Th Exeiec ofCMY andS Ntoa
M. P. Col
."UItraina eeomn Deatmn of Arclualcnmc
Papr o.3 icign tae nierit
198 Eas LiI anig Michigan 4882
MSU INTERNATIONAL DEVELOPMENT PAPERS
Eric W. Crawford, Carl K. Eicher and Carl Liedholm, Co-Editors
The MSU International Development Paper series is designed to further
the comparative analysis of international development activities in Africa,
Latin America, Asia and the Near East. The papers report research find-
ings on historical, as well as contemporary, international development
programs. The series includes papers on a wide range of topics, such as
alternative rural development strategies; non-farm employment and small-
scale industry; housing and construction; farming and marketing systems;
food and nutrition policy analysis; economics of rice production in West
Africa; technological change, employment, and income distribution;
computer techniques for farm and marketing surveys; and farming systems
research. While the papers mainly convey the research findings of MSU
faculty and visiting scholars, a few papers will be published by
researchers and policy-makers working together with MSU scholars on
research and action programs in the field.
The papers are aimed at teachers, researchers, policy-makers, donor
agencies, and international development practitioners. Selected papers
will be translated into French, Spanish, or Arabic. A list of available
papers and their prices may be obtained from: MSU International Develop-
ment Papers, Department of Agricultural Economics, Agriculture Hall,
Michigan State University, East Lansing, Michigan 48824, U.S.A.
Individuals and institutions in Third World countries and U.S. Agency
for International Development officials may receive copies free of charge.
FARMING SYSTEMS RESEARCH IN EASTERN AFRICA: THE
EXPERIENCE OF CIMMYT AND SOME NATIONAL
AGRICULTURAL RESEARCH SERVICES,
*This paper is published as part of the farming systems research
program being carried out under "Alternative Rural Development Strategies"
Cooperative Agreement AID/ta-CA-3 at Michigan State University funded by
the U.S. Agency for International Development, Bureau for Science and
Technology, Office of Rural Development and Development Administration.
**CIMMYT, Nairobi. The views expressed in the paper are not
necessarily those of the International Wheat and Maize Improvement Center
MSU is an Affirmative Action/Equal Opportunity Institution
All rights reserved by Michigan State University, 1982.
Michigan State University agrees to and does hereby grant to the United States
Government a royalty-free, nonexclusive and irrevocable license throughout the
world to use, duplicate, disclose, or dispose of this publication in any manner
and for any purpose and to permit others to do so.
Published by the Department of Agricultural Economics, Michigan State
University, East Lansing, Michigan 48824 U.S.A.
TABLE OF CONTENTS
PREFACE . . . . . . . . . . . . . . v
1. INTRODUCTION . . . . . . . . . . . 1
2. INTRODUCING FSR PROCEDURES IN EASTERN AND SOUTHERN AFRICA 8
3. FSR PROCEDURES DEVELOPED BY CIMMYT FOR CARRYING OUT
ADAPTIVE AGRICULTURAL RESEARCH . . . . . . .. 11
3.1 Identifying recommendation domains . . . . .. 12
3.2 Diagnosis of farming systems . . . . . .... .14
3.2.1 Step 1: collecting background information on
the circumstances of target farmers . . .. 16
3.2.2 Step 2: the exploratory (informal) survey . 19
3.2.3 Step 3: the verification (formal) survey . . 22
3.3 On-farm experiments . . . . . . . . 25
4. PROGRESS IN INSTITUTIONALIZING FSR PROCEDURES IN NATIONAL
AGRICULTURAL RESEARCH SERVICES . . . . . .... .27
4.1 Kenya . . . . . . . . ... .. .. . .27
4.2 Zambia . . . . . . . . ... .. .. . .33
4.3 Tanzania . . . . . . . ... . . .35
4.4 Zimbabwe . . . . . . . ... . . .37
4.5 Lessons from country experiences . . . . .... .40
5. CURRENT ISSUES IN FSR PROCEDURES . . . . . .... .42
5.1 A framework of recommendation domains as a link
between national and local priorities . . ... 42
5.2 FSR--carte blanche or a predetermined focus? . . .. 43
5.3 Can LDC research manpower handle a large number of
recommendation domains? . . . . . . .. 44
5.4 Is modelling a farming system worth the extra cost? 46
5.5 Outstanding issues in promoting FSR for agricultural
development . . . . . . . . ... 47
REFERENCES . . . . . . . ... ..... . .. .61
This paper by Michael Collinson is the third in a series of papers on
Farming Systems Research (FSR) in the Third World. The first two papers by
David Norman and Gilbert, Norman, and Winch discussed definitional, con-
ceptual, and methodological problems in organizing and carrying out farm-
ing systems research. The purpose of this paper is to report on CIMMYT's
experience over the 1976-81 period in helping institutionalize farming
systems research methods in national agricultural research services in
Eastern and Southern Africa. Subsequent papers in this FSR series will
report empirical findings by FSR teams in various Third World countries.
Michael Collinson has been a leader in microeconomic research in
Africa for two decades. His experience in farm management research started
20 years ago with a series of surveys in Tanzania in 1962. Collinson's
Farm Management in Peasant Agriculture (1972) is a standard reference on
how farm management studies can help meet the needs of extension workers
and local planners. We are pleased that he has taken time to share his
experiences over the past five years in Eastern and Southern Africa. The
theme of Collinson's paper is the need to develop cost-effective research
programs which are consistent with small pools of trained scientists and
limited budgets for recurrent expenses.
There are five major lessons from Collinson's experience which might
help administrators of national research services in other parts of Africa
and in other regions of the Third World. First, Collinson reports how
CIMMYT staff members introduced the idea of farming systems research by
conducting demonstrations of FSR which produced reports within 7-9 months
after the initiation of research. The results of these demonstrations
provided administrators of research services with tangible results for
evaluating FSR approaches. Second, Collinson describes the use of the
exploratory or informal survey as a cost-effective technique for gaining
information about farming systems and constraints on improving these sys-
tems. Collinson calls the exploratory (informal) survey the "pivotal"
step in the diagnosis of the farming system. CIMMYT's exploratory (infor-
mal) questionnaire is reprinted on pages 49-60 as a guide for researchers.
Third, he discusses the lessons learned in modifying a "strongly compart-
mentalized research establishment" in Kenya (by posting young economists
on research stations) and candidly describes how difficult it was for
junior economists to establish a meaningful dialogue with senior technical
scientists on these stations.
The fourth highlight of the paper is a discussion of how the lessons
from Kenya were used in drawing up a plan to help reorganize and strengthen
the agricultural research system in Zambia. In Zambia, a plan was drawn up
in 1980 to establish a two-level hierarchy--Commodity Research Teams work-
ing from research stations and Adaptive Research Planning Teams (Farming
Systems Research Teams). The Adaptive Research Planning Teams (ARPTs)
include a Farming Systems Economist, Farming Systems Agronomist, and a
Research/Extension Liaison Officer. The decision to strengthen Commodity
Research Teams and to establish ARPTs in Zambia will hopefully eliminate
the friction that might arise if young trainee economists are posted in
Commodity Research Teams which are dominated by senior technical scien-
tists. This two-level approach has also been adopted by ISRA in Senegal as
part of a multi-donor financed program to strengthen Senegal's national
agricultural research system.
Fifth, the paper points out the integral link between farming systems
research and in-service and short-term training workshops. For example,
CIMMYT's Farming Systems workshop in April 1981 was attended by 42 partic-
ipants (more agronomists than economists) from 10 African countries.
CIMMYT now plans to expand its FS training program in Eastern and Southern
Africa. CIMMYT's experience thus plays up the value of local training
programs which emerge from FSR experimentation within an agro-climatic
Carl K. Eicher
Professor of Agricultural Economics
and Co-Editor, MSU Rural Development Papers
Since 1976 CIMMYT-2/ has been developing and promoting Farming Systems
Research (FSR) in cooperation with national research services in Eastern
Africa. The use of FSR methods has brought benefits not only to farmers,
but also to researchers, by improving their efficiency in designing tech-
nology appropriate to the needs of small farms. This paper presents a
progress report on the experience that CIMMYT has gained in helping
introduce and implement farming systems research in Eastern Africa over
the 1976-81 period.
The paper is written from the perspective of CIMMYT, which has a
program aimed at developing, refining, and promoting FSR methods in order
to help improve the relevancy of national agricultural research and exten-
sion services. CIMMYT's farming systems program in Eastern Africa has the
(1) It is strictly operational and focuses on procedures appropriate
to agricultural research services in many LDCs.
(2) The aim is to improve relevancy of research, not to discover
(3) The program is not 'doing FSR'; its success or failure has to be
judged by its impact on national agricultural research proce-
dures and eventually on their relevance to the problems of farm-
(4) The FSR methods being prompted by CIMMYT are continually being
modified and improved as experience is gained through coopera-
tion with agricultural research services.
Throughout eastern Africa there is an increasing political and policy
concern for meeting the needs of small farmers, many of whom are extremely
isolated from the research process. A quotation from Kenya's 4th Five-Year
Development Plan underscores this concern:
"Research must be of increasing relevance to the farmer's situation.
This includes not only the physical environment that confronts him
but also the socio-economic setting of his farm activities" (Kenya,
1/I should like to thank all reviewers of the draft of the paper.
Some very useful comments were anonymous but those who owned up were:
Duncan Boughton, Derek Byerlee, Eric Crawford, Steve Franzel, Peter
Hildebrand, Roger Kirkby, Carl Liedholm, Don Plucknett, and Don
2/CIMMYT (Centro Internacional de Mejoramiento de Maiz y Trigo),
(International Maize and Wheat Improvement Center), Mexico 6, D.F.
Yet the sheer numbers and variability of small farms and their lack of
organization make it difficult for researchers to address their problems.
The isolation of smallholders from national research services not
only stems from difficulties of linkage but also from differences in per-
spective and rationale between them and larger farmers. Agricultural
professionals, in both private and public service, understand the commer-
cial goals of large farmers. These goals are generally consistent with the
modern agricultural orientation that researchers have received in their
university training. Small farmers, on the other hand, have goals and
decision criteria which may be regarded as irrational by researchers.
Smallholders who are operating close to the subsistence level and have poor
resource endowments place a high priority on a secure food supply for their
family. The aversion of small farms to the risk of losing their food
supply influences their decision making and gives rise to resource alloca-
tion and management strategies different from those of large farmers with
more favorable resource endowments. For instance, farmers' decisions as
to which mixture of maize and beans to select will vary with circumstances.
The final decision on the mixture to plant will depend on at least the
following three sets of factors:
(1) The relative potential of the two crops in the area.
(2) The preferred use of the two crops in local diets.
(3) The local market demand for: (a) maize, (b) beans, (c) both
crops, and (d) other crops grown in the system.
The results of agricultural research services have often not been
relevant to the management problems of small farmers for the following
(1) Researchers have historically placed heavy emphasis on biologi-
cal potential and yield as the dominant criteria upon which to
base recommendations for farmers. But farmers never seek bio-
logical potential for its own sake and never make decisions on
which crop and animal products to produce on the basis of yield
(2) Governments act on behalf of small farmers by deciding research
priorities and which findings will be promoted. Researchers
themselves select 'the best' results of experiments and pre-
scribe these as 'improved management practices' which the exten-
sion services should teach to farmers. 'I know what is best for
you' very much dominates the process.
(3) Researchers on experiment stations are isolated from small farm-
er clients. Reward systems provide little incentive for re-
searchers to understand the farmers they work for and this iso-
lation compounds the prescriptive nature of the recommendation
process. Moreover, natural resources and economic conditions on
experiment stations are different from those in which small
farms operate. For example, soil fertility and labor supply on
an experiment station are likely to be very different from those
on local farms.
Farmers' needs and circumstances are always specific to local situa-
tions. Relevance in research demands that local farmers provide their own
perspectives in planning experiments and their own decision criteria in
evaluating the results. FSR attempts to understand the way farmers make
decisions and it encourages farmers to participate in the research pro-
cess. FSR has the capacity to strengthen linkages between researchers and
their small farmer clients.
In order to implement FSR, two major developments in research organi-
zation and planning are required:
(1) Recognition of the need for both technical and adaptive re-
(2) Recognition of the role a Farm Systems Economist (FSE) can play
in the adaptive research cycle.
