Content uploaded by Bakary Sanou
Author content
All content in this area was uploaded by Bakary Sanou on Sep 18, 2017
Content may be subject to copyright.
[Jpn.
J.
Agric. Econ.
Vol.19,
pp.21-26,
2017]
Determinants
of
Adoption and Continuous Use
of
Improved Maize Seeds in Burkina Faso
Bakary
Sanou\
Kimseyinga Savadogo2 and Takeshi Sakurai1
*
This study aims to identifY socio-economic and institutional factors that promote the adoption and continuous use
of
improved seeds
in
maize
production in Burkina Faso. We use nationally representative panel data and confirm low rates
of
adoption
and
continuous use
of
improved maize seeds. Using a bivariate probit
model
with sample selection,
we
find
that
while
the adoption
of
improved seeds increases with access to credit, membership
in
farmer organization, and contact
with
extension services, its continuous use is
not
affected
by
such institutional factors after controlling for province level
fixed effects
that
capture agro-ecological variations
in
the country.
Key
words:
improved maize seeds, continuous use
of
technology, Burkina Faso
1.
Introduction
The use
of
agricultural technologies such as chemical
fertilizers, pesticides, and improved seeds, has long been
considered as an effective pathway to increase agricultural
productivity in Sub-Saharan Africa (Feder eta!., 1985; Minten
and Barrett, 2008; Saka and Lawai, 2009). Although combined
use
of
these technologies
is
often recommended (World Bank,
2008; Hailemariam et
a!.,
2013), improved seeds in particular
play an important role in this process as this input alone may
contribute to about 40% increase in yields (Bikienga, 2002).
Thereby the use
of
improved seeds is essential for the
transformation
of
subsistence farming that
remains
in many
African countries into market oriented ones.
In Burkina F aso agricultural productivity remains low in
general, especially that
of
cereal crops. Average yield
of
cereal
from2002 to 2012 was 1.04
t/ha
inBurkinaFaso, but it was over
6.50 t/ha in some developed countries like United States, 1.75
and
1.51
t/ha
respectively in Cote d'Ivoire and Ghana according
to FAOSTAT. Certainly this situation
is
strongly due to some
natural constraints such as climatic hazard and poor quality
of
soils. However, the weak adoption
of
new technologies
is
manifested to be one
of
the explicative factors. Indeed, irrigation
still overstays at an embryonic stage with only less
than
14%
of
potential irrigable lands effectively cultivated. Further, chemical
fertilizers are applied to less
than
31%
of
land area for cereal
production, and the application rate
is
estimated to be about
19
kg·'lm
(Combary, 2013), while it
is
over
145
kg!ha in developed
world (Diiro and Sam, 2015). Moreover, less than 15%
of
1
The
University
ofTokyo
2
Universite
Ouaga
2
Corresponding
author*:
atsakura@mail.ecc.u-tokyo.acJp
farmers use improved ll!aize seeds (CEFCOD, 2013).
Because
of
the significant role
of
new technologies in
agricultural production and their low adoption rates in the
number
of
developing countries including Burkina Faso,
technology adoption in agriculture has been empirically a topic
of
considerable interest by scholars. A wide literature on this
subject has identified categories
of
socio-economic, institutional
and environmental factors as main determinants
of
their
adoption in developing countries (Feder eta!., 1985; Saka and
Lawai, 2009) and in Burkina Faso as well (Savadogo et
a!.,
1998; Adeoti eta!., 2002; Combary, 2013). However, although
the low adoption rates are partly due to disadoption (i.e. farmers
who once adopted a new technology have stopped using
it)
for
example Moser and Barrette (2003) and deGraft-Johnson eta!.
(2014), only a few studies have been focused on the factors
affecting continuous or discontinuous use
of
adopted
technologies. Examples
of
such a few studies are Oladele (2005),
Neill and Lee
(2001
),
and Tura eta!. (2010).
