Title: Optimizing the process of making farmer recommendations in developing countries
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00102044/00001
 Material Information
Title: Optimizing the process of making farmer recommendations in developing countries
Physical Description: Book
Language: English
Creator: Alvarez, Jose
Copyright Date: 1982
 Record Information
Bibliographic ID: UF00102044
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.

Full Text

Reprinted from
journal of Agronomic Education
Vol. 11, 1982, p. 43'-50

Opt i mizi ng t he process of maki ng farmer recom mendat ion s

in developing countries'

Jose Alvarez2


Many development projects have failed because
of unsound recommendations. This paper looks at
how the process of making farmer recommenda-
tions can be improved in the small farm sector of
developing countries. The examples provided sug.
gest (a) the need to conduct ex-ante evaluations,
(b) the necessity of a multidisciplinary team work-
ing in an interdisciplinary manner with both bio-
logical and social scientists involved during all
phases of the project, and (c) the importance of
developing technology which is based on the
small farmer's goals and farming system, includ-
ing his resources and constraints. Farmmng sys-
tems research is proposed as the appropriate
framework conducive to meet the challenge of op-
timizing the process of making farmer recom.
mendations in developing countries,

Additional index words: Farming systems re-
search, Interdisciplinary research, Ex-ante evalu-
at ions.

SINCE the end of World War II, increasing attention
has been given to the process of economic develop-
ment of the less developed countries. There exists a con-
sensus on the need for sustained growth to bridge the
gap that separates the developing countries from the in-
dustrialized nations. Although the issue of overall devel-
opment strategies is still under debate, the key role that
the agricultural sector has to play is today widely ac-
The efforts of almost four decades, however, have
not been very successful. In his farewell speech to the
World Bank on 30 Sept. 1980, Robert S. McNamara
Widespread poverty is an open insult to the human dig-
nity of us all. For we have collectively had it in our
power to do more to fight poverty, and we have failed to
do so. Sustaining the attack on poverty is not an eco-
nomic luxury, something affordable when times are easy
and superfluous when times become troublesome. It is a
continuing social and moral responsibility, and an eco-
nomic imperative and its need now is greater than evel
(The Palm Beach Post, 1980).
Agricultural professionals from the United States
have always been concerned with the development pro-

Contribution of the Univ. of Florida, Inst. of Food and Agric.
Sci., Gainesville, FL 32611. Florida Agric. Exp. Stn. Journal Series
No. 3569. Invited paper for the 1981 annual meetings of ASA-CSSA-
SSSA on "Agronomy: increasing food--conserving resources, a
worldwide responsibility," Atlanta, Ga., 2 Dec. 1981.
'Area economist, Food and Resource Economics Dep., Univ. of
Florida, Agricultural Research and Education Center, Belle Glade, FL
33430. Comments received from Christina H. Gladwin, Peter E.
Hildebrand, A. E. Kretschmer, Jr., and two anonymous referees are
gratefully acknowledged.

cess. Just 5 years ago, the theme for the annual meetings
of ASA-CSSA-SSSA was "Agronomists and Food:
Contributions and Challenges," intended to "serve as
an inspiration and a stimulus to all who are truly con-
cerned about adequate food for the people of the
world" (Thorne, 1977). In terms of contributions, it
was said that ". . The agronomists, crop scientists, and
soil scientists who have contributed to so many past suc-
cesses in achieving greater food production should feel
justifiable pride" (Wharton, 1977). Concerning the
challenges, and anticipating McNamara's words and the
central theme of the 1981 Meetings, Wharton, (1977)
added: "When looking to the future and the role of
agricultural professionals, it is clear that the challenges
remain and may well become greater." It thus becomes
obvious that agricultural scientists, including those in
the biological and social disciplines, must share the re-
sponsibility of optimizing the process of making recom-
mendations for farmers in developing countries to bring
about the growth needed to alleviate widespread poverty
and accelerate the narrowing of the gap between the in-
dustrialized and the underdeveloped nations of the
This is an economically slanted paper dealing with the
small farm sector. It describes the importance of a
multidisciplinary approach within the "farming systems
research" framework. The examples provided are in-
tended to illustrate how agricultural scientists from
several disciplines can work together, in an interdisci-
plinary manner, to optimize the process of making man-
agement recommendations.


There is no doubt that the solutions to most of the
agricultural problems faced by developing nations have
to be the result of multidisciplinary efforts; i.e., agri-
cultural professionals from several disciplines working
on the same problem. Figure 1 outlines the steps of the
interdisciplinary approach (a method that integrates all
appropriate disciplines). The figure includes some use-
ful references for biological scientists working with agri-
cultural economists, since the author is an economist.