Technical research, on research stations, seeks new materials and
methods which are technically feasible in the environment of the country or
region served by the station. Results are additions to a body of knowledge
about potential management improvements. On-farm adaptive research se-
lects and tests components of the body of knowledge identified as ap-
parently relevant to the needs and circumstances of target groups of farm-
ers normally within, but occasionally even outside, that general region. A
Farm Systems Economist attempts to understand how a farmer allocates his
scarce resources of land, labor, and cash between crop, livestock, and off-
farm production in a manner which best satisfies his and his family's
priorities. This is essentially the economic problem. The FSE's profes-
sional task is to understand farmers' decision criteria, and to identify
how and why, in managing his farm, the farmer compromises on the optimal
technical management of any one enterprise in order to raise the productiv-
ity of the whole system.
The six stages in the research process are shown in Figure 1 and can
be illustrated as follows:
(1) The description and interpretation of the farmers' situations
and the identification of management problems and possible de-
(2) The identification of improved agronomic practices and farming
methods which, if used, have a potential for improving the wel-
fare of farmers.
(3) On-farm testing of potential improvements selected as relevant
and feasible for target group farmers, given the conditions they
face in production.
(4) The identification of unsolved technical problems of farmers and
the incorporation of these problems as priorities in the Techni-
cal Research process.
(5) Commodity and disciplinary research.
(6) Body of knowledge.
Locally based adaptive research teams carry out stages 1 & 3. Stages
2 & 4 are links to the technical research programs. The problems identi-
fied in adaptive research form the agenda for commodity and disciplinary
research (Stage 5). The output of experiment station research is the 'Body
of knowledge' (Stage 6) of materials and methods known to be technically
viable under experimental climate and soil conditions on the station and in
the region served by the station.
In summary, the perspective taken here is that FSR (on-farm adaptive
research) can potentially help a research station produce results which
are relevant to the conditions and circumstances of small farms.
In the early 1970's, following some adverse commentaries on the Green
Revolution, CIMMYT's Economics Program sponsored adoption studies in seven
countries in order to identify the most influential factors in shaping the
adoption of new maize and wheat technologies. The results of the studies
revealed that natural resources (e.g., agro-climatic zone) and economic
circumstances of farmers were major determinants of the choice of tech-
nology adopted by farmers. The studies also revealed that technologies
being recommended by experiment stations were often inappropriate to the
needs of small farmers. This mismatch frequently resulted in non-adop-
tion, partial adoption of some components of a technical package, or
Survey diagnosis of
(1) resource and environment
problems and development
Target Group Farmers
of a Recommendation
Domain in a Region'
Experiments on apparently
relevant materials (
and techniques under
Identification & Evaluation /
of materials and techniques
offering potential for problem
solution and the exploitation
problems and possible
(4) new practices and
to farmers' development
Body of Knowledge of
materials and techniques (6)
suitable for the climate
and soils of the Region
Commodity and Disciplinary
research, solving priority
technical problems and
new materials and practices
Interactions between Station-based Technical Research and
On-Farm Adaptive research.
distorted adoption--that is, adoption of the technology by those
subpopulations in a position to exploit it, such as large farmers who were
not the primary targets of government policy.
Following the adoption studies, CIMMYT's Economics Program developed
more relevant criteria to interpret agronomic experiments, using the same
criteria that farmers use when deciding whether a technology is a good
choice for them. A manual for the economic interpretation of agronomic
experiments was published by CIMMYT in 1976 (Perrin, et al.). The adoption
studies also revealed that technical scientists viewed ex post adoption
research as essentially negative historical commentaries. It became clear
to economists that they should shift their attention from ex post to ex
ante evaluation of prospective technologies. It was at this point that FSR
was seen to offer methods to describe and understand farmers' situations
and to bring this understanding to bear in carrying out ex ante studies.
A set of FSR procedures was drafted by members of CIMMYT's Economics
Program with three guiding principles in mind:
(1) FSR should have rapid turnaround time in order to fit in with the
annual planning and replanning cycle for experiments.
(2) FSR should get researchers on farms to improve the linkage be-
tween researchers and farmers.
(3) FSR should minimize professional manpower.
This second principle was consistent with a longstanding CIMMYT com-
mitment to carry out experiments on farmers' fields and to enlist the
participation of farmers in both sides of the Adaptive On-Farm Research
Cycle. The steps in the adaptive research cycle are listed in Table 1.
Steps 1-3 are preparatory steps which establish a framework for Adaptive
Research. Steps 4-17 describe the implementation of the cycle within a
Recommendation Domain; 4-11 are the diagnostic side of the cycle, 12-17 the
The time required, as set out in Table 1 for each step, assumes that
experienced professionals are implementing the procedures as part of an
ongoing program and that adequate logistical support, particularly trans-
port, is available. Where new Adaptive On-Farm Research programs are being
established and professional staff are undergoing training in procedures,
progress will be slower.
Procedural Steps in the Adaptive Research Cycle
1. Identification of the general region of the country with a priority
for adaptive on-farm research
2. Collation of secondary information on the natural and economic
conditions of the region
3. Identification of recommendation domains and target groups of
farmers operating a homogenous farming system within the region
4. Review of available background information on the recommendation
5. Informal survey
Discussions among farmers
6.. Design of formal verification survey
Sampling and fieldwork plan
7. Enumerator training and questionnaire testing
8. Formal verification survey--administration of the questionnaire to
a sample of target group farmers
9. Post coding, tabulation of data and specification of analyses
10. Data input and processing
11. Data interpretation and experiment planning
12. Selection of representative farmers and sites for on-farm experiments
13. Preparation and layout of experiments
14. Supervision and recording of experiments
15. Harvesting of experiments, measurement and
recording of yields
16. Statistical and economic interpretation of data
17. Planning for the next season's experiments
2. INTRODUCING FSR PROCEDURES IN EASTERN AND SOUTHERN AFRICA
CIMMYT has established four regional Economics Programs to promote
FSR procedures in cooperation with national agricultural research services
in the Third World. The first of these was started in Eastern Africa in
1976 with a headquarters in Nairobi and funding from the UNDP.
The steps in the Adaptive On-Farm Research cycle have uncompromising
(1) Adaptive On-Farm Research can only be done effectively with a
particular farm situation, and therefore an identified target
group of farmers, in mind.
(2) Disciplinary and commodity-based research institutions are often
not conducive to farmer-oriented experimentation. A compromise
is required on specialists' technical viewpoints if experiments
are to address farmers' priority needs.
(3) Experimental results must be achieved under conditions as close
as possible to those under which the farmer will implement them
to form the basis for sound recommendations.
The full organizational implications, presented cold, could be unaccept-
able to an existing research institution.
With this in mind, the full cycle of steps can be viewed as a
procedural and institutional reform package. CIMMYT's regional economists
have been free to develop their own strategies for diffusion. Promotion
has concentrated on those components of the cycle; first, which seemed most
appropriate to the circumstances of the national agricultural research
service concerned; second, which were best suited to the background and
experience of the individual economists.
CIMMYT's Eastern African Economics program has concentrated on the
promotion of FSR methods through a two-stage strategy. First, in countries
where agricultural research administrators expressed an interest in ex-
ploring new methods of bringing research closer to small farmers, a demon-
stration of the FSR diagnostic procedures was mounted in a maize or wheat
growing area of their choice. Usually an area was chosen where there were
no research recommendations, or where administrators felt the current
recommendations were not being accepted by farmers. CIMMYT worked with
local professionals in both the biological and economic sciences in
organizing and carrying out a demonstration of FSR diagnostic surveys. In
these demonstrations, the CIMMYT procedures were implemented within a
homogenous target group. An adaptive on-farm experimental program,
tailored to produce new techniques to solve problems and exploit opportu-
nities in the system, was drawn up based on the results of the demonstra-
The initial FSR demonstration in each country was written up as a
formal report to promote interest in the new procedures on the part of
research administrators and local professionals. Although CIMMYT has usu-
ally paid a major part of the costs of these initial FSR demonstrations, it
has established a policy whereby subsequent activities would have to be
funded by national sources, in an attempt to ensure that further coopera-
tion is more than a gesture. The Eastern African Economics Program's
annual budget of US $125,000 is regarded as 'pump priming'. In cases where
the administrators of national agricultural research services were satis-
fied with the outcome of the first stage--FSR demonstrations--then CIMMYT
was invited to pursue the second stage.
The second stage of the Adaptive On-Farm Research cycle was the place-
ment of economists in national agricultural research services. In coun-
tries in East Africa where administrators have recruited graduate staff as
trainee Farm System Economists, CIMMYT has helped to build up their profes-
sional capacity through in-service training programs supplemented by for-
mal training. In cases where students have been sent for overseas train-
ing, the Ford Foundation has provided the funding.
CIMMYT's strategy for introducing economists into agricultural re-
search services in Eastern Africa was guided by previous experience in
other parts of the world which had underscored two problems: First,
historically economists had carried out ex post studies of agricultural
technology--"coming back from the field" to inform biologists that they
had "got it wrong." Such an approach was not only unconstructive but also
built up antipathy in research establishments toward these "commentators."
As a result, initial CIMMYT strategies emphasize a positive approach,
presenting FSR as a methodology for improving the performance of the re-
search system rather than a remedy for its failure. Second, economists in
agricultural research establishments have often been isolated in their own
units or sections and economists have usually passed their operational and
methodological problems through three levels of decision making within the
technical establishment for decisions. At each level, an attitude of 'what
is this odd fellow doing in agricultural research anyway' often brought
decisions inimical to the effective operation of the economist. CIMMYT has
recommended two strategies to overcome the isolation of economists:
1) the avoidance of setting up separate Economics Units on research sta-
tions; and 2) the establishment of a direct professional link with senior
research administrators through the appointment of a senior economist
close to the Director of Research.
In mid-1978, the move began to introduce economists into agricultural
research institutions in several countries in Eastern Africa, hence avoid-
ing the establishment of "economics units." The following description of
the initial strategies followed is from the working paper to guide imple-
mentation efforts, written in 1978.
Initial objectives will be:
(1) To teach the FSE the tools of his trade and expose him to experi-
(2) For the FSE to gain credibility with the technical scientists on
research stations that he can improve the relevancy of the ex-
perimental programs and research recommendations to farmers'
needs and to national priorities.
(3) To foster the ideas of target group focused experimentation and
farmer participation in the research process.
The work program strategy will depend on the local situation. So far,
in Eastern Africa, technical scientists have little or no idea of how a FSE
can help the relevancy of their work. Indeed, relevancy, except as "a good
thing," is a hazy concept, a prescriptive mentality often still dominates
technical research work. On the whole, strategies can be divided positive
and negative. Positive strategies should normally be the basis of the
effort made at research stations. Unless used carefully, negative strate-
gies will enhance the barriers between technical and economic researchers.
However, if used carefully, negative strategies--essentially criticism of
past or present work--can increase the awareness of the technical scien-
tists of what is and what is not relevant. These strategies, with emphasis
to be placed where possible on the first positive one, are listed below.
(1) To interest individual technical scientists on stations in the
circumstances and priorities of farmers who will be offered the
results of their research. The FSE begins to feed information on
priorities and circumstances of target group farmers to selected
individual scientists outlining the implications for experimen-
tal content. Technical scientists who appreciate the need for
relevancy and express an interest in cooperation are the ideal.
(2) To review past research results and, by economic interpretation,
demonstrate how modified recommendations are more consistent
with farmer circumstances. Though a critical strategy, it has
proven positive in practice. It can give immediate benefits in
revising extension recommendations either by omitting components
which are unacceptable to target group farmers, or by modifying
treatment levels to bring them within the reach of farmers. The
more relevant improvement packages build credibility for the
economist with the extension service and with research adminis-
trators. It is less painful if the results and recommendations
reviewed are from the work of the past researchers and not pres-
(3) Questioning the relevance of ongoing research programs.
(a) By asking who, which target group of farmers, is the work
being done for. Making a strong issue of local specificity
(b) By asking whether target group farmers will be able to
absorb the level of changes implied in the management vari-
ables being tested.
Both types of questioning under (3) are immediately critical of current
research efforts and therefore of the station management and technical
scientists. The questions are justified if relevancy is a major issue and
sympathetic individual technical scientists cannot be identified. Both
types of questioning should be specific, and reviewing experimental pro-
grams and investigating target group circumstances will be a prerequisite
for the FSE before embarking on detailed discussions along these lines with
In outline, this is how CIMMYT Economics Program became involved with
3. FSR PROCEDURES DEVELOPED BY CIMMYT
FOR CARRYING OUT ADAPTIVE AGRICULTURAL RESEARCH
The FSR procedures developed by CIMMYT are low cost and have a rapid
turnaround in order to accommodate the scarce manpower situation that
exists in many research establishments in LDCs. The first step is to
identify target groups of farmers for whom the same research and develop-
ment effort will most likely be relevant. CIMMYT calls these target groups
"Recommendation Domains." The second step is to undertake a diagnosis of
the farming systems in the Recommendation Domains which have been selected
as a priority for intensified research and development (Byerlee, Collinson
et al., 1980). The diagnosis sequence concludes with a planned set of
on-farm experiments. In the second half of the Adaptive Research Cycle,
the experiments are carried out under the operating conditions of the
Domain to ensure that recommendations emanating from the work will be
accepted by farmers.
