ADOPTION OF AGRICULTURAL INNOVATIONS, CONVERGING NARRATIVES, AND THE ROLE OF SWEDISH AGRICULTURAL RESEARCH FOR DEVELOPMENT?

ADOPTION OF AGRICULT
URAL INNOVATIONS, CO
NVERGING
NARRATIVES
, AND THE ROLE OF SW
EDISH AGRICULTURAL
RESEARCH FOR

DEVELOPMENT
?




























Draft


Discussion paper, version 20
11
-
01
-
28


Johan Toborn


TABLE OF CONTENTS

Page

1. Introduction











1

2. Agricultural Innovations









2

2.1.
Ado
ption and
d
iffusion of agricultural innovations
–

theories and concepts


3


2.2.

Empirical adoption and adoption impact studies





4

3. Adoption of specific innovations illustrated







7


3.1. Illustrative key studies








8



3.1.1.

Embodied

exog
enous innovations





8




High Yielding Varieties







8




GM Crops








10




Fertilizers








14




Pesticides








15



3.1.2.

Packages of disembodied agronomic and managerial innovations


16




Conservation Agriculture






16




Integrated
Soil Fertility Management





20




Soil and Water Conservation






21




Integrated Pest Management






24




Rain Water Harvesting







27




Agroforestry








28




Low
-
External Input Technologies





30




Sustainable Agriculture/Integrated
Natural Resource Management

31

4. Adoption of agricultural innovations
–

conclusions






32

5. Perspectives and narratives on agricultural development





33


5.1 Historical development of agriculture






35


5.2 Markets and institutional fixes







35


5.3. Policy fixes










36


5.4 Livelihoods

is key









37


5.5 T
echnology
fixes









38



5.5.1. The Green Revolution in retrospect





38



5.5.2 The quest for sustainabl
e and multifunctional agriculture



39


5.6. Emerging
, complementary

narra
tives?






41



5.6.1. Convergence on technology, but with multiple pathways to intensification

41



5.6.2. Context matters, generalisation too





42



5.6.3. Towards integrative science






43



5.6.4. Putting knowledge into use






44

6. Concluding d
iscussion
–

new roles and binding constraints for Swedish

ag
ricultural development research?








46

References

Annex







ADOPT
ION OF AGRICULTURAL
INNOVATIONS
,
CONVERGING
NARRATIVES
,

AND
THE ROLE OF SWEDISH
AGRICULTURAL
RESEARCH FOR DEVELOP
MENT
?



1.

Introduction

This
paper
discuss
es

what contribution

Swedish

agricultural
science
,
in particular natural science
,

could or should make to
agricultural development in the south
a
part from the obvious business as
usual, i.e.
current
research issues
where we
presumably
have competence and comparative advantages.
Th
e latter

is certainly an important role of
Advanced R
e
search I
nstitutions in international agricultural
research.
Busine
ss as usual is in itself a position on agricultural research for development; though not
one that has emerged from deliberation
, but rather resulted from
the interests of individual scientists,
existing disciplinary and departmental divisions, academic rew
ard systems,
the few financial sources
available
,

and the research approaches they
sustain
.
Research projects making part of the position are
usually small, often acquired in com
p
etition through a peer review process, or as commissioned
research (notably P
h.D. training/research).
Research has, understandably, a biophysical
focus
and
usually allows limited opportunities to engage in dissemination and upscaling.
Competence can still be
p
ut to use in new contexts,
formats, and with new foci,
however.
From different entry points this paper
purports to provoke reactions on our position on agricultural research for development. The
provocations are subtle, however, and have no simple
responses
.
A marked deviation from business as
usual assumes that consid
erable
individual and collective
constraints have to be overcome,
and
that a
strong will to change is present.

This
is

a tall order, but r
eflecting on alternative use of competence
,

or
needs for
alternative c
ompetence
,

has a
n intrinsic

value and is seldom
done
.

In the best of worlds,
such
reflections could result in a position on how we view agricultural research for development, our role in
that context, and how we would like to change to fill that role.
A shared position on agricultural
research for devel
opment would increase the appreciation of how different disciplines and approaches
fit in the overall context and create synergistic value added

rather than existing in a state of internal
competition.


Efforts to develop agriculture are expected to resul
t in improved agricultural production;
“
improved
”

obviously having
multiple

interpretations.
Better technologies have to be generated and put into use.

Agricultural
scientists by training and tradition want to believe that
new

technologies

drive
agricultur
al development. Research findings

are passed through transformative
and communicative
stages and
finally
result in
improved production.

This default linear model is valid in some cases, and
utterly wrong in others.
How we

perceive

adoption and diffusi
on of

agricultural innovations

is
therefore a key element in our position of agricultural research for development
?

C
hapter 2

describes
alternative
theories and concepts of adoption and diffusion o
f agricultural innovations, and empirical
approaches to study

ad
option
and diffusion
.

For a private company
,

high adoption rates and diffusion
of its innovations

is a sign of success, presuming sound economics.

For public agricultural research,
adoption is a necessary
precondition

for assessing

if the
benefits generate
d by the
innovation

were

worth the
research
investment.
Benefits range from outcomes at adopter
-
level to community
-
level
changes in environmental, economic and social conditions and distributional impacts.
This
evokes the
intricate question

how identificat
ion, design, implementation and evaluation of agricultural research
and innovation diffusion can be improved.

For
composite natural resource innovations much more
sophisticated assessment and learning approaches are needed than represented by the default l
inear
model.


Chapter 3

illustrates

actual
adoption of agricultural innovations
.

Innovation is a deceptively simple
term, but in reality ranges from embodied external innovations (seed, fertilizers, pesticides, etc.) to
systems changes building on agronomic and managerial innovations.
This division is admittedly
artificial as technolog
y packages in agricultural development
often
include

both categories

and
simultaneous interventions of a non
-
technical nature.

For selected innovation
s under the two
categories

definitions and classification schemes are provided, selected studies of releva
nce
summarised,
extent of adoption/diffusion
commented on, and
illustrative
successful cases reported.
A
diffuse, yet
clear pattern emerges
.

Adoption of agricultural innovations, with exception
s

for early
Green Revolution
(GR)
success, has not progressed a
s fast as desirable and projected
, in particular not
in Sub
-
Saharan Africa
.
What to expect is per se methodologically complex to assess
; sometimes based
on mathematical models, sometimes rather airing disappointments
.
How do we explain the lack of
progress

and what could be done differently
?

2



Chapter 4
draws lessons from adoption studies

with implicit
implications for Swedish agricultural
research for development
.


Agricultural development is a study area of increasing complexity.
Our knowledge has increase
d,
c
onditions have changed

over time,

and a
dditional
dimensions

are

added by growing awareness of
e.g.
Man’s influence on ecosystems, water scarcity and climate change, and
effects of
globalisation.

To
make sense of a messy

reality
,

we often resort to narratives

or stories
.
There are several narratives that
from different perspectives
are
used

to
explain agricultural development and ultimately
adoption of
agricultural technologies.
Four

common
n
arratives
are
reviewed
in C
hapter
5
:
ma
rkets a
nd
institutional fixes; policy fixes
; livelihoods

is key
;
and
technology fixes with several subthem
es.
Each
narrative
comes in different

versions.
These narratives

are important building blocks of a

meta
-
narrative.
Then f
our

possible
emerging, compl
ementary narratives

seen relevant from a research
perspective

are discussed: c
onvergence on technology
, but with multiple pathways to intensification
;
c
ontext

matters
, generalisation too
;

t
owards integrative science,

and p
utti
ng knowledge into use
.
These
are desirable, not mutually
exclu
sive

narratives. Each narrative subtly provokes the business as usual
scenario.


Chapter
6

reflects

on the implications for Swedish
agricultural
research for development

beyond
business as usual.


2
.