Initiating the Research

The ultimate objective of any applied research project
is the development of recommendations for farmers.
Making farm recommendations is sometimes a difficult
task. Perrin et al. (1976) have developed a compre-
hensive publication intended for use by agronomists as
they make recommendations from agronomic data. It
explains how to:


Problems tobheAnalyzed Brady, 1977
individuals Involved -- -- -- + Heady, 1957
Research Environment Jhsn15

Experimental Desgn
*Selecting the population
*Choosing a functional form
*Planning the amount of

conlectona Dotero
-Analysisof Results



1. Identify the benefits associated with treatment al-
ternatives, and place values on the alternatives that
match farmers' goals.
2. Identify those inputs that change from treatment
to treatment and place values on those which match
farmers' goals.
3. Identify sources of variability which will make the
farmer uncertain about the net benefits he will get from
each treatment.
4. Derive recommendations from cost, benefit and
variability data that are consistent with the farmers'
desire to increase average income, with the farmers' de-
sire to avoid risks, and with the scarcity of investment
capital which is typical of most farm situations.
Although obviously written from an economic bias,
those four objectives identify, explicitly and implicitly,
the important categories before beginning any project
where a multidisciplinary team is going to work in an
interdisciplinary manner. They include the problems to
be analyzed, the individuals involved, and the research
Research in developing countries is typically problem-
oriented, falling in the category that Johnson (1954)
called "service research." Topics holding most promise
for a successful completion include those in which all
scientists involved identify the existence of a problem
that could be solved by their individual contributions to
the analysis.
The individuals involved are an important factor in
interdisciplinary research. They must be problem ori-
ented, highly qualified in their own fields, and able to
communicate with the rest of the team during all phases
of the research. All disciplines relevant to the project
should participate from the beginning. Including
economists, sociologists, anthropologists, or other
social scientists throughout the entire process can avoid
serious problems at the time of extending the results.
The Plan Puebla in Mexico illustrates this point: the
social scientists were called when administrators realized
that the farmers were not adopting some of the recom-
mendations included in the technological "'package."
The success or failure of any research project is highly
correlated with the environment provided by adminis-
trators. Support must go beyond "lip service." The de-
gree of complexity varies when the project takes a
product oriented (or sub-sectorial) approach, a func-
tional approach (extension or credit), a regional or
national approach. The bureaucracy also tends to in-
crease with the degree of sophistication of the project.
When the administrators do not deliver the needed sup-
port when most necessary, the researchers feel frus-
trated and disappointed. The development literature is
full of dramatic descriptions of project failures due to a
lack of a sound research environment.
The recipe for alleviating these problems must vary
from country to country. However, it is necessary to
train more individuals in administration, to minimize
red tape, to develop systematic operational methods,
and to decentralize the programs' administration (to
take into account both local capabilities and con-
straints) while Keeping a certain degree of control at a
higher level.

Heady, 1948, 1949
Headvand Pesek,1954
bach, 1953
----+Mason, 1957

Hedyand hraer 195

----+Perrin et al., 1976
Redman and Allen, 1954

Fig. 1. The interdisciplinary research approach, with
references mainly related to the work between econo-
mists and biological scientists.

Conducting the Research

Three areas deserve special attention here. They are:
the design of the experiment or research project, the col-
lection of suitable data, and the analysis of the results.
All members of the research team should participate
in the design of the experiment or project. Careful con-
sideration should be given to the understanding of the
underlying socioeconomic and physical relationships;
when ignored or separated, erroneous courses of action
could be taken. The plan and design of an experiment
determines the form of the statistical analysis. Three
problem areas here include the selection of the popula-
tion over which inferences are to be made, the choice of
a proper functional model to describe the form of the
response, and the planning of the economic amount of
Data problems encountered at the time of fitting a
model or analyzing the results are a consequence of
badly planned experimental designs. By working to-
gether from the early conceptualization of a project,
scientists are less likely to encounter data problems than
those working independently.
The author's experience a few years ago illustrates
this fact. For an Economics Handbook intended for
County Extension staff, I was assigned the section on
marginal analysis. It was frustrating to look at response
studies in Florida for the past 20 or more years and
being able to find only two or three good cases that
would help explain the theory with real world examples.
The reason was that the agronomists designing the ex-
periments were more interested in variance-type studies
than in determining the marginal quantities and the eco-
nomic optimal input use.
Figure 2 portrays a good example. In the upper