3.1 Identifying Recommendation Domains
Inevitably, there has to be a compromise between research for an indi-
vidual farmer, which is far too expensive, and for the heterogenous condi-
tions of the country as a whole, which is far too generalized. A framework
of Recommendation Domains forms the basis for research planning and for
identifying priority adaptive research Foci. In order to identify recom-
mendation domains, farmers are grouped into relatively homogenous Domains
according to their existing farming systems. The farming system is used as
a basis for an initial grouping for three reasons:
(1) The existing farming system is a manifestation of a weighted
interaction between natural, economic, and cultural circum-
stances and his own priorities and resource capabilities. The
present farming system reflects, better than any single criteri-
on, the balance of factors important in identifying homogenous
groups of farmers.
(2) Farmers operating a similar system have similar problems and
(3) The existing farming system is the starting point for develop-
ment--the base onto which productivity improvements have to be
A short questionnaire is developed to collect descriptive information
about farmers, with the smallest local administrative division as the
enumeration unit. A sample questionnaire is shown in Figure 2. It draws
upon the experience of educated local people. Where extension staff in the
Ministry of Agriculture are used, foreknowledge of their likely biases
should help ensure balanced information. A prior review of ongoing credit
and extension programs is useful for identifying biases. For example, if
the investigator knows in advance that 5 percent of farmers in the area
have received credit for the purchase of dairy cows this information can be
used to evaluate the validity of information supplied by extension agents.
A. Animal kept by 1. Two main types 1
most farmers of animals kept 2
2. If cattle main 1
purposes for 2
B. Foods grown (G) 1. Starch staples 1
or bought (B) by ranked in order
most households of importance
2. Relish crops to 1
flavour staples 2
3. Animal products 1
for food 2 Overall
C. Main cash sources 1. New cash crops 1 ranking
for most farmers and % growing 2
(1st, 2nd, 3rd) 2. Crop sales as 1
cash source 2
3. Livestock as a 1
cash source 2
4. Off farm 1
cash source 2
D. Land use methods 1. Years cultivated 1
and time of 2. Typical area (Ha) 1
most farmers 3. Main methods of 1
land preparation 2
4. Main months of 1
land preparation 2
E. Hire and purchase 1. Types of hired 1
of resources by labour & payment 2
2. Work done by 1
3. Main inputs 1
Figure 2. Farm System Zoning Questionnaire:
Central Province, Zambia
The aim in data interpretation is to identify different farming sys-
tems. The key step in interpretation is to identify dominant enterprises,
timing of activities, and husbandry methods, which play a significant role
in shaping the farming systems. This methodology provides a low-cost means
of identifying Recommendation Domains of relatively homogenous groups of
farmers. The questionnaire in Figure 2 was used over a population of 75
wards (the smallest administrative division) in Zambia. Seven Recommenda-
tion Domains were identified in a three-month period covering some 70,000
farm families. In terms of professional time, three stages can be distin-
guished in identifying recommendation domains:
(1) Preparation (10 man-days)
(a) Developing and testing the questionnaire.
(b) Arranging the program of visits.
(c) Preparing background material and maps.
(2) Data collection (10 man-days)
(3) Identification of Recommendation Domains (15 man-days)
(a) Interpreting the data.
(b) Specification of Domains, preliminary mapping of the geo-
graphical boundaries, and description of their hierarchical
It is important to emphasize that these are preliminary groupings. As
the diagnostic sequence is implemented within these identified target
groups, the characteristics of each domain will be more fully understood.
For example, variations in natural potential, such as amount of rainfall,
may have been overshadowed by other factors in the initial identification.
New technologies, however, may allow differential exploitation of this
variation, and the preliminary domains, based solely on the homogeneity of
the existing systems, will need further division.
3.2 Diagnosis of Farming Systems
After the recommendation domains have been identified, the next step
is to carry out a diagnostic sequence in selected priority domains in order
to identify key farmer problems and some preliminary ideas on how to help
solve these problems. This information will provide the content for Adap-
tive On-Farm Experiments in the second half of the cycle. An example may
help clarify the adaptive research process. In some semi-arid areas in
Eastern Africa, farmers have major problems in producing a reliable supply
of their staple foods such as maize. Researchers often recommend they
'should' plant before the rains start to give the crop full use of limited
moisture, but their oxen are often in such poor condition that the animals
cannot break the land. Hence, farmers frequently wait until the condition
of their oxen improves which occurs when the oxen feed upon the new grass
brought by the same rain which softens the land and reduces the power
required to break it. "Plant on time before the rains" addresses a symptom
and not the problem. Hence, farmers often compromise and plant the maize
crop late, a situation they are forced into because their animals are in
poor condition at the end of the dry season.
The 'Body of Knowledge' from technical research in a variety of disci-
plines can be tapped to try to solve the interrelated problems of the weak
condition of animals, delayed planting, and unstable maize yields. The
'Body of Knowledge' listed below provides examples of potential solutions
to these problems.
(1) Animal Production
(a) Improve the feed supply in the dry season
(i) By planting types of grass which will persist
throughout the dry season
(ii) By planting high bulk forage grasses in local low-
lying areas where there is residual moisture
(iii) By purchasing feed
(iv) By underplanting grain crops with a forage legume
(v) By selecting grain varieties and management prac-
tices which emphasize the use of residues for fodder
(b) Reduce stocking rates and thereby reduce competition for
dry season grass
(i) By providing artificial insemination services which
reduce the need to carry male animals
(ii) By opening up market opportunities for male calves
(iii) By encouraging the use of cows for draught purposes,
thus reducing the conflict between keeping oxen for
draught and cows for milk
(iv) By using A. I. services to improve the conformation
of cows for draught purposes.
(a) Yokes giving improved power transfer
(b) Equipment requiring less power to break the land
(3) Plant Breeding
(a) A shorter-term maize variety which can be planted late
(b) Alternative crops known to be less susceptible to drought
(a) Minimum tillage techniques which, with or without the use
of herbicides, will allow early planting
(b) Land preparation at the end of the previous rainy season
while the land is soft and animals are in good condition
These techniques, in contrast to unhelpful advice to plant early, are
positive and potential methods of solving problems of local farmers. Fur-
ther understanding of the farmers' present system is needed to determine
which of these apparent solutions have resource demands that are within
the endowment of target group farmers and are compatible with the resource
allocation requirements of competing enterprises.
Implementing some of the above suggestions will lead to the situation
where the classic output per unit of area criterion of researchers is
compromised in return for lower but more stable grain yields. For example,
underplanting grain with a forage legume, selecting grain varieties for
their fodder output, or using a shorter-term maize variety may improve the
reliability of yields but may actually reduce average grain yields. Simi-
larly, animal equipment requiring less power, minimum tillage techniques,
and winter ploughing may also compromise grain yields. But this is an
example of how a research system can address a real problem of farmers in
Eastern Africa--achieving a reliable supply of their staple food.
3.2.1 Step 1: Collecting Background Information on the
Circumstances of Target Farmers
Diagnosis of the farming system has three major steps. The sequence
acts like a sieve. It progressively sifts out unimportant facets of the
system and, with them, irrelevant research issues. Each step funnels the
diagnosis towards key circumstances of target group farmers, key problems
of their farming system, and the identification of appropriate development
opportunities. The first and second steps use cheap diagnostic procedures
to focus subsequent, more expensive procedures onto key issues.
As shown in Table 1, the first step for the Adaptive On-Farm Research
Teams is to collate secondary data on the natural and economic circum-
stances of the target farmers in order to identify management problems
posed by the local circumstances. The focus of enquiry is on the resource
allocation decisions of the farmers. The following types of secondary
information can be helpful to the research team:
(1) Natural circumstances
(a) Rainfall, amount, and reliability to determine
(i) The length and timing of the growing season
(ii) Uncertainties caused by periods of excess rain or
(b) Seasonal temperatures, to indicate
(i) Growth rates of plants
(ii) Frost incidence
(c) Soil characteristics and topography, to determine
(i) Potential erosion and soil fertility problems
(ii) Possible locational advantages and disadvantages,
flooding, and the use of residual moisture
(d) Pest and disease incidence as a source of uncertainty of
Rainfall uncertainty offers a very common example of the useful-
ness of this kind of review. If, in examining the monthly
rainfall probabilities, the researcher finds that the third
month of a five-month rainfall season is unreliable (e.g., it
has 4 out of 10 years below 70mm of rain), he has identified a
major management problem for local farmers. The management
strategies used by farmers to meet this uncertainty will be
examined in the second diagnostic step, the Exploratory (Infor-
(2) Institutional circumstances
(a) Types of marketing and supply channels
(b) Types and reliability of food distribution channels
(c) Extension and credit programs: the types of programs, the
numbers and types of participants
(d) Land tenure arrangements
(e) Farmer groups, whether voluntary, organized, official or
unofficial, and their planned and actual functions
This information describes the available physical infrastruc-
ture. Information on marketing and government intervention pro-
grams allows some assessment of the content and degree of pene-
tration of the local institutions, while information on land
tenure and farmer groupings describes part of the institutional
environment which will influence farmer decision making.
(3) Economic circumstances
(a) Population numbers and density, and the pattern of settle-
This information gives an indication of the scope for the inten-
sification of production. Different settlement patterns have
different implications for access to water and grazing.
(b) Available acreage and production figures
These figures give an idea of the relative importance of the
various enterprises in the Farming System.
(c) Marketed products
(i) Volume and trends for outputs sold and inputs pur-
(ii) Foods purchased, relative volume and trends over the
years and between seasons
(iii) Prices, trends over the years and over the calendar,
and marketing margins
Information on marketing helps to highlight how things are
changing in local farming communities.
The review of secondary information is a two- or three-day task for the
research team. Where data is limited, as it often will be, it should be
supplemented by extra time in the target group area discussing these same
aspects with businessmen and local officials. The review aids in under-
standing how the local production environment influences the management of
3.2.2 Step 2: The Exploratory (Informal) Survey
The Exploratory (informal) Survey is the pivotal procedure in the
three major steps in the diagnosis of the farming systems. The survey is
carried out by a team consisting of an economist, agronomist, and other
specialists who talk with target group farmers on their farms over a period
of ten to fifteen days. Through this process, the economist comes to
understand the farmer's perspective on production decisions and the bio-
logical researchers identify major shortcomings and compromises in manage-
ment that appear to under-exploit the biological potential of the area.
Interaction between the economist and biological scientists identifies new
management practices which would better exploit biological potential and
help farmers meet their priorities from within their resource endowments.
The Content of the Exploratory Survey
The first step in understanding the problems of target group farmers
is to realize that empirical research throughout the Third World has shown
that farmers are rational and purposive in their behavior to satisfy the
(1) The social and cultural obligations of their community
(2) A reliable supply of preferred foods day in and day out
(3) A supply of cash to provide additional basic needs
(4) Extra cash
Although the order of these priorities is fairly firm, the weighting of
each of the four will depend on how close to the subsistence level the
farmers are operating and the cultural milieu. At the subsistence level,
(1) and (2) will normally be very dominant and a farmer will pursue manage-
ment strategies to satisfy these priorities.
A systems perspective is used in the exploratory survey in order to
understand how farmers choose enterprises, growing period, and when and
how they use the product. This information gives the analyst a preliminary
weighting of farmers' priorities and an understanding of decision criteria
for each enterprise. There will be seasonal changes in the relative size
of enterprises, in the timing of enterprise activities, and in the use of
the output. Manipulating these factors gives farmers the flexibility to
react to seasonal changes in natural and economic circumstances.
The understanding of the priorities of farmers and the way these are
met is deepened by observation and discussion with farmers. Farmers have
limited amounts of resources at their disposal. The pattern of enterprises
over space and time, and description of the production methods used for
each, allow an assessment of how they have allocated their resources in
order to satisfy priorities. For this, the analyst needs a knowledge of
the labor and cash requirements for alternative techniques of production
for different enterprises. For example, hand hoeing may require 40-50 man-
days per hectare. The FSE needs this kind of input information to con-
struct approximate land, labor and cash use profiles for the farming system
by relating the size of each enterprise, the pattern of each enterprise
activity over the year, and the production techniques used in each enter-
prise activity. Each technique implies a resource requirement; the timing
of its implementation denotes the farmer's commitment of those resources
to that enterprise at that time.