Oladele (2005), in the case ofNigeria, uses an index number
for farmer's degree
of
discontinuation
of
adopting improved
maize varieties as the dependent variable and identifies variables
explaining the index by a single equation Tobit model. On the
other hand, both Neill and Lee (2001) and Tura eta!. (2010)
assume sequential decision making, where a farmer adopts a
new technology and then decides to continue using it or to
disadopt it because only adopters can be disadopters. In other
words, their analysis distinguishes never -adopters and
disadopters
of
a new technology. In order to deal with the two
22
decisions,
namely
to
adopt and
to
continuously
use
(or not
to
disadopt)
the
technology
in
question,
they
apply
a bivariate
probit model with
sample
selection
for
the
case
of
cover
crop
technology
for
maize
production
in
Honduras
(Neill
and
Lee,
200
I)
and
the
case
of
improved
maize
seeds
in
Ethiopia
(Tura
et
a!.,
2010),
and
show
the
importance
of
the
role
played by
institutional
factors
such
as
access
to
credit,
extension
services,
and membership
of
farmer
organization
in
farmer's decision
to
pursue using
one
technology.
There
are
three
points
of
common weakness
in
the
previous
studies.
First
is
small
sample
size:
60
in
Oladele
(2005),
370
in
Neill and
Lee
(2001
),
and
120
in
Tura
eta!.
(20
10).
None
of
them
can
claim that
the
study
uses
representative
data
of
a country
or
a
region/state.
Second
is
the
lack
of
time
dimension.
All
of
them
use
cross-section
data,
and explain adoption and continuous
use
of
a technology by current
and
time-invariant
variables
although
the
decisions
as
to
the
adoption
and
the
continuous
use
should
have
made
in
the
past Third
is
potential endogeneity
of
the
institutional
factors
used in
their
regression
analyses,
that
is,
unobservable
variables
may
affect
both
institutional
factors
and
technology adoption/continuous
use.
This
paper
focuses
on
the
low
adoption
rate
of
improved
technologies
in
Burkina
Paso,
especially improved
maize
seeds,
and
applies
a bivariate probit model with
sample
selection
like
Neill and
Lee
(2001)
and
Tura
eta!.
(2010)
to
identifY
socio-
economic and
institutional
factors
that determine
the
adoption
of
improved
maize
seeds
and
those
influencing farmers' decision
to
continuously
use
them.
However,
in
order
to
improve
the
existing
studies
and
to
make
academic
contribution,
we
use
representative data ofBurkina Paso and predetermined
variables
to
predict
decisions
to
be made
in
the
future.
In
addition,
since
our data cover
all
the
45
provinces
of
Burkina
Paso,
province
fixed
effects
are
used
to
control
for
observable/unobservable
province-level
variations
that
affect
both institutional
factors
and
technology
adoption/continuous
use.
2.
Methodology
1)
Analytical framework
Adoption and continued
use
of
one
technology
are
outcomes
of
two
interrelated
decisions.
Indeed,
continuous
use
is
subsequent
to
the
decision
to
adopt
the
technology and
can
be
observed only among
the
adopters.
However,
factors
explaining
the
two
decisions
can
be
different
for
a given
farmer,
although
some
factors
can explain
the
both.
Because
of
the
subordination
in
both
decisions,
it
is
necessary
to
take
into
account
of
potential
correlation between
the
unobservable
factors
captured by
the
error
terms,
and
a
bivariate
probit
model
with
sample
selection
is
suggested
(Neill
and
Lee,
2001;
Tura
eta!.,
2010).
2)
Bivariate probit model
with
sample
selection
We
assume
that
adoption
decision
is
motivated by
the
expected
utility,
which
is
a
function
of
the
expected profit
Thus,
a household
will
decide
to
apply
one
technology
like
improved
seeds
on
its
farm
if
he
predicts a
positive
profit
relative
to
the
case
without
adoption
estimated
by
available
information
including output
and
input
prices.
Then
in
the
following
year,
this
household
will
revise
the
prediction
of
profit based
on
the
performance
in
the
last year
and
decide
to
continue
its
use
ifhe
still
predicts a positive profit
The
bivariate probit
model
with
sample
selection
is
similar
to
the
Heckman's
sample
selection
model.
However
unlike
in
the
Heckman's
model,
the
outcome
equation
is
also
a probit model
in
the
bivariate
model
with
sample
selection.
