Public Disclosure Authorized

W'ORLDBANK TECHNICAL PAPERNUMBER 43 WTP43

August1985
FarmingSystems Research
A Review

Norman W. Simmonds
Public Disclosure Authorized

SECTORAL LIBARY
INTERNATIONALRANK

RECONSTRUCIONAND DEVELOPME,N.

17 985
SEP
Public Disclosure Authorized
Public Disclosure Authorized

.S9
S56 ~FILE COPYI
1985
c. 2
WORLD BANK TECHNICAL PAPERS

No. 1. Increasing Agricultural Productivity

No. 2. A Model for the Development of a Self-Help Water Supply Program

No. 3. Ventilated Improved Pit Iatrines: Recent Developments in Zimbabwe

No. 4. The African Trypanosomiases: Methods and Concepts of Control and Eradication
in Relation to Development

(No. 5.) Structural Changes in World Industry: A Quantitative Analysis of Recent Developments

No. 6. Laboratory Evaluation of Hand-Operated Water Pumps for Use in Developing Countries

No. 7. Notes on the Design and Operation of Waste Stabilization Ponds in Warm Climates
of Developing Countries

No. 8. Institution Building for Traffic Management

(No. 9.) Meeting the Needs of the Poor for Water Supply and Waste Disposal

No. 10. Appraising Poultry Enterprises for Profitability: A Manual for Investors

No. 11. Oppcrtunities for Biological Control of Agricultural Pests in Developing Countries

No. 12. Water Supply and Sanitation Project Preparation Handbook: Guidelines

No. 13. Water Supply and Sanitation Project Preparation Handbook: Case Studies

No. 14. Water Supply and Sanitation Project Preparation Handbook: Case Study

(No. 15.)Sheep and Goats in Developing Countries: Their Present and Potential Role

(No. 16.)Managing Elephant Depredation in Agricultural and Forestry Projects

(No. 17.)Energy Efficiency and Fuel Substitution in the Cement Industry with Emphasis
on Developing Countries

No. 18. Urban Sanitation Planning Manual Based on the Jakarta Case Study

No. 19. Laboratory Testing of Handpumps for Developing Countries: Final Technical Report

No. 20. Water Quality in Hydroelectric Projects: Considerations for Planning in Tropical
Forest Regions

No. 21. Industrial Restructuring: Issues and Experiences in Selected Developed Economies

No. 22. Energy Efficiency in the Steel Industry with Emphasis on Developing Countries

No. 23. The Twinning of Institutions: its Use as a Technical Assistance Delivery System

No. 24. World Sulphur Survey

( ) Indicates number assigned after publication.

(List continues on the inside back cover)
 WORLDBANK TECHNICALPAPERNUMBER 43 C

FarmingSystems Research
A Review

Norman W. Simmonds

The World Bank

Washington, D.C., U.S.A.
Copyright ) 1985
The International Bank for Reconstruction
and Development/THE WORLD BANK
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.

All rights reserved

Manufactured in the United States of America
First printing August 1985

This is a document published informally by the World Bank. In order that the
information contained in it can be presented with the least possible delay, the
typescript has not been prepared in accordance with the procedures appropriate to
formal printed texts, and the World Bank accepts no responsibility for errors. The
publication is supplied at a token charge to defray part of the cost of manufacture and
distribution.
The World Bank does not accept responsibility for the views expressed herein, which
are those of the author(s) and should not be attributed to the World Bank or to its
affiliated organizations. The findings, interpretations, and conclusions are the results
of research supported by the Bank; they do not necessarily represent officialpolicy of
the Bank. The designations employed, the presentation of material, and any maps used
in this document are solely for the convenience of the reader and do not imply the
expression of any opinion whatsoever on the part of the World Bank or its affiliates
concerning the legal status of any country, territory, city, area, or of its authorities, or
concerning the delimitation of its boundaries or national affiliation.
The most recent World Bank publications are described in the annual spring and fall
lists; the continuing research program is described in the annual Abstractsof Current
Studies.The latest edition of each is available free of charge from the Publications Sales
Unit, Department T, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433,
U.S.A., or from the European Office of the Bank, 66 avenue d'1ena, 75116 Paris, France.

Norman W. Simmonds is at the Edinburgh School of Architecture.

Library of Congress Cataloging in Publication Data

Simmonds, Norman W., 1914-
Farming systems research.