By relating resource allocation to endowments, the economist begins
to understand the limitations on the farming system and its development.
The limiting factors represent management problems to the farmer and also
leverage points for introducing new materials and techniques to improve
the system. The agronomist uses his disciplinary knowledge to understand
the ideal technical management of the key enterprises under local condi-
tions. He then identifies agronomic practices used by farmers and records
this information in his field notes.
The day-to-day interaction between the adaptive research team need
not result in the preparation of reports or even field notes. Whether the
sequence of collecting information is more or less formal, it has the same
essential steps: identification of problems of farmers, and the appraisal
of potential solutions by both the agronomist and the economist. The
economist and the agronomist may feel a need to call in specialists such as
Entomologists, Veterinarians, Animal Production Specialists, Anthropolo-
gists, and Sociologists. The information gained by the adaptive research
team forms the content for subsequent On-Farm Experimentation with a range
of new agronomic practices and materials including new varieties, fertil-
izer, pesticides, machinery and equipment, and even new enterprises. New
practices could cover one or any combination of the management components
involved in producing one or more enterprises: seedbed preparation, time
and method of planting, crop arrangement in the field, weeding timing,
frequency and intensity, and many other components relevant to crop pro-
duction. Innovations in livestock enterprises could include housing,
breeding, health, and feeding management.
Carrying Out the Exploratory Survey
The Exploratory Survey is carried out by the economist and agronomist
through unstructured interviews with target farmers within the Recommenda-
tion Domain. The interviews are best conducted during the growing season
and preferably one to two months before the harvest period. The discus-
sions with farmers represent a recursive learning process. Guidelines
setting out the areas of information which may need to be covered are
divided into "bite sized" sections. The researchers interview a farmer on
one or more sections and make field notes; they may interview the same or
different farmers on the same sections. At the end of the day, after
talking to perhaps three farmers and covering various sections of the
guidelines, researchers meet as a team and evaluate the information ob-
tained. This interaction between researchers at the end of the day is
essential in furthering their understanding of local farming. Farmers are
further interviewed on the same sections until researchers are confident
that they understand those aspects of local farming. At this point, each
researcher makes detailed notes on each heading of that section of the
guidelines. Gaps in the information are identified and filled in through
subsequent interviews. Wherever possible, interviews are carried out in
farmers' fields, with visible evidence of farmers' management before the
researchers. These visits present an opportunity for researchers to in-
teract with their clients.
The output of the Farming Systems Economist from the exploratory sur-
vey is in two parts. The first part consists of field notes organized in
the format of the Detailed Guidelines (see Annex. 1) which covers the
sections on systems. The second part is a "System scenario" which:
(1) Describes the enterprise pattern and output uses; sets out cal-
endars for crop production, human food use, and livestock food
use; and discusses variation in these features year to year;
(2) Specifies what farmers are trying to achieve with each enter-
prise and what decision criteria they use;
(3) Specifies the longer-term and the seasonal constraints on the
expansion of the farming system;
(4) Explains management strategies used by target group farmers to
achieve their priorities
(a) In the face of uncertain natural and economic circum-
(b) In the face of their resource limitations;
(5) Explains how these strategies influence production methods, in-
cluding the timing of production activities, for the major en-
(6) Specifies major points of leverage on the system in order to
guide the search for new materials and practices.
The agronomist's output is Field notes, organized in the format of the
Detailed Guidelines and covering the sections on husbandry, repeated for
the key enterprises. The notes should detail the compromises made between
present practice and what the agronomist judges as the ideal technical
practice, given local circumstances of climate and soil.
The joint output of the Exploratory Survey is a recommended set of on-
farm experiments which the agronomist believes will better exploit bio-
logical potential and which the economist perceives will be compatible
with the goals and resources of the farmer.
3.2.3 Step 3: The Verification (Formal) Survey
The Exploratory Survey is followed by a verification or formal survey
to determine whether the understanding gained from informal discussions
with selected farmers is indeed valid for the target population of the
Recommendation Domain as a whole. This survey may also seek deeper under-
standing of particular facets of the system and may measure those para-
meters crucial to the efficient planning or interpretation of experiments.
The initial facet to be verified is the homogeneity of the Recommenda-
tion Domain. Survey questions cover the variables by which the Domain has
been specified and demarcated. The responses are tabulated to show the
sample distribution across these variables. Hypotheses are explored for
subpopulations which emerge, and a redefinition of the boundaries of the
Domain may be necessary. Important facets to be verified are farmers'
priorities, decision criteria for each enterprise, resource limitations,
the degree of uncertainty, and the relationship of these facets to resource
Measurement of Inputs, Outputs, and Other Parameters
A major consideration in developing the sequence of FSR procedures
has been to stress rapid turnaround of information, and to keep costs low,
especially in terms of professional time. The need for measurement of
parameters has been minimized in the surveys carried out to date in Eastern
Africa (see CIMMYT, 1977). The main analyses are the computation of
frequency distributions across the population of those characteristics
being verified by the survey, and cross-tabulations to explore hypotheses
or differences among subpopulations. Usually the viability of a proposed
set of experimental treatments is contingent upon interactions within the
system. As a result, the measurement of some of these interactions may be
crucial for a proper interpretation of experimental results. Three exam-
ples of important types of interactions which may need to be quantified in
the verification survey are:
(1) The amount of seasonal labor required to implement improved man-
agement techniques in the experimental enterprise;
(2) The amount of cash needed to purchase inputs for the experiment-
al enterprise; and
(3) The impact of labor released by the use of a new practice--e.g..
herbicides for weeding--on the rest of the system.
The measurement of these interactions can be achieved by methods
consistent with the goals of low-cost and rapid turnaround. Cost route
data collection, based on frequent visits to the farm throughout the sea-
son, cannot be justified for the verification survey. For example, season-
al labor peaks are a major constraint on improving the farming system of
many farms in Eastern Africa. Such peaks limit either the area cultivated
or the standards of husbandry applied and thereby often limit the profita-
bility of purchased inputs. Quantification of the enterprise combinations
that lead to seasonal lead peaks enhances the planning and interpretation
of on-farm experiments. Researchers can use labor input coefficients from
cost route surveys carried out in comparable farming circumstances. If
credible labor input data is not available, it can be collected in the
course of the Verification survey. A questionnaire designed to collect
labor input data through a single visit to a sample of farms is shown in
CROP: ESTIMATED SIZE OF FIELD: HA.
Daily No. Days
Month Week Hours Taken for
Operation Done 1--4 Normal Work Group in Field Field
A Single Visit Crop Labour Input Questionnaire
The better farmers in any population already may have identified many
practices to improve their farming system. These practices can be identi-
fied in the verification survey. Farmers who appear to have developed
management strategies which relieve constraints should be followed up af-
ter the verification survey, with a view to understanding how their solu-
tions could be incorporated into on-farm experiments and subsequently dif-
fused to other farmers.
A single- or perhaps double-visit survey methodology is used for the
Verification Survey. Since a Recommendation Domain has a homogenous tar-
get population, it will only be necessary to include about 30 farmers in
the verification survey. To allow for the emergence of subgroups and the
possible need to redefine the Domain, fifty or sixty farmers will be safer.
The verification questionnaire is developed from the output of the Explo-
ratory Survey and it is completely location-specific and highly selective
in content. Many of the foci--for example, farmer priorities, resource
limitations, and management strategies--will be common across many target
groups, and standard question sequences can be employed to elicit the
information. On the other hand, the content, i.e., the enterprises fol-
lowed and the actual practices used, will vary across Domains.
As always with a single- or double-visit survey, it is important to
enlist the support of local leaders and ensure that the selected respon-
dents are participating in an activity which has the backing of the commun-
ity. Training of enumerators, careful protesting of the questionnaire,
and a level of supervision which evaluates each completed questionnaire
while still in the field, are all essential for achieving good quality
data. Hand tabulation and analysis of initial surveys will enable re-
searchers to gain familiarity with the data. But the formal survey lends
itself to easy computer processing.
3.3 On-Farm Experiments
The second half of the adaptive cycle, the On-Farm Experiments, are
planned on the basis of the information collected in the sequence described
above. The output from the verification survey is the verified content
which will be used to design a set of on-farm experiments. These experi-
ments will include management changes which the biologist believes will
improve the exploitation of local biological potential and, which, at the
same time, the economist believes will enhance the achievement of farmers'
goals. The range of levels over which these variables will be tested is
decided by the farmers' flexibility in managing the resource allocation
implicit in the changes. For example, where it is found thaf animal manure
is cheaply available to local farmers and the best way to maintain soil
fertility, the amount of manure to apply and the timing of the application
might be the two major variables studied.
The levels of farmyard manure to be applied will be dictated by the
availability of manure to target group farmers. The timing of application
will be decided by other factors, such as when the manure is to be applied
during the rotation, when labor is available during the cropping year (the
carrying and application of manure are labor-intensive), and the best
timing from the standpoint of the quality of the manure under present
methods of storage. All such information is fed from the exploratory and
verification surveys into the planning of on-farm experiments. Once the
use of manure emerges as a relevant experimental variable in the course of
the Exploratory Survey, further discussions with farmers will focus on
these and other factors that must be considered before on-farm experimen-
tation can take place. The level of non-studied variables is of equal
importance in planning relevant experiments. In traditional on-station
experiments, non-study variables are often fixed at levels which will not
inhibit the effects of the study variables; complete insect control, un-
limited labor for weeding, and optimal timing and method of planting are
some common examples. Clearly, these may be completely irrelevant to
Domain farmers. Hence, the variables to be studied in on-farm experiments
are selected with the farmers' management ability as a paramount consider-
Other variables have been rejected because the resource allocations
implied are either beyond farmers' endowments or would precipitate major
clashes with allocations in the present farming system aimed to meet farmer
priorities. The On-Farm Experiments must show whether productivity will
increase if the changes represented by the study variables are introduced
into farmers' management regimes. Thus, non-study variables and the con-
trol treatment in the experiments will be at farmers' existing management
levels; these are the only relevant baselines from which to measure im-
provements in performance. The Verification or Formal Survey will identi-
fy criteria to be used in selecting representative farmers and representa-
tive locations for the On-Farm Experiments. The farmers themselves will
best be able to determine the management of the non-study variables appro-
priate to the circumstances of the season and location of the experiments.
Similarly, the Formal Survey identifies the sources of costs and benefits
within the farming system which determine the viability of the management
changes for Domain farmers. Consideration of all relevant sources of costs
and benefits allows proper interpretation of the on-farm experimental re-
4. PROGRESS IN INSTITUTIONALIZING FSR PROCEDURES
IN NATIONAL AGRICULTURAL RESEARCH SERVICES
CIMMYT's Regional program was established in Nairobi in January 1976
with funding from the UNDP. Phase I covered the 1976-1979 period. Follow-
ing a program review in late 1978, a Phase II was funded to the end of 1983.
Demonstrations of the FSR procedures were the main program activities
carried out through mid-1979, when the emphasis switched to in-service
training of young FSR researchers. However, with a recent proliferation of
donor-financed FSR programs (at least twelve planned, from the Sudan south
to Botswana), CIMMYT's economists may increase its support for the plan-
ning and implementation of donor-financed FSR programs.
CIMMYT's Eastern African Economics Program was mandated in Phase I to
work in six countries: Kenya, the host country; Ethiopia; Uganda; Tanza-
nia; Zambia; and Malawi. Phase II extended this mandate to all interested
countries in Eastern, Central, and Southern Africa.
CIMMYT economists discussed FSR with senior agricultural civil ser-
vants, including research administrators, in Kenya, Uganda, Ethiopia, and
Tanzania in 1976, in Zambia and Malawi in 1977, and in Zimbabwe in 1980.
CIMMYT has provided assistance to national research services in Kenya,
Tanzania, Zambia, and Zimbabwe. Consultation and cooperation in training
has extended to 13 countries through the region. CIMMYT's experiences with
promoting FSR over the 1976-81 period are discussed below.