Let us
assume
(yk) k=l, 2
to
be unobservable farmer's
utility
perceived
from
adoption
of
improved
maize
seeds
(yi)
and
from
continuous
use
of
them
(yD
depending
on
a vector
X
of
explanatory
variables.
Suppose
that
(Yk) k=l, 2
are
binary
variables
taking
unity
in
the
case
of
adoption
of
improved
maize
seeds
and in
the
case
of
continuous
use
of
them
respectively
and
0
otherwise.
Then
the
standard
bivariate
probit
model
with
additive
error
is
specified
as:
* X'[J
{Y1
=
1,
if
Yi
>
0
y
1
=
1 1
+
£
1;
y
1
=
0,
otherwise
(1)
*
X'{J
{Y2
=
1,
if
Yi
>
0
andy~
>
0
Y
=
+E.
2 2 2 2'
y
2
=
0,
otherwise
(2)
where
X
1
and
X
2
are
a
vector
of
explanatory
variables
of
the
adoption
decision and
the
continuous
use
respectively,
{3
1
and
{3
2
are
parameters
to
be
estimated,
and
£
1
and
£
2
are
error
terms.
The
log-likelihood
function
of
the
model
is
given
by
the
following
equation:
lnL
=
If
{Yi1Yi2lnct>2
(X~f3v
X~f32,
P)
+Yi1
(1-
Yi2)
lnC/>2(X~fJv
-X~f32,
-p)
+(1-
Yi1)lnC/>1(-X~f31)}
(3)
where i=l,
2,
... ,
N.
In
this
specification,
cJ>
1
is
the
univariate normal
distribution,
and
C/>
2
is
the
bivariate
normal
distribution.
Yi
1
and
Yi
2
are
binary
variables
taking
unity
if
farmer
i
adopts
improved
maize
seeds
and
if
farmer
i
continuously
use
them
respectively
and
0
otherwise.
And
p
is
the
coefficient
of
correlation.
3.
Data and
Statistics
Analysis
1)
Data
and
variables
created
Our
study
utilizes
data
of
the
Agricultural
Permanent
Survey
Research Letters
23
collected by
the
Ministry
of
Agriculture
of
Burkina Faso from
2009/2010
to
2012/2013.
This
survey,
regularly conducted each
year since 2009/2010,
forms
a four-year panel
of
national
representative sample
of
4130 households drawn from
all
the
45
provinces,
of
which we select
2043
maize producers spread over
406 villages in 4
5 provinces
for
the
analysis.
The model
has
two dependent variables: adoption and
continuous use
of
improved
maize
seeds.
These variables have
been observed from the same farmers every year from 2009/10
to
2012/13.
We
take advantage
of
the
panel structure
to
define
"never-adopter", "disadopter", and "continuous user"
as
explained in the following paragraphs.
In
this paper we
classifY
maize seeds
into
either "improved"
or "conventional" depending on :funner's judgement because
they can tell which seeds
are
improved in most cases. Since the
survey does not record the name
of
varieties or their sources, we
cannot use such information
for
the
classification
of
maize
seeds.
1
l
Therefore, farmer's judgement
is
the most reliable
information available
for
us.
As
for
the
definition
of
adoption,
this
paper considers a farmer
as
an adopter
of
improved maize seeds
if
he/she planted
improved maize seeds (as defined earlier) even partially at least
once during the 4 years surveyed; and
as
a never-adopter
of
improved
maize
seeds
if
he/she never planted improved
maize
seeds during the 4 years surveyed.
Then,
adopters can be further
classified
as
either continuous users, disadopters, or others (cases
not possible
to
be classified): continuous users are farmers who
planted improved maize seeds every year during the 4 years
surveyed or farmers who did not plant improved maize seeds in
the first year surveyed
(i.e.
2009/10) but planted improved
maize
seeds during the last three years
(i.e.
from
20010/11
to
2012113);
and disadopters are
:funners
who experienced "disadoption"
at
least once during the 4 years surveyed "Disadoption"
is
defined
as
the case where a farmer adopted improved
maize
seeds in a
year and he/she did not adopt (or disadopted) them in the
following
year.