Region I IRegion nI Region m


Region I !Region

First, most statistical tests are geared to the 0.05 or 0.01
levels of significance. But farmers may be willing to ac-
cept lower levels of significance. For example, if variety
A yields 30 kg in an experiment, while variety B yields
40 kg, farmers may be quite happy to choose variety B,
because of the economics involved, even though this
yield difference is statistically significant only at the
0.15 level. Second, in some cases two treatment means
are not significantly different at any of several trial
sites, but the treatment means are different at the 0.01
level when the data are pooled. Furthermore, if the
treatment means are not significantly different, but an
economic analysis shows that one treatment is a better
recommendation than other, then a more careful
analysis of the recommendation, using net benefit
curves and marginal analysis, is in order. Thus, the
greatest value of statistical analysis is not in deriving
recommendations but in determining what is happen-
ing, biologically, in the experiments.
A good example to summarize this section is found in
Savoie and Kabay (1980). The authors state that farmer
recommendations should be preceded by thorough
agronomic research and economic analysis. Using a
production function approach with agronomic data
(from experimental plots and peasants' fields), simple
statistics, and elementary economic theory they esti-
mated the probability of success or failure for different
seed densities of 'Saxa' dwarf beans in Rwanda-a
country in east central Africa.
Their results show that, if actual seeding practice is 40
kg/ha, a recommendation to increase seed density to 90
kg/ha would have a good chance of succeeding: the
probability of achieving a greater profit than the maxi-
mum lower 95% confidence bound at 40 kg/ha is 92%,
while the likelihood of exceeding the expected profit at
40 kg/ha using 90 kg/ha is 69%. Furthermore, 71% is
the probability that profits at 90 kg/ha will be greater
than the expected profits at 140 kg/ha. Since these con-
fidence bounds were calculated for individual farmers
and not for the mean profit group of farmers, the
authors believe them to be conservative because the
average profit will have tighter confidence bounds.
Barring extraordinarily bad climatic conditions, the
farmers would quickly be convinced of the value of the
proposed seed density. This methodology also can be
applied to other crops and treatments such as fertilizers
and pesticides.


The existence of a multidisciplinary team working in
an interdisciplinary manner is not sufficient for the
success of a research project. Although the following
section emphasizes the need for ex-ante evaluations at
the farm level, the following two examples illustrate the
need to investigate the feasibility and implications of
'Ex-ante evaluations are those conducted before a large research
project is undertaken. The main objective is to try to determine the
probability of success of that project. These evaluations should be of a
short-term nature. Data are sometimes available, or could be gener-
ated by simulating farm conditions or by conducting farm trials on a
small scale.

Region a I



Fig. 2. Changes in an experimental design that would
yield data suitable for an economic analysis.

graph, the results of two treatments with five replica-
tions are shown. Since the treatments cover the three re-
gions of the production function, these data are worth-
less to the economist. However, in the lower graph
(which is drawn using the same scale), we see that, by
adding treatments and reducing replications, the econo-
mist can perform an economic analysis.

Extending the Results

The primary goal of applied research is the making of
farm recommendations from the data collected. A good
farm recommendation could be defined as a choice the
farmer himself would make if he had all the information
available to the scientists. When the new practice has to
show potential profits to be adopted, the input of the
economist is of utmost importance. It is not enough to
tell farmers in a particular region that they should use
200 kg of 6-8-6 fertilizer on their corn without con-
sidering the expected yield and the cost-price relation-
ships. The farmers would have no basis for determining
whether or not that is the most profitable amount to
Therefore, when potential profitability is the only re-
maining factor that could impede adoption (once the
sociological, anthropological, and other important
aspects have been considered), farmer recommenda-
tions should be the result of both statistical and eco-
nomic analyses. Most biological scientists are familiar
with the techniques available to determine whether or
not the means from two treatments in an experiment are
significantly different from one another. Some scientists
argue that non-significance eliminates the need for an
economic analysis. Perrin et al. (1976) gave exceptions.


profitability ratio for all grains produced. Data used
came from a small farmer credit survey conducted by
the Guatemalan Government and U.S. Agency for
International Development in 1974. The results of the
regression equations supported the conceptual model
(Fig. 3).
Traditional crops generally appear at near zero
income and farm size levels, while commercial crops are
cultivated when higher levels of income and farm size
have been attained. Elasticities of market supply for
traditional and commercial crops are high at low levels
of income and farm size. However, while commercial
crops still show some responsiveness at higher income
and farm size levels, the traditional crop response
becomes almost perfectly inelastic. This behavior is the
result of farmers becoming involved in the activities of
the market economy, once self-sufficiency has been se-
cured, and shifting into commercial crop production at
higher levels of income and farm size.
Thus, since traditional crops pervade the basic grains
production system in Guatemalan agriculture, little
hope prevails for the attainment of massive increases in
supply of basic grains. Although corn and rice seem to
have a slight potential for increased production in two
regions of the country, the resulting increases would fall
far behind the goal of the Guatemalan Government.