A
gricultural
innovations

“Many technologists beli
e
ve that advantageous innovations will sell themselves, that the obvious
benefits of a new idea will be widely realized by potential adopters, and that the innovation will
therefore diffuse rapidly. Seldom is this the ca
se. Most innovations, in fact, diffuse at a
disappointingly slow rate
” (Rogers 1995).


An innovation is
an idea, practice, or object that is perceived as new by an individual or other unit of
adoption.

In industrial and agricultural
innovation literature,
a distinction is made

between products,
processes, and

social/organisational innovations. Agricultural innovations, as traditionally studied, are
mainly to categorise as products, but with ele
ments of processes. Technology

is
us
ed synonymously
with innovat
ion
.


Agricultural i
nnovations
can be classified according to several parameters:

-

Genetic, mechanic and chemical innovations (private goods) and agronomic, managerial and animal
husbandry innovations (public goods);

-

Individual innovations (individual
adopter) and collective innovations (group of persons);

-

Continuous innovations, semi
-
continuous innovations, and discontinuous innovations with increasing
demands for new skills, knowledge and even investments;

-

Labour saving

innovations and land saving innovations
;

-

Process innovations and product innovations;

-

Endogenous and exogenous innovations (based on
Sonnino 2009)
.


A slightly different
categorization
is suggested by Sunding (1999):

-

Innovations embod
ied in capital
goods
or products
(“shielded” and “non
-
shielded”)
and innovations
not embodied
;

Innovations according to impact:

-

New products;

-

Yield increasing innovations;

-

Cost
-
reducing innovations;

-

Innovations that enhance product quality.

Innovations according
to form

-

Mechanical, biological, chemical, biotechnical, and informational innovations


For the purpose of this paper a distinction
is made
between

Embodied, exogenous innovations (EE
I
)
and packages of disembodied agronomic
and managerial
innovations (PDA
M
I)
.
In practice the two
categories are often combined.
The first category would mainly qualify as continuous or semi
-
3


discontinuous innovations, whereas the PDA
M
I category leans more to the discontinuous category, i.e.
more skill
-
intensive.


2
.1

Adoption

and diffusion of
agricultural
innovation
s

–

theories and concepts

Diffusion of innovations has been studied by many disciplines (e.g.

anthropol
ogy, sociology of various
brands
, education,
medicine, communication studies, marketing
,
business administration
, etc.).
From
an initial domination of sociology, economics has gradually taken over, possibly because of a stronger
emphasis on the theoretical basis for adoption
,

and its policy relevance.



The sociologist
Everett
Rogers’
seminal work on
diffusion of in
novations
(1995
)
is
a good starting
point into this area of study. An innovation according to Rogers is “an idea, practice or object that is
perceived as new by an indivi
dual or other unit of adoption”
.
Diffusion is seen as “the process by
which an innovat
ion is communicated through certain channels over time among

members of a social
system”. A

technological innovation usually has two components: a hardware aspect (the tool,
product) and a software aspect (how to use the hardware). For good reasons studies

of diffusion of
innovations have often addressed individual innovations, in practice innovations often come in
packages
–

clusters
–

and are interrelated and interdependent.


The characteristics of innovations explain their rate of adoption. Five such ch
aracteristics
of
importance
are discerned: 1) The relative advantage
reflects
how the innovation is subjectively
perceived superior to the previous idea; 2) Compatibility
reflects

how the innovation is perceived
“consistent with the existing values, past experiences, and needs of potential adopters
”
; 3) Complexity
reflects

the perceived difficulty to understand and use the innovation; 4) Tria
la
bility is
“the degree to
which an inno
vation may be experimented with on a limited basis
”
; and 5) Observability reflects how
the results of an innovation are visible to others.
An innovation can
further
be changed or modified
(re
-
invented) by a user.


Communication
,

through channels
,

provide
s

information to a social system with the purpose to
influence the knowledge and assessment of the innovation.
Mass media is ofte
n more effective in
creating
awareness of an innovation, whereas personal contacts are more effective in forming an
opinion abou
t a new idea. Such interpersonal communication is facilitated if
conveyors of information
are optimally similar to the receiver in certain attributes.


Time is a main factor in the decision
-
makin
g process, innovativeness and

an
innovation
’
s rate of
adoption. In the innovation
-
decision process
,

an individual passes through the stages
:

knowledge,
persuasion, decision, implementation (adoption) and confirmation (post
-
adoption assessment
)
.
Information is sought at the various stages to reduce uncertainty

about the usefulness of the
innovation. The decision stages result in adoption or rejection of the idea.


Innovativeness is an expression for how early an individual or ot
her unit of adoption is adopting

a new
idea

compared to
other members of the social
sy
stem.
A
dopter
s are divided into five

categories,
each
with its
o
wn characteristics
:

1) innovators, 2) early adopters, 3) early majority, 4) late majority, and 5)
laggards. Finally, rate of adoption is the relative speed with which an innovation is adopte
d by
members of a social system.


The social system with its interrelated units share
s

an interest in finding solutions to a common goal,
i.e. to improve their agricultural

system to enhance livelihoods.

Such a

system has a social and
communication structure that facilitates or impedes the diffusion of innovations in the system. Norms
,
being part of the social system,

are the established
behaviour patterns

for system members. Often
opinion leaders play a crucial

role in influencing system members. Change agents may have the
explicit role to influence members in a certain direction. Both opinion leaders and change agents are
central actors in diffusion of innovations.


Three main type
s

of innovation
-
decisions can
be distinguished: independent individual decisions
(adopt a HYV), collective decisions (soil conservation on hillsides), and authority imposed decisions.


4


The accumulated adoption over time, i.e. the diffusion, is frequently found to follow a sigmoid
distr
ibution. In marketing applications
,

this feature has often
been used

to predict and influence
diffusion.


Rogers’ account for innovation adoption and diffusion does not give theoretical explanations to how
adoption decisions are actually made. A classic ar
ticle by Feder (1985) is
a
frequent
departure
for
theoretical analysis of decision making.
This line of studies is mainly pursued by economists.
The
essence of his article and follow
-
up renderings on the subject include a number of complicating issues.


Of
ten

distinct technological options

are present.
Several
decision

proc
esses may then run
simultaneously or sequentially.
Farmers may
therefore

rather consider portfolios of innovations.
Further, i
nnovations may be divisible or of a lumpy character
,
presenti
ng a dichotomous choice,
which
could be a deterrent to

those
interested in trying on a small scale.

Lumpy investments may be only
partially recoverable and adoption decisions may at times be close to irreversible.
There may

be fixed
transaction or information costs associated
, that may
again
deter resource
-
constrained farmers.

Innovations
may be scale
-
neutral or contain economies of scale
, i.e. the innovation
may
favour better
resourced households
.
Fo
r divisible innovations
,

the intensity of use is of great interest (e.g. proportion
of land allocated, intensity of use per area unit).
Technologies may

show improved performance

over
time
, or become cheaper

due to
economies of scale,

and
therefore

gradually
become more attractiv
e to
farmers, ceteris paribus.

Diffusion of technologies is more complex than the spread of influenza.


Potential adopters
are uncertain
what
an innovation may
offer
.
Over time i
nformation from different
sources and from the farmer’s own experience
reduce
s

this
uncertainty
. A

better base is established for
adoption/rejection and intensity of use decisions.


The decision

maker is assumed to maximise

the utility of asset use over time, subject to various
resource constraints
, usually
assuming

a concave utility function
.

This can be
expressed
by

static
models, or by d
ynamic, sequential models
that

consider changing knowledge and conditions.

In a
dynamic model
,

new decisions depend on the results of previous decisions

and
their

effect on wealth
and income
,
and
revised
subjective
knowledge about the utility of the innovation, including production
outcomes, expected costs

and revenues. Farmers gradually learn how to ma
ke better use of the
innovation
. For management
-
oriented improvements, a better s
ystems performance may also
materialise over time. Hence parameters determining farmers
’

choice
are continuously updated.