(World Bank technical paper, ISSN 0253-7494

no. 43)
Bibliography: p.
1. Agricultural systems--Research. 2. Agriculture--
Research--On-farm. 3. Agricultural systems--Tropics.
I. Title. II. Series.
S494.5.S95S56 1985 630'.72 85-12181
ISBN 0-8213-0566-2
 ABSTRACT

Farming systems research (FSR) has become increasingly important as

an element of the research programs of the international agricultural
research centers, national research programs, and development projects with
agricultural research components. In this context, as the volume's foreword
explains, it is important to take stock of existing experiences with FSR, to
assess ongoing trends and future perspectives.

The report on the State of the art of farming systems research

reviews the history of this approach and attempts to define its scope, in
both a broad and a narrow sense. The paper describes various on-farm
research procedures, reports on experiences accumulated in this area in
different international agriculture research centers, discusses the
contributions of different disciplines to FSR, the relation of FSR to
agricultural extension, and raises some wider questions related to the
perspectives of, and needs for, carrying out FSR further in various
institutional settings.

- iii -
 CONTENTS

FOREWORD vii

PREFACE ix

ACKNOWLEDGMENTS x

LIST OF ABBREVIATIONS AND ACRONYMS xi

CHAPTER 1. INTRODUCTION

Context 1
Tropical Agricultural Systems 3
Innovation and Change 8
Anticipation of Argument to Follow 12

CHAPTER 2. FARMING SYSTEMS RESEARCH IN GENERAL (FSR sensu lato )

History 14
Small-farmer Characteristics 15
Systems 19
FSR (Farming Systems Research) sensu stricto 21
and OFR/FSR (On-farm Research with a
Farming Systems Perspective)
"Upstream" and "Downstream" 25
New Farming Systems Development (NFSD) 27
Genotype-environment Interactions 28

CHAPTER 3. THE NATURE OF ON-FARM RESEARCH WITH A FARMING SYSTEMS

PERSPECTIVE (OFR/FSP)

OFR (On-farm Research) Procedures 32

Training and Networks 47
Economists, Anthropologists, and Institutions 48
Relation to Extension 52

CHAPTER 4. WIDER QUESTIONS

Technological Change and the CGIAR 54

(Consultative Group on International
Agricultural Research/NARES (National
Agricultural Research and Extension
System) Context
Agro-forestry and the Neglect of Perennials 57
The Interests of the World Bank 61
CHAPTER 5. EXAMPLES

Beans in Highland Colombia 63

Maize Production Methods in Panama 65
NFSD (New Farming Systems Development) 65
and OFR/FSP (On-farm Research with a
Farming Systems Perspective) in Nigeria
Rainfed Farming in the Mediterranean 71
Mixed Upland Farming in Ethiopia 71
Cropping Systems in the Deccan of India 74
Rice-based Systems in Asia 76
Small-farmer Coconuts in the Philippines 77
Small-farmer Food-cropping in Indonesia 79

CHAPTER 6. SUMMARY

Introduction 82
Farming Systems Research in General 83
On-farm Research with Farming Systems 84
Perspective
Wider Questions 86

BIBLIOGRAPHY 88

ADDENDUM TO THE BIBLIOGRAPHY 96

- vi -
 FOREWORD

The growing interest in farming systems research (FSR) and the

substantialexpansion of FSR studies in many countries representsone of
the most significantrecent trends in the overall area of agricultural
research. Turning toward studying the farm as a "system"is a substantial
step forward -- from addressingonly its technicalor economic dimensions
towards capturing the tight interplay between the agrotechnical,economic,
sociological,managerial,and culturalvariables intrinsic to the farm
unit. At the same time, the orientationtowards FSR expresses a
recognitionof the enormous diversity that exists among farm and farmers
and an effort to translatethe understandingof these differencesinto more
precisely tailoredagriculturalresearchand advice fitting the needs,
constraints,and potential of various categoriesof farm systems.

The World Bank, in supportingagriculturalresearchthrough loans

and credits to national programs and grants to the ConsultativeGroup on
InternationalAgriculturalResearch,has a major interest in new ideas and
new approaches to agriculturalresearch in its borrower countries.
Increasingemphasis on the need for new technologyto support agricultural
developmentand the need to spread that technologyquickly,especially
among small farmers,has made it more than ever necessary to improve the
linkages between researchersand extensionworkers and between both these
categories of workers and the farming population. Farming systems
research, with its emphasis on determining the importantproblems of
farmers and on testing potential solutions at the farm level, can
strengthen these linkages and increase the efficiencyof both research and
extension systems,hence the Bank's interest in trying to synthesizeand
evaluate the many divergent views on the role and value of the farming
systems researchapproach.