New Sorghum Technology in Northeast Brazil

To assess the adoption potential of new sorghum vari-
eties in the northeast region of Brazil, Goodwin et al.
(1980) analyzed the relative importance of three vari-
ables (credit; policies to affect risk through reduction in
the income variance; and increased information to im-
prove perception of the distribution of returns from new

new agricultural policies at the regional and national

Guatemala's Basic Grain Policy

On 20 Jan. 1976, the Minister of Agriculture of
Guatemala announced that the Government was
launching a program to increase agricultural production
with special emphasis on basic grains (Diario La Tarde,
1976). The Institute of Agricultural Science and Tech-
nology (ICTA) of Guatemala was already working on
developing new technologies intended to generate in-
creases in productivity. The main objective was to aug-
ment the country's total supply of basic grains without
expanding the area committed to production.
A research project was developed.4 It was focused on
the traditional small farmer. Their contribution to over-
all production of basic grain was relatively important
since 55% came from farms under 7 ha (Waugh, 1975).
Since traditional farmers devote most of their basic
production to family consumption, a potential problem
could develop if these farmers used the new technology
to produce the same or even a smaller amount of grains
on less land, while diverting the newly available land to
the production of commercial crops. The investigation
of traditional and commercial supply response became
of utmost importance.'
The authors' research objectives were to estimate
market supply functions for each basic grain (wheat,
corn, rice, beans, and sorghum) or association (mixed
crops such as corn-bean) in the different regions of the
country; and to compute the corresponding income,
farm size, and price elasticities of market supply." A
model conceptualizing the small farmer's basic
economic system was developed. A surplus-output ratio
was estimated as a function of the product's farmgate
price, education of household head, total farm size, dis-
tance to the nearest market, quantity of the product de-
manded on the farm, total family income, and a relative

For detailed descriptions of the conceptual and statistical models
and results, the interested reader is referred to Alvarez (1977) and
Alvarez and Andrew (1977b). This section is mainly based on Alvarez
and Andrew (1977a).
The term traditional farmer is used to designate farmers who his-
torically ignore market stimuli and are not prepared to shift from one
crop to another; in general, the term traditional means any system
which has been used ror "a long time" and has not been "modern-
ized" particularly in the use of petroleum based products. The com-
mercial farmer is price responsive and has the means to shift between
crops; his farming is a business and he responds to market stimuli.
The difference between traditional and commercial crops is based on
the destination of the product and the utilization of labor in its pro-
duction. In traditional crops farmers tend to use about 80 percent
family labor and 20% contract or hired labor and, although some out-
put may be sold when a surplus occurs, production is mainly devoted
to family consumption. In commercial crops the characteristics are al-
most exactly the reverse (The discussion is based on personal com-
munication with Peter E. Hildebrand, Coordinator, Socioeconomics
Program, ICaTIA t aremala to 7h responsiveness of a dependent vari-
able to changes in an independent variable. For example, (dq/df)
(f/q), where dq/df is the derivative of surplus/output with respect to
farm size at a point on the surplus/output curve, gives the farm size
elasticity of market supply.

Fig. 3. Traditional and commercial income and farm
size-quantity relationships in developing agriculture.
QT, QM, and OH TepfOSent total quantity produced,
marketed and consumed at home, respectively, while
c and t represent commercial and traditional crops,


technology) on the small farmers' decisions. A linear
programming model was developed for a representative
small farm of the region using data from a 1973 survey
and complemented with subsequent field interviewing.
The objective was to simulate the choice of cropping
activities and evaluate the effect of a new sorghum tech-
nology. Results from the simulation model showed that
new technology should be adopted by farmers highly
averse to risk, even with an overestimation of the vari-
ance of returns. (Overestimation in the model resulted
when farmers were assumed to over-estimate the down-
side risk and the extension agent to overstate the po-
tential profitability.) For nonadopters, improved
estimates of the distribution of returns will occur over
time from the demonstration effect. Therefore, the
authors concluded that crop research programs may
need to consider several alternative technologies, as the
model results for the sorghum technology adopted were
highly influenced by the farmers' risk-aversion coef-
ficient. In summary, "ex-ante technology evaluation by
simulating farm conditions appear to be useful for
ascertaining the feasibility of new technology and the
farm level constraints to its introduction" (Goodwin et
al., 1980).
The first of the former examples identified the need
for a "bottom-up" approach while the second one
pointed out the necessity to consider several alternative
technologies. The following section describes an
appropriate framework for dealing with these and other
problems in the small farm sector.