Risk has been included in many models. Production, incomes and costs are not deterministically
known. Farmers have their
subjective
perception of risks involved, and consider not just the expected
mean outcome but also the distribution of risks around the mean.
The subjective perception of risk
may well deviate from the objective reality.
It is often assumed that farmers are risk avers
e with the
extent depending on several characteristics. To the farmer
,

the riskiness of an innovation compared to
the old idea then matters; also whet
her the risk varies together with
risks in
other parts of the system or
moves in the opposite direction. S
ome models suggest safety
-
first decision behaviour,
implying that
farmers have to be assured of a minimum result
,

and not base their decision on expected results.


Theoretical models of adoption behaviour have looked into variables that may explain the de
ci
si
on to
adopt or the intensity of adoption. Such factors include farm size, credit and information access,
personal traits of the decision
-
maker, tenure arrangement, etc.


Theoretical models for the aggregate adoption complement indiv
idual adoption model
s. Alternative
assumptions regarding individual adoption behaviour usually result in S
-
shaped curves. Cochrane’s
technological treadmill suggests
diminishing

gains over time due to price declines following increased
production due to adoption.


2
.2

Empirical

adoption
and adoption impact studies

A vast literature of e
mpirical studies
has

attempted to test the relationship of key variables to adoption
behaviour. The theoretical foundation for selection
of variables
is sometimes weak; which is
understan
dable as
theoretical
models often point in different directions. Early adoption studies had a
heavy emphasis on the
GR

packages, following the seminal studies of improved varieties in the US.

5



Unders
tanding past

adoption

Empirical studies attempt to
understand and explain adoption. It is an ex post perspective. Obviously,
technology research has to be guided by early
analysis of likely adoption of a technology at some
stage of development. Such ex ante analysis may include partial farm budgets to show

the economic
attractiveness of the technology, constraint and risk analysis. Should the innovation be selected for
dissemination, the analysis may be repeated when early signs of adoption are available and the trends
extrapolated, constraints focused, etc
. The first part of this section deals primarily with ex post studies.


Generalizations

(with many exceptions) by Ruttan (in Feder 1985) from early
GR

technology studies

are illustrative of the
possible
conclusions from such

studies:

-
The new HYVs were ado
pted at exceptionally rapid rates in areas where they were technically and
economically superior
;

-

Neither farm size nor tenure has been a serious constraint to th
e adoption of new HYVs of grain;

-

Neither farm size nor tenure has been an important source

of differential growth in productivity
;

-

The introduction of HYVs has resulted in an increase in the demand for labour
;

-

Landowners have gained relative to tenants.


Feder in h
is article summarises findings o
n individual adoption
with respect to

seven
m
ajor
explanatory variables: farm size, risk and uncertainty, human capital, labour availability, the credit
constraint, tenure, and supply constraints.
Considered important at the early stages of adoption, they
may become less significant in later stages.


His conclusion on the significance of farm size is illustrative and with a bearing on the other factors:

“The wide variety of empirical results, interpreted in the context of the theoretical literature, suggests
that size of holding is a surrogate for a l
arge number of potentially important factors such as access to
credit, capacity to bear risk..., access to scarce inputs (water, seeds, fertilizers, insecticides), wealth,
a
ccess to information, and so on”
. Since the influence of these factors

varies in di
fferent areas and
over time, so does the relationship between holding size and adoption behaviour. Because the
theoretical literature and analytical interpretation of empirical results suggest that several intervening
factors lie at the root of observed fa
rm
-
size/adoption relationships, the remainder of this section turns
to consideration of the observed role of such factor”.


On risk
,

Feder (ibid.) concludes that ...” most of the empirical work on the role of subjective risk is not
yet rigorous enough to a
llow validation or refutation of available theoretical work”.


Adoption research has moved on in the last 25 years. Still
,

Feder’s concluding comments should be of
concern. In the words of Doss (2006):

“... research was needed in
five areas: (i) examining
the intensity of adoption (not just dichotomous
choices); (ii) addressing the simultaneity of adoption of different components of a technology
package; (iii) analysing the impact of incomplete markets and policies on adoption decisions; (iv)
contextualizin
g adoption decisions within social, cultural and institutional environments; and (v)
paying attention to dynamic patterns of changes in landholdings and wealth accumulation among early
and later adopters” (p

208).


Doss argues that progress has been made i
n the first two fields (e.g. econometric techniques have
become increasingly sophisticated to deal with issues of endogeneity and
simultaneity of decisions
)
.
However, she maintains that “...some of the concerns ... remain unanswered, especially the issues
of
how institutional and policy environments affect the adoption of new technologies and how the
dynamic patterns of adoption affect the distribution of wealth and income”.


To Doss technology adoption
research
has three current foci: econometric and mode
lling
methodologies to understand adoption decisions, studies of learning and social networks in adoption
decisions, and continued local micro
-
level studies to understand adoption for policy purposes.
Generic
weaknesses of micro
-
studies of technology adopt
ion, according to Doss, are the lack of dynamics
originating in using cross
-
sectional data, and the lack of variation within samples. The latter can be
6


rectified by larger (and more expensive) samples or through meta
-
studies. The latter, in turn, assumes
c
onsistent defini
tions of variables are used.


For policy purposes we may be interested in
how
incomplete access to credit or cash, information and
labour markets may affect the adoption of technologies.
Doss

illustra
tes how
the measures that have
been
construed to measure such variables differ markedly between studies, do not necessarily measure
the core contents of the variable, make comparisons between studies cumbersome, and interpretation
of results less effective for policy design.


Ex post impact
assessment

Adoption studies have usually been accompanied by assessments of how adopters benefited. This
emphasis has been accentuat
ed over time
. Agricultural research has been resource
-
constrained in
recent decades

and r
esearch investments have to
progres
sively
justif
y

their returns
. R
esearch outcomes
and impacts have to be demonstrated to f
unders
.
A second purpose of ex post impact assessment is to
provide learning opportunities.
These
growing

demands
have
gradually
extended

to
consider

not only

technical

and economic production benefits,
but also

effects
on household incomes,
other household
asse
ts, vulnerability,
equity,
consumption, nutrition, food security, poverty, and environment, etc.
G
one are

the
also
days when agricultural research in practice mea
nt embodied ex
ogenous innovations.
A marked shift to NRM and policy research is noted, with telling evidence seen in the CGIAR system
(Renkow 2010). Outcomes and impacts of NRM and policy research are exceedin
gly more
complicated to assess.

Adoption studie
s have become part of a much more sophisticated cycle of
impact evaluation.


How to go about this challenge has occupied the Standing Panel on Impact Assessment of CGIAR’s
Science Council.
Its

S
trategic guidance for ex post impact assessment
(epIA)
of agricultural research
(Walter, T. 2008)
is a revealing treatment of how to refin
e

epIA

for accountability and learning
purposes. The guideline presents a typology of
epIA

composed of the primary objective to document
productivity and profitability, or s
elective high order impacts and the level of assessment (macro or
micro)
.

The principal cases are: 1) aggregate economic rate of return, 2) disaggregate economic rate of
return, 3) aggregate multi
-
dimensional impacts, and 4) disaggregate multi
-
dimensional
impact.
Central
to CGIAR’s rendering of the topic is the notion of impact pathways.
Planning of research projects to
be included in CGIAR’s portfolio include describing the
most plausible impact pathways from
problem identification to intended ultimate goa
ls of the CG. A generic impact pathway includes:

Inputs (research investments)
;

Outputs (first/immediate results of a research project)
;

Outcomes (the external use, adoption or influence of a project’s outputs by next or final level users
that results in a
dopter
-
level changes needed to achieve the intended impact)
;

Impacts (the ‘big picture’ changes in economic, environmental and social conditions that a project is
working toward. Within the CG System, project impacts should be in line with the center’s mis
sion
and vision statement
,

and with the CGIAR goals)
.