In order to assess the progress to date of farming systems

research, its contributionto agriculturalresearchand extension,and its
likely perspectivesof further development,the World Bank commissioned
Dr. Norman W. Simmonds to prepare a state-of-the-artreview of FSR. The
resulting paper is publishedin this volume. It is based on an examination
of the literature,discussionswith many people involved in this research
area, and visits to a number of national and internationalprogramswhere
farming systems researchis being practiced. As the report makes clear
there are many different perceptionswithin the scientificcommunityof
what constitutes"farmingsystems research." This state of affairs is
obviously to be expected in a research field that is comparativelynew,
though to one or another extent it has been practicedunder differing
titles for some years prior to acquisitionof its present label, shape, and
content.

- vii -
 As a relativelynew topic, we are convincedthat farming systems
researchwill continueto generatedebate as to its role, its research
methodologyand its organization. Obviously there is no single "best way
to do farming systems researchand such debate will certainlycontinue.
The World Bank hopes that this paper, which representsthe views of the
author and not necessarilythose of the Bank, will provide a useful
contributionto that debate and to the continuingevolutionof new ideas on
researchfor small and resourcepoor farmers.

Agriculture & Rural DevelopmentDepartment

The World Bank

- viii -
 PREFACE

The appearanceover the last decade or so of widespreadinterest

in, and a profusionof publicationson, farming systems research (FSR) has
been very evident, both in the InternationalAgriculturalResearchCentres
(IARC) and in national agriculturalresearchand extension systems
(NARES). Indeed, commitmentto at least some elementsof FSR in
agriculturalR & D programs in the tropicshas become so general that one
World Bank officerhas referredto the phrase as having "incantatory
value."

So it has, and many differentmeaningshave been attached to it

in what is now a large and heterogeneous(but erraticallypublished)
literature. No unified view of the whole field has appearedand it was in
the hope of achievingsuch that the World Bank invited me, early in 1983,
to prepare this paper. The Bank's interest in the field is, of course,
two-fold: it is a major donor to the IARCs by way of the Consultative
Group on InternationalAgriculturalResearch (CGIAR); and it funds
agriculturaldevelopmentprojects throughoutthe tropics,often with at
least a nominal FSR component. As current spendingunder these two heads
is about US$19 million and US$3,000 million, quite small proportionalFSR
componentsmust be of concern.

My terms of referencewere given by the Agricultureand Rural

DevelopmentDepartment of the World Bank, but it was accepted,so confused
is the subject, that the structure of my report would have to be
determinedmore by the logic of what I could discover than by prior
instruction. In practice I have found it necessaryto depart fairly
widely from my terms of reference (as to order, though not as to content)
and think of FSR, in its various guises, not as a more or less isolated
set of related activities,but rather as a componentof innovationin
tropicalagricultureat large.

Besides valuable briefing by Bank officersin Washington,I also

had the benefit of discussions,in Britain,with a number of people having
much experienceof the matter. My travels in Latin America,East and West
Africa, India, and South East Asia took in visits to several national
agriculturalresearchsystems (NARS) committedto FSR proceduresas well
as to IARCs. The coverage,however,was far from completebut could
hardly have been fuller in the time available (about four months). The
treatment in this report is thereforebroad and inevitablysomewhat
superficial. I doubt, however,whether a more extended survey would have
revealed any new principles.

- ix -
 ACKNOWLEDGMENTS

The views expressed in this report are mine, and the Bank is
not, of course, committed to them. Nevertheless Bank officers, both in
Washington and elsewhere, had a substantial part in shaping them and I am
most grateful for their guidance, particularly to Dr. J.K. Coulter and his
immediate colleagues in Washington.

I add the name of Dr. M.H. Arnold and thank him also for several
lively and informative talks. Grateful acknowledgment is also due to Dr.
G.A. Watson and the Cambridge University Press for a quotation from
Experimental Agriculture in Chapter 4 under "Agro-forestry and the Neglect
of Perennials."