The Descriptive (Diagnostic) Stage

Elliot, 1977
Glbertdt I).,19eo
"' iffean,s s o

Desciptive (Dlagnostic) - - -+Hildebrand, 1981

Design ~ -- -+Glain, 1979

Testing ~ -- - +Grldn, 1981

Evaluation -----+Chanchillo~nd Hlldebrand, 1979
Glad win, 1976, 1981

Extension ~ -- --+ a4,6

Fig. 4. The five successive stages of the farming sys-
tems research approach, with references describing
the framework and the appropriate methodologies at
each stage,

The purpose of this stage is to unveil the farming sys-
tem practiced by the farmers in an area where new tech-
nology or cultural practices are being considered. Em-
phasis falls on the farmers' goals as well as the con-
straints they face.
A very useful methodology for this phase has been de-
vised. The "sondeo" (a team rapid survey approach)
has been defined as "a modified survey technique de-
veloped by the Guatemalan Institute of Agricultural
Science and Technology (ICTA) as a response to budget
restrictions, time requirements, and the other
methodology utilized to augment information in a re-
gion where agricultural technology generation and pro-
motion is being initiated" (Hildebrand, 1981). Its
purpose is "to provide the information required to
orient the work of the technology-generating team. The
CTOpping or farming systems are described, the agro-
socioeconomic situation of the farmers is determined
and the restrictions they face are defined so that any
proposed modifications of their present technology are
apprOpriate to their conditions" (Hildebrand, 1981).
The sondeo is usually a 6-day operation by the team
as a unit, generally a 10-person team composed of dif-
ferent biological and social scientists. The objective is to
determine the most important cropping or farming sys-
tem and begin to search out the limits of the


Experience has shown that it is ill-advised to assume a
standard pattern of development for all countries and
that trying to directly transfer technology from the de-
veloped countries to the developing world does not
render satisfactory results. Dictating development
strategies from the top levels of government has also
proved unsuccessful. Furthermore, solutions for large
farms frequently do not work for small farms. Some de-
velopment specialists have devised a new way to deal
with agricultural development. A new context was
outlined during the 1970's and is being presently applied
in many countries. Mainly aimed at the small farmer, it
has been broadly defined as one "of developing tech-
nology that not only provides for increases in produc-
tivity but does so in a way that is wholly useful, usable,
and acceptable to the small farmer, given his goals and
farming system, and including the resources and con-
straints he faces" (Norman, 1978b).
Farming systems research identifies the contributions
of both biological and social scientists as a sine qua non
for its success. It has been stated that "the necessity of
recognizing and focusing on the interaction of the tech-
nical and human elements requires a multidisciplinary
team working in an interdisciplinary manner, with the
social scientist playing an ex ante rather than simply the
traditional ex post role characteristic of the 'top down'
approach" (Norman, 1978b). Five successive stages
have been identified (Fig. 4).


homogeneous system. (Delimiting a homogeneous area
to develop farmer recommendations cannot be done
without agronomists working with anthropologists,
agricultural economists, and other social scientists.)
Perhaps the most important aspect of the sondeo report
is to orient the first year's research and to locate future
collaborators for the farm trials and for the farm record
projects where "hard" data will be generated. A de-
tailed description of this methodology can be found in
Hildebrand (1981).

The Design Stage

During this stage, the information collected during
the descriptive (diagnostic) stage is analyzed while de-
signing the experiments. Final decisions about the ex-
perimental design are made after considering the range
of alternatives available given the constraints that have
been identified. The most promising strategies are then
selected. The next step consists of conducting the ex-

The Testing Stage

The testing of the new strategies should be conducted
in two parts. The first one, defined by Gladwin (1981) as
"the generation-of-technology stage", consists of
experiments conducted by the technical team mostly on
farmers' fields, although some experiments of the com-
modity programs are highly controlled trials on the re-
gional experiment stations. The second part is a repeti-
tion of the first but in this case the farmers themselves
are entirely responsible for the experiments. This is very
important since farmers frequently refuse adopting a
new practice recommended by agricultural experiment
stations arguing that conditions on the farm are suf-
ficiently different from experimental conditions. The
new practice, of course, is compared to check plots by
researchers and farmers.
Gladwin has correctly identified this phase as the
starting of the process of technology diffusion and
transference. Data gathered during the two parts of this
phase provide valuable information to the team and the
farmers about traditional versus new technology. At
this point, decision "tree" methodology can be used to
look at the decision process of early adopters with the
objective of evaluating the probability of success of the
new technology. A decision tree is "a sequence or series
of discrete decision criteria, all of which have to be
passed along a path to a particular outcome or choice"
(Gladwin, 1979). An example will be used later in this

