ORGANIZATION AND MANAGEMENT
AGRICULTURAL RESEARCH SYSTEMS
CARSOM* WORKING PAPER
INSTITUTE OF INTERNATIONAL AGRICULTURE
*Committee on Agricultural Research System Organization and Management
MICHIGAN STATE UNIVERSITY
East Lansing, Michigan 48824-1039
MSU is an Affirmative Action/Equal Opportunity Institution
LINKAGES BETWEEN AGRICULTURAL RESEARCH AND
EXTENSION IN LESS DEVELOPED COUNTRIES
by Nicholas W. Minot
This paper was published by the Committee on Agricultural Research
System Organization and Management, Institute for International Agriculture,
Michigan State University under the direction of Dr. Darrell Fienup and with
financing from the Title XII Strengthening Grant AID/DSAN-XII-G-0126, U.S.
Agency for International Development.
The Committee on Agricultural Research System Organization and
Management (CARSOM) was established in September 1981 as an interdepartmental
unit under the auspices of the Institute of International Agriculture. Its
purpose has been to coordinate and facilitate research and teaching
activities concerning the issues and problems in improving the organization
and management of agricultural research systems in developing countries.
CARSOM has three principal objectives.
1) Improve the services provided by MSU in it's overseas
project work and advisory missions.
2) Strengthen the quality of the training received by
professionals from developing countries studying at MSU.
3) Contribute directly to the analysis and development of
agricultural research policymaking through research
One of the most serious problems in the generation and adoption of new
technologies in developing countries has been the failure to form strong
working relationships between agricultural research and extension. Better
incentives must be found to help research and extension realize they have
common goals which can only be achieved through more effective integration
of their respective activities. The only reason for each service to exist
is to serve farmers and help promote national policy goals.
In this working paper, Nicholas Minot has reviewed Third World develop-
ment strategies since the 1950's as they pertain to the emphasis given to
agricultural research and extension and the linkages between them. There
is a comprehensive analysis of the constraints to establishing better
linkages as well as the current "state of the arts" in dealing with these
issues. Several country case studies are examined to illustrate some of
the more innovative approaches to the problem.
Mr. Minot developed this paper while he served as Research Assistant
to CARSOM during 1982/83. He is a graduate student in the Department of
Agricultural Economics at Michigan State University.
Darrell F. Fienup
The author would like to express his appreciation to Darrell Fienup,
as CARSOM Chairperson, for his support and guidance in the development of
this paper and to David Rohrbach for his extensive editorial and substan-
tive suggestions. In addition, the author is grateful for comments from
the following reviewers: Harold Riley, Russell Freed, Patricia Bonnard,
and Winston Oberg. Finally, credit is due to Cindy Spiegel for her
proficiency and patience in typing the manuscript.
TABLE OF CONTENTS
PREFACE . . . . . . . . . . . . . . .
ACKNOWLEDGEMENTS . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . .
BACKGROUND OF RESEARCH AND. EXTENSION STRATEGIES . . . .
AGRICULTURAL RESEARCH: GENERATING APPROPRI
Evaluation Criteria . . . . .
Agro-Climatic Conditions and Topography
Farm Practices . . . . . .
Explanations for Conventional Research
AGRICULTURAL EXTENSION: REACHING THE FARME
RESEARCH-EXTENSION LINKAGES . .
Conceptual Framework . . .
Empirical Studies . . .
Aproaches to Solutions . .
Establishment of a Liaison
Cooperative Activities .
Training and Visit System
Farming Systems Research .
ILLUSTRATIVE CASES . . . .
Botswana . . . . . .
Guatemala . . . . .
Taiwan . . . . . .
SUMMARY AND CONCLUSION . . .
. . . . . . 13
. . . . . 14
. . . . . . 15
. . . . . . 17
R . . . . . 19
. . . . . . 25
. . . . . . 26
. . . . . . 30
. . . . . . 33
. . . . . . 34
. . . . . . 36
. . . . . . 39
. . . . . . 45
. . . . . . 51
. . . . . . 56
. . . . . . 57
. . . . . . 60
. . . . . . 63
. . . . . . 67
It is often argued that the generation and diffusion of agricultural
technology ought to be considered as a seamless process rather than as the
two discrete stages commonly associated with the terms "research" and
"extension" (McDermott, 1976; Cummings, 1981). This follows from three
observations. First, there are no clear dividing lines between the two
stages; some activities cannot easily be classified as one or the other.
Second, research and extension share a common goal--the dissemination of
improved agricultural technology. Third, the productivity of each compo-
nent is dependent on that of the other. It is frequently observed
that "there is no use having an extension system unless there is something
useful to extend". By the same token "research without extension is not
of much use and can even be a waste of resources" (Uphoff, 1981:1).
In reality, however, the generation and diffusion of agricultural
technology in many developing countries is not a "seamless process." Instead,
distinct research and extension institutions operate with a large degree of
independence. Communication and direct working interaction between these
institutions is often surprisingly limited. In part, the weak linkage can
be traced to the administrative separation of the two agencies. But this
is not a sufficient explanation. Formal and informal interaction between
research and extension could compensate for the administrative separation.
This is frequently impeded, however, by differences in status and education
between researchers and extension personnel, by incentive structures within
each organization which do not encourage contact, and by time and resource
This paper will provide an overview of the issues involved in strength-
ening research-extension linkages in developing countries in terms of the
goals of improving the relevance of the agricultural technology generated
and accelerating the diffusion of that technology. This discussion will
emphasize consideration of the difficulties of serving the smallholder
sector. The development of this sector has been and continues to be a major
donor agency and developing country priority. Despite this, research and
extension responsiveness to smallholder needs remains deficient. This paper
explores the role of linkages between research and extension programs in
dealing with this problem and reviews several strategies currently being
employed to strengthen these linkages.
The next section provides a brief historical review of the evolution of
agricultural research and extension strategies in developing countries and
outlines the importance of research-extension linkages. Then, two sections
are devoted to a review of the constraints to effective agricultural
research and extension programs. After this foundation has been laid, the
nature of research-extension linkages is discussed more directly, including
a consideration of several approaches to strengthening these ties. Three
innovative national systems are used to illustrate the need to adapt solu-
tions to local circumstances. Finally, the results of the paper are
summarized and the major conclusions reviewed.
BACKGROUND OF RESEARCH AND EXTENSION STRATEGIES
Development strategies in the field of agricultural research and
extension have reflected prevailing strategies of economic development, the
experiences of previous efforts, and, of course, economic and historical
forces. In the 1950s, the dominant view of rural development held that
"modernization" consisted of the change in peasant attitudes from
"traditional" to "progressive." In particular, the peasant was seen as unre-
sponsive to economic incentives, and hence, resistant to the changes
required to increase agricultural production. Research on the adoption of
innovations focused on sociological factors such as the education, social
position, and personal characteristics of the potential adopter. The
implicit assumption was that appropriate (i.e. profitable) technology
existed and that poor diffusion was the primary obstacle to adoption. This
accordingly became known as the "diffusion model" of agricultural develop-
ment. The resultant rural development strategies emphasized agricultural
extension, community development, and credit programs. Extension research
generally involved comparisons of alternate communication forms with which
to extend the message, e.g. individual and group meetings, demonstrations,
radio, and so on.
The community development programs used village-level generalists to
stimulate self-help projects and facilitate the provision of social
services meeting self-defined needs of the local people. Following the
initiation in 1952 of an ambitious community development program in India
with funding from the U.S. foreign assistance agency and the Ford Foundation,
the United States and the United Nations helped encourage and finance the
rapid proliferation of this strategy. By 1959, the U.S. was assisting
twenty-five countries with such programs and another thirty or so had
developed with alternative funding. U.S. aid focused on Asia with India,
Pakistan, and the Philippines receiving almost half the total amount
(Holdcroft, 1978). The experience of the community development programs
revealed that 1) more technical assistance was needed to stimulate food
production, 2) social stratification within the village prevented the
achievement of the stated goal of creating harmonious communities and mass
participation, 3) efforts to work through local elites introduced a
paternalistic bias and impeded widespread distribution of benefits, and
4) friction and opposition from agricultural specialists was generated
The popularity of agricultural extension did not rise and fall as
dramatically as did that of community development, but the attention given
to it has fluctuated over the years. Through the 1950s, U.S. efforts
helped establish and strengthen extension systems in Latin America, Asia,
and in northern Africa. The general approach was to transplant the U.S.
Land Grant System, integrating research, extension, and university educa-
tion. Strongest emphasis was placed on extension based on the assumption
that it was the weak link of the three (Barker, 1981; Rice, 1974). Although
there was support for agricultural research, especially by the Ford and
Rockefeller Foundations, Moseman (1977:370) writes that "the primary
emphasis in technical assistance for agricultural development...was on the
introduction and testing of materials and practices from agriculturally
Meanwhile, British, French, and Dutch colonial administrations established
commodity-based institutions which combined research and extension activities
in Africa and Asia. The research institutions concentrated heavily on
export crops, and the extension work included important regulatory functions
to maintain military/administrative control (Anthony et al, 1979:223-261
describes the evolution of research and extension in Africa). It is interest-
ing to note that in the 1920s and 1930s the Japanese colonial administration
in Taiwan and Korea conducted similar programs to develop rice production
for export to Japan (Hayami, 1974).
In the 1960s, growing doubt regarding the assumptions behind the
"diffusion model" led to a reevaluation of this strategy. In Transforming
Traditional Agriculture (1964a), T.W. Schultz argued that small farmers were
economically rational given their scarce resources and precarious livelihood.
This line of thought had important implications for extension.
Wherever agricultural extension programs have been launched,
based on the assumption that these farmers are necessarily
inefficient in using (allocating) the agricultural factors
at their disposal, it is highly probable that the programs
have not contributed and cannot contribute to economic
growth...[they] are attempting to induce farmers to adopt
and use one or more new agricultural inputs that simply are
not productive enough to make it worth while for farmers to
introduce them. (Schultz, 1964b:196-197)
Similarly, Moseman (1970:71) wrote:
This "extension bias" [in development efforts] met with
only limited success because of the paucity of applicable
indigenous technology and the general unsuitability of U.S.
temperate zone materials and practices to tropical
This view that traditional farmers are responsive to economic incen-
tives was dramatically confirmed by the rapid rates of adoption of high-
yield varieties (HYVs) in certain areas. In Kenya, the area cultivated by
smallholders with improved varieties increased from zero to almost 47,000
hectares between 1963 and 1967 (USAID, 1979: Appendix C, page 2). Rice
varieties developed by the International Rice Research Institute (IRRI)
spread rapidly where irrigation and available fertilizer made them profit-
able, primarily in Asia. Rapid rates of adoption were also observed for
semidwarft wheat varieties developed by the International Center for Corn
and Wheat Improvement (CIMMYT), again primarily in Asia and, to a lesser
degree, in Latin America.
Much debate has centered on the fact that, in some areas, HYVs spread
more rapidly among large farmers and in well-endowed regions contradicting
equity objectives. There is disagreement over the scope and impact of this
trend and whether it is attributable to the biological technology itself or
to unequal access to complementary inputs, such as water, fertilizer, and
credit. But there is little doubt that small farmers adopted HYVs where
it was feasible and profitable for them to do so (Dalrymple, 1979). As a
result, slow adoption rates are now discussed primarily in terms of economic
and institutional constraints rather than in terms of irrational resistance
to change among tradition-bound farmers.
The successes of IRRI and CIMMYT led to what Esman (cited in Barker,
1979:495) called the "big science" model of agricultural research. National
agricultural research systems were seen as too poorly funded and staffed to
make significant technological advances. The international agricultural
research centers (IARCs) seemed to be a more cost-effective way of using
scarce research manpower and funds. 'Support for national agricultural
research declined in the second half of the 1960s, while support of the
IARCs expanded. Evenson (1978:226) estimates that annual bilateral and
multilateral aid to national agricultural research declined by $20-30
million (approximately 30%) in real terms over 1965-1971. The Consultative
Group on International Agricultural Research (CGIAR) was formed in 1971 to
coordinate multilateral funding for the four existing IARCs. Six additional
IARCs were established and incorporated into the system within three years
More recently, a number of factors have diminished the appeal of the
"big science" model and contributed to renewed interest in supporting
national systems to complement the IARCs. First, though the IARCs continue
to play an important role in the generation of agricultural technology,
dramatic results on the scale of the "green revolution" wheat and rice
varieties have not been replicated. Second, a substantial body of litera-
ture has accumulated indicating that national, as well as international,
agricultural research yields high economic rates of return, typically between
30 and 70%.1 The World Bank (1981:19-20) has responded by stating that
underinvestment in agricultural research appears to be
substantial, and considerably more money could be invested
in this activity with the expectation that returns would
exceed both the opportunity cost of capital and the returns
from most feasible alternatives in rural areas.
Third, there is evidence that the ability to "transfer" technology from
other countries (and thus from the IARCs) depends heavily on the existence
of an indigenous research capacity. Studies of the diffusion of the green
revolution varieties show that agro-climatic factors and topography were most
important in determining the usefulness of HYVs to farmers, and thereby the
rate of adoption (Perrin et al, 1976 review such studies). The greater
the environmental sensitivity of the technology, the greater the need for
national agricultural research capacity. This explains the fact that the
majority of improved varieties in use today are crosses of improved HYVs
and local varieties. Evenson and Kislev (1975:58-77) used data for more
than forty countries over a 20 year period to construct an econometric
model of wheat and maize yields as a function of indigenous research output
and estimated "borrowed research." 2 This study and an expanded, more
sophisticated follow-up study are summarized as follows:
1Schuh and Tollini (1979) discuss methodological issues; Evenson,
Waggoner, and Ruttan (1979) provide a table summarizing 32 such studies
starting with the pioneering work of Griliches (1958); and for an opposing
point of view, see Pasour and Johnson (1982).
2The number of internationally abstracted articles was used as a
proxy for indigenous research output, and borrowed research was estimated
assuming that the flow of borrowed knowledge was a function of time and the
"stock" of available knowledge in that country's agro-climatic zone. In
spite of-the highly aggregated nature of the study, the coefficients were
significant with the expected signs. A zero intercept indicated that "no
borrowing takes place in the absence of indigenous research work" (Evenson
and Kislev, 1975-70).
if investment in international research centers is to bear
fruit in countries which at present have minimal research
capacity, high priority must be given to the support of
national research systems. (Evenson and Binswanger,
Agricultural research and extension strategies have also recently
expanded to encompass support for farming systems research (FSR) programs.
Though the term has been indiscriminately used due to the popularity of
the concept, most sources define FSR as a location-specific, farmer-
oriented, interdisciplinary type of agricultural research using rapid
diagnostic survey techniques, on-farm testing, and an iterative process
of technology development and testing. It was conceived to produce simple,
useful recommendations for small farmers recognizing that the complex
cropping systems and multiple-objectives of these farmers call for a
holistic approach to research (see Shaner et al, 1982 and Rohrbach, 1981).
The potentially important role of FSR in strengthening linkages between
research and extension will be considered later.
The Training and Visit Extension System represents an alternative
agricultural development strategy. This entails an attempt to improve the
effectiveness, discipline, and technical competence of the extension service
through the use of regular in-service training of agents, rigid farm visit
schedules, and organizational reforms (see Benor and Harrison, 1977). The
planned and apparent role of the T and V system will also be discussed.