Thus, in order
for
a farmer
to
be a disadopter,
he/she must have adopted improved maize seeds in either
2009/10, 2010/11, or 2011/12 and disadopted them in the
respective following
year.
We
use this definition
of
disadoption
since our concern
is
the
stability or instability
of
technology
1 )
In
Burkina
F
aso
both
open-pollinated
and
hybrid
varieties
are
available
as
improved
maize
seeds.
While
open-pollinated
improved
seeds
can
be
recycled,
conventional
varieties
may
sometimes
be
sold
in the
market.
Therefore,
the
sources
of
seeds
are
not
important
information
for
the
classification.
2)
Since
we
have
4
years,
the
number
of pennutations
of"!
=adopf'
and
"O=non-adopt"
is
16.
Out
of
the
16
pennutations,
the
excluded
adoption regardless
of
current status
of
technology
use.
There
are
two cases that are neither continuous users nor disadopters
(i.e.
others).
One
is
the case where a farmer did not adopt
in
the first
two years and adopt in
the
last two
years.
The
other
is
the
case
where a farmer did not adopt in the first three years and adopted
in the last
year.
Since we cannot
classifY
them based on our
definition, we
drop
such cases from the
analysis.
2l
By definition,
the
classification
of
never -adopter, disadopter,
and continuous user
is
a
fixed
household characteristic over
the
survey period, or in other words cross-sectional variation. Thus,
our empirical strategy
is
to
estimate equation
(3)
cross-
sectionally using only predetermined variables
at
the beginning
of
the
survey period, namely in 2009/10. Since
the
latent
variables in equations
(1)
and
(2)
depend on
the
expected profit,
explanatory variables should be ones that can affect
it
Hence,
as
the explanatory variables, we consider
sex,
age,
marital
status,
and literacy
of
household
head,
household
size,
livestock
ownership, dry season cropping, quality
of
roof and wall
of
the
house, access
to
credit
3
l,
membership in farmer organization, and
access
to
extension
services.
The last three
(credit,
membership,
and extension)
are
regarded
as
institutional variables in
this
paper,
which can potentially
be
endogenous in the
model.
In
addition, we
use
dururny variables
for
all
the
45
provinces
in order
to
control
for
observable and unobservable province-
level fixed
factors
such
as
mean yield, production risk, expected
input/output prices, and market/infrastructure development
4
l
Although our
data,
just
like
Neill and Lee (2001) and Tura et
al.
(20
I
0),
.do
not allow
us
to
calculate profit realized in the previous
year,
which
as
discussed above will influence farmer's decision
to
continue using improved maize
seeds,
we use province
dururnies
as
proxies
of
expected yield, production risk, and
input/output prices unlike Neill and Lee (200
1)
and Tura et
al.
(2010).
With respect
to
the
yield and production risk, the
45
provinces are comprised
of
10
provinces in
the
Sudanian zone,
25
provinces in the Sudano-Sahelian zone, and I 0 provinces in
the Sahelian
zone.
The three agro-ecological zones are
distinguished by annual rainfall:
the
Sudanian zone located in
the
south-western part
of
the
country with about
1000
mm
of
annual
rainfall in
five
months;
the
Sahelian zone in the north-eastern
part with
less
than 400 mm
of
annual rainfall in three months;
permutations
are
(0,
0,
I,
1)
and
(0,
0, 0,
1).
The
number
of
such
households
are
78
and
129
respectively.
3)
Access
to
credit
means
an
acquisition
of
credit
during
the
latest
twelve
months.
4)
For
the
regression
analysis,
one
of
the
province
dummies
is
excluded
to
avoid
multicollinearity.
Barn
province
in
the
Sahelian
zone
is
excluded
and
is
served
as
the
reference
category.
24
and
the
Sudano-Sahelian
zone
occupying between
the
two.5l
2)
Descriptive
statistics
Right half
of
Table
I
is
for
the
descriptive
statistics
of
the
explanatmyvariables.
It
stands
out that
763
(37.5%)
of
the
2033
sample
maize
producers
have
at
least
once
used improved
maize
seeds
(i.e.
adopters),
and
that continuous users
constitute
only
248
(12.2%)
of
the
sample
maize
producers.
These
results
reflect
the
low utilization
of
improved
seeds
in
maize production
in
Burkina
Faso,
as
shown
in
CEFCOD
(2013).
In
addition,
the
descriptive table
shows
37.1%
of
adopters
received
credits.
Credit
access
is
higher
with
continuous users
(60.5%)
than
with
disadopters
(9.9%).
Regarding
farmer
organization,
53.0%
of
adopters
belong
to
a
farmer
organization
for
crop
production including cotton
production.
While only
20.9%
of
disadopters
are
members
of a
farmer
organization
for
crop
production, 73.4%
of
continuous
users
are
members
of
such
an
organization.
Extension
services
from
cotton
companies
shows
a similar
tendency.
18.5%
of
adopters
received extension
services
from
cotton
companies.
Receivers
of
extension
services
are
much higher
among
continuous
users
(29.4%)
than
disadopters
(4.6%).
Thus,
those
institutional
factors
seem
to
contribute
to
continuous
use
of
improved maize
seeds.
4.
Estimation
Results
and
Discussions
Left part
of
Table
1 presents
the
results
of
the
estimation.
Considering
the
potential endogeneity
of
institutional
variables,
we
estimate
4
models:
model
1
has
no institutional
variables,
model2
includes
credit,
model3
includes
membership
in
farmer
organization,
and
model
4
includes
extension
services.
By
comparing
the
4
models,
the
inclusion
of
the
institutional
variables
affect
little
the
estimation
results
of
other
variables.
Thus,
we assume that province
dummies
control
for
unobservable/observable
factors
well
and
we
interpret
the
estimation results
of
the
institutional
variables
as
they
are.
The
p parameter
is
not significantly
different
from
zero
in
all
the
models,
implying
that
the
residuals
of
the
two probit
equations
are
not significantly
correlated.
However,
since
simultaneous
estimation by maximum likelihood
can
be
more
efficient than
separate
estimation
of
each probit
equation,
we
accept
the
estimation
results.
1)
Role
of socio-economic
variables
of
the
model
One
of
the
socio-economic
factors
drawing our
attention
is
household
size:
it
appears
to
be
decisive
in
the
both
decisions
of
5)
Average
maize
yield
was
1.45t!ha, 1.26t!ha,
and
0.88t!ha
and
its
standard
deviation
was
0.21t!ha,
0.28t!ha,
and
0.36t!ha
in
the
Sudanian,
the
Sudano-sahelian,
and
the
Sahelian
zones
respectively
from
2009110
to
2012/13
(Direction
Gent\rale
des
Etudes
et
des
adoption
and
continuous
use
of
improved
maize
seeds.
In
rural
area of Burkina Faso
where
labor
market
is
not
well
developed,
household
size
strongly
determines
labor
availability.
Thus
the
adoption
of
improved
maize
seeds,
which may require
more
financial
investment
in
seeds
and
chemicals
than
otherwise,
is
done
under
an
assurance
oflabor
availability;
this
may
denote
its
importance
in
increasing
the
likelihood
of
continuous
use
of
improved
maize
seeds.
Neill and
Lee
(2001)
reached a similar
result by highlighting that
availability
of
family
labor
is
positively linked
to
adoption
decision of"mucuna", a cover
crop
for
maize
production,
in
northern
Honduras.
Literacy
and
household
assets
like
tin
roof
and
thatch
wall
are
found
to
have
influence
in
technology
adoption,
as
is
common
in
the
literature.
But literacy
has
no
effect
on
continuous
use.
As
for
household
assets,
relatively rich
households
(with
tin
roof)
are
indifferent
to
continuous
use,
but very poor
households
(with
thatch
wall)
are
likely
to
disadopt
improved
maize
seeds.
2)
Roles
of
institutional
factors
ofthe model
As
for
institutional
factors
models
2,
3,
and
4
show
that
all
the
institutional
variables
except
for
government extension
services
have
significantly positive
influences
on
farmers' adoption
decisions.
In
Burkina
Faso,
where
most
of
rural
households
are
faced
with liquidity
constraints,
credit
is
very essential
for
the
acquisition
of
improved
maize
seeds.
This
result
is
consistent
with many previous
findings
regarding
the
role
of
credit in
technology adoption including
Tura
eta!.
(2010).
Membership
in
furmer
organization
is
known
to
facilitate
farmers'
access
to
technology such
as
fertilizers
and improved
seeds.
In
Burkina
Faso,
it
is
quite
common
that
farmer
groups
serve
as
guarantees
before
acquisition
of
input
credit.
Extension
services
have
also
a
positive impact
on
the
adoption
of
improved
maize
seeds,
although those
from
the
government
is
not
statistically
significant.
Thus,
infonnation about new
technologies
is
important
for
farmers
to
adopt
them.
In
spite
of
the
significant effect
of
the
institutional
variables
on
the
adoption,
none
of
them
influences
significantly
the
continuity
of
using improved
maize
seeds.
The
findings
are
not
consistent with
the
above
mentioned
existing
literature.
As
discussed
in
the
previous
section,
the
descriptive
statistics
in
Table
1
indicate
strong
association
of
the
institutional
variables
with
continuous
use
of
improved
seed.
But
the
estimation
results
imply that
the
association
is
due
to
ago-ecological
conditions
probably
such
as
expected yield
and
risk
in
each
province.6J
Statistiques
Sectorielles,
2014).
Almost
all
maize
is
grown
under
rainfed
condition
in
Burkina
Faso,
but
some
maize
is
planted
in
garden
with
manual
watering
from
well
and/or
pond.
Table 1. Estimation results with descriptive statistics
Modell Model2 Model3 Model4 Descriptive statistics
(mean)
Explanatmy
Variables
Adoption
Cont.
use Adoption
Cont.
use Adoption
Cont.
use Adoption
Cont.
use Adopters Disadopters
Cont.
users
Sex ofhousehold head
0.38
1.07' 0.48' 1.09'
0.36
0.85
0.30
0.83
0.978
0.950 0.996
(l=male,
O=female)
(0.25) (0.56)
(0.25) (0.57)
(0.23) (0.61)
(0.24)
(0.69)
Age ofhousehold head -0.01"'
-0.00
-0.01'
-0.00
-0.01'
-0.00
-0.01'"
-0.00
II
47.9
50.8
46.6
(years)
(0.00) (0.01) (0.00)
(0.00)
(0.00) (0.01) (0.00)
(0.01)
Household
size
0.03"' 0.02"' 0.02"' 0.02" 0.02"' 0.02' 0.03"'
0.00
II
11.7
10.2 13.5
(head
cmmt)
(0.01) (0.01) (0.01)
(0.01) (0.01) (0.01)
(0.01)
(0.01)
Marital
status
ofhousehold head
-0.03
0.03
-0.06
-0.05
-0.06
O.Dl
0.00
0.08
II
0.942 0.926 0.956
(1
=manied, 0 otherwise) (0.22)
(0.26) (0.21) (0.28) (0.21) (0.26) (0.22) (0.27)
Literacy ofhousehold head 0.23"
0.00
0.24"
0.04
0.20"
-0.05
0.22"
-0.08
II
0.320 0.240
0.331
(l=yes, 0 otherwise)
(0.10)
(0.13) (0.10) (0.15)
(0.10) (0.13)
(0.10) (0.14)
Household having livestock
0.13
0.26
0.10
0.28
0.10 0.29 0.16
0.25
II
0.972
0.969 0.976
(1
=yes,
0 otherwise) (0.21) (0.34) (0.24)
(0.29) (0.22) (0.34) (0.21) (0.34)
Quality ofhouse roof 0.29"'
0.13
0.26"'
0.15
0.25"' 0.04 0.31'"
0.06
0.549
0.441
0.589
(1
=tin,
0 otherwise) (0.09) (0.17) (0.09)
(0.17) (0.09)
(0.18) (0.09)
(0.18)
';>;!