The Evaluation Stage

Evaluation takes place one year after the farmers
have tested the new technology. The purpose is to check
if farmers are still using it, and if not, why not. The
ICTA has developed an "index of acceptability" which
is computed for each recommendation in the tech-
nological "package" (Chinchilla and Hildebrand,

tertlizr e times
R* 3
B 3

\Don'ttiedtilize 34 ses~
A*I Success
R.0 rate*97%

Fig. 5. Decision to fertilize twice, at planting and the
second weeding, in the Plan Puebla, Mexico. Source:
Gladwin (1976).

1979). When the figure is low, the technical team may
drop one or more recommendations of the new
technology before they become "official" recom-
mendations of the institute.
The above procedure, however, is not sufficient. Pro-
gram planners need to know why farmers are doing
what they are doing or why they are not doing what the
technical team recommends (Gladwin, 1981). Based on
her vast experience with modeling small farmers' de-
cision processes, Gladwin (1976) has used decision
"tree" methodology to elicit farmers' reasons for non-
adoption of new "improved" technology.
This methodology was used to elucidate why farmers
were not adopting some of the four recommendations
included in the technological "package" of the Plan
Puebla in Mexico during 1973-1974. By developing and
testing a decision "tree" for each of the recommenda-
tions (which included profit, risk, capital, and knowl-
edge as decision criteria) the author was able to pinpoint
the main factor limiting adoption of each recommenda-
Figure 5 depicts the decision "tree" model tested for
one of the recommendations in the existing soil types.'
Although farmers in the area fertilize only at the first
weeding, the recommendation included fertilizing both
at planting and at the second weeding. The decision
model states that two applications can occur if farmers

Soils type A include fields without irrigation but with enough
moisture in the soil if plowed correctly after the preceding harvest so
that the farmer can plant early in April. Type R represents fields with
irrigation. Type B includes fields without irrigation and without
moisture mn April so that the farmer must wait for the first "regular'
rain to plant, which may occur in April or May or as late as June.
'The one error means one farmer who passed all the criteria on the
path but did not fertilize twice.


consider that fertilizing at planting is profitable and
they can pass the following constraints: risk of loss of
plants, risk of loss of costs of fertilizer, capital, and
credit. The sample included 34 farmers. The results
identified the main factors limiting adoption with a 97%
of success in predicting farmers' decisions concerning
the recommendation." In this case, the reason was non-
profitability on type A soils. The obvious policy de-
cision is to drop the recommendation on those soils.

The Extension Stage

Agricultural extension can be defined as "the provi-
sion of increased knowledge and skills necessary for
farmers to be able to adopt and apply more efficient
crop and animal production methods to improve their
productivity and living standards" (Russell, 1981). The
main objective of extension work is to help traditional
farmers to progress beyond subsistence farming by be-
coming more productive and therefore more involved in
the activities of the market. In a recent article, Russell
(1981) puts it in a farming systems research framework
when he states that
.the solution rests largely in gearing local agricultural
research to develop inputs and practices that are ac
ceptable and usable in a particular context. This comes
down to working with the local people to produce local
answers to local questions.
It has been stated in this paper that the process of
technology diffusion and transference really starts dur-
ing the testing stage. Once a new technology is evaluated
and recommended, it will have more chances of being
adopted since it has been built with the farmers' co-
operation and based on their resources and constraints.
An important characteristic of the farming systems re-
search approach is that of maintaining a constant feed-
back to and from research. This two-way flow of in-
formation helps make the system dynamic during all
How the former is done will depend on many factors.
Setting aside other relevant factors (like developing
effective or using effective marketing channels), the ex-
tension work will have to be adapted to local conditions.
The structure of the administration, the characteristics
of the area where the extension specialist is going to
work, the number of visits, etc., are variables to
consider if the extension stage is to be successful. But it
is necessary to emphasize again the "need to combine
the difficult study of farm systems with the longer itera-
tive approach of greater participation by farmers in
planning and evaluating as well as implementing further
improved practices that are already being developed
(Russell, 1981).