Impact pathway analyse
s
are

part of ex ante planning but
,

when well prepared,
also

play an important
role in e
pIA

assessments. The Guidance explores the accumulated experience and best practice of
eco
nomic rate of return studies (where obviously adoption is a first step to assess) and
multidimensional impact studies.
Generally a range of models and analytical tools have to be deployed
in a judicious manner.
Although there has been methodological progre
ss, in particular
multidimensional impact studies still require improvement and additional emphasis.


L
ivelihood approaches
are used with increasing frequency
in multi
-
dimensional impact assessments
(e.g. Adatao and Meinzen
-
Dick 2007). Policy and
NRM

rese
arch
epIAs have

a less impressive record
.
They

are for good reasons more difficult and resource
-
demanding to conduct.
How to address NRM
epIAs has

long
been a concern with
the CGIAR (e.g. Fujisaka and Whi
te 2004).
Two
special
difficulties
in implementing e
pIAs relate to attribution of impact and
the
counterfactual evidence.
Attribution is about how benefits and impacts can be
casually linked to
the research in question
,

when
several stakeholders have been involved in various capacities
,

and confounding fact
ors may have
played a part. The counterfactual evidence asks what would have happened in the absence of the
research project.
Counterfactuals are cumbersome to construe
,

and seldom look
at
the merits of
7


alternative research investments, as
would be

standard operating procedure in investment analysis.
Moisture stress, for instance, can be addressed through short maturing
and

water efficient varieties, but
also through rainwater harvesting or combinations of both. Such counterfactuals may fall under
d
ifferent research budgets and/or institutional domains, and are
therefore, disregarding methodological

problems, in practice precluded.
Both these difficulties are more cumbersome in policy and NRM
research.
As a guideline, up to 3% of a research institute
’s budget should be set aside for impact
assessment with epIAs constituting a sizable share of that amount.
I
t is underlined that epIAs should
be considered part of science, not tack
-
ons to fulfil donor requirements.


E
x ante
impact assessment

Just as epIA
s have become more important, ex ante studies are increasingly recognised as vital to
improve allocation of scarce resources to activities that contribute to the development objectives of the
research organisation.
A blend of models and tools are required
to secure data answering four basic
questions: 1) wher
e is the impact likely to occur
; 2)
by whom will the impact be felt
; 3) wh
ich impacts
will
be generated
; and
4)
what is the value of these impacts?” (Thornton et al 2003).


Some 15 approaches to ex ante

assessment approaches are illustrated below:

Village workshops/discussions, stakeholder consultations, key informant interviews; community
-
level
formal surveys; community
-
level formal surveys for looking at adoption and impact; financial and
economic ana
lyses of the production effects of new technologies; transect walks, aerial photography;
spatial analysis, GIS, satellite imagery; market studies; economic surplus methods; in
-
depth
anthropological/sociological and characterization studies, farmer assessme
nts; participatory nutrient
flow diagrams; follow the technology, participatory technology development; hard biophysical
simulation models of component processes and interactions; softer biophysical models of component
processes and interactions; multiple
objective mathematical programming models of the household;
rule
-
based (softer) models of the household

(based on Thornton 2003).



The constant assessment
–

learning cycle

Integrated Natural Resource Management
(INRM
,

see 5.6.3
)
research is radically different from
traditional technology research.
INRM
tries to build the capacity of natural resource managers to
manage change in sustainable ways. This is an inherently indeterminate and complex process.
Douthwaite et. al.

(2004) des
cribe the central role of monitoring and evaluation
(M&E)
through the
different stages of research planning and implementation.
Evaluation would change from a focus on
accountabilit
y to support learning and adapt
ive management of stakeholders involved in a

project. “…
more emphasis should be placed in the use of evaluation to improve, rather than prove, on helping to
understand rather than to report, and on creating knowledge rather than taking credit” (ibid). Ex post
impact assessments
would not quantify i
mpact based on inappropriate economic models but rather use
evidence from many sources that the intervention has contributed to impact. Effective M&E, based on
a shared impact pathway vision, “will identify and describe incipient processes of knowledge
gen
eration and diffusion, the emergence and evolution of innovation networks, and the creation of
organizational capabilities” (ibid). Future impact assessment will have to convincingly show how this
growth in processes and capabilities
contributed to wider
-
s
cale development changes. New methods
have to be applied to follow the process, including livelihood approaches, simulation modeling and
indicator combinations.



3
.

Adoption
of specific innovations
illustrated

This
chapter
illustra
tes

studies and
findings

relating to

adoption of
specific
innovations
,
when possible
from
Sub
-
Saharan Africa. A total coverage is precluded, whic
h has to be emphasised.
Recall is made
of the previous distinction between e
mbodied, exogenous innovations (EE
I
) and packages o
f
disembodied agronomic and managerial innovations (PDAMI).
T
his is a fussy and floating distinction.
Under PDA
M
I there is considerable
conceptual overlapping
.
Some
innovation
concepts
have arisen in
response to the sustainability discourse, others through

integrating over disciplinary boundaries

to
address broader research questions
, still others as marketing branding
.

Under the
respective PDAMI
concept

is often
an array of different designs and specifications of the respective design.
This feature
is also

a major obstacle in
measuring

diffusion of these approaches compared to the more distinct
embodied external innovations.


8


PDAMI approaches selected are those frequently figuring in the current discourse. Some of those have
been there for extended period, but with emphasis shift through the years.


Additional approaches could have been included. Organic farming in much relies

on agronomic and
managerial principles shared with CA, ISFM, SWC, IPM, RWH, AF, LEIT, SA/INRM, except for the
restrictions on embodied, exogenous innovations. Certified organic farming offers opportunities of
niche products for domestic consumption and ex
port, with associated questions of certification
procedures and costs, premium prices, and self
-
organised market links or acceptance by larger value
chains. Integrated livestock systems is an important concept, but would add unduly to the paper and in
a se
nse replicate the crops account: breeds, feed, animal health, alternative management regimes, etc.
Studies of integrated crop/livestock systems innovations is a relatively neglected field

where

synergies
and constraints posed by different production enterp
rises are less known.


The
innovation
concepts
covered

are the following:


Embodied, exogenous innovations

Packages of disembodied agronomic
and

managerial
innovations

High Yielding Varieties (HYV)


Conservation A
griculture

(CA)

GM crops




Integrated Soi
l Fertility M
anagement

(ISFM)

Fertilizers




Soil and Water C
onservation

(SWC)

Pesticides




Integrated Pest Management

(IPM)






Rain Water H
arvesting

(RWH)






Agroforestry

(AF)






Low External Input Technologies

(LEIT)






Sustainable A
griculture

(
SA)
/INRM








3.1
. Illustrative key studies


3.1.1.
Embodied

exogenous

innovations

These innovations

formed the backbone of the original
GR
.
It is thought
-
provoking to view them in
the light of Rogers
’

innovation characteristics important for adoption

and try to make a similar simple
analysis of PDAMI innovations
.

The utility or relative advantage
is here primarily understood as the ability to deliver higher yields or
reduce losses

in a single season
. This in turn should result in improved economic re
sults, given that
output/input price ratios are favourable.
For fertilisers and pesticides
,

labour savings may materialise

compared with other means to provide nutrients and control pests
.

The
risk
downside of the innovations possibly include
higher

yield and profit variat
ion caused by
weather and price changes
,
sudden
lack
of or
uncertainty
of timely access to inputs,
uncertain credit
prospects;

poor, varying or inappropriate input quality

and traits; incorrect use of inputs
, and
consequences
in cas
e
of crop failure and economic losses.
Non
-
access to inputs and credit may be
killing assumptions.

Similar reflections can be made on
the other characteristics: compatibility, complexity,
trialability
, and
o
bservability
compared to PDAMI cases.