Almost inevitably, more effort has been put into Kenya, the host
country, where there is a strong political and policy emphasis to increase
the prosperity of the mass of the rural population. FSR procedures were
demonstrated in two areas in 1976: one in Siaya District on the shores of
Lake Victoria, the other in Kwale District on the Coast. Since Kenya did
not have any farm economists in its Scientific Research Division of the
Ministry of Agriculture, economists from Egerton College and biologists
from the research services helped carry out the FSR demonstrations. The
Siaya demonstration was written up as a formal report and was well received
by senior research administrators in 1977. A one-day seminar for senior
research staff at the Faculty of Agriculture, University of Nairobi, in
June 1977, provided an opportunity to discuss the procedures and the con-
cepts behind the demonstrations. Other demonstrations were undertaken and
two examples show how the use of FSR procedures help to orient research
(1) In Siaya, Western Kenya, hand cultivation predominates and maize
plantings are spread over a period of three months. Plantings
made over the last half of this period are susceptible to failure
due to the ending of the rains. An earlier maturing variety,
although not exploiting biological potential as effectively,
would be a more flexible management tool for farmers, given
their resource constraint on seedbed preparation. An earlier
maturing variety would also enhance the reliability of produc-
tion of their staple grain--a major priority of small farmers
(2) Vihiga, Western Kenya, is an area with small holdings and two
growing seasons a year. Over 75 percent of farmers were esti-
mated to be growing the recommended maize hybrid (H 614C) in the
long rainy season. With a 180-day maturity period, this hybrid
was often kept in the field until August or September. Because
land is scarce, the second rains crops must be planted on the
same land. Even the shorter-term local maizes cannot be estab-
lished and grown properly in the second rains unless they are
planted by early August, due to dry weather in January. The
focus of the FSR surveys was to determine how to increase maize
production from the system as a whole. A possibility is to grow
a shorter-term hybrid in the long rains and harvest it in July to
allow timely preparation for the second season and enhance the
viability of using purchased inputs for this second crop. At
present, this short rains crop is considered too unreliable to
risk the use of purchased inputs. A second possibility is to
retain the higher potential, longer-term variety for the long
rains and to experiment by relay planting it with the second
rains crops (Lihanda, 1978).
Following the FSR demonstrations and review of the findings in early
1978, the Government of Kenya asked CIMMYT for help in building a capacity
to use FSR procedures for planning research experiments within the
agricultural research services. An Economic Advisor to the director of Re-
search was appointed and he cooperated with CIMMYT in developing an in-
service training program. Six agricultural graduates, two of whom had a
Master's degree in Agricultural Economics, were recruited in August 1978
and six more were recruited in August 1979. It was planned that the
agricultural graduates would spend 12-18 months on research stations to
gain experience with diagnostic survey procedures and some familiarity
with experimental methods. During this period, they were helped in plan-
ning their work programs and in implementing field work. The trainees also
attended two week-long residential workshops on FSR methods, mounted
jointly by CIMMYT and the SRD. After this in-service training period,
those holding first degrees were sent for Master's degree training in
When a Farm System Economist (FSE) trainee was posted to a research
station, he was introduced at a meeting of the station scientists and his
possible contribution to the work of the station was discussed. During the
meeting, agronomists were identified who expressed an interest in improv-
ing the relevancy of research for small farmers. The program planned for
the FSE trainee was aimed to dovetail with the ongoing experimental program
of an agronomist. The trainee's initial program of work was designed to
increase his ability to understand farmers' situations and to incorporate
this information into the planning and interpretation of experiments. Em-
phasis was placed on Exploratory Survey work, with some limited effort in
Domain identification and in Formal Verification surveys.
On the whole, good progress has been made in building up an FSR
capability among trainees. The best trainees have a good grasp of the
spectrum of methods available for data collection and their relative suit-
ability for various needs and situations. Most trainees are now able to
contribute to improving the planning and interpretation of experiments.
Two examples illustrate how this process worked:
(1) Exploratory Surveys of farmers growing intercropped maize and
beans in Eastern Kenya threw new light on the interpretation of
experiments in alternative mixture patterns. The surveys iden-
tified an acute labor shortage during crop establishment and
showed that returns to labor required to establish the crop
mixture would be a key criterion in appraising experimental
results. Labor data which was collected from large plot experi-
ments designed as a randomized block with five replicates pro-
duced results which are shown in Table 2.
Returns per Unit Area and to Planting Labour
Yield kg/ha Yield Value Labour US$/hr
System Maize Beans US$ per ha Hours/150 sqm Plant. Lab.
A: pure maize 4339 -- 542 3.58 2.24
B: pure beans -- 1723 754 9.39 1.13
C: recommended 3444 768 766 17.15 .63
D: farmers 2800 331 495 3.11 2.34
E: D modified 3231 371 566 5.48 1.55
Source: Gathee, 1981.
Recommended planting patterns for maize/bean mixtures (Treatment
C) require five times more planting labor than simulated farmer
pattern (Treatment D) which gave almost four times the return to
the planting labor used. In the farming systems of target groups
growing maize/bean mixtures, which have a short rainy season and
where land is not limited, there are often intense labor peaks at
the time of crop establishment. For such target groups, return
to seasonal peak labor used may be a more appropriate criterion
than return per unit area in comparing results from experimental
(2) Exploratory and Verification Survey work was carried out in an
area of Western Kenya with high population density and an acute
scarcity of land. The results revealed a marked interaction
between crop and livestock enterprises in the use of crop resi-
dues as byproducts for feeding local animals kept for milk. The
data are shown in Table 3.
Sources of Feed for Local Dairy Animals
% of Ranked in Importance (%) Less
Source of Feed Farmers 1 2 3 4 Important
Grazing: own farm 100 100 -
: commonland 79 25 4 4 46
: other farm 67 67
Maize : green stalks 96 42 42 8 4
: dry stalks 83 4 29 17 33
: leaves 25 25
Banana leaves & stems 42 42
Sugarcane tops 50 21 29
Source: Wangia, 1980.
The dominance of maize as a source of cattle feed, both green and
dry material, led to proposals for two adaptive experimental
programs which were designed to examine:
(a) What increase in maize plant population would be possible
so that fodder production could increase without penalizing
grain yields, in both the long and short rains, and
(b) The effects of alternative timing of picking the leaves and
tops of maize, on grain and fodder yields.
The second major leg of program strategy was to build up the credibil-
ity of the Farm Economist with technical researchers, particularly agrono-
mists. Here the program has had limited success. Many of the problems
encountered in establishing a close working relationship were features of
the research organization, particularly the strong compartmentalization,
upheld by everything from disciplinary loyalty to parallel compartmentali-
zation in the layout of government estimates and fund votes.
The authority of Research Station Directors, both to influence re-
search planning within disciplines and to control operations, was badly
underestimated. It had been foreseen that the cost of on-farm trials would
be an obstacle to the introduction of this component, but budgetary factors
also proved to be a barrier to moving Farm Economists and biologists from
research stations to work among farmers. Station Directors were loathe to
release transport and travel funds in disproportionate amounts to new,
junior research officers. Despite their involvement in briefing sessions
on all stations at which economists were introduced, Directors were often
not clear on or convinced of the role of the trainees. At the same time,
they identified 'money related' jobs on the stations to be the responsi-
bility of an economist. These jobs ranged from costing the station dairy
herd or vegetable production to assessing future market prices for a new
crop currently under observation. While an economist can carry out these
responsibilities, the danger is that he will be detracted from his role of
helping carry out adaptive research.
The marriage of the economist and the biologist has proved to be an
extremely difficult one to consumate in Kenya for the following reasons:
(1) The specialist orientation of the technical scientist was per-
ceived to be diluted and his professional peer group status
threatened if he cooperated with economists. Similar penalties
exist in working close to the farmer, where science is "less
(2) Technical researchers are locked into a series of experiments
which are programmed and budgeted over a period of years. Ad-
ditional commitments are difficult to reconcile with these pro-
grams and supplementary budgeting is an awkward procedure.
(3) When any series of experiments is concluded, the established
mechanism for deciding new commitments comes into operation.
Attempts by the economist trainee to override this creates con-
flicts within established channels. Hence, his interventions,
given his junior status, are usually squashed.
(4) The junior status of the trainee economist strengthens the im-
pression of "unjustifiable intrusion" in two ways: (1) Coopera-
tive efforts with "senior" biologists are awkward because the
trainee is the underdog, and (2) the trainee has a poor command
of his new profession and cannot put his case in a convincing
Overall, the introduction of FSR procedures into research planning
and interpretation seems to be seen as an attempt to promote the agronomist
and economist, perceived traditionally as playing service roles to disci-
plinary researchers, to central and pivotal roles. The establishment
feels threatened and the social scientist, seen as the intruder, is reject-
While these problems of implementation have occurred at the local
station level in the last two years, major institutional changes in agri-
cultural research have taken place in Kenya. In 1979, a reorganization
divided the Ministry of Agriculture into two ministries, with the new
Ministry of Livestock Development taking over responsibilities for live-
stock economists. A bill was passed by the parliament in 1979 to establish
the Kenya Agricultural Research Institute (KARI), a parastatal to carry
out all crop, animal, and forestry research. The new institute has not yet
taken up the full role intended for it. At the same time, not all the major
personalities presently in charge are convinced that social scientists and
FSR procedures have any role in agricultural research. Little progress has
been made in expanding the adaptive research capacity since early 1980.
When CIMMYT economists made contact with Zambian agricultural admin-
istrators in 1977, the administrators expressed a strong concern for the
need to assist small farmers because the historical emphasis on copper and
large commercial farms was not meeting food needs and the political needs
of the country. Administrators were aware that small farmers operated
under different circumstances than large farmers and that a new orienta-
tion to research and extension was necessary. The government was interest-
ed in CIMMYT's interdisciplinary approach and a demonstration was imple-
mented near Serenje, a small town in the maize growing area in the Central
Province 300 kms north of Lusaka. The demonstration was carried out by an
economist from the Rural Development Studies Bureau--a research unit at-
tached to the University of Zambia--and biological scientists from the
Central Research Station at Mount Makulu. The field work was carried out
in the first half of 1978 and the final report was accepted by a meeting of
the Program Steering Committee in December 1978. The Committee then
charged CIMMYT with the responsibility to develop a capacity for imple-
menting FSR within the research services. Two new graduates were recruited
for the program and joined the research services as trainee FSEs in
July 1979. Throughout, CIMMYT has been responsible for helping to plan and
supervise their work programs and has sponsored their participation in the
in-service training workshops mounted in association with the Government
of Kenya. At the request of the Steering Committee, a further demonstra-
tion was implemented during 1979 which divided Central Province, one of
nine major administrative units in Zambia and covering some 116,000 square
kilometers, into Recommendation Domains. The final report showed the
potential uses of a Recommendation Domain framework in research and devel-
During 1980, a plan was drawn up to establish a two-level hierarchy to
strengthen agricultural research in Zambia--Commodity Research Teams and
Adaptive Research Planning Teams (ARPT). Under this plan, Commodity Re-
search Teams, working from established research stations, would be comple-
mented by Farming Systems teams which were called Adaptive Research Plan-
ning Teams. A Senior Farm Economist was appointed at the end of 1980 to
coordinate the development of the ARPT Program. Currently, three agrono-
mists and one trainee economist are attached to the ARPT program. The
current pre-occupation of the ARPTs is to identify Recommendation Domains
for two more of the nine Provinces and to implement Exploratory surveys to
be used in planning on-farm experiments for three Domains during the
1981/82 crop season.
The decision to adopt a two-level hierarchy--Commodity Research Teams
and Adaptive Research Teams--was influenced significantly by the problems
CIMMYT experienced in Kenya in trying to modify a strongly compartmen-
talized research establishment. A two-level hierarchy provides an oppor-
tunity to build an adaptive research effort through young professionals
trained from the start in this approach. Under this system, it is hoped
that conflicts between senior technical scientists and trainee economists
will be eliminated and there will not be a need for the new discipline of
economics to break into ongoing disciplinary programs. Since all the young
professionals will be learning their roles simultaneously, this will avoid
the imbalance created when a new economist has not yet accumulated the
experience to demonstrate his usefulness to the established biologist.
A major problem in developing Adaptive Research capacity in Zambia is
the scarcity of agricultural graduates. While continuing to seek trainees
from the ranks of agriculturalists graduating from the University of Zam-
bia, the Government of Zambia is asking donor agencies to provide technical
assistance personnel to strengthen commodity research teams and to help
The FSR procedures were presented in a proposal to the National Crop
Research Planning Committee of Tanzania in October 1976. The Committee
gave CIMMYT permission to approach Directors of Research Stations and to
seek their interest in a demonstration of FSR procedures. Two Directors
were approached, one at the Central Research Station, Ilonga, and one at
Uyole Agricultural Centre, Mbeya; and subsequently, FSR demonstrations
were mounted in cooperation with the staff from both stations in 1977.