It is important to note that the trends in agricultural research and
extension discussed thus far reflect strategies largely employed in donor-
agency funded projects. They are not always manifested, however, in the
pattern of overall national expenditure for extension and research. This
is due both to intervening historical factors such as decolonization and to
the widely varying levels of national commitment to research and extension.
The first major attempt to compile data allowing the comparison of
national support for agricultural research and extension was made by Evenson
and Kislev (1975a,b). In their expanded and updated version of this study,
Boyce and Evenson (1975:11) identified several funding patterns. 1) Expendi-
tures on agricultural research as a proportion of agricultural product
increased by a factor of two to three over the 1951 to 1974 period. 2) The
proportion of expenditures on agricultural extension grew during the fifties
and sixties, but increased only in the poorer countries in the early 1970s.
3) High-income countries, when compared to low-income countries, spent higher
proportions of their agricultural product on research but lower proportions
on extension. For example, in 1974 countries in the highest income category
(over $1750 per capital in 1971) spent 2.5% of their agricultural product
on research but 0.6% on extension, whereas the countries in the lowest
income category (under $150 per capital in 1971) spent 0.67% on research but
1.82% on extension. In a 1981 study, Oram and Bindlish confirmed these
general patterns. Boyce and Evenson (1975:13) interpret this as evidence
that "decision-makers are acting as though they believe extension resources
to be highly substitutable for research resources." The authors view this
perception as mistaken. They claim the relatively large investment in
extension services by low income countries results in part from a "bias in
favor of projects with quick results" (p. 118). Pinstrup-Andersen (1982:67)
notes this allocation pattern may also occur because the benefits from
agricultural research are less visible than those from extension and
because research is more expensive per person employed.
In summary, the growth of the view that the traditional farmer is
rational given his or her resource constraints has led to greater attention
to technology generation relative to technology diffusion. In addition,
increasing awareness of the high degree of environmental sensitivity of
agricultural technology has highlighted the importance of a strong national
agricultural research capacity to adapt technologies to local agro-climatic
conditions. The IARCs are no longer viewed as a "solution" to the limita-
tions to agricultural research in developing countries, but as an important
complement to national research efforts. Similarly, extension is no longer
seen as sufficient for technology transfer, but rather as one component in
the process of technology generation and diffusion. With the growing
acceptance of the idea that both institutions are necessary to deliver
suitable agricultural technology, there has been increasing interest in
improving the linkages between the two functions. In other words, instead
of being seen as substitutes for each other, there is greater focus on the
complementarity between the two and on ways to improve that complementarity.
In this paper, "linkages" will refer to permanent channels of communi-
cation and mechanisms for cooperation between institutions. Linkages
between research institutions and extension agencies facilitate the flow
of technical information and recommendations from research to extension.
They also must encompass the flow of information from extension to research
concerning farmer practices, constraints, and response to previous
This definition seems simple. Yet the failure to appreciate the proper
nature of this function is broadly apparent in continuing citations of its
weakness. Evaluation reports from donor agency and developing country
analysts have consistently called attention to the significance of this
constraint. For example, it has been mentioned as a problem in the cases
in Ethiopia (Gebre, 1980:33), Tanzania (Nelson, 1979:10), Uganda (Ferguson,
1971), Upper Volta (Ames and Busch, 1979:206), Kenya (ISNAR, 1981a:71-72),
Swaziland (USAID, 1980:68), and Nigeria (Idachaba, 1980:28) in Africa.
In addition, this deficiency has been cited in Egypt (Oteifa, 1980:152),
Iran (Karami and McCormick, 1982), India (Rao, 1972), Sri Lanka (Golden,
1974:266), Indonesia (ISNAR, 1981b:16), and in Brazil (Lopes, 1980:152),
as well as in other Latin American nations (Rice, 1974). Yet few of these
assessments do more than simply cite the existence of "poor coordination"
or "weak linkages" and recommend that they be strengthened.
What is needed is more specific prescription of the procedures, train-
ing, and organizational framework which will strengthen this linkage.
As Chambers (1975:2) notes:
this word 'coordination' is an admirable means of evading
clear and detailed thought and prescription... Furthermore,
it is often accepted as so unquestionably good that more of
it is always desirable.
The point is that coordination is achieved through specific actions which
entail costs. We must seek optimal coordination rather than maximum
coordination. This requires stepping beyond the easy response of abstract
conceptualization and developing well-defined strategies to meet well-
defined circumstances. Poor linkages between research and extension are
only a problem to the extent that they contribute to specifically definable
constraints. Perhaps too much of the technology being developed is not
appropriate to farmer needs. Perhaps extension agents do not recognize
or cannot effectively transfer the relevant technology that exists. Each
of these problems might call for a different response. This paper calls
attention to the fact that research-extension linkages must be considered
in the context of the actual problems in technology generation and diffusion
to which they relate.
AGRICULTURAL RESEARCH: GENERATING APPROPRIATE RECOMMENDATIONS
One of the most frequent criticisms of agricultural research in develop-
ing countries is that it does not produce recommendations which are useful
to the farmer, especially the small farmer. .This criticism comes both from
extensionists who often feel unfairly blamed for the failure of farmers to
adopt "improved" methods and crop varieties and from outside evaluators and
agricultural institutions. For example, a survey of six research institu-
tions and eight extension services in Latin America revealed that
82% of the research scientists felt that extension personnel
made too little effort to learn about and transfer the
technologies [while] seventy-five percent of the extension-
ists felt that the new technology was not acceptable to
smallholders. (Fernandez, 1981:62)
The World Bank (1981:74) confirms the latter view of the case of Africa
concluding that "perhaps the most important problem [for African extension
services] is that, in many instances, the staff does not have a valid
message to extend." Similarly, Morss et al (1976a:103) conducted a review
of 36 rural development projects in eleven countries and reported:
We began our study on the assumption that the technology
necessary for improved output and income for small farmers
was available... Instead, we found that 61 percent were
inadequate for one reason or another.
Since applied agricultural research generally involves the testing
of a few input levels and practices under specific conditions and using
certain criteria, we can categorize the potential problems of representa-
tiveness as follows: 1) the criteria used to evaluate technologies may not
agree with those used by farmers, 2) the agro-climatic factors (uncontroll-
able conditions) may not correspond to those faced by farmers, and 3) the
inputs and practices held constant in the experiment (controllable condi-
tions) may not correspond to those actually used. These will be considered
Agricultural research in developing countries has been criticized for
its focus on narrow technical criteria, primarily the maximization of
average yield. Collinson (1982:2) writes:
Researchers have historically placed heavy emphasis on
biological potential and yield as the dominant criteria
upon which to base recommendations for farmers. But
farmers never seek biological potential for its own
sake and never make decisions on which crop and animal
products to produce on the basis of yield alone.
An important dimension frequently neglected by researchers evaluating
a technology is the cost and availability of the recommended inputs relative
to the value of the output as perceived by the farmer. Anthony et al
(1979:257, 259) notes that in Africa, "standard recommendations for
fertilizer application are frequently not economical" and some practices
are promoted as "good farming" without any consideration of their economic
value to the farmer.
A more subtle example is given by Belshaw and Hall (1972:49) who
criticize the following statement of a research director:
cotton yields for this area are optimum when sown about
mid-June. Therefore, the preceding crop itself must be
an early maturing, though possibly lower-yielding variety
They remark that such a judgement can only be based on the relative costs,
prices, and yields of each crop under alternative planting dates. It
must also consider the relative value of each crop to the farmer. This
example reflects the frequent deference researchers show to national policy
goals, which place priority on cash crop production, despite the fact that
farmers are likely to place greater emphasis on securing a minimum level
of food production.
In addition, researchers commonly fail to consider the importance of
labor costs to the farm family. The labor requirements of a new variety
or practice is often one of the farmers' greatest concerns, especially
given the frequent severity of seasonal labor bottlenecks at planting,
weeding, and/or harvest time (see Lele, 1975). The World Bank (1979),
in its policy paper on agricultural research, recommends greater focus
on net returns to labor and reduced attention to returns to land. Similar-
ly, research recommendations have in some cases been rejected because they
did not take into account the value of by-products such as stalks which
may be used as animal feed, bedding, or construction material. By failing
to account for this, researchers may overestimate the benefits of an
"improved" variety over the existing one.
In addition to economic returns, another criterion which agricultural
research has historically neglected is risk. High-yielding varieties
generally increase production risk. They may contain a lower resistance to
disease and pests. They also require greater use of purchased inputs with
fluctuating prices and, at times, questionable availability (Stavis, 1979:14).
Risk is particularly important to the near-subsistence farmer for whom
failure to earn a minimum income or produce a minimum amount of food could
Many other criteria could probably be added to this list. The argument
should be clear, however, that the more narrow and purely technical the
criteria used by agricultural researchers, the greater the divergence
between what they recommend as "improved" and what the farmer will consider
Agro-Climatic Conditions and Topography
Second, research results may be inappropriate to farmer needs due to
the differences in agro-climatic conditions and topography between the
research site and the target farming region. Not only does this limit the
effectiveness of international diffusion of technology, as noted earlier,
but it limits the appropriateness of recommendations across relatively
small agro-climatic sub-zones within a single country. Through political
interference or simple error, a research station may be located in a region
highly atypical of the area of cultivation of the crop or crops for which
it is responsible (see, for example, Arnold, 1976). More commonly, a
single research station may be called upon to make recommendations serving
several agro-climatic zones or entire country. For example,
Kenya has a very wide range of ecological conditions and soil
types... Yet, technical advice for Kenyan agriculture is
most often given in one or two national packages. (Leonard,
Additionally, research stations tend to be located on level, well-
drained, sites with relatively fertile soils. Naturally, technology
adapted to these conditions may not be appropriate for the hilly, infertile,
or drought conditions of many farms.
Finally, research recommendations may not be appropriate to farmers
if the actual farm practices differ greatly from the controlled conditions
on the research station, i.e. the physical inputs, labor inputs, and cultural
practices which are held constant in the agricultural experiments.
Improved varieties may be superior only with unrealistically high
levels of fertilizer application; the optimal date of planting will vary
depending on the crop which precedes it; and fertilizer response will
probably be different on the irrigated research plot than on the rainfed
farm. Furthermore, research trials generally consist of single crops
planted in rows, whereas broadcast planting and intercropping are common
on small farms throughout the developing world.
If recommendations are based on careful cost accounting of all inputs
and are made as a complete package, the results from such experiments may
be valid. Yet experience has shown that such radical changes in a farming
system are unlikely to be widely adopted. More commonly, however, partial
recommendations are derived from inappropriate testing conditions which
encompass intensive management, row cropping of pure stands, careful weeding
and water control, substantial chemical application, and even tractor
ploughing (Belshaw and Hall, 1972).
While awareness of this problem has grown in the last several years,
a recent description of research in Upper Volta states:
several bilateral and multilateral organizations were at
work testing varieties of hybrid seeds utilizing high-cost
technology which were unavailable even to the wealthiest
farmers. Even though their research may be of high quality,*
it will certainly remain unused for years. (Ames and
Even the relatively advanced research system in India suffers from this
Since a large proportion of the recommendations were
developed on research station under irrigated conditions
without financial constraints, there was often a gap
between the inputs and practices required by the recom-
mendations and what a rainfed farmer with limited funds
could reasonably be expected to do. (World Bank, 1981:3)
Thus, persistent reluctance to adopt research recommendations fre-
quently indicates differences in evaluation criteria, in agro-climatic
conditions, and/or farming practices between the research trials and actual
farm conditions. Some of the more commonly unadopted recommendations
involve more than one of these divergences. For example, row cropping
and the use of pure stands have been resisted by small farmers in many
areas. The advantages of intercropping became publicized by the results
of the Puebla Project in Mexico.
Although subsistence farmers in Mexico and other countries
have traditionally grown maize and beans in association,
agricultural scientists have generally rejected this
cropping system, and dedicated their resources to investi-
gating the two crops alone. [Nevertheless] the net income
from the maize-bean association receiving both chemical
fertilizer and chicken manure...was 2.28 times that derived
from maize planted alone. (Jimenez and Laird, 1974:292)
The potential benefits of intercropping also include reduced labor
requirements for weeding, better soil and water conservation, and extended
harvest period to reduce seasonal labor bottlenecks and provide food over
a longer time, and reduced risk of total crop failure due to crop variety
(see Grimes, 1963 for an early reference and Ruthenberg, 1971). Neverthe-
less, until recently, research on intercropping has been neglected (for
example, see Nweke, 1979:28 and USAID, 1980).
Other examples of consistently rejected recommendations include
"optimal" planting dates for cash crops which interfere with food production,
high levels of chemical application which may be uneconomic or require peak-
season labor use, and crop spacing recommendations which ignore the labor
"costs" of weeding.
Explanations for Conventional Research
It is useful to ask why this type of agricultural research is so wide-
spread if it does not produce recommendations useful to much of its clientele
Three possible explanations will be mentioned here. First, many of the
methodologies of conventional agricultural research were transplanted from
developed countries where they are more appropriate to local farming condi-
tions. In particular, the bias toward monocropping research is partly based
on training in developed countries. Whyte (1981:21), in writing about the
initial inclinations of the Puebla Project scientists, observes that
They simply followed the customary [U.S.] practices without
recognizing that the logic of U.S.-style row spacing without
interplanting depended upon the use of the tractor, which was
not appropriate for small farmers with little capital and
Second, the continuing reliance on conventional agricultural research
methodologies may in part, be attributed to their simplicity. Intercropping
research incorporating multiple objectives requires more complex experi-
ments, greater data collection and more refined statistical analysis.
Interdisciplinary research is frequently more difficult to manage, and on-
farm trials introduce a variety of methodological complications and addi-
tional costs. Unfortunately, these costs and complications are more obvious
than the hidden, but substantial costs of inappropriate technology.
Third, in concentrating on high-input monocropped trials to maximize
yields, conventional agricultural research does respond to the needs of
one important though limited research clientele. The large farmer
is able to come closer to the conditions existing on the
research station because he has better access to all the
factors of production. He can more readily add labor to
change planting dates and risk changing part of his
cropping pattern, as he has either other enterprises or
resources as an insurance in the event of failure.
(Russell, 1981 :31)
Official government policy may favor large commercial farmers believing
that they alone are capable of providing surpluses for export, for agro-
industry, or for urban consumption. Alternatively, in the absence of
explicit research policy, large farmers may simply be more able to
communicate their needs and exert political pressure to direct the course
of research. The current challenge of these institutions is to broaden
their focus and revise their techniques to serve the small farmer's
multiple-objectives, to carry out limited-resource trials, and to investi-
gate intercropping systems.
AGRICULTURAL EXTENSION: REACHING THE FARMER
There is little doubt that the extension services in many developing
countries are demoralized internally and discredited externally.3 In
part, this is a legacy of the "extension bias" in development efforts of
the 1950s and 1960s which established and expanded extension services in
spite of the shortage of viable technical packages to extend. However,
extension organizations have had significant internal problems as well.
These limit their ability to disseminate what appropriate recommendations
do exist and reduce their capacity to contribute to the research process.
Seven commonly cited deficiencies of extension organizations will
be discussed below, but it is useful to remember that all of these are
only problems to the extent that they affect 1) the number of farmers
which are reached, 2) the quality of contact with those farmers, and 3) the
distribution of benefits which result from the pattern of contact. Specif-
ically, the deficiencies tend to contribute to a pattern of very few
contacts, relatively ineffective contact, and a significant pattern of
bias toward the larger and more progressive male farmers.