H
IGH YIE
LDING VARIETIES

In Asia HYV diffusion was impressive.

In Africa
,

the success has not been repeated. Africa’s
ag
roecological heterogeneity puts

greater demands on locally suited varieties. Africa also relies on
many
more
staple crops
–

orphan crops
-

neglec
ted by seed breeders

who long concentrated on the
main cereals
. GR varieties were originally bred to respond to fertilizer application, which
with some
exceptions
has not materialised (see next section). Part of the discouraging performance lies with weak
seed breeding, industry and distribution capacity,
sometimes weak performance of new varieties,
combined with credit constraints.


Studies
:


The diffusion and impact of high
-
yielding varieties have been extensively studied.
Hazell
et al (1991)
and Evenson (2003)
are typical studies.


9


Extent of adoption
:

The table below summarises

area
cover
age,
variety releases
,
share of area planted to modern varieties,
CGIAR share of modern variety area, and the contribution of crop genetic improv
ement to yield
growth by crop and region.

Crop

Area
(million ha)
in
developing
countries,
1998

Estimated
number of
variety releases,
1965
-
1998

Share area
to modern
varieties,
1998

Share modern
variety area to
any CGIAR
ancestry, 1998

C
rop genetic
improveme
nt
contribution to
yield growth,
1965
-
1998
(%/year)

Wheat

120

2188

0.82

0.64

0.96

Rice

150

1484

0.64

0.58

0.79

Maize

97

1494

0.87

0.55

0.67

Sorghum

39

363

0.44

0.38

0.50

Millet

36

123

0.44

0.85

0.57

Barley

20

105

0.49

0.80

0.49

Lentils

3

49

0.23

0.90

0.28

Beans

23

642

0.18

0.90

0.21

Cassava

17

252

0.15

0.93

0.22

Potatoes

9

458

0.88

0.17

0.74

All

535

7246

0.65

0.60

0.72

Region

Area
(million ha)
in
developing
countries,
1998

Estimated
number of
variety releases,
1965
-
1998

Share area
to modern
varieties,
1998

Share modern
variety area to
any CGIAR
ancestry, 1998

C
rop genetic
improvement
contribution to
yield growth,
1965
-
1998
(%/year)

Latin America

57

3146

0.51

0.55

0.66

Asia

337

2229

0.83

0.57

0.88

Mena

49

715

0.56

0.81

0.69

SS Africa

92

1157

0.23

0.62

0.28

All

535

7246

0.65

0.60

0.72

Source: Evenson, cited in Renkow (2010)


The table demonstrates the
past

concentration on the three major cereals.
In Sub
-
Saharan Africa, u
se
of improved varieties
and crop genetic improvement contribution
to yield growth 1965
-
1998 are

much
lower than in other regions
. Later studies indicate an increased interest in orphan crops in SSA. There
are also reports on use of improved maize in West and Central Africa reaching some 60% of the area.
Improved dual pur
pose cowpea is progressing in dry savannah regions of West Africa. New varieties
of the common bean have been adopted on about half the total bean area in East, Central and Southern
Africa

(Renkow 2010)
. The so
-
called New Rice for Africa is commented on be
low.


Successful examples
:

The New Rice for Africa (
NERICA
) has been praised as a miracle crop for Africa and an illustration
of an emerging African
GR
. The interspecific cultivar of rice, developed by the now Africa Rice
Center (WARDA)
, was created
1994
b
y crossing African rice with high
-
yield Asian varieties of
Orza
sativa
.
Originally developed for upland cultivation, there are now 18 upland Nericas and 60 low
-
land
Nericas. Field tests of the varieties show promises for higher yield under varying conditio
ns, more
protein, shorter growing season, and a greater resistance to pests and diseases.


Rice is a neglected crop in
Africa with consumption by far exceeding production. The 2008/9 global
food
-
price crisis triggered production increases in SSA of 16
-
18%

and a further 4.5% in 2009. Across
the S
a
hel performance was even more impressive
–

44%
,

and
a staggering
241% in Benin. These
increases are attributed to the price surge and the success of Nerica dissemination. Success stories
include Nigeria with close to 200

000 ha of Nerica and Uganda with 35

000 ha in 2007.
Progress has
also
been reported for Burkina Faso
, Ethiopia, Guinea
, Mali, Sierra Leone and Togo (
Warda 2010).

Nerica cultivation in uplands of West Africa is estimated at 6.7% (CGIAR/Science Council 2008).


10


Limited studies of adoption have been carried out in e.g. Benin, Cote d’Ivoire, Guinea, and the
Gambia.
As new Nerica variety diffusion and adoption is not random, the Average Treatment Effect
framework has been applied to remove non exposure and selection

bias (WARDA 2008).
Adoption
rates differ between the countries
, as should be expected
. Some findings on determinants of adoption
of Nerica tell that in:

Cote d’Ivoire
:
growing rice partially for sale, household size, growing upland rice, past participatio
n in
Participatory Variety Selection, and living in a PSV
-
hosting village had a positive impact
,

whereas
age
of farmer and having a secondary occupation had a negative impact;

Guinea
:

positive impacts were noted for participation in a training program and
living in a village with
NGO SG2000 activities;

Benin
:

land availability, living in a PSV
-
hosting village, varietal attributes such as swelling capacity
and short growing season were important determinants of adoption; and
;

The Gambia
:

living in a village
where dissemination by WARDA, contac
t
s with the NARI, access

to
credit, and experience in

upland rice farming had a positive impact (WARDA 2008).


Whether the diffusion

of Nerica is a success is a matter of judgement.
The review panel assessing
WARDA label
led the
spread record sobering. It also highlighted the importance of ava
ilability of
high
-
quality seeds

at cheap prices
,

and termed lack of seeds a major factor behind disadoption. A
technical innovation obviously needs a supporting infrastructure. Furthe
r, the advantages of Nerica
may not be higher yields per se, but early maturity, tolerance to water stress, good taste and flavour
,

and short straw. Effect on yields and determinants of adopti
on seem

to be quite heterogen
e
ous

over
countries and farmer cat
e
gories
(CGIAR/Science Council 2008, pp 95
-
99)
.
Questions have also been
raised about the solidity of claimed Nereica yield benefits: “still, even in places where Nerica has been
introduced, overwhelming evidence from the field of substantial yield benefits

are slim (Larsson et al
2010).


Massive partnerships have been formed for research, training, seed production and dissemination of
the Nerica lines. Dangers exist
,

that top
-
down initiatives crowd out local initiatives and oversell the
story (GRAIN 2009, C
GIAR/Science Council 2008, p 82).


“
For institutions that rely only on donor funds to survive, the

temptation is strong to oversell potential
products and breakthroughs to donors. Breakthroughs

are by definition one
-
time shots and it is
difficult to
maintain the level of interest of donors over a

long period. Overselling research activities
have immediate benefits in terms of donors
’

support

that reward success stories, but it has a long term
cost, which can be the loss of trust of the

scientific comm
unity if research results do not back up the
initial claims.


Other CGIAR Centers seem, in retrospect, to have succumbed to this temptation, perhaps

inadvertently. The Panel thinks that WARDA too needs to be cautious with the NERICA story

and the
way it is

sometimes reported, probably by excess enthusiasm
” (
CGIAR/Science Council 2008, p 82)
.


GM

CROPS

“It is high time that the heroic simplification of the ‘GM crops are good for the poor’ st
o
ryline i
s
finally laid to rest” (Glover

2009).


The role of GM

crop
s in agricultural development and their benefits for poor farmers are hotly
debated. Gene transfers, health hazards, narrowing genetic base, etc. is one line of arguments.
Patenting and increasing farmer dependence on
an
increasingly concentrated biotech i
ndustry
,

and a
limited capacity of public sector research to innovate in fields of less commercial interest is
another. A
third line of
critcism

argues that there are alternative ways to bring around the effects biotech can
deliver, or more moderately, that approaches should be combined (e.g. Bt GMOs and IPM).