Emphasis in this paper falls on small, traditional
farmers because the development literature, after years
of indifference, is now focusing on them. Several
factors that would help optimizing the process of mak-
ing recommendations for small farmers in developing

countries have been identified. They are:
1. The relevance of the farming systems research ap-
proach. Although there are still implementation prob-
lems (Norman, 1978b), this approach appears to be an
appropriate framework to generate solutions since it has
been successfully tested in many areas of the world. It
allows a multidisciplinary team to work in an interdisci-
plinary manner with all scientists involved during all
phases of the project.
It is necessary to emphasize that the existence of a
farming systems research program is not going to gener-
ate successful solutions by itself. Other factors (like
credit, marketing channels, land availability, etc.) also
need to be present for the recommended practices to be
adopted and farmers' income to be increased.

2. The need for ex-ante evaluations. A few examples
have been provided. Several methodologies are appro-
priate, depending on data availability and the objective
of the evaluation. Even when some time must be spent
on data collection, it is a worthwhile investment when
compared to the years involved in doing research with
little or no probability of adoption of new practices.
Ex-ante evaluations are particularly important be-
cause they may point out the direction the research
should take, or identify a complete reorientation of
efforts where the probabilities of success are greatest. In
plain words, they can serve the purpose of telling re-
searchers whether or not the derived recommendation
"will fly."
3. The importance of the extension stage. A farmer
recommendation derived from a farming systems re-
search approach contains the input of all scientists in-
volved. It is tailored to the farmers' goals and the con-
straints they face since it has been previously tested by
the farmers themselves. During the extension stage it is
extremely important to treat each recommendation
separately. The critical factor or intervention point in
the decision to adopt one recommendation in the
"package" is not necessarily the factor limiting adop-
tion of another recommendation.
Two decades ago, a development specialist wrote
.. What we tend to forget, however, is that the es-
sential aspect of an "underdeveloped" economy and the
factor the absence of which keeps it "underdeveloped"
is the ability to organize economic efforts and energies,
bring together resources, wants, and capacities, and so
to convert a self-limiting static system into creative, self-
generating organic growth (Drucker, 1964).
That challenge is still there for all agricultural scien-
tists interested in the developing world. It has been
recently stated that
The worldwide food crisis is real. It is serious, and it will
probably worsen. Worldwide famine in all probability
will be a recurring theme in our lifetime. But it is not in-
escapable, provided we hold fast to the following article
of faith: While the world's resources may be limited, we
have yet to discover the bounds of human creativity
(Wharton, Jr., 1977).
By working together, development scientists can play
a major role in getting closer to the bounds of human
creativity. Optimizing the process of making farmer


recommendations will help the self-generating growth
become a reality in the developing countries.


1. Alvarez, J. 1977. Traditional and commercial farm sup-
ply response in agricultural development: the case for
basic grains in Guatemala. Ph.D. Thesis, University of
2. ----, and C. O. Andrew. 1977a. Contradictory
policies in developing agriculture: Guatemala's basic
grains, a case in point, Latinamericanist 12:3-4.
3. -,and ----. 1977b. Supply response by
traditional and commercial producers of basic grains in
LDCs. South. J. Agric. Econ. 9:157-162.
4. ----, and B. E. Melton. 1981. The role of agricul-
tural economists and physical scientists in IFAS inter-
disciplinary research. Univ. of Florida, Food and Re-
source Economics Dep. Staff Paper No. 171.
5. Brady, N. C. 1977. The role of agronomists in interna-
tional agricultural development. p. 95-108. In M. D.
Thorne (ed.) Agronomists and food: contributions and
challenges. American Society of Agronomy, Madison,
6. Chinchilla, M., and P. E. Hildebrand. 1979. Evaluacion
de la aceptabilidad de la tecnologia generada para el
cultivo de maiz en Quezaltenango, 1977-1978. ICTA,
7. Diatio La Tarde. 1976. Ano VI, No. 1614, Martes 20 de
Enero, Guatemala.
8. Drucker, P. F. 1964. Marketing and economic develop-
ment. p. 333-338. In R. J. Holloway and R. S. Hancock
(ed.) The environment of marketing behavior--selections
from the literature. John Wiley and Sons, N.Y.
9. Elliot, H. 1977. Farming systems research in Francophone
Africa: methods and results. Paper presented at the
Middle East and Africa agricultural seminar held in
Tunis, Tunisia, 1-3 February.
10. Gilbert, E. H., D. W. Norman, and F. E. Winch. 1980.
Farming systems research: a critical appraisal. Michigan
State Univ., Dep. of Agric. Economics Rural Dev. Paper
No. 6.
11. Gladwin, C. H. 1976. A view of the Plan Puebla: an
application of hierarchical decision models. Am. J. Agric.
Econ. $8:881-887.
12. ----. 1979. Production functions and decision
models: complementary models. Am. Ethnol. 6:653-674.
13. ----. 1981. Contributions of decision-tree method-
ology to a farming systems program: the case of ICTA.
Univ. of Florida, Food and Resource Economics Dep.
Mimeo Rept.
14. Goodwin, J. B., J. H. Sanders, and A. D. de Hollanda.
1980. Ex-ante appraisal of new technology: sorghum in
northeast Brazil. Am. J. Agric. Econ. 62:737-741.
15. Gotsh, C. H. 1977. The concept of farming systems in the
analysis of agricultural research and development pro-
grams. Paper presented at the Middle East and Africa
agricultural seminar held in Tunis, Tunisia, 1-3 February.
16. Heady, E. O. 1948. Elementary models in farm produc-
tion economics research. J. Farm Econ. 30:201-225.
17. ----. 1949. Implications of particular economics in
agricultural economics methodology. J. Farm Econ. 31:
18. ----. 1957. Organization activities and criteria in ob-
taining and fitting technical production functions. J.
Farm Econ. 39:360-369.