Although Tanzania has a history of farm economists on its research
stations, there was no economist working at the Central Research Station in
1977. The FSR demonstration, in the drier areas (less than 1000 mm per
year) of Morogoro and Kilosa Districts, was carried out by biological
scientists from the station and farm economists from the Faculty of Agri-
culture of the University of Dar es Salaam, Morogoro. The Exploratory and
Formal Surveys were completed in early 1977. A draft report and proposals
for experimentation were discussed in October 1977 and the final report
was presented to the Chief Research Officer in the Ministry of Agriculture
in December 1977. The major recommendation was to begin experiments on
short-term maize selections which required about 45 days to tassle, 20-25
days less than varieties presently used in the area. The report concluded
that a maize selection tasselling in 45 days would have the following
benefits for target farmers:
(1) The most reliable period of rainfall is early March to mid-May, a
period of 70-75 days. A variety tasselling in about 45 days,
planted on the first good March rains, would make the best use of
this period of rain and would increase the reliability of their
(2) The shorter-term maize variety would enhance the farmers' abil-
ity to react to contingency situations--such as late starting
rains or a marked gap in the rains or rains finishing in early
May. These contingency situations were mentioned as key hazards
by a high proportion of farmers.
(3) Maize is already used to supply food early in the new season. A
variety tasselling in about 45 days would produce food 3 weeks
earlier than present varieties.
(4) The shorter variety planted earlier in these areas would free
the wet low-lying locations some 3 weeks earlier than the exist-
ing varieties, which has several possible ramifications:
(a) An increased probability of getting a bean crop from these
(b) An increase in the number of low-lying areas where a bean
crop could be attempted, and
(c) The possibility, in the better low-lying areas, of growing
other crops with a slightly longer growing season than
(5) The introduction of a short-term maize crop, capable of giving a
reliable supply of the favored staple from an early or mid-March
planting, will have two more important and interacting repercus-
sions on the farming system, both of which should improve the
scope for cash crop production by target farmers.
(a) The later planting date will spread the demand for labor
into late March/early April (weeding the maize planted).
(b) Increased reliability of the maize crop will reduce the
need to plant sorghum as an insurance. Labor will be freed
from sorghum establishment and weeding in the January/Feb-
ruary peak labor season.
Both situations will improve the flexibility of target group
Allowing them to reduce their cash outlay on labor and
tractors needed to give them the required area of crops
(2/3 of all farm expenditures), thus permitting the use
of cash for other purchased inputs;
Or allowing an increase in the area cultivated by main-
taining their present outlay or hired labor and/or trac-
Or allowing more timely planting over a larger area of
cotton or other crop grown for cash.
The second demonstration in Ufipa District in Southwest Tanzania was
carried out by an economist and technical scientists from Uyole Agricul-
tural Centre, a parastatal with responsibilities for agricultural research
throughout the southern highlands of Tanzania. Fieldwork took place in
late 1977. The final report, delayed due to an abortive attempt to develop
a computer processing link with the University of Dar es Salaam, was dis-
cussed with all Centre scientists in mid-1979.
Since then, little progress has been made in obtaining a commitment by
national agricultural research services in Tanzania to utilize the demon-
strated FSR procedures. While the procedures have been used extensively by
Uyole Agricultural Centre for diagnostic survey work in the Southern High-
lands, the linkage with On-Farm Experimentation has been limited. One
ongoing research program at Uyole is the diagnosis of the same farming
system by three techniques of data collection: the Exploratory Survey,
the Exploratory plus Verification Survey, and a frequent visit cost route
survey. This research is designed to quantify the incremental costs and
benefits of the three techniques. In addition, the Crop Science and Rural
Economy Departments at the Faculty of Agriculture, Morogoro have cooperat-
ed in using FSR procedures to plan experiments in a program funded by the
International Development Research Centre (IDRC). The new national re-
search parastatal, Tanzania Agricultural Research Organization (TARO), is
planning to use an FSR approach to research, although implementation has
not yet started.
With the independence of Zimbabwe in 1980, contact was made with
Research and Specialist Services (R + SS) in the Ministry of Agriculture.
Both R + SS and the Department of Agricultural Development (DEVAG),
charged with the development of agriculture in the Tribal Trust land,
showed an interest in using FSR procedures to plan appropriate experi-
ments. A demonstration of FSR procedures was undertaken in the Chibi
District of the Victoria Province, some 400 km south of Salisbury, in the
first half of 1981.
The demonstration highlighted an increasingly common phenomenon in
Africa: increasing population densities are demanding more arable land
and are imposing pressure on grazing land. In the Chibi District of
Zimbabwe, these pressures are jeopardizing the ability of farmers to main-
tain their cattle, which they value as a source of draft power, milk, and
as a last resort, cash. The reduction in grazing particularly exacerbates
the dry season (4-5 months) feeding problems, and hence the animals are in
poor condition for draft work by the beginning of the new rainy season.
Farmers in the area have already adopted the following strategies to
counter these pressures:
(1) New maize varieties are giving better results on the low fertil-
ity soils than the millets or their traditional maizes;
(2) Manure is being used to help maintain soil fertility;
(3) Winter ploughing uses animals when they are in their best condi-
tion after the rains and minimizes draft requirements in October
and November when animals are weak;
(4) Maize stover, groundnuts haulms and, in a few cases, grass are
harvested and preserved to maintain a satisfactory condition
through the end of the dry season;
(5) Cows are being used for draft purposes;
(6) Some farmers establish maize or millet late, from January to
March, even after harvesting their groundnuts, to provide extra
dry season fodder.
These strategies bring their own problems. Because crop residue is
now heavily used for dry season feed, it is scarce for making manure or for
ploughing back in the fields. The poor condition of cows suffering from
low nutrition and overuse as draft animals for as much as five months of
the year leads to low fecundity, poor calf nutrition, and increasing mor-
tality. Indeed, many more farmers in the area will find themselves without
their own draft animals, unless the downward spiral is arrested. In the
extreme, hoe cultivation will be established in many parts of the area, as
arable encroachment finally absorbs the grazing area. Even now there is
increasing migration to towns in the area, as people's expectations can no
longer be realized from the land and available animals.
This key problem area of increasing the productivity of animals in the
farming system is one of four foci identified for priority research at both
technical and adaptive levels. Within this focus, three thrusts were pro-
(1) Increased dry season feed
(a) Raise maize density to increase fodder
examine fodder/grain balance
examine the interaction with drought tolerance
(b) Test bulk fodders or a forage legume, or both intercropped,
to plant after groundnuts (the earliest harvested crop)
examine the possibility of a shorter season groundnut
examine the possibility of an earlier harvest of existing
varieties, with stocking to avoid sprouting. Either
would give late fodder plantings a longer period with
(c) Undersow the maize crop with a forage legume
examine alternative planting patterns for maize to en-
courage forage legumes
(2) Improvement of cows as draft animals
(a) Timing of service and calving alternatives to give better
calving intervals, with cows used in draft from September
to November and March to May
(b) Service cows by A.I. or standing bulls to give them better
conformation for their ploughing role, which also removes
the need to hold male calves and reduces grazing pressures
(an offtake for male calves would be an incentive)
Both measures to improve cows as draft animals were felt to be
contingent upon the farmer's ability to recognize heat and on a
stronger inclination on their part to control service. Present-
ly, with low nutrition affecting both bulls and cows, farmers
are happy to have their animals serviced under any circum-
(3) Improve stover management
(a) Examine the time of cutting of maize stover, balancing the
full maturity of grain with the nutritive value and palat-
ability of the stover
(b) Examine methods of holding and feeding of both groundnut
haulms and maize stover
With the demonstration of FSR procedures recently completed in Zim-
babwe, it remains to be seen whether the Ministry of Agriculture will
accept such procedures as a means to improve relevancy in research for
Tribal Trust farmers.
4.5 Lessons From Country Experiences
The emphasis on in-service and short-term training has preoccupied
CIMMYT staff members since the middle of 1979. By late 1980, Kenya trainee
economists and significant numbers of agronomists had attended various
training programs. With a growing demand for FSR training throughout East,
Central, and Southern Africa, it was decided to broaden the scope of the
training and cover both the Diagnostic Survey sequence and the On-Farm
Experimentation sides of the Adaptive Research Cycle. Workshop time was
increased from one to two weeks. The first workshop with a regional
emphasis was held in September 1980 and covered the FSR procedures for
Diagnostic Survey work. The second workshop with a regional emphasis was
held in April 1981 and covered procedures for the planning, management,
and interpretation of On-Farm Experiments. There was an overwhelming
demand for this workshop; it was attended by 42 participants, more agrono-
mists than economists, from ten African countries. On the basis of our
experience to date, we think that CIMMYT's Regional Economics Program
should expand its training activities and also work closely with FSR-
oriented programs sponsored by various donors.
Lessons have also been learned from working with national research
services. The most important is the recognition of the need for a flexible
and pragmatic approach to different institutional situations and to the
personalities involved in each situation. A major strategy is to focus on
research services where there is already a strong awareness that research
relevance is a problem. Within such establishments, if authority is
strong, it may be helpful to introduce FSR procedures. Where direction is
weak or conservative or where organization is poor, new procedures can be
seen as an added source of confusion--a nuisance. In such circumstances,
only a bottom-up approach, working through the station and with individual
scientists, seems feasible. Ideally, top-down authority and a bottom-up
approach working through individual researchers can be complementary.
A clear distinction has emerged between technical and adaptive re-
search. Technical research is the solution of technical problems on re-
search stations organized along disciplinary and commodity lines. Adap-
tive research is a selection and testing, from the range of potentially
relevant technical solutions, of a partial or whole solution to a particu-
lar problem that has been established as a priority by a target group of
farmers. A revised implementation strategy then is to establish adaptive
on-farm research teams, whose members build up their experience together,
drawing on both the existing body of knowledge and on older disciplinary
oriented specialists for potential solutions to identified systems prob-
lems. Once established, adaptive teams begin to channel unsolved techni-
cal problems back to the specialists. This process continues until prob-
lems identified on farms preoccupy both adaptive and technical researchers
in the research hierarchy.
5. CURRENT ISSUES IN FSR PROCEDURES
This section sets out some current issues in the'use of FSR procedures
for improving relevancy in agricultural research. It moves from concepts
through procedures to promotion strategies. The first issue is the need to
reconcile national and local priorities to plan effective research and
5.1 A Framework of Recommendation Domains as a Link Between
National and Local Priorities
A framework of Recommendation Domains of farmers provides an inter-
face between national and local priorities. The framework of recommenda-
tion domains can assist in developing more effective development programs
by linking top-down policy objectives with local needs as revealed by the
exploratory and verification surveys. The four common steps in planning
agricultural development programs are:
(1) National policy objectives dictate that a commodity should be
promoted. For example, government policies dictate that cotton
should be grown for local industries as a substitute for import-
ed raw materials.
(2) Areas are selected with suitable natural growing conditions.
(3) Areas are selected which have infrastructure and institutions
onto which programs and marketing services can readily be graft-
ed. (Alternatively, the necessary infrastructure and institu-
tions form part of the development program.)
(4) A package of practices is specified that aims at attaining opti-
mal technical management of the commodity under the climate and
soil conditions of the chosen area.
There is now substantial evidence in Eastern Africa that this se-
quence fails to give adequate attention to both the priorities and the
economic circumstances of local farmers. Yet these factors are crucial to
the success of agricultural development programs because recommendations
which are incompatible with the priorities of farmers will be unacceptable
Some device for coordinating the efforts of policy makers, planners,
researchers, and extensionists is long overdue. A framework of
Recommendation Domains has the potential of helping select research prior-
ities, and promoting communication between farmers and the bureaucracy and
between departments in the bureaucracy. On the basis of using FSR proce-
dures in Eastern Africa, we believe that the zoning questionnaire, back-
ground information, and the exploratory and verification surveys can pro-
vide information that will help evaluate the suitability of each Domain for
programs designed to meet one or several national policy objectives. In
addition, the diagnostic sequence can be used to identify needed changes in
policies and institutions to support local development programs.
5.2 FSR--Carte Blanche or a Predetermined Focus?
The issue is whether, in trying to identify farmers' needs, the diag-
nostic survey sequence should or should not be pre-focused onto one enter-
prise such as maize.
It is a major issue. FSR proponents themselves base their case
against traditional experimentation on compartmentalized research estab-
lishments and heavy disciplinary specialization. The breeder believes
variety is the main problem, the soil scientist fertility, the entomolo-
gist insects, and so on. The farmer will have his own list of problems,
many interrelated. If research effort addresses those well down on his
list, missing say his top three, he will regard such changes as irrelevant
to his priority needs and will be unwilling to commit his limited resources
to their adoption. The question is does a systems approach which is pre-
focused onto one enterprise fall into this same trap of treating, what may,
for the farmer, be secondary issues.
It will almost always be found that one enterprise offers the best
leverage on system problems and forms a focus for experimentation and
development. But it is clear that investigation from a whole system
orientation may focus research on quite a different enterprise from one
which is pre-determined. Ideally, the investigation and planning process
should be initiated carte blanche from a whole system orientation with the
intention of focusing experimentation on key problems and the enterprise,
or enterprises, offering best leverage for the solution of those problems.