Risks and IPR/market concentration are not expounded on here.


Farmers’ adoption of y
ield enhancing GM

crops (e.g. more nutrient efficient or drought tolerant)

would in
many
respects not differ from HYVs in general.
Crops with increased insect

or insecticide
resistance
require more knowledge and experience gathering to fit overall IPM regi
mes. Bio
-
fortification
, in particular if multiple traits to counter vitami
n A, z
inc and iron deficiency are
11


combined, presumes some household understanding of the deficiencies, its effects and cures, and
observational skills to judge the benefits of new cr
ops. There are also likely implications for
processing, storing and distributing the food to individual family members that have to be known and
adhered to.


The general debate on
GM
s

has to varying extent filtered down to farmers

and consumers
, affecting

their initial uncertainty and subjective perception of
risks involved.
Farmer risks
may include new
aspects

with respect to e.g. seed price (monopoly pricing and transfer fees/rent), produce price
(consumer reactions), and
export potentials
affected by

th
e regulatory framework
s

in other countries.


Future crops will multiple traits may present even more complex decisions in a production sy
stem and
livelihood context.
Such
GM

crops
may,
from an adoption point of view
,

be more complex than
traditional HYVs.


Studies
:


A growing literature follows adoption, diffusion and impact of GM
s
. Smale (2006)
in a review of 106
ex ante and ex post peer reviewed articles draws interesting conclusions. Most of the studies looked at
farm impact; others covered farm/industry, consumer, consumer industry, industry and trade. From the
review she
deduces that due to t
he limited field research in terms of locations, crops, traits, small
sample size, and scholars involved, findings cannot be generalised. A number of methodological
weaknesses are noted, but are gradually
being
rectified. Like in many technology studies, t
he initial
enthusiasm has been superseded by a more cautious weighing of pros and cons. This is in line with
traditional variety studies: 1) where performance will vary across location and time; 2) the net
economic impact on society is not easily measured;

3) the length of the time period of observation
matters
(discontinuities in adoption due to changes in external factors); and 4) the institutional and
social context is often more significant for impacts than any particular variety trait. Future studies h
ave
to look deeper into impacts on labour, health, environment, equity and poverty.


It is stressed, however, that GM

crops

differ from traditional HYVs
e.g. in being more knowledge
intensive in development, putting a regulatory framework in place, and mak
e farmers understand the
technology. Pest and disease resistance traits will yield better returns when farmers know better how to
manage secondary pests and disease resistance.


Partially building
on
and extending Smale’s

review of studies of impacts of

b
iotech adoption, Glover
(2009) questions whether these technologies are really pro
-
poor

-

a never dying narrative in her words.

Her scepticism is founded in the adoption and impact
a
ssessment made of Bt cotton in China, India,
and South Africa
,

and the ori
entation and methodological

flaws that characterise them.
Such claims
about future benefits are not uncommon

in technology development in g
eneral
. To Glover
the pro
-
poor GM narrative is a classical construct
, starting from statements on the poverty, hunger

and food
insecurity of the world and the multidimensional
challenges of

feeding the world. Then a long list of
potential applications of GM technology

follows
. Typically, no account is taken of technical
difficulties to be overcome, and social and institu
tional contexts to be considered for the benefits to be
realised, even if the technologies are technically efficient.


“
It is high time that the heroic simplification of the ‘GM crops are good for the poor’ storyline is
finally laid to rest. Only when we h
ave driven it out can we hope to give due attention, more calmly
and carefully, to the other aspect of biotechnology’s undying promise
–

undying, in this case, because
it has never been given the chance to live. The extravagant hype of the GM crop advocat
e
s (and not
only the alarmism of

anti
-
GM campaigners) has unfortunately suffocated debate about this important
new technological field. It is a field which, in truth, does indeed hold the potential to help address
some important developmental challenges of
the twenty
-
first century, whether thro
ugh genomic
techniques, marker
-
assisted selection or indeed some transgenic applications. But, to realise this
potential, it is not enough to pay lip
-
service that GM crops will not be a silver bullet against hunger
and

poverty, while simultaneously designing impact assessments around the implicit assumption that
such a magical effect is indeed possible. We need to think about how technologies may work in the
dynamic and complex agricultural systems and institutional fra
meworks of the real world. We need to
understand how farmers actually use technology. And we need to focus on problems to be solved and
12


challenges to be overcome, in all their complexity, rather than focusing on particular types of
technologies and looking

for opportunities where they might be deployed. Hopefully, the promise of
biotechnology is really not dead. But a realistic assessment of both its promise and its pitfalls requires
a new set of research questions, different research methods and a rigorous

focus on the problems to be
solved rather than a fascination with a quick technological fix
”

(ibid, p 43)
.


Extent of adoption
:

The International Service f
or Acquisition of Agri
-
Biotech A
pplications annually publishes its Global
status of commercialised biotech/GM crops. It reports by country
how commercialised
biotech crops
have diffused in absolute terms
,

in proportion to the crop area
,

draws conclusions on the temporal r
ate
,
and narrate
s other findings and events of importance
.

ISAAA is sponsored by the industry
. Its

figures
have a
t

times been questioned.


Global area of
biotech crops in 2009: by country (million hectares)

Country

Area

Biotech crops

USA

64.0

Soybean, maize, cotton,
canola, squash, papaya, alfalfa, sugarbeet

Brazil

21.4

Soybean, maize, cotton

Argentina

21.3

Soybean, maize, cotton

India

8.4

Cotton

Canada

8.2

Canola, maize, soybean, sugarbeet

China

3.7

Cotton, tomato, poplar, papaya, sweet pepper

Paraguay

2.2

Soybean

South Africa

2.1

Maize, soybean, cotton

Uruguay

0.8

Soybean, maize

Bolivia

0.8

Soybean

Philippines

0.5

Maize

Australia

0.2

Cotton, canola

Burkina Faso

0.1

Cotton

Spain

0.1

Maize

Mexico

0.1

Canola, soybean

Chile

<0.1

Maize, soybean, canola

Colombia

<0.1

Cotton

Honduras

<0.1

Maize

C
zech Republic

<0.1

Maize

Portugal

<0.1

Maize

Romania

<0.1

Maize

Poland

<0.1

Maize

Costa Rica

<0.1

Cotton, soybean

Egypt

<0.1

Maize

Slovakia

<0.1

Maize

25 countries

134



Diffusion
of

some crops is impressive:

“For the first time, more than three
-
quarters (77%) of the 90
million hectares of soybean grown globally were biotech; for cotton, almost half (49%) of the 33
million hectares were biotech; for maize, over a quarter (26%) of the
158 million hectares grown
globally were biotech; and, finally for canola, 21 % of the 31 million hectares were biotech
”

(
ISAAA
2009)
.

Individual crop/country performance is also impressive, like Bt cotton being cultivated on 87%
of India’s 9.6 m has of co
tton.


Progress in Africa, with the exception of
South Africa, is less impressive
so far
. Bt cotton was planted
over 115,000 has in 2009 (29% of cotton area) and enough seed produced to cover 70% o
f cotton area
in 2010. Egypt planted approx
imately 1,000 ha

of a hybrid Bt yellow maize in 2009.


New varieties with multiple traits are under development. The Golden rice ventures into
biofortification. Drought tolerant
,

more water

and nitrogen
-
use

efficient
, and salinity resistant varieties
are also in process;

possibly with greater relevance for African contexts.
Orphan crops and p
erennial
cereals are

other line
s

of exploration.


13


Successful examples
:

Nutrient deficiencies

are common in developing countries. Staple food b
iofortification

by
conventional breeding or using GM technologies could complement improved diet or supplement
s
.
Such use of GM crops are by many seen as more relevant than the herbicide or insect tolerant
approaches originally developed for large scale western agricultu
re.