19. ----, and J. Pesek. 1954. A fertilizer production sur-
face with specification of economic optima for corn
grown on calcareous ida silt loam. J. Farm Econ. 36:
20. ----, and W. D.Shrader. 1953. The interrelationship
of agronomy and economics in research and recom-
mentations to farmers. Agron. J. 45:496-502.
21. Hildebrand. P. E. 1977. Generating small farm technolo-
gy: an integrated multidisciplinary system. Paper pre-
sented at the 12th West Indian agricultural economics
conference held in Antigua, Guatemala, 24-30 April.
22. ----. 1981. Combining disciplines in rapid appraisal:
the sondeo approach. Agr. Admin. 8:423-432.
23. Ibach, B. 1953. Use of production functions in farm man-
agement research. J. Farm Econ. 35:938-952.
24. Johnson, G. L. 1954. Major opportunities for improving
agricultural economics research in the decade ahead. J.
Farm Econ. 36:829-840.
25. Mason, D. D. 1957. Statistical problems of joint research.
J. Farm Econ. 39:370-381.
26. Norman, D. W. 1978a. Farming systems research to im-
prove the livelihood of small farmers. Am. J. Agric.
Econ. 60:813-818.
27. ----. 1978b. The farming systems approach: rele-
vancy for the small farmer. Paper presented at the
CENTO seminar on increasing the productive capacity of
small farmers held in Lahore, Pakistan, 17-21 December.
28. Perrin, R. K., D. L. Winkelman, E. R. Moscardi, and
J. R. Anderson. 1976. From agronomic data to farmer
recommendations--an economics training manual. Inter-
national Maize and Wheat Improvement Center Informa-
tion Bull. No. 27, Mexico.
29. Redman, J. C., and S. Q. Allen. 1954. Some interrela-
tionships of economic and agronomic concepts. J. Farm
Econ. 36:453-465.
30. Russell, J. 1981. Adapting extension work to poorer agri-
cultural areas. Fin. Dev. 18:30-33.
31. Savoie, P., and M. Kabay. 1980. Choosing optimum ap-
plication rates in developing countries. Am. J. Agric.
Econ. 62:734-736.
32. Schmehl, W. R., P. F. Philipp, and W. W. Shaner. 1981.
Farming systems research methodology. Am. Soc.
Agronomy Abstracts, p. 47.
33. Simpson, J. R. 1980. Determining optimal types of cattle
for tropical and subtropical dairy operations. Univ. of
Florida Food and Resource Economics Dep. Staff Paper
No. 153..
34. Swanson, E. R. 1957. Problems of applying experimental
results to commercial practice. J. Farm Econ. 39:382-389.
35. The Palm Beach Post. 1980. West Palm Beach, Fl.,
Wednesday, 1 October, p. A-4.
36. Thorne, M. D. (ed.). 1977. Agronomists and food: con-
tributions and challenges. American Society of Agrono-
my, Madison, Wis.
37. Waugh, F. V. 1953. Applicability of recent developments
in methodology to agricultural economics. J. Farm Econ.
38. Waugh, R. K. 1975. The institute of agricultural science
and technology of Guatemala (Instituto de ciencia y tec-
nologia agricola) ICTA--Four years of history. ICTA,
39. Wharton, C. F., Jr. 1977. Ecology and agricultural de-
velopment: striking a human balance. p. 1-8. In M. D.
Thorne (ed.) Agronomists and food: contributions and
challenges. American Society of Agronomy, Madison,

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs