smallholder 2012. “Farmer’s Perception and Adoption of New
Farmers’ Fertilizer Use in Nairobi. Kenya: AERC. Aquaculture Technologies in the Western Highlands of
Onumah, E. E., and H. D. Acquah. 2010. “Frontier Analysis of Cameroon.” Tropicultura 30 (3): 180–184.
Aquaculture Farms in the Southern Sector of Ghana.”World Wongnaa, C. A., D. Awunyo-Vitor, A. Mensah, and F. Adams.
Applied Sciences Journal 9 (7): 826–835. 2019. “Profit Efficiency Among Maize Farmers and
Oppong, B. A., E. E. Onumah, and S. Asuming-Brempong. 2016. Implications for Poverty Alleviation and Food Security in
“Technical Efficiency and Production Risk of Maize Ghana.” Scientific African 6: e00206.
Production: Evidence from Ghana.” Asian Journal World Bank. 2010. Annual Report on Implementing Agriculture
of Agriculture Extension, Economics and Sociology 11 (3): for Development. Washington, DC: World Bank.
1–9. World Bank. 2015. The World Bank Annual Report 2015.
Salifu, H., A. Alhassan, and K. Salifu. 2015. “Determinants of Washington, DC: World Bank.
Farmers’ Adoption of Improved Maize Varieties in the Wa Yiadom-Boakye, E., E. Owusu-Sekyere, P. K. Nkegbe, and K.
Municipality.” American International Journal of Ohene-Yankyera. 2013. “Gender, Resource Use and
Contemporary Research 1 (4): 27–33. Technical Efficiency Among Rice Farmers in the Ashanti
Sharma, A., A. Bailey, and I. Fraser. 2011. “Technology Region, Ghana.” Journal of Agricultural Economics 2 (3):
Adoption and Pest Control Strategies Among UK Cereal 102–110.