Around 3 billion people are dependent on rice for their caloric intake.


“Among cereals, rice has the highest energy and food yield but lacks essential amino acids and
vitamins needed for normal body functions. It lacks beta carotene, the precursor
of Vitamin A needed
for sight and cell differentiation, in embryonic development of mammals, and in functioning

of the
immune system and of body mucosal membranes. Vitamin A deficiency (VAD) is a
nutritional
problem
in the developing world afflicting 127 m
illion people and 25% of pre
-
school
-
children.
Currently around 250,000 to 500,000 become blind annually, 67
%
of whom die within a month, or
around 6,000 deaths of children day, equivalent to 2.2 million a year” (ISAAA 2009).


Conventional breeding hence c
annot be used to enhance beta carotene, contrary to iron and zinc
deficiency (affecting 57% and 71% respectively in South East Asia).


The history of the Golden Rice
(GR
I
)
dates back to 1984.
An extensive partnership involving
foundations, international a
nd national research institutes and private companies has developed
t
ransgenes
,

first from daffodil

GR
I
, later replaced by maize transgenes

(GR2G)
.
The choice of
varieties to be introgressed with the GR2G event will be based on their popularity and accepta
bility in
the respective country.
In 2012 GR is expected to be introduced in the Philippines and Bangladesh, to
be followed by India, Indonesia and Vietnam.
So far GR
I

has thus

not been disseminated. However, it
is interesting to review ex ante anticipatio
n of future adoption.


How well GR
I

will perform under field conditions still has to be explored. Be
ta

carotene levels have
been substantially increased since the first events and may be even higher under field conditions.
Beta
-
carotene content and stabili
ty, bioavailability, bioconversion and bioefficacy studies are being carried
out. Studies are also made to measure how typical way
s of preparing rice affect

the retention of beta
-
carotene

(e.g. Tang et al 2009)
.
Possible impact on taste and agronomic perfo
rmance make
another
part of the necessary st
udies
.

Summary findings claim that “GR could probably supply 50% of the
recommended dietary allowance of vitamin A from a very modest amount
–

perhaps a cup
–

of rice, if
consume
d

daily. This amount is well within the consumption habits of most children and their
mothers”
(
ASN 2009)
.


The developed transgenes are donated to the Global Rice Humanitarian Board that oversees the
Golden Rice Project and the institutions involved in the

Golden Rice Network. GR seeds will therefore
be available to poor farmers in countries with adequate biosafety regulations at no license fee
,

and
with the right to retain seeds.

Ex ante studies confirm substantial and cost
-
effective benefits from GR
intro
duction, also compared
to

alternative measures (
Dawe

and Unnevehr 2007
)
.


Illustrations of initial future adoption indicate 7
to
7.5 million ha planted to GR
I

in the Philippines,
Bangladesh and India
through introgressing the transgenes into the most popu
lar improved rice variety
(3 in t
he case of India). This assumes, among other things,
that traits of the new varieties are
considered equal or better than the existing ones by farmers and markets, and that the value of the
nutritional improvement is apprec
iated.


The GR
I

has stirred an intensive debate. Proponents argue that unjustified and impractical legal
requirements have delayed dissemination of GR
I

and caused unnecessary and considerable suffering
and deaths

(
Potrykus 2010)
. Responsibility for inform
ing about GR
I

has to be shared with the health
sector when

biofortification through production now has to be combined with traditional dietary
improvement and supplements.
Cooperation over between sectors has to transpire.
Opponents of GM
crops raise conce
rns about biosafety, the presumed presence
and interests
of large corporations, and
the suspicion that GR
I

is
a commercial conspiracy
–

a Trojan horse
–

to gain acceptance for GM crops

in general
.

14



On paper GR
I

could become a successful innovation with
fur
ther
potential to be crossed to also
address iron and zinc deficiency.


Despite several decades of public debate and information campaigns, producers and consumers in
Europe are still at a loss how to view risks and benefits associated with GM food and fe
ed. For farmers
in developing countries with less access

to information, future decisions to use GR
I

can go either way.
Strong persuasion can be made to bear from agricultural and health ministries, agricultural research
institutions, extension services and the seed industry
, once biosafety regulations and GR
I

approval are
established nationally. Large a
nd small NGOs will demonstrate a variety of
different stands
–

for
differing

reasons
–

but are most likely not to come out strongly
in favour of

GR
I
.
It will be an
interesting study field to follow how farmers will make sense of central and local informati
on for and
against GR
I

to gather knowledge, form an opinion, and make decisions on whether to try

GR
I

or not.
It is a more complex setting than when the
GR

was launched
. C
entral mass media
is likely to
play

a
decisive role.


F
ERTILIZERS


Diffusion of ferti
lizers has been extensively studied in the era of the
GR
. In Africa the success has not
been
sustained

over longer periods.

The limited use of adapted HYVs is one explanation

to low SSA
consumption figures
. Production risks are
also
more prominent in Afric
a with degraded soils where
fertilizer has limited response and
larger weather variability and
moisture stress that deter
s

risk
-
averse
farmers.
Farm gate fertilizer prices are much higher in Africa, and profitability of fertilizer use is often
less convinc
ing
. Access to, availability and timeliness of fertilizer delivery discourage adoption.
Elimination or varying and uncertain subsidisation of the input have impacted consumption. Due to
small markets and sales volumes, economies of scale have not been expl
oited in fertilizer procurement
and distribution.
This adds to the importance of using the right fertilizer composition
, amount,

and
application;
requirements that have not always been met.


It is not surprising that many initiatives promote higher
fertiliser use in Africa, i.e. the proposed
increase to 5
0 kg/ha in 2015 from the 2007

9.6
kg/ha
(The Abuja Declaration on Fertilizer for African
Green Revolution 2006).

There are those who claim this is a golden opportunity not to
get dependent
on fertili
zer use. This is missing the point, however, as also made clear in many of the background
documents to the Fertilizer Summit. How to increase fertilizer consumption AND
simultaneously
use
agronomic and managerial innovations to restore/improve the natural
resource base is the unresolved
dilemma.


Extent of adoption
:

Annex 1 shows fertiliser consumption by country
,
region
, and by

developed
/
developing nations.
African fertilizer consumption is extremely low, and few countries deviate from the pattern; Egypt
b
eing the no
table exception. Figures

show annual use of fertilizer per hectare
,

which may be
distributed

over two crop seasons. In some countries commercial farming of high value/export crops
significantly contribute to the national figure.

Average consumpt
ion figure
s

therefore tell little about
the proportion of adopti
ng

farmers and intensity of use

within a country.


Successful examples
:

Two interrelated successful examples are the introduction of fertilizer micro
-
dosage and
improved
profitability and
affordability from subsidies, using vouchers
, small packs of fertilizer,

and a
warrantage system. Micro
-
dosage is claimed to increase the technical efficiency of fertilizer but is
often promoted in combination with subsidies, voucher
s, small fertilizer
packs,

and warrantage

(collective grain storage with inventory credit enabling farmers to obtain better produce prices through
later sales,
see
ICRISAT 2009).


Micro
-
dosage entails precision application of fertilizers to seeds placed in holes. Rates are ty
pically 4
-
6 grams per hole
,

or equal to a full bottle cap or a three
-
finger pinch.
Good yi
eld increments have been
reported in West Africa (Burkina Faso, Niger, Mali) for sorghum and millet (44
-
120%
,
ICRISAT
2009
,
see also
Tabo 2007 and 2008, Mateete 2010)
. Similar results

are recorded for Zimbabwe,
15


Mozam
bique, and South Africa (Twomlow

2010).
Even rates as low as 0.3 gram/hole are repo
rted
t
o
yield good increments and cost
-
benefit ratios (Aune 2007). These

rate
s

are

substantially below
recommendations aime
d at maximiz
ing yields or profits, and seek

the best return to the small
investment farmers can afford.
Initial success is believed to stimulate to and finance further
intensification.