First, extension organizations by their very nature must cover a
large number of often dispersed and isolated farmers with relatively few
agents. In spite of the wealth of literature on the use of mass media such
as written materials, audio-visual aids, radio, and television, the vast
majority of extension activity still involves individual and group contact
between the extension agent and farmers. The ratio of farm families to
extension agents is generally between 500 and 1500 to 1, but this varies
greatly between and within countries (ECA/FAO, 1971:66, Hornik, 1982:46-49).
3See World Bank (1981:1) for the case of India, Rice (1974) and
Benalcazar (1978) for Latin America, De Vries (1976) for Tanzania,
Chambers (1975), and Lowdermilk (1981:25).
Since extension agents cannot reach every farmer, some kind of selec-
tion is necessary. Some extension services concentrate on a single commodity.
These are often employed by a commodity-based research institute (e.g. the
Kenya Tea Development Authority and the Colombian National Federation of
Coffee Growers). Others are organized under regional development agencies
(as in. Senegal and the CADU project in Ethiopia). A third approach is
to direct attention toward particular types of farmers. The most common
among these strategies focuses attention on the most progressive farmers
(e.g. the Focus and Concentrate Program in Ghana evaluated in Atsu, 1974).
Explicit use of the progressive farmer strategy has been criticized,
however, for being inequitable (Atsu, 1974; Ascroft et al, 1972; Kazee,
1981:174; and Mbithi, 1973:89). Anthony et al, (1979:239) notes that
this strategy is based on the assumption that these progressive farmers
will act as a positive reference group among their neighbors,
i.e. neighbors will copy their superior farm practices. The
weakness of this argument is that it ignores the privileged
position of the progressive farmer, his access to institu-
tional and technical resources which are not available to his
In fact, the demonstration effects have been limited.
Even in the absence of an explicit progressive farmer strategy, exten-
sion agents tend to work with larger or better endowed farmers for several
reasons. First, these farmers are more receptive to new technology and
are more likely to initiate agent contact given their successful applica-
tion of 'past recommendations'. Second, they can provide more hospitality,
and possibly economic and political favors to the agent. Third, the
urban-educated agent may feel more at ease socially with the wealthier
farmers. Fi-nally, extension system targets for fertilizer sales or improved
variety adoption may be achieved more quickly by serving a few large farmers
rather than many small farmers. This tendency both limits the total impact
of the extension effort and contributes to an inequitable distribution
of research benefits. It also underscores the importance of carefully
aligning the incentive structure for extension agents with specific program
A third problem, also affecting the distribution of benefits resulting
from investments in extension is the fact that extension agents are pre-
dominantely male. Yet in much of Africa and other developing region, women
perform at least half of the agricultural tasks (Boserup, 1970). Given
cultural obstacles to communication between the sexes, a serious bias is
introduced, a bias which is frequently cited but only rarely effectively
challenged. As increasing priority is placed on food production, this bias
is particularly important since cultural roles regarding the sexual division
of labor often give the responsibility of food production to women.
Fourth, extension organizations are notoriously underfunded relative
to the number of their personnel. This is a common bureaucratic problem,
but in the case of extension it is exacerbated by the geographic distance
(hence political invisibility) of extension activities from urban decision-
makers, as well as by the weak effective demand for extension services.
Salaries and the consequent status of agents tend to be extremely low. It
might be noted that simply raising staff salaries, however, may draw more
educated urbanities into the service and enlarge the social gap between
farmers and agents (Stavis, 1979:37).
One of the most serious shortages, mentioned frequently in the litera-
ture, is that of vehicles and fuel necessary for agents to visit farmers.
An extreme example is Ethiopia where, in 1968, there were 27,000 farm
families for each agent, 60% of which lived at least half a days walk from
the nearest all-weather road, and yet the extension service had only fifteen
vehicles at its disposal (ECA/FAO, 1971:33, 66; Stommes and Sisaye, 1979:229)
This and other material shortages limit the reach of agents both directly
and indirectly through their contribution to the demoralization of extension
Fifth, institutional characteristics of extension organizations are
often cited to explain poor performance. Benor and Harrison (1977:6) believe
that the lack of a single, direct line of technical support and administra-
tive control is "the most fundamental problem in most extension services"
and criticize the lack of specific, realistic, and measurable goals for
agents. Similar deficiencies are mentioned by Rice (1974) in Latin America
and Leonard (1977:216-217) in Kenya including frequent transfer, authori-
tarian management style, insufficient support and monitoring of subordinates,
and the lack of clear, realistic work targets. In India, internal communi-
cations within the extension organizations are
mainly used for discussing the quality of the work of sub-
ordinate staff, and even here, praise for good performance
is out of character. Inadequacies are emphasized while
successes are passed over...impersonality and distance in
communication relations indicate status differences...(Rao,
Leonard notes that in both the Kenyan and Nigerian extension systems a gap
in status and salary exists in the hierarchies which is a vestige of the
colonial separation of European and African positions. This gap consti-
tutes a barrier to communication and promotion which leads to demoraliza-
tion of junior extension personnel and, at times, their group resistance.
Such institutional problems cannot but affect both the number of farmers
reached and the quality of contact.
Sixth, extension agents are often expected to carry out a variety of
tasks only peripherally related to their teaching mission. They frequently
become the "footmen" for whatever agricultural project or campaign is
currently being promoted. Within bounds, this expansion of responsibilities,
beyond those originally called for in the U.S. model, represents a necessary
adaptation to the developing country environment. The strictly applied U.S.
extension model was originally criticized because of its dependence on the
fact that 1) adoption of new technology often depends heavily on the
availability and timeliness of inputs and credit and on the existence of a
market for the output, and 2) frequently these markets are weak or non-
existent. Lowdermilk (1981:15) explains
One of the main reasons the large investments in extension
in the 1950s and 1960s were less than successful was the
false assumption that markets, roads, credit, facilities,
agro-business, stability in prices, and other services were
available. Without these support activities, extension
could. do little, whatever the model.
Thus, agents in many developing countries were called upon to supply, or
at least facilitate the purchase of recommended inputs and credit.
These responsibilities, however, were not well distributed. In addi-
tion, the extension agent was further burdened with administrative and
data collection responsibilities. Chambers, in describing extension in
East Africa and South Asia, notes that much of the agent's efforts go
toward regulatory functions, "guessing and reporting data," input supply,
and "special projects and programmes which may or may not involve farmer
visits." The regulatory functions of extension have declined in Africa
since independence, though the involvement of agents in input supply and
market-related activities persists. A study of the Chilean system in the
1960s revealed widespread use of "production contracts" in which management
and supplies and provided in return for adoption of the complete package,
the latter being enforced by extension workers (Brown, 1970). Similarly,
when extension is responsible for the provision of credit, the agent often
serves as debt collector, a role which may impede his ability to disseminate
technical information. The massive amounts of paperwork sometimes required
of agents is reviewed by Whyte (1981:5-6). Mbithi (1973:81) criticizes
the sporadic demands of ad hoc campaigns to disseminate specific practices.
Each requires full extension.support despite their overall lack of consis-
tency and continuity. In India,
the programmes in reality, become for the field workers another
set of targets where failures could be disastrous to their
careers. Transference of technology is only incidental to this
pattern, and more often than not, a casualty (Rao, 1972:
The final problem discussed here is that of the qualifications and
abilities of the extension agent. Although this factor has been de-
emphasized in recent literature in favor of institutional explanations of
poor extension performance, there is little doubt that it is important.
Raising the educational requirements is often impractical in the short
run given the salary levels and the limited availability of more educated
candidates. It may even be undesirable in light of the finding of Leonard
(1977:126-127) that Kenyan extension agents educated at the secondary
level were less effective and less satisfied with their position than those
less educated. In-service training is "universally endorsed, but it seems
to be minimally implemented" with very few effective programs in the
developing countries (Hornik, 1982:55). More often, technical support is
provided by subject-matter specialists within the extension agency who
answer questions on an ad hoc basis. Many of these specialists and many
extension workers themselves lack practical farming experience (Whyte,
1981:5; Leonard, 1973).
In summary, seven deficiencies of extension programs in developing
countries were discussed in terms of their impact on 1) the number of
farmers reached, 2) the quality of the contact with these farmers, and
3) the distribution of benefits resulting from the pattern of contact.
These include a limited capability to serve a large and broadly dispersed
population, a bias toward larger farmers, a neglect of women farmers, a
lack of operating funds and materials (in particular, transportation
facilities), management characteristics, the proliferation of non-technical
responsibilities, and the lack of training and experience of extension
Thus far, a list of common deficiencies in agricultural research and
extension has been provided, with little explicit discussion of research-
extension linkages. First, this follows from the belief that the research-
extension linkage is not an end in itself, but rather a means to alleviate
the above deficiencies. That is, better communication between researchers
and extension personnel is only useful to the extent that it improves the
relevance of research results or the diffusion of those results to farmers.
Second, some of the strategies for strengthening these linkages involve
the improvement of broader aspects of the process of technology generation
and diffusion. Thus, it is necessary to describe the components of these
two fundamental challenges, making research results appropriate and
disseminating them to farmers, before the linkages between research and
extension themselves can be discussed.
Having delineated the problems specific to research and to extension,
we will now consider research-extension linkages and various ways that the
quality of this linkage affects the severity of the above problems. First,
several conceptual models of research-extension linkages will be considered.
Next, there will be a review of a number of studies that have looked
specifically at communication between research and extension. And third,
some approaches to strengthening these linkages will be discussed.
Nagel (1979) proposes a conceptual framework encompassing a "research
subsystem", a "dissemination subsystem", and a "user subsystem." He
Historically, the first linkage, however weak, was that
between researcher and user. Out of necessity, the
dissemination subsystem has evolved as an intermediary.
Ideally, the flow of knowledge between the subsystems
includes forward linkages between all three...as well as
feedback. (p. 137)
He goes on to describe the appropriate goals and institutional framework
for each subsystem. This approach takes as given the research-extension-
farmer trichotomy. It focuses on information dissemination, appearing to
exclude the flow of materials, such as improved seeds, from the model.
McDermott (1976) makes the important distinction between scientific
research which abstracts from empirical knowledge and technology develop-
ment which synthesizes knowledge to produce a useful product. He divides
the process of technology generation and diffusion into six stages:
scientific research, technology development, testing, and adaptation,
technology integration (i.e. incorporation into the farming system), and
technology diffusion. He argues that dividing this single process into
"research" and "extension" creates a conceptual and practical gap between
technology testing and technology diffusion where "we simply cannot
distinguish research from extension" (McDermott, 1981:2). This conceptual
gap is said to be reflected in the weakness of these intermediate activities.
Similarly, Lionberger (1974) accepts that organizational specialization
between research and extension is required, but argues that the "inclina-
tions to professionalization of sub-specialties" can lead to excessive
differentiation. He adds "this is exemplified by researchers who write
reports primarily for their own kind and by extension workers who are
mostly concerned with following their own subsystem's instructions"
Even without a trend toward over-specialization, the administrative
separation of research and extension could create some misunderstanding
regarding the allocation of responsibilities for those tasks which are
neither pure research nor pure extension. Examples of these intermediary
activities might include on-farm testing and local testing of varieties,
the adaptation of recommendations to location-specific factors, and various
aspects of commercialization of inputs or seed necessary to facilitate
implementation of the recommended technology. Especially where resources
available to the two organizations are limited, it is natural that both
extension and research avoid these intermediary tasks and concentrate
resources on activities which are strictly within their "jurisdiction."
This reduces the relevance of the technology generated and reduces the
opportunities for cooperation and interaction between research and
extension which, as will be seen later, is a potentially important channel
Another important aspect of weak research-extension linkages is the
absence of "upward" communication from farmers to researchers. As discussed
earlier, there is growing acceptance of the rationality of the farmer.
This is accompanied by an increasing awareness of the complexities intro-
duced by the multiple objectives of farmers and the variety of agro-climatic
and resource constraints farmers face. Upward communication, whether
through formal data collection or informal contact, is important in screen-
ing technology according to the criteria, resources, and physical constraints
of farmers. In addition, it facilitates identification of the patterns of
and reasons for non-adoption. Upward communication does not have to
involve the extension agent: it may be accomplished by field work and
data collection by researchers. But historically, the extension organiza-
tion has been expected to serve as an intermediary between farmers and
Rolings (1975; after Havelock, 1975), reflecting a common theme
in the literature, distinguishes between standard "one-way" agricultural
development model and the "linkage model." In the former, technical
information is passed down from research to extension to the progressive
farmer who, in turn, disseminates the technology by demonstrating its value.
In the latter, technology is transmitted from research to extension to
farmer, but at the same time the needs and opportunities perceived by
farmers are transmitted "upward" and incorporated into the research strategy.
A crucial requirement of the linkage model is that the researchers and the
extension workers be "utilizer-oriented;" that they attempt to provide a
service (technological improvement as defined by the farmer) rather than
a particular product (say, an improved seed). Rolings cites three alternate
orientations: 1) agency and/or government orientation, in which national or
bureaucratic goals are promoted, 2) channel orientation, in which the exten-
sion media is perfected to the neglect of the message, and 3) message
orientation, in which there is "hammering on a vision or idea, regardless
of the relevance of the message" (p. 117).
An alternative, perhaps complementary, perspective on research-extension
linkages is the concept of vertical coordination which might be borrowed
from the industrial organization literature. Vertical coordination refers
to the harmonization of product and information flow between two industries
which are vertically related stages in the production of a commodity.
That is, what are the mechanisms for establishing terms of purchase between
say, orange growers and shippers, or between processors and wholesalers.
In the study of U.S. agricultural commodities, the interest in vertical
coordination has led to the development of the subsector approach in which
the entire vertical chain of related industries is taken as the unit of
analysis (French 1974). Important issues include: how do product charac-
teristics and the patterns of supply and demand influence the terms of
vertical coordination, what are the incentives for vertical integration
of various functions into one firm, how do institutional mechanisms reduce
transactions costs, and what structures allow information regarding changing
demand to filter back to the producer.
If agricultural technology is considered the "product," extension the
"seller," and research the "producer," all of the issues above are relevant
to the analysis of the research-extension system. Clearly, there are some
important differences, primarily that the "producer" and "seller" have
non-commercial motivations and that the exchange of the product is not
regulated through the price mechanism. Nevertheless, there are many
parallels between the operation of a research-extension system and the
marketing of a good among consumers exhibiting a complex and varied set
At the least, such a model suggests the importance of upward flow
of information regarding "demand." Furthermore, it may be helpful to look
at the structure of subsectors which have evolved under the discipline of
the market in order to understand and improve public organizations facing
similar tasks. Finally, it suggests that the responsiveness of the
research-extension system may be improved by institutional forces which
"simulate" market discipline. For example, farmer advisory boards and
local funding are said to be an important factor in maintaining extension
accountability in the U.S. system.
In summary, current conceptual models of research and extension tend to
emphasize certain characteristics. First, research and extension are
essentially a single process which is conceptually divided due to bureaucra-
tic requirements of specialization. Second, this conceptual and administra-
tive separation tends to result in a neglect of intermediary activities
which are not clearly the jurisdiction of either institution. It also
leads to weakened communication between the two functions. Third, communica-
tion between research and extension is important not just in transmitting
technical information down to the farmer but also in passing information
concerning farmer criteria, resources, and constraints back up to the
Earlier, the literature on problems particular to research or extension
was briefly reviewed. A smaller number of studies have closely examined
the actual linkages between agricultural research and extension using
surveys of farmers, extension personnel, and researchers. Although
differences in perspective and emphasis make generalization difficult,
the results suggest some common problems.
Lakoh and Akinbode (1981) surveyed farmers, extension agents, and
researchers in Sierra Leone and found that the extension agents were a
relatively minor source of ideas for researchers and that researchers
seemed more concerned with scientific recognition than in working with the
agents on practical problems. Actual communication between researchers
and extension personnel was most frequently through research reports and
pamphlets, while face to face contact generally occurred only at the request
of the researcher. Researchers confirmed, at least verbally, the importance
of communication with and feedback from extension personnel, but felt that
this was hindered by the absence of a formal coordinating body and "poor
means of communication" (p. 296).
The author's survey of extension agents further documented the weakness
of this linkage. A majority primarily relied on information from other
agents and from supervisors. Over half did not use information from the
research stations in spite of positive attitudes toward this source. They
attributed this to transportation problems, the hoarding of publications,
and the absence of a forum for contact.
The situation is more complex in India with its elaborate network of
regional research stations, small interdisciplinary institutes run by the
Indian Council of Agricultural Research, agricultural universities, and
extension services which operate with considerable variation among states.
Rao (1972) found vertical communication limited, especially upward communi-
cation of farmer needs.
The system should pass information on farm needs vertically
upward, to both the higher layers in extension and to the
research community. In reality, however, the bureaucratic
value system has made it an outward-directed medium rather
than a source of stimulus for new research. (p. A-162)
Even this outward (i.e. "downward") communication is weak. Personal contact
is obstructed by status differences, the lack of a unit responsible for
screening and distributing research results to the extension service, and
research publications which do not consider the practical problems of
converting research results into a suitable package for extension. Unlike
Sierra Leone where linkages were seen as potentially constructive, Rao
reports that researchers in India were aware of the gap but remained
unconcerned. They felt it was the responsibility of the extension agent
to come to them and that "in any case, rarely do the problems of farmers
need new research. Researchers believed that the solutions are there and
it is up to the farmer and extension groups to bring them up" (p. A-164).
A letter survey of extension agents and specialists in Iran concen-
trated on extension problems, but it is clear that many of them are attrib-
utable to weak research-extension linkages. As much as 64 % of the extension
personnel rated extension services as "fair" or "poor," citing lack of
resources, irrelevance to farmer needs, and "lack of a close and constant
liaison with the agricultural colleges where research is carried out"
(Karami and McCormick, 1982).
A similar survey of extension agents in Nigeria revealed that the
most frequent sources of technical information were the research stations,
extension specialists, and training course notes. The importance of
training course notes as a source of technical information indicates that
these courses are useful and could be employed to increase the technical
competence of the agents. The extension workers also reported that it took
over eight months for details of announced programs to reach the local level
and that it generally took years to obtain a government loan. Furthermore,
the "bottlenecks in getting research information to the farmers without
undue delay are thought to be in large part within the extension administra-
tion" (Iwueke and Findlay, 1980:68). The agents expressed greater satis-
faction with the administration of the research stations.
Neither of the latter two studies mention the value of institutionalized
contact between research and extension personnel. Nor was there discussion
of upward flow of information concerning farmer needs. In general, these
case studies confirm the patterns outlined in the previous section. They
highlight the extent to which status differences between research and
extension personnel, and even within the extension hierarchy, contribute
to poor communications, exacerbating the effect of administrative separation
In addition, the case studies reveal that although extension agents often
perceive the need for greater contact with researchers, the latter may feel
such communication would be unconstructive. And lastly, the case studies
serve to remind us that while the more abstract discussion of linkages may
emphasize two-way communication, the unfortunate fact is that even downward
flow of information may be quite limited.
Approaches to Solutions
In many cases, weak research-extension linkages can be attributed to
deficiencies within each organization. Lack of confidence in the capabili-
ties of either organization can lead to a vicious circle of mutual distrust
and reduced contact. If research does not incorporate the multiple objec-
tives of farmers into its trials, test technology under farm conditions,
and consider the labor and resource constraints faced by the small farmer,
the extension agent may be justified in feeling that contacting research
personnel is unproductive. Similarly, the organizational problems and
demoralization of the extension service discussed earlier would naturally
discourage the researcher from communicating with extension personnel.
Thus, internal improvement of both the extension and research organizations
will, in many cases, weaken this vicious circle and contribute to more
On the other hand, one can argue that linkages between research and
extension may foster some of the reforms thought to be necessary in each.
Greater contact with the research community can make the extension service
more effective by developing the technical and problem-solving skills of
the extension workers, assuming a "valid message" exists to be transmitted
to the extension service. Similarly, having to confront the actual farm
problems presented by extension personnel can introduce needed focus and
discipline to make applied research more relevant to real farm problems.
Five types of approaches to strengthening the linkage between research
and extension will be considered: integration of the two functions into
one organization, the establishment of a liaison unit, the promotion of
cooperative activities for research and extension personnel, the use of
frequent in-service training in the Training and Visit Extension System,
and Farming Systems Research which attempts to focus research work on
farm-level problems. These are not mutually exclusive approaches. In
particular, on-farm research is discussed as an example of a possible
cooperative activity and as a component of farming systems research. But
they illustrate the range of alternative strategies.
The most obvious institutional arrangement to promote close communica-
tion and cooperation between research and extension functions is to locate
them within the same organization. Indeed, semi-autonomous commodity-based
research institutes often have their own extension service. An exceptional
example is the Kenya Tea Development Authority (KTDA) which provides a
variety of services to tea growers, particularly smallholders. It engages
in technological research, extension, road building, the provision of credit
and inputs, and the marketing and processing of the tea. This requires
considerable coordination since tea production demands strict quality con-
trol and rapid, timely processing. The services provided are tied to
adoption of recommended practices using legally enforceable contracts.
There is, however, a system of committees composed primarily of representatives
chosen by growers which channel feedback from farmers upward, as well as
distributing policy decisions downward (Lele, 1975:73). Extension agents
serve a specific set of farmers, all of whom are registered with the KTDA.
Explicit job targets are set for these agents, and the records of farm
visits are collected and tabulated twice a year. The result is that the
KTDA has been very successful in expanding tea production and export,
in serving increasing numbers of smallholders, and in increasing the
incomes of those farmers.
The advantages as well as the disadvantages of integrating research and
extension functions in a single administrative unit are shown by the case
of the KTDA. On the positive side, a single institution theoretically has
broader goals of promoting production and serving the target population
rather than simply generating new varieties or disseminating a particular
message. By assuming responsibility for the supply of inputs and the
marketing of the commodity, economic incentive is created to promote pro-
duction and reduce marketing costs by whatever means possible, whether by
adjusting recommended input levels, changing cultivation practices, or
building roads. In addition, by specializing in one commodity, extension
agents are more likely to be technically competent and respected by farmers.
This pattern of greater farmer confidence in specialized extension agents
is confirmed by the experience in Nigeria (Ekpere, 1973), Kenya (Leonard,
1973:141), Chile (Brown, 1970:207), and Colombia (Adelman, 1981:460).
There are several limitations to this strategy, however. The success-
ful examples of this type of institutional arrangement are almost exclusively
ones which specialize in a major export crop. This simplifies the message,
facilitates self-financing through marketing or export taxes, and assures a
return of scarce foreign currency. The intensity of effort usually devoted
to these high-return crops would be difficult to replicate on a national
scale. The KTDA spent $18 per farmer for extension activities in the late
1960s, a sum which if spent on a nation-wide basis would have exceeded the
entire government expenditure on agriculture (Lele, 1975:69). In addition,
expanding such an institutional structure to cover all important crops
runs the risk of creating an organization with a mandate so broad as to make
administration and technical coordination impossible. Ministerial reorgani-
zation is an apparently attractive idea and easy to recommend, as Chambers
(1974:2) notes, but the difficulties of implementation are often under-
Some observers appear to accept the pattern of separate research and
extension institutions as a necessary evil. Others, such as McDermott
(1976:5), argue convincingly that the types of linkages between the two
functions is more important than the formal integration of the two organi-
It makes little difference whether all of the functions are
encompassed within a single administrative structure...
However, it makes a very great difference whether "in some
ways and on some occasions" they "may be viewed as a single
Establishment of a Liaison Unit
A less drastic organizational reform is the formation of a unit to
act as a liaison between the extension service and research institutions.
In Nigeria, the Agricultural Extension Research Liaison Service (AERLS)
was established to strengthen the linkage between the Institute for
Agricultural Research (IAR), affiliated with Ahmadu Bello University, and
ten of the state extension services (in Nigeria, extension is the responsi-
bility of the states). AERLS is composed of subject-matter specialists who
participate in the research committees of the IAR, presenting problems from
the field and evaluating research recommendations. The AERS conducts in-
service training for field extension staff, produces extension materials
and radio programs, and adapts research recommendations to local conditions
(Okigbo et al, 1981a). To give an idea of its relative size, AERLS is
allowed a recurrent budget one sixth that of the IAR and employs about one
fifteenth as many people. A national review and evaluation of the agricul-
tural research system noted the strengths of this institution: its autonomy
allows an independent position on matters related to research recommenda-
tions and extension procedures, it is closely involved with research
activity, and it has a well-defined role in support of the state extension
services. The evaluation recommended that other AERLSs be established
for each of the six multi-crop research centers and each unit be limited to
covering three or four states (Okigbo, 1981b:32-33).
In Kenya, extension-research communication occurs through joint meet-
ings between provincial extension officers and research personnel. An
evaluation by the International Service for National Agricultural Research
(ISNAR), however, recommended the creation of a Department of Agricultural
Extension Research Liaison and Communication within the Ministry of
Agriculture. This would be responsible for facilitating communication
(a) within the research system, (b) between the research system,
the extension services, and other user organizations, and (c)
for feedback of problems from users to guide research (ISNAR,
More specifically, this unit would work with the agricultural research
ensure constant interaction with the extension services and the
agricultural industry. Important constraints in the agricultural
systems would be identified, areas of researchable problems
would be defined and brought to the attention of research
scientists, and the impact of improved technologies would
be assessed through farmer situations. The department would
also provide a point of contact of research with the extension
service and the users of research results, [organizing] train-
ing workshops and seminars to expose different levels of
extension officers to new techniques developed in research
Similar suggestions were made in ISNAR evaluations of agricultural research
in Indonesia (ISNAR, 1981b:16) and Rwanda (ISNAR, 1982:62, 69) though not
described in detail.
A still more limited form of institutional linkages is the use of one
or more Extension Liaison Officers attached to each research institution.
Such an arrangement is in use in Brazil, though a recent World Bank report
has questioned the effectiveness of this tie (Lopes, 1980:152). Writing
about Tanzania, Albrecht (1977:20) notes that an Extension Liaison Officer
was not successful in drawing attendance at joint meetings of research and
extension personnel or in fostering cooperation. This illustrates the
limitations of organizational reform without accompanying changes in
attitudes. In addition, any liaison officer will owe first loyalty to
his or her own agency, and it is difficult to see such an arrangement
leading to effective two-way communication. Perhaps this has been
recognized in Tanzania for Liwenga (1980:61) reports three years later in
a research/extension liaison unit is to be established in
each research institute. The proposed organization will have
a separate budget with a more reliable funding and an improved
scheme of service that will attract, retain, and motivate
In summary, we may state three conclusions. First, complete organiza-
tional integration of research and extension agencies can be very effective
when focused on a commodity or region, but is probably impractical for
the whole range of crops on a nation-wide basis. Second, extension liaison
officers attached to research institutions may not be independent enough
to provide feedback from extension staff regarding problems with research
recommendations. Third, an independent liaison unit may be useful in
coordinating-the two-way flow of information, but its success depends on
the quality of communication flows within extension and attitudes of
research personnel. Although these arrangements have not yet proven them-
selves, they hold some promise and merit further experimentation.
It has been suggested that joint activities involving research and
extension personnel provide an opportunity for informal contact and infor-
mation exchange. Diagnostic data collection and on-farm research trials
constitute two areas for possible collaboration.
Diagnostic data collection, like on-.farm research, can facilitate the
upward flow of information which stimulates greater research responsiveness
to farm-level problems. In 1966, Mellor wrote:
Systematic survey of farmers is an alternative device for
discovering farmers' problems and thereby directing the
research process. Since in many low-income countries, it
is in practice difficult to establish close rapport between
farmers and biological science research personnel because
of gaps in background, education, and culture, it falls
more heavily on systematic survey to point out the problems.
Such a survey cannot be a perfect substitute for informal
contact between researcher and farmer, but it can at least
mitigate the problem. (Mellor, 1966:283)
Since that time, recognition of the complexities of farming systems and
farm-level decision-making have underscored the importance of this view.
However, the record of using extension agents in data collection has
been mixed. As mentioned earlier, extension agents have often been burdened
with the tasks of collecting general statistics, monitoring various on-going
projects, and providing numerous progress reports. Under pressure to make
estimates of figures he cannot possibly know and faced with a heavy work-
load, inaccurate and possibly falsified data is almost inevitable, especially
when the data measures the progress of a project for which the agent is
responsible (Chambers, 1975 and Benor and Harrison, 1977).
Partly in response to these problems, recent efforts have been made to
develop streamlined methodologies that collect only the minimum data
necessary to direct research efforts and plan projects. Under the rubric
of "rapid rural appraisal," a number of methods have been tested including
informal interviews with farmers by researchers (Hildebrand, 1981), surveys
of extension field staff (Collinson, 1982), and various combinations of
exploratory and more formal survey techniques (Gilbert et al, 1980:45-50).
Rapid rural appraisal methodology is associated with farming systems
research (FSR), thus it tends to be "researcher-intensive." There are two
reasons for this: 1) the FSR approach promotes direct contact between
researchers and farmers as an important learning experience for the former,
and 2) some of the methods, such as informal surveys, involve qualitative
skills which may be difficult to teach to enumerators or extension agents.
However, as these methods are refined and institutionalized, the incorpora-
tion of extension personnel for certain tasks could take advantage of their
local knowledge of farming practices and constraints, reduce researcher-
intensivity, and facilitate communication between research and extension
personnel. These responsibilities must replace, rather than add to,
existing data collection tasks and, in any case, must not consume a large
portion of the agent's time over the year.
On-farm research is another potential area of joint activity between
research and extension. This can contribute to the relevance of agricul-
tural research and provides an opportunity for more consistent research-
extension collaboration. Given the range of current interest in these
activities, this strategy will be discussed in greater length.
Shaner et al, (1982:111) distinguished between three types of on-farm
trials. 1) Researcher-managed trials take place on rented or borrowed
portions of the farmer's parcel. These are useful for untested or other-
wise risky technologies and for trials with complex experimental designs.
Although agro-climatic conditions may vary across trials, the control of
management variables (levels of input and practices) allows good initial
assessment of many of the factors of greatest concern to the farm clientele.
2) Superimposed trials generally involve single factor treatments or
portions of the farm field otherwise managed by the farmer. 3) Farmer:
managed trials are, as their name suggests, fully managed by the farmer.