The examples given in the paper featuring starch grains and livestock show
how crucial the system interactions are in dictating the relevant develop-
ment opportunities for a particular enterprise. They also show,
incidentally, that FSR procedures for diagnosis handle crop and livestock
with equal facility. Though whether livestock or indeed tree crops can be
handled effectively on the On-Farm Experiment side of Adaptive Research
cycle is an unanswered question. In CIMMYT, the mandate for work on maize
and wheat has demanded a compromise from a full system orientation. This
has been made in Eastern Africa by locating demonstrations in areas where
maize is the major enterprise in the local farming system. With subsist-
ence and semi-subsistence farmers, the major starch staple is almost in-
evitably their priority crop, and, as the one which absorbs the most
resources, is the crop which gives best leverage for the development of the
whole system. At the same time, the funnelling nature of the diagnostic
sequence allows confirmation at the Exploratory survey stage, before costs
are significant, that the pre-focus on maize or wheat is consistent with
effective leverage on the whole system.
Operationally, when trying to introduce FSR procedures into an essen-
tially commodity-oriented policy and planning environment, a pre-focused
approach is more readily compatible with existing institutions and proce-
dures. A whole system approach may be much more difficult for incumbent
professionals to accept.
5.3 Can LDC Research Manpower Handle
a Large Number of Recommendation Domains?
It was emphasized earlier that research planning is a compromise.
Ideally, research and development efforts would be tailored to the circum-
stances of each individual farmer. This is clearly impractical. It is
equally impractical to mount the same R & D effort for all farmers in a
country, as their circumstances are so diverse. We have argued in this
paper that one must search for homogenous farming systems or recommenda-
tion domains as a working compromise. But let us see whether it is a
compromise within the grasp of the significant number of LDCs with limited
research resources. First, what is the likely size of a recommendation
domain? In Zambia, we established an average of 10,000 farms per Domain,
ranging from 3,000 to 23,000 farms in the seven Domains identified. The
numbers will vary according to the variability of natural and economic
circumstances and with the density of population. In Zambia, the climate
and soil were fairly homogeneous; economic factors, particularly access to
markets, were variable; and population density was very low, at 3 or 4
people per square kilometer. Extrapolating the 10,000 farms per Domain in
Central Province gives some 60 or 70 Domains for the Zambia agricultural
sector as a whole. The Zambian Research Service plans to field an ARPT
team of an agronomist and economist in each of the 9 Provinces for a total
commitment of some 18 professionals. This implies that each team would be
responsible for some eight Recommendation Domains.
The next step is to make assumptions about the workload for the
agronomist and economist to determine whether each team has the ability to
cover eight Domains. The main responsibility of the agronomist is to
manage, supervise, and interpreted the on-farm experiments. He will also
help the economist carry out the Exploratory Survey work. The agronomist's
workload is assumed to be the following:
(1) With support staff he will be able to supervise 24 on-farm sites
(2) He must cover six sites across each Domain to obtain 'Domain
(3) He will need three years' experimentation at each site.
These assumptions would allow the agronomist to gradually cover the
eight Domains of his region over a six-year period.
The Diagnostic Survey sequence is the main responsibility of the
economist. He must also participate in discussions with the agronomist,
extension staff, and farmers at the experimental sites. He will be respon-
sible for the economic interpretation of the experimental results, and the
collection and collation of economic data from experiments on Domain
farms. To match the rate of coverage of the agronomist he will need to
implement the Diagnostic Survey sequence in three new Domains every two
Under these assumptions, the Adaptive Research cycle would require
six years before returning for a second round, including an evaluation of
the extent to which first-round recommendations were adopted by Domain
farmers. The estimated rate of coverage seems adequate in Zambia. With
some exceptions, the manpower commitments of two adaptive agricultural
research professionals per 80,000 farms seem within the reach of the coun-
tries of East, Central, and Southern Africa.
The above figures are useful for purposes of arithmetic but should not
obscure the flexibility in organization and management that is inherent in
an Adaptive On-Farm Research program. For example, a breeding effort
cannot be justifiable for the needs of one Recommendation Domain. But as
the needs of many Domains are accumulated, this information can be trans-
mitted to the breeders. Likewise, fertilizer trials may range across
similar soil types and cover many target groups; yet the interpretation of
the results will be geared to the needs of each target group.
5.4 Is Modelling a Farming System Worth the Extra Cost?
The FSR procedures promoted by CIMMYT in Eastern and Southern Africa
have been designed with major problems of LDC institutions in mind: limit-
ed funds and limited manpower. The procedures allow researchers to gain an
understanding of the farming system of a target group of farmers over a 2-
or 3-month period. Costs are low and turnaround time is rapid. Among farm
economists working in LDCs, a major controversy centers upon the intensity
of investigation and analysis required to understand the systems. A spec-
trum of data collection methods is available, from single-visit surveys to
cost route methods in which farmers are visited regularly over the agricul-
tural calendar. Similarly, a spectrum of analytical tools is available,
ranging from simple cross-tabulations to sophisticated variants of mathe-
matical programming. The CIMMYT procedures are close to the rapid and
cheap end of the collection and analysis continuum, with a turnaround time
of 2 to 3 months for any one target group of farmers, compared to 12 to 24
months for frequent-visit data collection and programming analysis.
Cheap and rapid FSR procedures can gain an initial understanding of 4
to 12 target groups in the time period that a single target group can be
studied by frequent-interview research methods. Where research manpower
is scarce it is assumed that the benefits from this initial understanding
of 4 to 12 target groups will be much higher than those obtained from an
increased understanding of one target group through detailed data collec-
tion and the use of sophisticated modelling techniques for system analy-
sis. Furthermore, mathematical programming cannot effectively represent
real life situations, and, in practice, the level of control of enumerator
and respondent error in cost route data collection techniques is usually
poor. The conclusion is that the sophisticated methods of data collection
and analysis at the far end of the spectrum are not cost effective and
useful in serving thousands of small farmers in the typical country in
Eastern and Southern Africa. Certainly, increments in understanding
gained from modelling farming systems are acquired at a very high cost,
particularly in terms of the understanding of other systems foregone due to
the scarcity of farm economists working at the micro-level in LDCs. This
is the rationale behind CIMMYT's emphasis on low-cost rapid procedures for
farming systems diagnosis.
CIMMYT's experiences in implementing the diagnostic sequence raise
questions about whether even the Formal Verification survey is necessary.
This is the most complex and expensive part of the survey sequence, absorb-
ing 60-70 percent of the turnaround time. Verification Surveys in Eastern
Africa have never seriously contradicted the findings of the Exploratory
Surveys. The Formal Verification Survey results have been used for two
(1) To provide numbers to support the diagnosis of the Exploratory
or Informal Survey, which is important where the FSR procedures
are being demonstrated to new institutions.
(2) To provide data to improve the detailed planning of experimental
treatments and the fuller interpretation of experimental re-
It has become clear from CIMMYT's experience to date that the Explora-
tory Survey is the key stage in diagnosis, bringing a systems perspective
to bear on a farming situation, and the biological and social scientists
into direct contact with the farmers. And it is probable that the collec-
tion of data from Domain farmers during the period of on-farm trials can be
used to verify or modify the findings of the Exploratory Survey and to
develop the experimental program. If this change were made, the sequence
would bear a much closer resemblance to that developed at the ICTA in
Guatemala (Hildebrand, 1976).
5.5 Outstanding Issues in Promoting FSR for Agricultural Development
Some outstanding issues arise in the wider mobilization of FSR pro-
cedures for agricultural development. These could usefully fill further
papers in this series.
FSR procedures which are carried out as part of an Adaptive Research
cycle offer an excellent device to bring farmers into the technology de-
velopment process and to close the gap between research and extension. The
use of FSR procedures by research establishments would be enhanced if the
motivation of both technical and adaptive research scientists could be
closely related to the benefits their work brings to farmers.
Training for would-be researchers, and perhaps more generally for
Third World agriculturalists, is a second urgent issue. All professionals
working in smallholder agriculture need a systems perspective. A major
output of CIMMYT Economics programs is manuals and study reports now widely
used as training material. The operational procedures of the manuals would
be much more meaningful if the concepts had already been digested in formal
diploma and degree courses. All diploma and Bachelor degree training
should include courses which give students an understanding of how small
farmers operate their systems. Not only would this be a foundation for
building relevance in research, program planning, and extension, but it
would also help dissipate the attitude among agricultural professionals
that small farmers are irrational. Fieldwork is an essential complement to
learning concepts in the classroom. Fieldwork could include both surveys
and on-farm experimentation, again not only to teach methods, but to incul-
cate a healthy attitude toward working with farmers in their fields. It is
difficult to think of a better way to equip Third World professionals for a
down-to-earth role in agricultural development, a role often shunned with-
in the existing educational value system. Finally, an awareness should be
fostered at senior research levels, and even at senior political levels, of
the narrowness of traditional criteria used to select technology and of the
opportunities opened by a farming systems perspective. Political exhorta-
tion itself would be a more effective engine of development if the content
were relevant to the needs of the exhorted. There is a scope for promo-
tional work at the highest levels to provide leverage on recalcitrant
personalities and institutions. In this same, wider context, FSR proce-
dures can serve as a link between local and national priorities and between
the farm and its infrastructure and can provide insights to develop im-
proved policies and programs to help small farmers.
GUIDELINES FOR EXPLORATORY (INFORMAL) SURVEYS
These guidelines are quite long. A F are sets of questions for
discussion with farmers, each of which should occupy 1-1 1/2 hours on the
farm. In the course of an Exploratory survey, three or four farmers may be
asked each set of questions. However, for some the researcher may feel
clear and confident about those facets after talking to two farmers, other
sets may benefit from discussions with five or six farmers. It is profit-
able, after a set of questions has been discussed with two farmers, to
write up detailed notes in the same numbered format as the Guidelines.
Gaps in the information for that set can be filled by questioning further
farmers and their answers will complete the notes on that set.
It is useful to decide on several specific, introductory, questions
to establish that the new farmer is indeed in the Recommendations Domain
investigated. The questions will refer to key characteristics of that
It will also be useful for researchers to interact after a day's
interviewing and combine impressions of sets of questions covered during
the day. They may have visited the same or different farmers. It is
important that both production agronomist and farm economist are in the
field together for this informal survey work. The two disciplines will
interact strongly in covering sets E and F.
SET A: Questions seek to describe the farming system through the
enterprise pattern and use of products and to identify recent changes in
the relative importance of the enterprises farmed. Questions (1) and (2)
are preliminaries for all respondents.
ENTERPRISE PATTERN, OUTPUT USE, AND SYSTEM TRENDS
(1) List the crops grown and livestock kept by local farmers. Note
whether each enterprise is farmed by most or just a few, and, if
only a few, what is special about those few.
(2) Obtain rough estimates for each enterprise.
a) the number of fields devoted to it; b) the total area in-
volved; and c) the output the farmer would usually expect from
(3) For each major crop, list the varieties grown. Make it clear
whether each farmer grows more than one variety.
(4) For the major crops, crop varieties, and animals, list the end
uses to which they are put. In the case of crop varieties, this
includes the fruit and any other part of the plant used as a
byproduct. Animal products and byproducts are equally impor-
(5) For each identified end use, detail the sequence from the field
through harvesting, storage, preparation, processing, and use,
including timing; when it is taken in the life-cycle of the plant
and the timing of stages through to use.
(6) (a) Note crops, crop varieties, or livestock that used to be
widespread among farmers of the area but are now disappear-
ing. Assess why such enterprises are losing popularity.
(b) Note crops, crop varieties, or livestock that are becoming
increasingly widespread recently. Assess why such enter-
prises are gaining popularity.
SET B: Questions seeking further description of the farming system
through specification of the enterprise calendar, food preferences, and
facets of the food calendar.
ENTERPRISE CALENDAR, FOOD PREFERENCES, AND FOOD CALENDAR
(7) For each crop and, where different varieties are grown, for each
crop variety establish:
(a) The usual time of first cultivation
(b) The usual time of planting
(c) The latest possible time for a viable planting
(d) Reasons for varying the planting time of the crops
(8) Assess the major reasons why local farmers use different varie-
ties, and, if relevant, plant them at different times.
(9) For each crop, and crop varieties where appropriate, establish
the usual time for direct harvest for use from the field (if
relevant) and the usual time of final harvest.
(10) For each livestock type establish:
(a) The usual timing of births
(b) The usual period producing milk and dry
(c) Any special periods for the consumption of meat
(11) Detail the main dishes eaten by farm families in the area, the
preferred constituents, and the preferred state of each constit-
(12) What dishes, with what constituents, are eaten when preferred
foods are not available?
(13) Using a table like the one below, list preferred and substitute
foods in the left-hand column. Against the calendar show:
(a) Months when supplies of each food are readily available
every year Shade out the months
(b) Months when supplies of each may be uncertain in some sea-
sons cross hatch these months
(c) Months when supplies are never available leave blank
MAJOR LOCAL FOODSTUFFS J F M A M J J A S 0 N D
(14) Identify whether any new foods are becoming popular and replac-
ing traditional ones. Assess why old foods are receding and why
new ones are gaining popularity.
(15) Are there main foods which are purchased by many families:
a) All the year round; b) at certain periods (specify when)
(16) Are there main foods which have to be purchased at certain
periods in poor years? If so, for each food:
a) How frequently does this occur; b) what is the usual reason;
c) when did this occur last for most families in the area?
(17) What are the prices of the major foods just before and just after
the main harvest?
SET C: Detailed questions on variations in output and its causes on
TWO major products on the farm. Questions on the main methods used to
produce these two products. Take one at a time.
CAUSES OF OUTPUT VARIATIONS, MAIN PRODUCTION METHODS
(For each selected crop in turn questions 19, 20 & 21.)
(18) With which two of your crops is the loss of output in a season
most serious for you and the family?
(19) In Q. 2 you said you would normally expect this level of output.
(a) What is the lowest output you remember?
(b) What year was this?
(c) What factors caused this low output then?
(d) Can other factors also strongly influence output?
(e) How soon in the season can you tell output will be poor?
(f) When you see this, do you take any action to look for other
sources of food/cash?
(g) If output is poor for example in the year you mentioned -
how did you manage for food/cash?
(20) In Q. 2 you helped us estimate what you usually grow.
(a) Does this area change much from year to year?
(b) Which year was the smallest area you grew?
(c) Why was the area small that year?
(d) Which year did you grow a lot more than usual?
(e) Why was the area grown large that year?
(21) What methods do you mainly use in growing this crop?
(a) Cultivation; (i) How many are done and what method is
used for each?
(ii) Do you ever use other cultivation
methods? Under what circumstances?
(iii) When did this happen last and how do
you assess it?
(b) Weeding; (i) How many are done and what method is
used for each?
(ii) Does the number vary from year to
(iii) What factors cause the number to
(iv) Have you ever used any other method?
How do you assess it?
(c) What items do you usually buy for this enterprise (seed,
fertilizer, herbicide, insecticide, fungicide, hired
labor, or machinery, etc.)?
(i) Source of purchase
(ii) Quantity purchased, price paid
(iii) Year started purchases for this crop
SET D: These questions elicit information on the level of endowment
of land, labor, and cash, and seek clues to factors constraining system
LAND, LABOR, AND CASH AVAILABILITY
(22) Can you get more land if you feel you need it?
(a) How would you get more land?
(b) Would it be near or far away?
(23) Will all the crops you grow do well on all your fields?
(a) Detail the characteristics of soil and/or location most
suitable to your main crops.
(b) Do you change crops depending on the season?
(24) Do you grow the same crops in the same places year after year?
If not, what rules have you about moving crops to different
(25) (a) How many people do you have in the family?
(b) How many work on the farm?
(c) Which work on which crops is done only by:
(d) Are there special responsibilities for:
(iii) Cattle herding and feeding
(26) (a) Which is the busiest month of the year for farming in this
(b) What work, on which enterprises, is going on at this time?
(c) Is there a second busy period?
(d) What main work goes on then?
(e) Are these two busiest periods every year? If not, how do
(27) (a) Do you and other local farmers hire any labor during the
(i) Permanently hired
(ii) Casual hired
(iii) Customary labor
(b) When casual labor is hired, or customary labor used, what
months are these, and what work is mainly done?
(28) (a) Do you and other farmers hire any machinery during the
(b) If so, which crop/operations is it mainly used for and at
what time of the year is it hired?
(29) What are the main sources of cash income for local farmers?
(30) Do many families have members working off the farm
(a) Permanently what are the locations and types of work?
(b) Temporarily what kind of work is temporarily taken up? Is
it taken up at this time because
(i) Opportunities arise then
(ii) Farmers need cash then
(31) What is the most difficult period of the year for cash for local
families? Is it because income is scarce or expenses are very
high? If expenses are high, what are the major items absorbing
This is the end of the initial phase of investigation and initial
analyses are now made.
1. Derive labor allocation constraints.
2. Identify resource constraints.
3. Identify farmers' priorities and decision criteria for each en-
4. Identify farmers' management strategies.
5. Identify potential points of leverage in the farming system,
major factors contributing to low productivity, and specify the
enterprise and production activities involved.
6. Work out approximate levels of return to present cash outlays.
SET E: The following questions provide information on the compro-
mises in production methods and timing of production activities on crop
enterprises which are identified as points of leverage in the farming
(32) How does the farmer decide where he will plant his crop? What
factors does he consider in the decision?
(33) Land preparation:
(a) What is the method of land preparation?
(b) When does the preparation start in relation to the start of
the rains and to planting time?
(c) What sequence of work is involved if there is more than one
(d) How does the farmer work; does he prepare a whole field
before planting, or prepare and plant a bit the same day, or
(e) What is the final form of seedbed?
(f) Are there alternative methods of land preparation?
(a) What is the arrangement of plants in the field, the main
crop, and any mixtures?
(b) Where other crops are mixed in it will be important to
describe in what sequence all the crops are put in the
(c) How do farmers plant in relation to rainfall; dry planting
before rain, the same day as rainfall, within a limited
period after rains?
(d) Do farmers just make one planting of the crop each season or
are there usually several?
(e) Do farmers commonly have to replant?
(f) What is the method of putting the seed in the ground, and
how many seeds are put per hole?
(35) Weeding and Thinning
(a) What implement or implements are used for weeding?
(b) How soon after planting is the first weeding done? Does the
timing vary very much with conditions; if so, how much and
(c) How many weedings will normally be done? Will this vary
with the date of planting, the weather, or the soil in the
(d) Do they thin the plants either in the row or from each
planting hole? If so, at what age? Do they use the thin-
ning for cattle feed?
Pest control, if any
(a) Major pests for which control is sought
(b) Timing and method of control
(c) Assessment of proportion of local farmers using pest con-
Use of fertilizer, if any
(a) Type of fertilizer, source
(b) Usual rate, method, and time of application
(c) Assessment of proportion of local farmers using fertilizer
(38) Use of leaves, top, and stalks for cattle feeding
(a) Proportion of local farmers using
(b) Method of feeding to animals
(c) For leaves; number of pickings made, number of leaves
taken, and the timing in relation to plant growth
(d) For tops; stage of plant growth that the top is taken. Is
this critical time for cattle feed?
(39) Method and timing of harvesting and storing
(a) At what stage does harvesting begin?
(b) What method is followed in picking, shelling, and disposing
(c) How is the crop stored; is any preservative used?
(40) Seed selection and preservation
(a) Do the farmers usually select seed in the field or from
their stored harvest? If from stored, when is it selected?
(b) What criteria do local farmers use when they choose next
year's seed from their own crop?
(c) Do they process and preserve the chosen seed in a special
(41) Is the crop treated in any other way, either while in the field
or in the household? Are there resources used in growing the
crop in a way not covered here?
SET F: These questions seek to detail the compromise in production
methods and the timing of production activities in livestock enterprises
which are identified as points of leverage on the farming system.
(42) (a) Ask what the composition of the herd is and relate this to
major uses of the livestock enterprise.
(b) Does he own his animals or are they loaned to him?
(c) In a crisis, when they need food or cash, which type of
animal is sold first?
(43) (a) What are the main ways of disposing of animals; consump-
tion, sales to neighbors, loans to neighbors, sales to but-
(b) What disposals from the herd were made over the last year?
(c) What types of animals and through which channels?
(d) Are some categories of animals difficult to sell?
(44) (a) Has the farmer had any animals die over the last year?
(b) What are the main causes of death?
(c) Does he do anything to prevent these problems?
(d) What steps did he take to treat?
(45) (a) Which month of the year do your animals usually calve?
(b) Is it the same month every year? If not, what does the
month of calving depend on?
(c) Does calving at this time bring you any problem:
(i) With the calves?
(ii) With the mothers?
(d) What period of the year would you prefer to have calves?
(e) What benefits would this bring:
(i) To the calves?
(ii) To the mothers?
(f) Do you exercise any control over the time when your cows are
(46) Discuss with the farmer the calving history of his female ani-
(a) Age at first calving
(b) Number of calves born
(i) Died as calves
(iii) Disposition of survivors
are his calves reared?
What access do they have to their mother?
Up to what age do they continue to suckle?
Is any special food or treatment given to encourage wean-
Do the calves run with the herd?
is the milking regime?
Who milks the animals?
How many times a day?
(c) What period of the year are animals dry? Are they all dry
together? If not, then what means are used to maintain a
continuous supply of household milk?
(49) (a) Where is the source of water for the animals? Does this
vary with the season?
(b) Are the animals taken to water or is water brought in?
(c) How frequently are animals watered in the dry season? How
far must they be walked to water?
(d) If water is brought in, does the farmer give more to some
animals than others; if so, which type?
(50) (a) What are the major types of feed used over the year (e.g.
grazing on commonland, grazing on own farm, grass or other
fodder carried to animals)?
(b) Does the community exercise any control over the grazing of
(c) Which are the most difficult months for feeding the ani-
(d) How far will animals move to grazing in the dry season?
(e) Are these feeds given to all animals or do some types get
(51) (a) Who herds the animals during the day?
(i) Are all animals herded together or is the herd split
(ii) Is there any communal herding?
(b) Who decides where the animals will go for grazing at vari-
ous times of the year?
(c) How are the animals housed at night? Are they all together?
(52) (a) Is the cattle manure used on the fields?
(b) Is it mixed with any material in the night shelter?
(i) What materials?
(ii) When are they mixed in?
(c) At what time of the year is the manure moved out of the
(d) Is it handled or 'managed' any way prior to application to
(e) Is the night shelter a reasonable site for storing manure
without too much loss of quality?
(The detailed description of management practices and the comparison
of present with technically ideal practices will have identified compro-
mises and provided points of potential intervention in the System. Possi-
ble new materials and methods to exploit these leverage points are pre-
screened for viability in the system and their acceptability to the farmer.
As a final part of the Informal Survey, farmers' attitudes to new materials
and methods are tested. Unless such tests show antipathy to the ideas put
forward, these materials and methods will form the content of local adap-
Byerlee, D.; M.P. Collinson; et al. (1980). Planning Technologies Appro-
priate to Farmers: Concepts and Procedures. Londres, Mexico:
CIMMYT. (1977). "Demonstrations
Planning Adaptive Agricultural
Economics Programme. Research
of an Interdisciplinary Approach to
Research Programmes." Eastern African
Report No. 1. Nairobi: CIMMYT.
Gathee, J.W. (1981). "Research Workers Look at Intercropping from the
Farmer's Point of View." Kenya Farmer (Jan.).
Hildebrand, P.E. (1976). Generating
A Multidisciplinary Methodology.
Technology for Traditional Farmers:
Guatemala City: ICTA.
Kenya, Republic of. (1979). Development Plan 1979 83, Part 1. Nairobi:
Lihanda, F.C.L.M. (1978). "Adaptive Research Planning; Vihiga Division,
Kakamega District, Kenya." (Mimeograph).
Perrin, R.K.; D.L. Winkelman; E.R. Moscardi; and J.R. Anderson. (1976).
"From Agronomic Data to Farmer Recommendations: An Economics Train-
ing Manual." Information Training Bulletin No. 27. Londres, Mexico:
Dairy Farmer Survey Report, Parts of Kakemega
Scientific Research Division, Ministry of Agri-
MSU INTERNATIONAL DEVELOPMENT PAPERS
Carl K. Eicher and Doyle
A Critical Survey," 1982
C. Baker, "Research on
in Sub-Saharan Africa:
Eric W. Crawford, "A Simulation Study of Constraints
on Traditional Farming Systems in Northern Nigeria,"
1982 (136 pp.).
M.P. Collinson, "Farming Systems Research in Eastern
Africa: The Experience of CIMMYT and Some National
Agricultural Research Services, 1976-81," 1982
Copies may be obtained from: MSU International Development Papers,
Department of Agricultural Economics, Agriculture Hall, Michigan State
University, East Lansing, Michigan 48824, U.S.A. Individuals and
institutions in Third World countries and U.S. Agency for International
Development officials may receive copies free of charge.