Advantages may be more pronounced in drought years

due to better deve
loped root systems.
The
required s
kills can be acquired through short courses/demonstrations
, e.g. by agro
-
dealers
.
L
abour
demand

may increase slightly
.
Further development of the technology may include fertilizer tablets for
easier and more precise applic
ation and seed coating.
Possible soil mining in the absence of addition
of organic matter has to be further investigated

and could be a possible downside.


Evidence of

actual adoption is patchy

so far
. Some 25

000 farmers have participated in trials in We
st
Africa and several organisations are involved (ICRISAT, AGRA, USAID, WACARD, FAO)
. There
are plans
to increase this figure to 500

000.


ICRISAT
-
Zimbabwe has for several years been involved in relief and recovery work in Zimbabwe.
More than 160

000 farm
ers have received free fertilizers, micro
-
dose instructions
,

and access to paired
demonstrations. Yield results have been encouraging. Whether this translates into continued use of
micro
-
dosage in the absence of free fertilizers is yet to be seen (Twomlow
,

2010).



Rather a question of how markets and institutions can stimulate fertiliser use, subsidies for many years
ruled out under structural adjustment programmes, have again been tried in Malawi with follower
schemes in countries like Ghana, Kenya and Ta
nzania.
The Malawi case has been intensively studied
and debated. It involved large
-
scale distribution of heavily subsidised fertilizers and seeds despite
the
stand of several international organisations. Vouchers were issued, using varying targeting princ
iples

(on vouchers, see Minot and Benson 2009)
. Not surprisingly
,

maize production increased to the point
where a former national
maize
deficit was turned into an exported surplus. The fertilizer programme
cost as
a share

of MinAg budget, national budget,
and GDP in 2008/9 reached 74%,
16.2% and 6.6 %
respectively
due to the
volume

of the programme, the level of subsidy, and the soaring fertilizer prices.
Since then the level of ambition has been scaled back, but the economic, financial and political
sustai
nability has been questioned (Dorward and Chirw
a

2010).



The programme has also been criticised for crowding out other needed long
-
term investments in
infrastructure, agricultural research and extension. In essence, the fertilizer push should be combined
with other integrated
soil fertility measures to get

better grain/fertilizer efficiency and less reliance on
fertilizer (GRAIN 2010
).


P
ESTICIDES

Increased u
se of pesticides has been less an objective than e.g.
for
fertilizers and HYV. Interest was
early f
ocused on using pesticides to limit economic injury rather than spray
ing

by calendar in a move
towards IPM, safe use of pesticides, and use of less toxic substances. Recently
,

adoption of
biopesticides
,

and possible reductions from introduction of insect
r
esistant
and herbicide tolerant
crops
have been the subject of many studies.


Pesticides can be classified by target organism, chemical structure, and physical state. More than 1000
active ingredients are registered as pesticides, and over 16000 pesticide products are being marketed in
the United States. It is
hence

more difficult t
o follow use and adoption of pesticides than other
embodied
exogenous innovations.


Extent of adoption
:

Annex 1 shows the pesticide use in African countries. For many countries figures are not available,
which is symptomatic.
Such summary figures do not re
veal the toxicity of the substances used, nor
how use and use intensity is distributed between user categories.




16


3.1.2.
Packages of disembodied agronomic
and managerial
innovations

Such packages are proposed in general terms from agricultural textbooks w
ith its headings (e.g.
soil
preparation and conservation, nutrient management, pest management, water management
) and
hierarchies (e.g.
nutrient management
–

cropping systems
–

farming systems
)
. Packages are also
suggested based on e
merging problems or con
cerns
that emphasise conservation

(Soil and Water
Conservation, Conservation Agriculture
, rainwater harvesting
),
sustainability

(Sustainable
Agriculture), or
l
ess
reliance on
external inpu
ts

(Low
-
external
-
input
-
technologies). Several concepts
use
integrated

as a prefix to signify a coherent and comprehensive approach to package formulation
(IPM, INRM, ISFM, etc). So
me packages are given selling
labels to
appeal to authorit
ies, donors or
end users (e.g. Evergreen agriculture, F
ertilizer trees; all b
eing part of i.a. agroforestry).


Disregarding

overlapping between concepts, different roots, trajectories and connotations, there is
nowadays a proliferation of and

possible competition between proposed entry points

and pathways
to

intensification of agricultural production.


Some of the approaches figuring in recent

debates

are looked at from a diffusion and adoption
perspective:

Conservation Agriculture, Integrated Soil Fertility Management, Soil and W
ater
C
onservation,
Integrate
d Pest Management, Rainwater Harvesting, Agroforestry, Low
-
External I
nput

Technologies
,
Sustainable A
griculture
/
Integrated Natural Resource Management
.
One could obviously
ask how the concepts relate and are ordered.
Although appearing to address different

issues, concept
definitions and classification schemes frequently show considerable overlapping.
In practice, specific
innovation packages can fall under several of the concepts.
Concept entry points differ: soil fertility

management
, conservation, agrofo
restry, water harvesting, pest management
, etc
;
are
all essential
components of
sustainable land management.
Dis
aggregation
of systems may be an expression

for
what is managea
ble, but the current discourses

beg the question:
what
’
s in a name?


Some of the
concepts come in more sophisticated versions, including multiple scales of intervention,
collective action, social innovations, and combining modern and traditional knowledge (e.g. RWH and

SA/INRM).


These
concepts

entail

methodological difficulties when s
tudying adoption. Each
concept

comprises
numerous
site
-
specific interpretations of general principles. Approaches generally have several
components
,

which can be adopted in different combinations, sequence
s

and
degrees
.
M
ultiple years
of practice are often

required
to yield full benefits. In the interim
,

external and household internal
conditions
may

change
, including experiential learning of the household
. Decision
-
making is made
sequentially
. Studies of how the ap
proaches have been received
therefore
ofte
n
contain contradictory

evidence of

success and failure

(aggravated by disadoption when
promotion projects
wind up)
,
and
difficulties to generalize and compare between cases.



C
ONSERVATION AGRICULT
URE

“Despite the publicity claiming widespread adoption of

CA, the available evidence suggests virtually
no uptake of CA in most SSA countries, with only small groups of adopters in South Africa, Ghana
and Zambia” (Giller 2009).


Definition
:

“
CA is a concept for resource
-
saving agricultural crop production that s
trives to achieve acceptable
profits together with high and sustained production levels while concurrently conserving the
environment. CA is based on enhancing natural biological processes above and below the ground.
Interventions such as mechanical soil t
illage are reduced to an absolute minimum, and the use of
external inputs such as agrochemicals and nutrients of mineral or organic origin are applied at an
optimum level and in a way and quantity that does not interfere with, or disrupt, the biological
pr
ocesses. CA is characterized by three principles which are linked to each other, namely:

1.

Continuous minimum mechanical soil disturbance.

2.

Permanent organic soil cover.

3.

Diversified crop rotations in the case of annual crops or plant associations
in case of perennial
crops

(FAO)
”
.


17


More specific versions of CA are available for instance Zambia where
key pra
ctices for ox farmers
and hoe

farmers

are specified. (Conservation Farming Unit 2007).



Studies:

Knowler &

Bradshaw (2007) draw interesting conclusions from a review and synthesis of recent
research on conservation agriculture. Based on 130 financial analyses of conservation agriculture and
other soil and water conservation in Sub
-
Saharan Africa and Latin Amer
ica/Caribbean, the authors
conclude that the former produced
positive
net present value

in 90% of the cases, the latter in 58 % of
the cases. Since adoption of CA is still low with some notable exceptions, what other factors are at
play when farm finance i
mpact seems positive. Selecting 31 published ex post studies of adoption of