These are used for a final assessment of the farmer's response to the
technology. Adoption rates and possible adpatations in the mode of tech-
nology application are examined. These tests also serve as demonstration
trials. Generally, only the most successful technologies identified
by researcher-managed trials are selected for testing in farmer-managed
An important contribution of on-farm trials to the improvement of
research-extension linkages is that they help ensure the relevance of
technological recommendations. The availability of a valid message is a
precondition for extension personnel to have any interest in communicating
Reviewing the conditions necessary for relevant research discussed
earlier, the strengths of on-farm testing are obvious. First, given the
complexity of the criteria used by farmers to evaluate alternative
cropping systems, it is clear that one solution is simply to let farmers
evaluate the technology themselves, as in farmer-managed trials. ICTA
uses an Acceptability Index: the product of the proportion of collaborators
in farmer-managed trials who actually use the technology the following
season multiplied by the proportion of their land on which the technology
is used (USAID, 1982:9). Alternatively, a widely distributed CIMMYT manual
on on-farm research analysis recommends the use of incremental net benefits,
marginal rate of return, and two simplified measures of risk, criteria
which would be helpful in the pre-testing of a technology in research-
managed trials. CIMMYT emphasizes that the costs and returns measured must
be those actually faced by participating farmers (Perrin et al., 1976:41).
Second, the variety of agro-climatic conditions under which a recommended
technology might be applied is more thoroughly sampled by scattered on-farm
trials than by station-based experiments. This is particularly true if
rocky, sloped, or poorly drained plots typical of the region are included
in the on-farm trials since these conditions were generally not found on
Third, the non-treatment experimental variables will be more likely
to correspond with the environmental constraints and management practices
of the representative farmer. Specialized management practices such as
extra weedings or additional applications of water, so often a part of
station-based work, may simply be impossible.
Technical scientists frequently object to the sloppiness of on-farm
testing and the difficulty of drawing valid conclusions with so many
uncontrolled variables. Part of this derives from legitimate scientific
concern over the statistical credibility of the results. Evidence suggests,
however, that valuable information can be derived from less than statisti-
cally perfect trials. Leonard (1973:144) expresses this well.
Scientists are unlikely to make educated guesses, however, unless
field officers put considerable pressure on them or start making
them themselves. A good guess about a way to increase the pro-
fitability of the farmer's operation this season is more valuable
to him than a carefully proven recommendation five years later.
At issue is the fact that excessive concern for internal validity, available
only from highly controlled experiments such as those done on-station, may
sacrifice external validity, which can be achieved only be testing and
adapting technology to the "sloppy" conditions of the real world.
On-farm research can additionally promote research extension linkages
if extension personnel can be incorporated into the on-farm research process.
In addition to improving the cost-effectiveness of on-farm research (a
point which will be discussed later), on-farm research provides an oppor-
tunity for interaction and communication between research and extension
personnel. Ideally, the scientists would gain an appreciation for the
complexity of farming systems and the difficulty of making constructive
recommendations to a diverse clientele. The extension agent could gain
a fuller understanding of the strengths and limitations of the research
process. In all probability, such collaboration would provide extension
agents with the training and practical experience they are often said to
need. Exposure to scientific method might heighten their ability to draw
conclusions from what they observe and hear in the course of their inter-
action with farmers.
Furthermore, to the extent that the extension agent understands how
recommendations are generated, he may find them more believable. The
recognition that such information is useful can stimulate greater interest
in its communication. The agent may also be more able to identify those
situations under which the recommendation is or is not appropriate. The
experience in Sri Lanka is noteworthy.
It has been a matter of debate as to why extension personnel
should be involved in carrying out trials. The Extension
Worker in contact with the farmer should himself be convinced
of the performance of and must accept a new variety and its
associated management practices...before he can convince
farmers to do so. Our experience has been that getting the
instructor to lay out these trials not only keeps him in
contact with the problems of cultivation but also involves him
directly in the program and sets up a dialog between him-and
the research staff on the one hand and with the farmer on the
other. (Abeyratne, 1973, cited in Chambers, 1975:6)
Thus, diagnostic surveys and on-farm testing are two activities which
contribute to researcher knowledge of farm-level constraints and provide
opportunities for collaboration between research and extension staff.
Improved communication between the two bodies is accordingly facilitated.
Two final points should be made. First, a major obstacle to such
joint activities is the administrative separation of research and extension.
Senior administrators in both organizations are loathe to lend manpower
to an activity they do not control. Some observers (Whyte, 1981) have
argued that eventually only a unified administration of research and
extension activities will resolve these difficulties. Others (McDermott,
1976) maintain that it is enough that they act as if they were part of a
single system. In this case coordinating institutional mechanisms, such
as those discussed in the previous section, may suffice.
Second, the proportion of extension agent time allocated to data
collection, on-farm trials, and advisory services depends in part on the
level of quantitative and qualitative knowledge of farm-level conditions
held by researchers and on the existence of valid recommendations to
extend. In the absence of these, the collection of relevant production-
related data is an important priority and the opportunity costs of extension
participation are relatively low. Where valid recommendations do exist,
more emphasis on the provision of technical assistance is clearly justifi-
able. On-farm trials lie somewhere in between, being an important tool
for agent training and technology evaluation over almost the entire
Training and Visit System
The Training and Visit (T & V) Extension System, developed by Benor
and Harrison (1977) and the World Bank, is an attempt to deal with many
of the above discussed problems of extension organization and research-
extension linkages. Its main features are a rigid farm visit schedule,
regular in-service training, and a focus on agricultural advisory services.
The World Bank has established T & V Extension Systems in India, Turkey,
Burma, Nepal, Sri Lanka, and Thailand, and more recently in several African
nations. Some are national efforts whereas others are organized on a
project basis in selected areas. The characteristics of the T & V System
can be illustrated by considering its response to the seven problems
common to extension programs listed above.
First, the T & V System attempts to maintain relatively intensive
coverage of farm families without concentrating on one crop. Farmer-agent
ratios of 500:1 to 800:1 are recommended depending on the population density
and complexity of cropping patterns. Benor and Harrison (1977:35) insist
that administrators must resist the temptation to spread the agents more
thinly. They argue that the area of coverage should be expanded only after
the pilot area achieves some success and after sufficient resources become
Second, the progressive farmer bias is rejected, at least in the
official formulation of the Systems. Contact farmers, who are to teach the
techniques to several neighbors, are supposed to be imitable.
They should not be regarded the community's most progressive
farmers who are usually regarded as exceptional and their
neighbors tend not to follow them. (Benor and Harrison, 1977:13)
No mention is made of the third pattern, that of the neglect of female
farmers by predominantly male extension workers. This is a serious
commission in view of the number of female farmers in many developing areas.
The shortages of resources in extension agencies is not something
which can be solved by organizational reform, but Benor and Harrison
(1977:36) do insist that every village extension worker have at least a
bicycle for which salary advances are to be provided if necessary. Of
course, the T & V Systems established thus far have enjoyed large-scale
World Bank financing. Benor and Harrison argue that as T & V extension
proves itself, greater national funding will be forthcoming to maintain
the system when the World Bank withdraws.
The fifth problem is the institutional and management problems common
to many extension organizations. Benor and Harrison place much emphasis
on the importance of having a "single, direct line of technical support and
administrative control" (p. 6). All extension personnel are merged into a
single extension service with possible exception "where an area's agricul-
ture is dominated by a crop which is the raw material for a vertically
integrated industry (e.g. rubber, tobacco, tea)" (p. 11). The span of
control throughout the hierarchy is kept to 5-8 supervisees per supervisor.
In addition, a fortnightly (or weekly) schedule is established for farm
visits by the extension agent in which the agent visits the same farm(s) on
each day of the cycle.
Sixth, the responsibilities and paperwork of the extension agent are
kept manageable by excluding almost all non-advisory tasks. Coordination
with input suppliers is called for but the extension service is not directly
Seventh, the technical qualifications of the agents are developed
through frequent (weekly) sessions for in-service training and discussion
of problems in the field. One day every two weeks is spent with
subject-matter specialists (SMS) and another with the immediate supervisor.
The T & V System is also designed to make recommendations appropriate
to the resources and constraints of farmers. Initial recommendations
involve changes in management practices. Only later are recommendations
requiring purchased inputs made. To reduce risk to the farmer, recommenda-
tions are to be incremental and tested on a portion of the farmers land.
Above all, "nothing should be recommended which will not increase farmers'
income" (p. 14). In addition, the structure provides for feedback to
researchers through the subject-matter specialists.
In India, the T & V System has been implemented in stages since 1977.
A report by the World Bank (1981), noted that the impact of the T & V
System was difficult to quantify but clearly positive in terms of staff
morale, extension credibility to the farmer, and accelerated adoption.
It is worth reproducing a passage from the report describing how T & V
extension can develop research-extension linkages.
The most encouraging and, in the long run, the most important
result of the introduction of training and visit extension is
the development of a dynamic link between farmers, extension
staff and research workers. Frequent and systematic contact
between research and extension is built into the training and
visit system: direct, frequent and repeated contact between
farmers and VEWs [village extension worker] (supplemented from
time to time by SMS) forces the extension service to focus on
technically and financially feasible recommendations for
improving farm production; the VEWs' regular contact with
research through the SMS gives them the opportunity to raise
immediate production problems; and to respond effectively to
such questions, SMS need to work with researchers to resolve
field problems. Through this sequence, pressure is placed
upon researchers to concentrate on the practical field problems
of the average farmer. There are signs that such a linkage is
emerging. (World Bank, 1981:8).
Again, it should be noted that such a linkage is useful to the extent that
it contributes to more appropriate research recommendations or to their
dissemination. Thus, such institutional development is an investment which
is assumed to produce these concrete results with time.
An econometric study of the Hooghly District in West Bengal compared
farming patterns before and after the establishment of the T & V System.
Cropping patterns tended to move toward the optimum (as defined by a linear
program) and there was a significant increase in the adoption of recommended
practices, but the effect on yield was mixed. In general, the system
effects were more pronounced for farms smaller than 5 acres than those
larger (Ray et al, 1979).
The T & V extension system has not been, however, without problems
and criticism. Administrative norms and procedures have made it difficult
to reverse the centralized, top-down style in India (World Bank, 1981)
and Thailand. In the latter case, it was reported that messages from
subject matter specialists in extension to a research were routed verti-
cally to the Director General of Extension, to the Director General of
Research, and then down to the appropriate researcher (Sarikaphuti, 1981:24).
The consolidation of extension staff into one service may also intro-
duce problems. In India, transferred agents resorted to legal action to
block the reorganization (Howell, 1982:282). In Thailand, extension
personnel were removed from the Rice Department which had previous inte-
grated research and extension functions for this commodity. Thus, accord-
ing to the Director General of the Department of Agricultural Extension,
the research-extension linkage for rice was weakened, at least in the short
run (Sarikaphuti, 1981:24).
Another problem has been the persistent tendency for the contact
farmers to be traditional village leaders and larger, more successful
farmers. This has been the experience in India (World Bank, 1981:13),
Nigeria (Howell, 1982:280), and Brazil (Russell, 1981:32).
Several problems in implementation result from the fact that the T & V
System was developed in South Asia and, in its original form, may not be as
appropriate for Africa and Latin America. First, the intensity of extension
coverage and the use of contact farmers may be more realistic in areas of
high population density common in Asia than in more sparsely settled parts
of Africa. Second, the absence of women extension agents would present a
more serious bias in Africa and Latin America where women play a larger
role in production decisions than in India (Russell, 1981:33). Third, if
the research system cannot provide a stream of appropriate recommendations,
strengthening the extension service and its linkage to research will be
unproductive. In India, with its relatively advanced network of agricultural
research institutes, and agricultural colleges, this did not appear to be a
problem (World Bank, 1981:16). In Thailand, on the other hand, the effec-
tiveness of a T & V project was hampered by the lack of appropriate research
results (Sarikaphuti, 1981). Russell confirms that this is a problem,
especially in Africa, but argues that the discipline of the research
extension-farmer linkage in the T & V System is the most effective way of
developing a responsive research system. Fourth, one of the most frequent
criticisms of the T & V System is that it restricts the activities of the
agent to agricultural technical assistance. This narrow view of extension
excludes a number of activities commonly assigned to extension agents:
establishment of cooperatives and other farmer organizations, the facilita-
tion of input and credit distribution, the expansion of.marketing outlets,
and more general rural development activities. In India, these functions
were absorbed by the private sector or other agencies (World Bank, 1981).
Where input and credit markets are weak or non-existent, it is self-defeating
to concentrate extension activity on technical assistance if the technology
requires some of these inputs.
However well trained the staff may be, and however good their
links with research, without a supply of inputs and a market-
ing infrastructure, the credibility of the extension worker
will soon be lost (Russell, 1981:33).
This point is also made by Rice (1974) in the case of Latin America.
Russell (1981) describes the strategy of several World Bank projects in
which these services are provided through a state organization. As will
be discussed in a later case example, the extension system in Botswana
resembles the T & V System in many respects except this one. Kingshotte
(1980b:321) cites long distances, poor communications, and the cost of
establishing other mobile field networks to carry out these functions.
We also felt that if the Demonstrator were a source of both
advice and guidance, at least on how to obtain practical
help, his advice would be more readily taken... [this system]
is, we believe, that best suited to the needs of the late
1970s and 1980s, just as the T & V System is that best suited
to some countries in Asia (Kingshotte, 1980b:321).
In summary the Training and Visit System contributes to the dissemi-
nation of appropriate research results by concentrating the agents' atten-
tion on agricultural technical assistance and by establishing a rigid
schedule of farm visits and in-service training. It attempts to strengthen
the supervision and support of the field agents and to clarify their goals
and responsibilities. Regular contact between subject matter specialists
and extension agents is supposed to prompt subject matter specialists to
communicate regularly with researchers. The case of Thailand, shows that
the effectiveness of the T & V System may be reduced, however, by 1)
obstacles to communication between subject-matter specialists and researchers,
and 2) the lack of appropriate research results to be disseminated. In
addition, it demonstrates that the goal of establishing a unified extension
service may conflict with the goal of strengthening research-extension
linkages if integrated research and service institutions exist for selected
commodities. Experience in Africa, and to a lesser degree Latin America,
confirms that the T & V System must be adapted to conditions in those areas.
Farming Systems Research
Farming systems research (FSR) represents a step beyond on-farm research,
in the emphasis it places on multidisciplinarity and the design of adaptive
research trials to solve location specific problems.4 This research approach
essentially encompasses the advantages of on-farm research and diagnostic
survey work discussed above. The following discussion more fully describes
several specific positive and negative attributes of FSR efforts, however,
in view of the broadening interest in assessments of the methodology's
FSR is most commonly conducted by a small interdisciplinary team
including an agronomist, an economist, and as needed, the representatives
of other disciplines. Their activities fall into the following stages:
1) the selection of a target area (or recommendation domain) with similar
geo-climatic conditions and farming patterns, 2) the study of local
resources and constraints using secondary sources and one or more diagnos-
tic surveys, 3) the planning and implementation of on-farm trials, 4) the
analysis of those trials and formulation of recommendations, and 5) the
extension of the recommendations to farmers (Shaner, et al, 1982:27-29).