Quality ofhouse
wall
-1.15"' -5.10"' -1.27"' -5.80"'
-1.09***
-5.14"' -1.23'" -5.17"'
~
0.003
0.020 0
(1)
(1
=thatch,
0 otherwise)
(0.37) (0.35)
(0.39) (0.47)
(0.35) (1.86)
(0.33) (1.98)
~
-0.02
-0.15
-0.09 -0.24 -0.10 -0.26 -0.09 -0.27
:::>"
Maize growing
in
rainy season
0.973
0.967
0.972
l'
(1
=yes,
O=no,
in dry
season)
(0.26) (0.27) (0.27) (0.26)
(0.26) (0.28)
(0.25) (0.33)
~
Access
to
credit 1.21"'
0.73
~
(1
=yes,
O=no)
(0.17) (0.50)
0.371
0.099
0.605
Membership in crop prodnction 0.87'"
0.32
II
0.530
0.209
0.734
organization
(1
=yes,
O=no)
(0.11)
(0.46)
Membership
in
other production 0.32"
0.19
II
0.056
0.061
0.044
organization
(1
=yes,
O=no)
(0.16) (0.33)
Extension services
from
NGO 0.45"
0.23
II
0.063 0.048
0.060
(!=yes,
O=no)
(0.19) (0.37)
Extension services
from
government
0.04
0.41
II
0.039
0.037 0.048
(1
=yes,
O=no)
(0.14) (0.28)
Extension services
from
cotton companies 1.13"'
0.22
II
0.185
0.046 0.294
(l=yes,
O=no)
(0.20) (0.41)
Province Dummies
Yes Yes Yes Yes
Yes Yes
Yes Yes
-2.13"' -7.08"' -2.42'" -8.21"'
-2.06***
-6.38' -2.09"'
-5.86
Constant
(0.40) (1.33) (0.43) (0.89)
(0.41) (3.41)
(0.39) (4.02)
Wald
test
for
p=O
0.05
(0.59)
0.36
(1.22)
-0.34
(0.80)
-0.38
(0.85)
Number
ofSam_Eles
2033 2033 2033
2033
II
763
515
248
Note:
Absolute standard
errors
are
in parentheses. They
are
heteroskedasticity robust and clustered by province.
***
**
and ' indicate significance level at
1%,
5% and
10%
respectively.
~
26
5.
Conclusion
This
study
aims
to identify socio-economic and institutional
factors facilitating the adoption and the continuous use
of
improved seeds in maize production in Burkina Faso. A bivariate
probit model with sample selection
is
used to identify these
factors. The results indicate that the adoption
of
improved maize
seeds increases significantly with socioceconomic factors such
as literacy
of
household head, availability
of
family labor,
household assests as well as with institutional factors namely
access
to
credit, membership in farmer organization, and access
to extension services. Thus, such institional factors are
considered to provide farmers with "easy" conditions to access
to this technology.
However, none
of
these institutional factors contributes to
continuous use
of
improved
maize
seeds. Except for some
household socio-economic factors like family size and assets,
the continuous use is largely determined
by
province level
observable/unobservable factors such as expected yield and its
variability. This
findings
are the contribution
of
this paper that
uses nationally representative data unlike existing literature on
this
topic. The results may imply that the observed frequent
disadoption is caused
by
farmers' trial use
of
improved seeds
without good knowledge about the suitable conditions for
growing new varieties.
This
study does not consider the price
of
improved maize
seeds in comparison with that
of
conventional ones because such
information is not available in the dataset. Making the model
dynamic incorporating realized profit in previous year is a
necessary extension
of
this
study.
References
Bikienga,
I.
M.
(2002)
Une
Evaluation
des
Secteurs
des
Engrais
et
des
Semences
au
Burkina
F
aso,
African
Trade
Investment Program.
CEFCOD:
Centre
d'Etude,
de
Formation et
de
Conseil en
Developpement
(2013)
Situation
de
Reforence
des
Principales
Filieres
Agricoles
au
Burkina
Faso,
Minisrere de !'Agriculture et
de
Ia
Securite
Alimentaire.
Combary,
0.
S.