Multidisciplinarity, in particular the use of economists along with
agricultural scientists, permits consideration of a broader range of factors
In fact, the distinction between FSR and on-farm research programs
is not always clear. In particular, a continuum exists between simple
efforts to move station-based research into farmers' fields to attempts
to analyze the farming system involving little or no technology testing.
in farm-level decision-making. In addition, the divergence of geo-climatic
conditions between the research site and the farm is reduced or eliminated
by the FSR's location-specificity. The approach aims to generate inexpen-
sive and relatively simple recommendations which will easily fit into
existing farming systems.
Since FSR is designed to bring the researcher closer to the farmer it
might be questioned whether extension would be necessary in a country with
a full-scale farming systems research program. In short, the answer is that
FSR may improve the relevance of research recommendations, but sheer numbers
make it impractical as a diffusion institution. In fact, FSR programs have
been frequently criticized for requiring the application of an inordinate
amount of research resources to solve limited location specific problems.
Collinson (1982:44-46) has attempted to prove the feasibility of
extensive FSR programs in Africa. Following a set of initial surveys in
Zambia he designed recommendation domains of 10,000 farms. He estimated
With some exceptions, the manpower commitments of two adaptive
agricultural research professionals [a team] per 80,000 farms
seem within the reach of the countries of East, Central, and
Southern Africa (Collinson, 1982:45).
Each team must cover eight domains. If about six on-farm research
sites per domain are needed to provide a "domain-wide" recommendation,
Collinson suggests each team administer 24 sites (four domains) and
alternative groups of domains every three years.
At a minimum, it may be questioned whether 24 on-farm research sites
per team is feasible. Experience in Honduras has indicated that:
Between 10 and 20 individual trials on five or more farms
is about the maximum load that a conscientious FSR team of
two or three researchers could manage with one vehicle
Of course, there may be differences in geography, complexity of trials and
support staff, but the important point is that with a minimal staffing
commitment (one team per 80,000 farms), the time of FSR scientists is
already stretched thin. It is clear that a nation-wide FSR program may
substantially improve the research community's knowledge of farm-level
problems, but it cannot substitute for an extension service in bringing
the results of research to the farmers. The participation of extension
in testing and disseminating technology improves the cost effectiveness
of FSR by making it less researcher-intensive and by reducing the critical
time between the initial survey and the adoption payoff.
FSR can play an obvious role as intermediary between station-based
research and extension. To date, however, the strength of this link has
been limited. Evidence indicates that FSR programs have generally not had
very strong linkages with conventional agricultural research. Norman
(1980:66) notes that research institutions may not accept FSR activities due
to resource limitations, objections to FSR philosophy, the threat to their
professional investment in conventional methodologies, the administrative
complication of multidisciplinary teams, and the time required to develop
an FSR program. In addition, FSR programs are often instituted in
cooperation with one of the IARCs or a donor agency which may give it the
stigma of being an imported methodology. Such problems have been found
in Honduras (Galt, et al, 1982:32-34), Guatemala (Gostyla and White,
1980:17), Senegal (Crawford, personal communication), and East Africa
(Collinson, 1982). Hence, FSR is often conducted on a project basis or
in some other administratively independent unit, in spite of almost
universal endorsement of the idea that FSR should be integrated into the
FSR linkages with the extension service have remained even weaker.
This has been identified as a major constraint in three research systems
in the process of developing nation-wide FSR programs: Senegal, Guatemala,
and Honduras. The Unite Experimental project in Senegal is one of the
earliest FSR efforts in the developing world. Yet, the extension service
was not involved until the research effort is completed.
The diffusion activities include testing the proposed system
under farm conditions, initially under the direction of
researchers and involving a few receptive farmers. Later,
during the demonstration and pre-extension phases of testing,
the trials are managed by many farmers on a large scale.
Finally, techniques or sets of practices that successfully
pass through the above sequence are transferred to farmers
by the extension system (Gilbert, et al, 1980:114-115).
The program has since been superceded by a new donor funded project which
attempting to resolve this problem. The research institute in Guatemala
(ICTA) has been successful in integrating on-farm participatory research
into its research system, but the traditional structure and approach of
the extension service has hindered th6 development of cooperation and
communication between ICTA and DIGESA, the extension agency (Whyte,
In Honduras, greater involvement of extension in a similar on-farm
research strategy has been achieved, but many of the working relationships
established are only informal and depend on the motivation of the regional
research coordinator or the inclination of individual agents. It was
observed that "implementation and acceptance of FSR was invariably faster
when extension was involved from the beginning" (Galt, et al, 1982:35).
An evaluation of an FSR project in northern Honduras came to a conclusion
which might be applicable to a number of FSR projects. A critical
the apparently overly intensive application of scientific
talent to a limited number of research sites. It would
probably have been more cost-effective to have spread the
principal researcher's energies over more trials by
utilizing low-level research assistants, extension
specialists, and extension agents more imaginatively.
Shaner, et al (1982) provides an outline of how FSR and extension
could be integrated. Their proposal identifies various tasks at each
stage of the FSR methodology in which extension workers could participate,
from target area selection to on-farm testing. It is suggested that the
extension agent play a significant role in on-farm testing, problem
identification and development of an information base. The authors also
recommend the use of an "extension specialists in farming systems" (ESFS).
The responsibilities of the ESFS include:
learning FSR & D procedures by working closely with the researchers
familiarizing researchers with extension's capacity, needs, and
educating extension personnel in FSR & D philosophy and methods
training extension personnel in the application of the new tech-
generally coordinating the activities between research and exten-
sion, especially during multi-locational testing and the pilot
production programs (Shaner, et al, 1982:148-150).
The use of such a position is being tried in Zambia where a team
includes a Farming Systems Economist, a Farming Systems Agronomist, and
a Research/Extension Liaison Officer (Collinson, 1982:vi). In Senegal,
an AID project is under way to establish Production Systems Research
(PSR) teams attached to several agricultural research stations. A
Specialist in Extension/Research Liaison, was proposed who
will be helping the extension staff of the [Regional Development
Agency] in disseminating the new technology. He is the only
member of the PSR team who will have a staff function with RDA
management... (USAID, 1981:23).
However, after three years, this position has not been filled.
FSR has been relatively successful at promoting the flow of informa-
tion upward to the researcher. However, it cannot substitute for an
extension service in bringing the results of research to the farmer. With
the opportunities for cooperation in data collection and on-farm testing,
it would seem that FSR could be both an important bridge between conven-
tional research and extension. But in practice, FSR's linkages with each
have been difficult to establish. Collaboration with extension and with
farmer organizations is crucial in extending the scope of FSR research, and
thus, in improving its cost-effectiveness. The use of a research/extension
liaison specialist on the FSR team has been proposed but not yet adequately
In the previous section, a variety of approaches to improving research-
extension linkages were discussed. It is clear that the success of these
approaches frequently depends on the strength of other components of the
research-extension system. For example, the Training and Visit Extension
System, a relatively comprehensive strategy, relies to some degree on the
quality of agricultural research institutions and on the effectiveness of
credit and input markets. Furthermore, it was apparent that the appropri-
ateness of any given approach depends on the national context: existing
agricultural institutions, national policy, infrastructure development, and
operational environment. In view of this, it seems useful to examine
several national systems and the mechanisms that have evolved to link
research and extension. The three examples, Botswana, Guatemala, and
Taiwan, demonstrate ambitious approaches to strengthening research-extension
linkages and the importance of adapting these approaches to local conditions
The case of Botswana represents an attempted African adaptation of the
Training and Visit Extension System. A brief review of farming systems
research in Guatemala highlights the potential value of this alternative
strategy to strengthen research-extension linkages. Lastly, the Taiwan
research-extension system is discussed as an example of long-term evolution
of a series of institutions designed to ensure accountability of the
extension service to farmer needs.
The case of Botswana is interesting because until the mid-1970s, the
extension service demonstrated many of the structures and deficiencies
common to African systems. With the organizational reform of the Ministry
of Agriculture in 1975, a system similar to the Training and Visit
Extension System has emerged, but with some important adaptations to local
conditions. (This description is based largely on a pair of articles by
Kingshotte, 1980a and 1980b).
Before 1975, the Ministry of Agriculture was divided into ten depart-
ments and divisions including one for agricultural extension. Extension
activities were also carried out by the Animal Production, Land Use,
Research, Cooperative Development, and Veterinary Departments and Divisions.
There were complaints of inefficiency due to overlapping jurisdictions,
conflicting recommendations, the over-use of specialists at the field
level, and frequent deferral of responsibilities by agents to the often
unavailable specialists. The Pupil Farmer Scheme (PFS) and the Pupil
Stockman Scheme (PSS) concentrated up to 90% of the attention of the
Demonstrators (field agents) on advising approximately 7% of the farmers.
Membership in the PFS was limited to those farmers with draft oxen, thus
excluding the bulk of poorer farmers. As occurred with the progressive
farmer strategies in other countries, the "demonstration effect" on non-
members was less than expected.
The reorganization of the Ministry of Agriculture, implemented after
considerable consultation at various levels, consolidated the units into
five divisions: Agricultural Research, Veterinary Services, Agricultural
Field Services, Cooperative Development, and Planning and Statistics.
The Department of Agricultural Field Services contains a Field Section
including extension, farmer training, and special projects, as well as
several specialist support units.
The most essential change was that all farmer contact,
extension work, and schemes to help farmers were being
organized by, or through, the officers with territorial
responsibilities--for example the Demonstrator or [his
supervisor]. (Kingshotte, 1980a:198)
This was seen as necessary given the costs of reaching the sparsely settled
and remote population. Also, because of the generally low level of agri-
cultural technology, it was not felt necessary to use subject-matter
specialists at the field level. The Demonstrator, it was argued, being a
"specialist" in the people and resources in his or her area, would best
be able to coordinate the various projects and services.
Considerable attention has been paid to supporting the Demonstrator.
The workloads were reduced and equalized with a farmer-Demonstrator ratio
of 275:1, although this varies with population density and the available
transportation. Job descriptions were made explicit, and the span-of-
control ratio (that is, the ratio between supervisees and supervisors) was
limited to about 5:1 at most levels.
Linkage to the agricultural research system is provided primarily
through a team of subject-matter specialists at the Regional Level (there
are forty Demonstrators per Region) who provide technical support. They
design and test technologies and develop extension programs on the basis of
research results and previous experience in the Region. They help prepare
mass media extension materials and provide agricultural inputs or arrange
for their delivery to the Field Section. In addition, they help prepare
a loose-leaf Demonstrator's Handbook with basic agricultural data, current
recommendations, and resources available for farmers, which is updated
regularly. Lastly, they participate in Monthly Management Meetings at the
District Level (there are ten Demonstrators per District). At these meet-
ings subject-matter specialists train the Demonstrators in agricultural
techniques and management, and both groups discuss monthly plans, current
programs, and problems faced by Demonstrators. For more senior staff,
there are annual two-week in-service training courses (Kingshotte, 1980a:
208 and 1980b:318).
Other reforms include 1) the termination of the Pupil Farmer and Pupil
Stockman Schemes in favor of a more broadly-based extension effort using
farmer committees, 2) the use of "Management by Objectives" with partici-
patory planning at all levels, and 3) the initiation of simple farm records
kept by Demonstrators and used for planning.
This system is similar in many respects to the T & V Extension System:
there is a unified extension agency, frequent in-service training, tighter
supervision, more specific and realistic job responsibilities, and the use
of subject-matter specialists to support the field agents. The system in
Botswana is different, primarily, in that the agents' responsibilities are
broader than just the provision of agricultural technical information.
Long distances and weak input markets mean that agents are necessarily
involved in the arrangement of input supplies and credit, communal fencing
projects, dam building, livestock helath programs, and some administrative
A brief review of the case of research and extension in Guatemala is
useful because it evolved from an extension-dominated strategy based on the
model promoted by U.S. technical assistance in the 1950s. The extension
system was established and initially administered by U.S. technical assis-
tance. In 1958, the administration of extension was transferred to local
control, but it continued to follow the U.S. model which accepted the
following tenets: the extension service should be a distinct institution;
it should focus on general rural welfare rather than production per se; it
should use individual contact with farmers rather than group methods; it
should be strictly educational and should not engage in the provision of
inputs or credit; and it should be administratively independent from the
research system, though not isolated (Rice, 1974:117-120). In three surveys
of Latin American extension agents conducted by Rice (1974:143) in 1971,
the Guatemalan extension service was consistently ranked one of the weakest
in the region.
Until the early 1970s, research was located in DIGESA, a general
services agency which also contained the much larger extension division.
Growing concern over the weakness and traditional orientation of agricul-
tural research culminated in 1973 with the creation of the Instituto de
Ciencia y Tecnologia Agricola (ICTA), an autonomous institute responsible
for all agricultural research. ICTA adopted a strong applied research
approach involving commodity-based research, some on-farm testing, and
closer contact with farmers (Fumagali and Waugh, 1977). In 1975, the
Socioeconomic (SE) unit was created to evaluate the impact of new tech-
nologies. In its first assignment, the SE unit confirmed the complaints
of farmers in one region that the fertilizer they were required to use in
order to obtain credit was not worth its cost. As a result, the fertilizer
requirement was dropped from the package (Whyte, 1981).
Although ICTA was conducting on-farm trials, the SE unit pushed for
more research using the criteria, input levels, and practices common among
SE recommended that scientists substitute bullocks for tractors
in the experiments conducted at the [experiment station], that
they undertake experiments on rocky slopes typical of the kind
many farmers had to cultivate, and that they limit input levels
close to those that farmers were using. (Gostyla and Whyte,
Despite disagreement with many station-based technical scientists, the
ICTA leadership allowed the SE unit to broaden its responsibilities to
include some on-farm agronomic research. A strategy with many of the
components of FSR developed: informal diagnostic surveys to identify
recommendation domains, on-farm testing of alternative cropping systems,
and close collaboration with farmers.
Slowly, this multidisciplinary approach became more integrated into
ICTA's research activities with the use of trained graduates of agricul-
tural high schools ("peritos") representing the SE unit at the regional
level, and the collaboration of agronomists and economists in two-person
survey teams and in the implementation of the farm record program.
Technical scientists take primary responsibility for
generating technology at the center and in [researcher-
managed] field trials, and for evaluating that tech-
nology's agronomic performance. SE in turn is responsi-
ble for orienting and carrying out diagnostic agro-socio-
economic studies of new work areas, working with inte-
grated multi-disciplinary teams... The farmer, of course,
is a participant in this process: initially, as an
information source, then as an experimental practitioner
in trials, and finally as the primary evaluator in the
[farmer-managed trials]. (Gostyla and Whyte, 1980:30)
Information regarding farming patterns and response to new technology was
available to researchers from the surveys, from the results of farmer-
managed trials, and from nearly 500 farmers participating in the farm
record program. This information has created greater awareness of regional
needs, greater research focus on problems affecting farmers (even among the
commodity programs) and more attention to collecting and evaluating native
crop varieties. This confirms the capacity of FSR methodology to promote
upward flow of information to improve the relevance of research.
Unfortunately, the case of Guatemala also provides an example of the
difficulty of linking FSR to extension activities to accomplish the equally
important objective of rapidly disseminating research results. DIGESA has
continued to operate under the traditional assumptions that farmers are
irrational and that extension problems derive from failures in the medium
rather than inadequacies of the message. It has maintained its conventional
emphasis on supervised credit and individual contact with farmers.
There are, however, several promising signs. First, in 1978 a Letter
of Understanding between ICTA and DIGESA was signed to delineate the
responsibilities of each and affirm the need for cooperation (Shaner et al,
1982:360-61). Second, an evaluation of the Small Farmer Cropping Systems
research project in Guatemala by the Tropical Agricultural Research and
Training Center (CATIE) reported that DIGESA agents were:
restive in their primary role of farm credit advisors
and looked toward CATIE's work with the ICTA as a possible
break-through that would draw them increasingly into the
action through more "field day" type demonstrations of
the better cropping systems alternatives. (USAID, 1980:E-8)
In fact, DIGESA was recently relieved of the responsibility for supervised
credit (USAID, 1982:B-9). And third, ICTA has begun to experiment with
small research-extension projects, using farmer-leaders to carry out trials
and to travel to other villages to explain the results.
The eventual goal, however, must be to use DIGESA agents to conduct
trials and organize paraprofessionals. A start in this direction has been
made with the training of agents by ICTA in on-farm research methodologies.
Whyte (1981:63) argues that resolution of the problem will eventually
require structural change to bring ICTA and DIGESA under the same leader-
ship, but it is clear that substantial progress has been made simply by
developing institutional mechanisms of coordination.
The agricultural research and extension system in Taiwan is one of the
most effectively integrated in the developing world. Research and extension
were initiated during the early 1900s under Japanese occupation in order
to provide rice for Japan's growing economy. After World War II, control
of Taiwan reverted to China and starting in 1949 it was governed by the
Nationalist Party which had declared independence from the mainland.
As part of the government's commitment to economic development,
spurred by political and symbolic competition with the People's Republic of
China, large investments have been made in agricultural development. A
major land reform program in the 1950s, significant improvement of the
rural road network, the development of effective institutions to serve
agriculture, and other improvements have contributed to the doubling of
net farm family income over 1952-1972. (Stavis, 1974).
Responsibility for agricultural research is divided among agricultural
colleges which perform basic research, research institutes which carry out
basic and applied research on a commodity basis, and improvement stations
which are involved in applied research and extension for their particular
region. In addition, government-owned corporations conduct research and
extension activities for sugar, pineapple, and several other export crops
(Lionberger and Chang, 1970:35-40).
Most extension activity is carried out by a dual structure involving
government public offices at the township level and the farmers' associa-
tions. The former are located within a hierarchy of township, county, and
"provincial" governments. The latter represents perhaps the most distinc-
tive aspect of Taiwan's agricultural institutions.
The township-level farmers' associations are self-financing legal
entities owned and controlled by their membership. This control is,
however, subject to considerable supervision by county and provincial
farmers' associations and by the government itself. They regulate employee
qualifications and salary, the fees charged for services and credit,
expenditure patterns, and hiring practices. The associations are financed
by fees collected from the government for collecting and storing taxes
(paid in rice), through the sale of farm and general supplies (including a
virtual monopoly on fertilizer sales), and by marketing farm products.
Funds are also raised by providing credit, backed up by the deposits of
other members (Wang, 1981:185). As a measure of the importance of these
associations, it should be noted that their membership includes almost 70%
of the farm families. Furthermore, they handle about 30% of all marketed
agricultural product and supply about one third of all rural credit (Stavis,
By law, 70% of the association profits must be spent on agricultural
extension with each association hiring (and firing) its own extension
workers. There are generally between one and four per township. Farmer
discussion groups which meet about once a month are the primary means of
agent-farmer contact; these groups include approximately 15% of the
"Provincial" refers to all Taiwan since the government considers
itself a province of China.
association members (Axinn and Thorat, 1970:78-79). Agent salaries are
a function of the profitability of the association, which in turn is
partially determined by the agricultural productivity of the township.
Thus, job security and salary are influenced by the performance of the
extension worker as perceived by the farmers, introducing a measure of
accountability rare in extension systems (Stavis, 1979:40). This provides
the agent with the incentive to be familiar with research recommendation,
to seek out answers to pressing agricultural problems, and to regularly
provide the farmers with information they consider useful. Clearly, this
is an important strength of the Taiwan agricultural research-extension
system. The extension agents generally have little formal education.
Eighty percent are engaged in farming operations, and 96% reported that
they would not recommend practices obtained from research without evidence
of its local adaptability (Lionberger and Chang, 1970:339). This naturally
contributes to the credibility of extension recommendations.
There is no legal or direct administrative relationship between the
extension offices of the farmers' associations and the agricultural improve-
ment stations. But in practice they work closely-together. With assistance
from the extension agents of the farmers' association, the improvement
stations establish demonstration plots.
The farmer who owns the land is given seed, fertilizer,
and pesticide by the crop improvement farms. In addi-
tion, the crop improvement farms may pay land rent and
labor expenses. The farmer manages the field and keeps
records of production. Generally speaking, one-half of
the townships will have such a demonstration plot for
each crop. (Stavis, 1974:84)
Extension agents also distribute improved seeds and extension materials
provided by the improvement station.
A study by Li' and Change (1970) focused on communica-
tion pimong the fane extension organizations at the town-
ship, ind provincialss, and the three research systems
mentio'. The study that:
[vel] superviboth sides had many contacts
cultural rese service agencies. Those in
ts' associate! more frequent with the
dearch source-e likely to be interactive
trectional in Farmers' association super-
ve more likely. information directly from
rources, part from agricultural improve-
mons, while [level] government extension
s-s were more o obtain information through
[ channels, both were characterized by
frequent use farm information sources.
(r and Chang,4).
That in flows upwadicated by the high proportion of
improvtion technic report that current experiments are
in parn ideas provextension personnel or farmers. That
relevacal advice i:ng the farm is confirmed by the con-
fidencarmers place extension agents and the research
], Taiwan hased a research-extension system with
stroncctive linkagut the formal legal or administrative
integien proposed.s possible through the gradual evolu-
tion (sticated set:ultural institutions, in particular,
the fisociations. "e several important factors in the
succe, system. Fiinsion agents are hired by, thus
respol township fatsociations. They actively test and
adaptral technolo9commendation to their farmer/clients.
!e elaborate structure of extension services pro-
videstion of flexind farmer input through the farmers'
associations and top-down coordination to maintain responsibility to
Third, research-extension linkages are promoted by joint activities
of personnel from different agencies, the demand for technical information
by farmers and extension workers, and the existence of regional improve-
ment stations whose central purpose is applied research.
Population growth-, agricultural diversification, urban migration, and
factional disputes within the farmers' associations threaten the effective-
ness of these institutions (Wang, 1981). In addition, observers have
questioned the transferability of this model pointing at the unusually
homogenous agrarian structure, the Confuscian tradition of deference to
authority, and the high level of individual and governmental commitment
to agricultural development. At the very least, however, the case of
Taiwan demonstrates the potential of developing agricultural institutions
to channel information upward, and ensure research and extension responsive-
ness to the needs of farmers.
SUMMARY AND CONCLUSION
There is considerable agreement that linkages between research and
extension organizations are weak in many developing countries. This con-
ceptual and administrative separation of agricultural research from exten-
sion, breaking up what is essentially a single process of technology
generation and diffusion, creates a number of problems. It allows each
organization to focus on the tasks considered within its domain to the
neglect of the broader goal of making useful agricultural technology
available to farmer. Researchers tend to focus on the type of work which
offers professional and social reinforcement from their scientific col-
leagues. This usually encourages more basic research or research under
strictly controlled conditions which allow a higher degree of statistical
accuracy at the expense of representativeness. Extension organizations
maintain internal incentive systems which, at best, reward emphasis on
communication techniques and skills and, at worst, encourage work avoidance
Neglected are the important tasks which are not clearly the responsibility
of either agency: the field testing of agricultural technology, the
adaptation of recommendations to local conditions and geo-climatic factors,
and the coordination of information flow between research and extension
As a result, relations between researchers and extension personnel
have not been constructive. Researchers criticize the extension service
for its disorganization and its inability to transmit improved technology
to farmers. Extension personnel accuse the research establishment of
pursuing esoteric research and neither understanding nor caring about
the actual problems of farmers. In many countries the need to compete
for annual budget allocations and for donor agency funding can increase
This conflict has been reflected in the literature and in the trends
in funding by donor agencies and national governments. Interest in
extension and community development was greatest in the 1950s and early
1960s. During the late 1960s and early 1970s, there was a shift in focus
to the potential of the international agricultural research centers. Since
the mid-1970s, interest in strengthening national agricultural research
capacity has grown substantially. As discussed earlier in this essay,
these shifts were manifestations of changes in the underlying assumptions
regarding the process of agricultural transformation and the nature of
agricultural technology. The primary changes were the growing consensus
concerning the rationality of the traditional farmers and greater awareness
of the location specificity of many types of agricultural technology.
Just as evaluations of extension programs in the 1950s and 1960s
rarely questioned the value of the message being delivered to the farmer,
current evaluations of national research programs do not often consider
how the message is to be delivered. Frequently, they merely mention the
"need to strengthen linkages" between research and extension. Nonetheless,
growing attention should be directed toward developing strategies to
improve communications between research and extension organizations.
This paper has taken as a starting point the idea that research-
extension linkages are only a means to improve the relevance of the tech-
nology generated and the effectiveness of its diffusion. For this reason
a review was made of problems "internal" to research and extension
organizations. A number of common deficiencies of developing country
research systems were categorized into those relating to 1) the use of
evaluation criteria for new technology which do not reflect those used
by farmers, particularly the neglect of economic return and of risk,
2) the performance of research on sites for which agro-climatic factors
are not representative of the range of farm-level conditions, 3) the
design of experiments in which the controlled variables, such as agronomic
practices and input levels, do not reflect actual farm conditions for the
majority of farmers.
With respect to extension, there was a brief review of seven related
problems faced by extension services in developing countries: reaching
large numbers of dispersed farmers, the bias toward large farmers, the
bias toward male farmers, shortage of operating funds relative to salaried
staff, management deficiencies, excessive workloads assigned to agents,
and the weak technical and practical background of the agents themselves.
Research-extension linkages, no matter how well developed, cannot
deal with all of the problems discussed above. However, to the extent
that regular communication with extension workers provides researchers
with professional and personal reward for having useful information to
offer, it can cultivate the interest and willingness to deal with farm-
level problems. Insofar as this linkage provides researchers with infor-
mation on farm problems and farmers' responses to previous recommendations,
as well as an improved understanding of farm resources and constraints,
it will increase the ability of researchers to make more appropriate
recommendations. And finally, to the extent that such linkages entail
active cooperation between the two agencies they can facilitate more
systematic data collection and widespread on-farm testing of agricultural
technology. The most direct benefit to extension is that their recommenda-
tions are likely to be more valuable to their clientele. Farmers may begin
to seek out such advice. The resulting increased confidence of the govern-
ment in the service can foster higher levels of consistent funding. It is
clear, however, that the benefits of this linkage to extension depends on
the practical value of research results.
Empirical research on linkages between research and extension has not
been extensive. It consistently identifies several key obstacles, however,
to better communication and cooperation. These include 1) the lack of
formal mechanisms, such as regular meetings, to exchange ideas, 2) status
differences between the extension personnel and the more educated scientists,
and 3) mutual suspicion of the capabilities and usefulness of the work of
the other organization. Such difficulties may be self-reinforcing. The
latter two problems, for example, may be exacerbated by the lack of formal
and informal contact between extension agents and researchers.
Three general approaches to improving research-extension. linkages
(institutional integration, the establishment of a liaison unit, and coopera-
tive activities) and two specific strategies (the Training and Visit Exten-
sion System and farming systems research) were considered. Efforts to
develop formal linkages through institutional structures range from complete
integration of research and extension to the creation of a liaison officer
attached to one of the organizations.
Complete organizational integration is an attractive solution at first
glance and is in fact a model followed by a number of successful institutions
which focus on major export commodities. However, these successes may be
attributable to concentrated effort on a highly profitable crop, a pattern
probably not replicable on a nation-wide, multi-crop basis. At the other
extreme, liaison officers attached to research institutions may not be
independent enough to provide feedback from farmers and extension workers
concerning the applicability of research results. A separately funded
liaison unit is more likely to foster two-way communication, but it was
noted that uncooperative attitudes on the part of research personnel and
internal obstacles to communication within the extension service can defeat
Cooperative activities were discussed as a way to institutionalize
informal and formal channels of communication between research and extension
personnel. Although there are administrative and bureaucratic complications
to inter-agency cooperation, high-level political support for this can
result in significant returns. In particular, on-farm trials of new
practices and varieties and diagnostic surveys are two areas for possible
cooperation. In addition to promoting communication, these activities can
yield benefits even if (or especially if) there are currently no valid
recommendations to offer.
The Training and Visit System and farming systems research are
strategies which, in a sense, approach similar goals from opposite
directions. The T & V System attempts to impose on extension agents a
rigid but realistic work schedule, putting heavy emphasis on in-service
technical training and restricting the agents role to that of providing
technical assistance in agricultural production. Ideally, research is
focused on applied problems by the discipline of regular sessions with
extension personnel to update recommendations, answer questions, and
respond to feedback from agents. Thus, the linkage occurs through in-
service training. However, in practice the T & V System has been tested
primarily in countries with relatively advanced research systems and rural
market infrastructure. Attempts to adapt it to African conditions are
currently underway. Here its capacity to stimulate more effective research
will be tested.
Farming systems research is a strategy to focus research efforts on the
problems and resource limitations of small farmers. Researchers are involved
in data collection and field testing of incremental improvements in existing
farming systems. In a sense, this implies that the extension service is
bypassed as the intermediary between research and extension. In practice,
FSR projects have rarely had strong linkages with conventional research or
with the extension service. It has been argued, however, that institution-
alization of FSR will require greater collaboration with the extension
service if the FSR teams are to be spread thinly enough to cover a diverse
nation-wide set of agroecological zones.
Two other factors make FSR more amenable to research-extension
linkages. First, researchers are involved in field work to a greater degree
than in conventional agricultural research, thus broadening the scope for
potential collaboration with extension personnel. Second, in a successful
FSR program each has more information to offer the other so there is more
incentive to communicate: FSR researchers have a more immediate need for
quantitative and qualitative information about farming systems and
constraints in the region and they can generate more directly applicable
information to offer extension personnel.
Just as the effectiveness of the T & V System depends on a steady flow
of appropriate agricultural technology from the research system, long-term
success of FSR depends on the efficient operation of the extension agency.
If the extension agency is unable to organize effective dissemination of
the technology, or develop a close working relationship with research
organizations, then FSR may remain an interesting but impractically inten-
Two other comparisons between FSR and the T & V System can be made.
First, a major challenge for FSR is to move beyond the project approach and
become more institutionalized. The T & V System, on the other hand, involves
the reorganization of existing extension systems, and accordingly tends to
be institutionalized almost from the beginning. Second, the FSR approach
is more flexible and its assumptions more universal. Because the T & V
System is more structured and depends to a greater degree on the agricultural
and institutional characteristics of the country, direct international
transfer is less feasible and more adaptation is required.
The importance of adapting institutions to local conditions was
highlighted by the review of three innovative research and extension
systems which have made progress in integrating the process of generating
and diffusing agricultural technology.
The case of Botswana demonstrates that with an effective civil
service and participatory reform, an extension system similar to the T & V
System can be adapted to African conditions. The principal changes were
1) that extension workers were given broader roles to compensate for weak
rural markets and 2) that the frequency of in-service training was lower
in response to more limited resources and low population density.