(2013)
Decisions d'adoption
et
d'intensification
de
!'utilisation
des
engrais chimiques
dans
Ia
production
cerealiere
au
Burkina Faso,
Revue
CEDRES-Etudes
56:
22-35.
deGraft-Jolmson,
M.,
A.
Suzuki,
T.
Sakurai,
and K Otsuka (2014)
On
the
Transferability
of
the Asian Rice Green Revolution
to
Rainfed
Areas
in sub-Saharan
Africa:
an Assessment
of
Technology
Intervention in Northern
Ghana,
Agricultural
Economics
45(5):
555-
6) Although the results
are
not
shown,
if
the
province dummies
are
grouped
for
each agro-ecological
zone,
the
zone dummies
are
significant in both equations, and the most favorable zone
(i.e.
the
570.
Diiro
G.
M. and
G.
M. Sam
(2015)
Agricultural Technology Adoption
and
Non:funn
Earnings
in
Uganda:
A Semiparametric Analysis,
The
Journal
of
Development
Area
49(2):
45-162.
Direction Generale
des
Etudes et
des
Statistiques
Sectorielles
(2014)
Annuaire
des
Statistiques
Agricoles
2012,
Misnistere
de
I' Agriculture
et
de
Ia
Securire
alirnentaire.
Doss,
C.
R
(2006)
Analyzing Technology Adoption Using
Microstudies:
Limitations, Challenges, and Opportunities
for
Improvement,
Agricultural
Economics
34:
2
07-219.
Feder,
G.,
R
Just,
and
D.
Zilbennan
(1985)
Adoption
of
Agricultural
Innovations in Developing
Countries:
A
Survey,
Economic
Development and
Cultural
Change
33(2):
255-298!
Hailemariam,
T.,
K Meoale, and
S.
Bekele (2013)Adoption ofMultiple
Sustainable Agricultural Practices
in
rural
Ethiopia,
Journal
of
Agricultural
Economics
64(3):
597-623.
Minten,
B.
and
B.
C.
Barrett
(2008)
Agricultural
Technology,
Productivity,
and Poverty in
Madagascar,
World
Development
36(
5):
797-822.
Moser,
C.
and
B.
C.
Barrett
(2003)
The
Disappointing Adoption
Dynamics
of
a
Yield-increasing,
Low External-input
Technology:
the
Case
of
SRI in
Madagascar,
Agricultural
Systems
76(3):
1085-1100.
Neill,
S.
P.
and
D.
R Lee
(2001)
Explaining
the
Adoption and
Disadoption
of
Sustainable
Agriculture:
The
Case
of
Cover
Crops
in
Northern Honduras,
Economic
Development
and
Cultural
Change
49(4):
793-820.
Oladele,
0.
(2005)
A
Tobit
Analysis
of
Propensity
to
Discontinue
Adoption
of
Agricultural Technology Among Farmers
in
Southwestern Nigeria,
Journal
of
Central
European
Agriculture
6(3):
249-254.
Saka,
J.
0.
and
B.
0.
Lawai
(2009)
Determinants
of
Adoption and
Productivity
of
Improved
Rice
Varieties
in
Southwestern Nigeria,
African
Journal
of
Biotechnology
8(19):
4923-4932.
Savadogo,
K,
T.
Reardon, and K Pietola
(1998)
Adoption
of
Improved
Land Use
Technologies
to
Increase Food Security
in
Burkina
Faso:
Relating Animal
Traction,
Productivity,
and Non-Farm
Income,
Agricultural
Systems
58(3):
441-464.
Tura,
M.,
D.
Aredo,
R.
La
Rovere,
G.
Tesfahun,
W.
Mwangi, and
G.
Mwabu
(201
O)Adoptionandcontinued use
of
improved
maize
seeds:
Case
study
of
Central
Ethiopia,
African
Journal
of
Agricultural
Research
5(17):
2350-2358.
World
Bank
(2008)
World
Development
Report:
Agriculture
for
the
Development,
World
Bank.
Sudanian zone)
is
the most influential in
the
adoption and
the
continuous use
of
improved
maize
seeds
as
expected.