In Guatemala, the institutionalization of FSR was a central issue.
In particular, the development of communication and cooperation between
the reoriented research institute and the traditionally oriented extension
service was a gradual but essential task in the improvement of the overall
The last case, that of Taiwan, is quite different from the others in
that the research and extension system is the result of long-term institu-
tional development rather than a recent reform. Local political and
cultural factors may make this a less transferable model. Nonetheless, it
serves as a reminder that a long-term goal in the development of agricul-
tural institutions should be the evolution of mechanisms which give farmers
some degree of control over research and extension personnel, whose purpose
is to serve the farmer.
Adelman, A. (1981). "Colombian friendship groups: Constraints on a rural
development acquisition system." Journal of Developing Areas 15 (April)
Albrecht, H.A. (1979). "Report of a study of research priorities,
Research and Disease Control Section, Crop Development Division."
Working paper for Agricultural Research and Organization Conference.
21-28 February, Arusha, Tanzania.
Ames, G. and L. Busch. (1979). "Agricultural and institutional develop-
ment in the Sahelian countries of West Africa: A real challenge in
technical cooperation." Agricultural Administration 6 (3):199-209.
Anthony, K.R.M. et al. (1979). Agricultural Change in Tropical Africa.
Ithaca: Cornell University Press.
Atsu, S.Y. (1974). "The focus and concentrate program in the Kpanda
and Ho Districts: Evaluation of an agricultural extension program."
Legon: University of Ghana, ISSER, Technical Publication Series No.
Ascroft, J. et al. (1972). "Does extension create poverty in Kenya?"
East African Journal 9 (3):28-33.
Axinn, G. and S. Thorat. (1970). Modernizing World Agriculture: A
Comparative Study of Agricultural Extension Education Systems. New
Delhi: Oxford & IBH Publishing.
Benalcazar, R. (1978). "New techniques, agricultural extension services,
and credit facilities as instruments of economic development." in
Dams, T. and K. Hunt (eds.). Decision Making and Agriculture.
Proceedings of the 16th International Conference of Agricultural
Economists. 26 July-4 August 1976, Nairobi. International
Association of Agricultural Economists, Agricultural Economics
Institute. Lincoln: University of Nebraska Press.
Barker, R. (1981). "Establishing priorities for allocating funds to
rice research." in Johnson, G. and A. Maunder (eds.). Rural Change:
The Challenge for Agricultural Economists. Proceedings of the 17th
International Conference of Agricultural Economists. 3-12 September
1979, Banff, Canada. International Association of Agricultural
Economists, Institute of Agricultural Economics, Farnborough, U.K.:
Benor, D. and J. Harrison. (1977). Agricultural..Extension: The Training
and Visit System. Washington, D.C.: World Bank.
Boserup, E. (1970). Women's Role in Economic Development. New York:
St. Martin's Press.
Brown, M. (1970). "Agricultural extension in Chile: A study of
institutional transplantation." Journal of Developing Areas 4 (2):
Chambers, R. (1975). Two Frontiers in Rural Management: Agricultural
Extension and Managing the Exploitation of Communal Natural Resources.
Brighton, U.K.: Institute of Development Studies.
Collinson, M.P. (1982). "Farming systems research in Eastern Africa:
The experience of CIMMYT and some national agricultural research
services, 1976-1981." East Lansing: Michigan State University,
International Development Working Paper No. 3.
Cummings, Ralph, Jr. (1981). "Strengthening linkages between agricultural
research and farmers: An overview." Prepared for the Workshop on
Linkages between Agricultural Research and Farmers in Developing
Countries. 13-14 May, OECD, Paris.
Dalrymple, D. (1979). "The adoption of high-yielding grain varieties
in developing nations." Agricultural History 53 (4):704-26.
Ekpere, J. (1973). "A comparative study of job performance under two
approaches to agricultural extension organization in the Midwestern
State of Nigeria." Ph.D. Madison: University of Wisconsin.
Evenson, R. (1978). "Organization of research to improve crops and
animals in low-income countries." in Schultz, T.W. (ed.).
Distortions in Agricultural Incentives. Bloomington: Indiana
Evenson, R. and Y. Kislev. (1975a). "Investment in agricultural research
and extension: An international survey." Economic Development and
Cultural Change 23 (3):507-521.
(1975b). Agricultural Research and Productivity. New Haven:
Yale University Press.
Evenson, R. and H. Binswanger. (1979). "Technology transfer and research
resource allocation." in Binswanger, H. and V. Ruttan (eds.).
Induced Innovation: Technology, Institutions, and Development.
Baltimore: Johns Hopkins University Press: 164-210.
Evenson, R., P. Waggoner, and V. Ruttan. (1979). "Economic benefits
from research: An example from agriculture." Science 205 (14):
Ferguson, L.C. et al. (1971). Agricultural Research in Uganda: A
Survey Evaluation and Recommendations. Columbus: Ohio State
Galt, D. et al. (1982). "Farming systems research (FSR) in Honduras,
1977-81: A case study." East Lansing: Michigan State University,
International Development Working Paper No. 1.
Gebre, H. (1980). "Agricultural research in Ethiopia." in Bengtsson, B.
and G. Tedla (eds.). Strengthening National Agricultural Research,
Report from a SAREC Workshop, Part I: Background Documents.
Stockholm: Swedish Agency for Research Cooperation with Developing
Gilbert, E.; D. Norman; and F. Winch. (1980). "Farming systems research:
A critical appraisal." East Lansing: Michigan State University,
Rural Development Paper No. 6.
Golden, W.P. (1974). "Sri Lanka's agricultural extension service." in
Rockefeller Foundation. Strategies for Agricultural Education in
Developing Countries. New York: Rockefeller Foundation: 238-75.
Gostyla, L. and W. Whyte. (1980). "ICTA in Guatemala: The evolution of
a new model for agricultural research and development." Ithaca:
Cornell University, Rural Development Committee, Special Series on
Agriculture Research and Extension No. 3.
Grimes, R.C. (1963). "Intercropping and alternative cropping of cotton
and maize." East Africa Agriculture and Forestry Journal 28:161-163.
Havelock, R. (1975). "What we should know about knowledge utilization."
Paper presented at the Agricultural University, Wageningen, Holland.
Hildebrand, P. (1981). "Combining disciplines in rapid appraisal: The
sondeo approach." Agricultural Administration 8 (6):423-432.
Hornik, R. (1982). "Communication and agriculture." Palo Alto:
Stanford University, Institute for Communication Research.
Howell, J. (1982). "Managing agricultural extension: The T & V System
in practice." Agricultural Administration 11 (4):273-284.
International Service for National Agricultural Research. (1981a). Kenya's
Agricultural Research System. The Hague: ISNAR.
(1981b). The Agency for Agricultural Research and Development of
Indonesia. The Hague: ISNAR.
(1982). Le Sisteme National de Recherche Agricole du Rwanda.
The Hague: ISNAR.
Iwueke, C. and E.W. Findlay. (1980). "Factors related to the effective
administration of extension in agriculture and rural development in
Nigeria." Agricultural Administration 7 (1):59-69.
Jimenez, L. and R. Laird. (1974). "Mexico: The Puebla Project." in
Rockefeller Foundation. Strategies for Agricultural Education in
Developing Countries. New York: Rockefeller Foundation: 276-330.
Karami, E. and R. McCormick. (1982). "An appraisal of the agricultural
extension service in Iran as perceived by extension specialists
and extension agents." Agricultural Administration 10 (2):145-149.
Khan, A.H. (1978). "Ten decades of rural development: Lessons from
India." East Lansing: Michigan State University, Rural Development
Working Paper No. 1.
Kingshotte, A. (1980a, 1980b). "Organization and management of agricul-
tural extension and farmer assistance Botswana, Parts 1 & 2."
Agricultural Administration 7 (3):191-209 and 7 (4):297-322.
Lakoh, A.K. and A. Akinbode. (1981). "Agricultural research delivery
system in Sierra Leone." Agricultural Administration 8 (4):289-305.
Lele, U. (1975). The Design of Rural Development: Lessons from Africa.
Baltimore: Johns Hopkins University Press.
Leonard, D.K. (ed.). (1973). Rural Administration in Kenya. Nairobi:
East Africa Literature Bureau.
(1977). Reaching the Peasant Farmers: Organizational Theory
and Practice in Kenya. Chicago: University of Chicago Press.
Lionberger, H. (1974). "Organizational structures for productivity in
agricultural extension." Journal of Developing Areas 8:395-408.
Lionberger, H. and H. Chang. (1970). Farm Information for Modernizing
Agriculture: The Taiwan System. New York: Praeger.
Liwenga, J.M. (1980). "Agricultural Research in Tanzania" in Bengtsson,
B. and G. Tedla (eds.). Strengthening National Agricultural Research,
Report from a SAREC Workshop, Part I: Background Documents. Stockholm:
Swedish Agency for Research Cooperation with Developing Countries: 55-61
Mbithi, P.M. (1973). "Agricultural extension as an interventionist
strategy: An analysis of extension approaches" in Leonard, D.K. (ed.).
Rural Administration in Kenya: A Critical Appraisal. Nairobi: East
Africa Literature Bureau: 76-96.
McDermott, J.K. (1981). "Linking research and extension to serve the
small farm operator." Presented at the symposium Research for Small
Farmers. 15 November, Beltsville Agricultural Research Center,
Mellor, J. (1966). The Economics of Agricultural Development. Ithaca:
Cornell University Press.
Morss, E. et al. (1976). Strategies for Small Farmer Development: An
Empirical Study of Rural Development Projects in the Gambia, Ghana,
Kenya, Lesotho, Nigeria, Bolivia, Colombia, Mexico, Paraguay, and Peru.
(2 vols.) Washington, D.C.: Westview Press.
Moseman, A.H. (1970). Building Agricultural Research Systems in the
Developing Nations. New York: Agricultural Development Council.
(1976). "Organizational structures for national agricultural
research systems" in Southeast Asian Regional Center for Graduate
Study and Research in Agriculture, Agricultural Research Management,
Asia, Volume I. Laguna, Philippines: SEARCA: 45-69.
Nagel, V.W. (1979). "Knowledge flows in agriculture: Linking research,
extension, and the farmer." Zeitschrift fur Auslandische
Landwirtschaft 18 (April-June): 135-150.
Nelson, R.S.M. (1979). "Background paper for Tanzania Agricultural
Research." Prepared for the Agriculture and Livestock Research
Workshop. 26 February-3 March, Arusha, Tanzania.
Norman, D. (1980). "The farming systems approach: Relevancy for
the small farmer." East Lansing: Michigan State University, Rural
Development Paper No. 5.
Okigbo, B. et al. (1981a). Report of the Research Institutes Review Panel
1980/1981, Volume I: Background Papers and Evaluation of Individual
Research Institutes. Ibadan, Nigeria: Green Revolution National
Committee, Moor Plantation.
(1981b). Report of the Research Institutes Review Panel
1980/1981, Volume 2: Recommendations and Observations on Various
Issues. Ibadan, Nigeria: Green Revolution National Committee, Moor
Oteifa, B.A. (1980). "Status of agricultural research in Egypt" in
Bengtsson, B. and G. Tedla (eds.). Strengthening National Agricultural
Research, Report from a SAREC Workshop, Part I: Background Documents.
Stockholm: Swedish Agency for Research Cooperation with Developing
Oram, P.A. and V. Bindlish. (1981). Resource Allocation to National
Agricultural Research. Washington, D.C.: ISNAR/IFPRI.
Pasour, E.C. and M.A. Johnson. (1982). "Bureaucratic productivity: the
case of agricultural research revisited." Public Choice 39:302-317.
Perrin, R.K.; D.L. Winkelmann; E.R. Moscardi; and J.R. Anderson. (1976).
From Agronomic Data to Farmer Recommendations: An Economics Training
Manual. Londres, Mexico: CIMMYT Information Training Bulletin No. 27.
Pinstrup-Anderson, Per. (1982). Agricultural Research and Technology in
Economic Development. New York: Longmans.
Ray, A.K. et al. (1979). "Quantitative and qualitative impact of Training
and Visit System on different groups of farmers: A case study of the
Hooghly District, West Bengal." Indian Journal of Agricultural
Economics 34 (4):20-25.
Rohrbach, D. (1981). "Issues in developing and implementing a farming
systems research program." Washington, D.C.: USDA, Office of
International Cooperation and Development.
Roling, N. (1975). "Knowledge brokerage for increasing the relevance of
agricultural research in African smallholders." Rural Africana No. 27:
Russell, John. (1981). "Adapting extension work to poorer agricultural
areas." Finance and Development: 30-33.
Ruthenberg, H. (1971). Farming Systems in the Tropics. Oxford, U.K.:
Ruttan, V. (1982). Agricultural Research Policy. Minneapolis: University
of Minnesota Press.
Sarikaphuti, Y. (1981). "Bridging research and extension gap through
development communication in Thailand" in Food and Fertilizer
Technology Center, Support Communication for Rural Development
Programs. Tapei, R.O.C.: FFTC: 21-29.
Schuh, G.E. and H. Tollini. (1979). "Costs and benefits of agricultural
research: The state of the arts." Washington, D.C.: World Bank
Staff Working Paper No. 360.
Schultz, T.W. (1964a). Transforming Traditional Agriculture. New Haven:
Yale University Press.
(1964b). "Economic growth from traditional agriculture." in
Moseman, A.H. (ed.). Agricultural Sciences for the Developing Nations.
Washington, D.C.: American Association for the Advancement of
Stavis, B. (1974). "Rural local governance and agricultural development
in Taiwan." Ithaca: Cornell University, Special Series on Rural
Local Government No. 15.
(1979). "Agricultural extension for small farmers." East Lansing:
Michigan State University, Rural Development Working Paper No. 3.
Stommes, E. and S. Sisaye. (1979a, 1979b, 1980). "Administration of
agricultural development programmes: A look at the Ethiopian
approach Parts 1, 2, and 3." Agricultural Administration 6 (3):
221-239, 6 (4):253-267, and 7 (1):1-23.
United Nations Economic Commission on Africa and the Food and Agriculture
Organization. (1971). A Comparative Analysis of Agricultural
Extension Systems of Eight East African Countries With Suggested
Guidelines for Improvement. Addis Ababa: UN ECA/FAO.
United States Agency for International Development. (1980). Kitale
Maize: The Limits of Success. Washington, D.C.: AID Project
Impact Evaluation Report No. 2.
------ I "Project Paper: Agricultural Research and Planning
Project23." Dakar: USAID/Senegal.
--- ( Guatemala: Development of the Institute of Agricul-
tural Sland Technology (ICTA) and Its Impact on Agricultural
esrLicfarm Productivity. Washington, D.C.: AID Project
Impact Eion Report No. 30.
Wang, Y.T. ( "The roles of farmer associations and agricultural
developmograms in Taiwan." in Johnson, G. and A. Maunders
(eds.). Change: The Challenge for Agricultural Economists.
Proceedirthe 17th International Conference of Agricultural
Economists2 September 1979, Banff, Canada. International
AssociatiAgricultural Economists, Institute of Agricultural
Economics borough, U.K.: Gower: 158-166.
'---- -. (19 Agricultural Research: Sector Policy Paper.
Washington.: World Bank.
----(1984ccelerated Development in Sub-Saharan Africa: An
Agenda forn. Washington, D.C.: World Bank.