Before leaving the U.K. and while traveling, I had the benefit
of advice and guidance from many people. I should especially like to
acknowledge, with many thanks, the help of the following. In the U.K.:
Dr. R.K. Cunningham, Mr. K.R.M. Anthony, Mr. R.W. Smith, Prof. C.R.W.
Spedding, Dr. I.D. Carruthers, Dr. D. Gibbon, Dr. C.C. Webster, Dr. A.G.
Smith, Dr. R. Matthewman. In Colombia: the CIAT FSR workers, Drs. J.
Woolley, J. Davis, A. Bellotti, A. van Schoonhoven, C. Sere, R. Howeler,
J.M. Toledo and, of ICA, Sr. Hiriam Tobon. In Costa Rica: the CATIE FSR
workers, Drs. M. Avila, G. Budowski, E. Escobar, M.A. Esnaola, and L.
Navarro. In Mexico: Dr. D. Winkelmann and his FSR colleagues at CIMMYT
and, at INIA, Dr. Antonio Turrent. In Ethiopia: Mr. G. Gryseels and his
FSR colleagues at ILCA. In Kenya: Drs. M.P. Collinson and P.
Anandajayasekeram (CIMMYT), Dr. R.A. Kirkby (IDRC), Dr. J.B. W. Matata
(NARL), Drs. B. Lundgren, and P. Huxley (ICRAF), Dr. P.K. Kusewa and
colleagues at Katumani Research Station, Machakos. In India: Dr. S.M.
Virmani and FSR colleagues at ICRISAT; Dr. R.P. Singh and colleagues at
AICRPDA, Hyderabad; Dr. B.R. Hegde, Dr. K.V. Sampath and Mr. M. Jyoti and
colleagues of the Karnataka State Agricultural Department. In the
Philippine Islands: Drs. D.M. Wood and V.M. Carangal (IRRI); Dr. E.C.
Quisumbing and FSR colleagues (P.I. Ministry of Agriculture); Drs. Ramon
Valmayor and Deli Capasin (PCARRD); Dr. E.L. Rosario (UPLB); Mr. F.
Quero (Ministry of Agriculture, Tacloban, Leyte); Drs. K. Takase, S.T.
Senewiratne and A.K. Auckland (Asian Development Bank); Drs. D. Clark and
T. Hobgood (USAID). In Indonesia: Dr. S. Sadikin and Mr. Inn G. Ismail
(AARD); Mr. S. Draper and Mr. H.D. Matheson (World Bank). In Nigeria:
Drs. C.H.H. ter Kuile, G. Wilson, H.J.W. Mutsaers, P. Ay, Malik Ashraf,
and P. Juo (IITA); Dr. D. Sumberg (ILCA); Mr. D. Feldman (FACU); Mr.
P. Reid (Ilorin ADP); Dr. C. Harkness (KNADP).
 LIST OF ABBREVIATIONS AND ACRONYMS

AARD: Agency for AgriculturalResearch and Development

ACSN: Asian Cropping Systems Network

AFSN: Asian Farming Systems Network

AICRPDA: All-India Co-ordinatedResearchProject for Dryland

Agriculture

CATIE: Centro Agronomico Tropical de Investigaciony Ensenanza

CGIAR, ConsultativeGroup on InternationalAgriculturalResearch

CG System:

CIAT: Centro Internacionalde AgriculturaTropical

CIMMYT: Centro Internacionalde Mejoramientode Maiz y Trigo

CIP: InternationalPotato Center

CSR: cropping systems research

FACU: Federal AgriculturalCoordinatingUnit (Ibadan)

FSAR: farming systems approach to research

FSP: farming systems perspective

FSR/E: farming systems research and extension (Hildebrandin

Sheppard, 1981)

FSR sensu lato: farming systems research in the broad sense of more or
less anything to do with farming systems,real or
invented.
FSR sensu
stricto: farming systems research,here used in the narrow sense
of study of existing farming systems

IADP: Ilorin AgriculturalDevelopmentProject

IARC: InternationalAgriculturalResearchCentre (of the CG

system qv)

ICAR: Indian Council for AgriculturalResearch

- xi -
ICARDA: International Centre for Agricultural Research in Dry
Areas

ICRAF: International Council for Research in Agro-Forestry

ICRISAT: International Centre for Research in the Semi-Arid

Tropics

IITA: International Institute for Tropical Agriculture

ILCA: International Livestock Centre for Africa

IRRI: International Rice Research Institute

ISNAR: International Service for National Agricultural Research

KNADP: Kano Agricultural Development Project

NAES: National Agricultural Extension System

NARES: national agricultural research and extension system

NARS: national agricultural research system

NFSD: new farming systems development; phrase invented for the

purpose of this report

OFAR: on-farm adaptive research (=OFR)

OFE: on-farm experimentation (IITA literature, the experimental

phase of OFR)

OFR/FSP: on-farm research with farming systems perspective (Byerlee

1982), the CIMMYT terminology, adopted here

R & D: research and development

USAID: US Agency for International Development

- xii -
 CHAPTER 1. INTRODUCTION

Context

Since the early 1950s, the population of the world has

approximately doubled and food supplies have approximately kept in step.
These processes have been very unevenly distributed, however, because the
rich, temperate countries have grown slowly in population but rapidly in
food production, whereas the poor tropical countries have, many of them,
more than doubled in population and have maintained, at best, a modest
food surplus. At worst, as in much of tropical Africa, population
doubling times are of the order of 20 years (compound growth rates of 3 -
4 percent) and there is chronic and worsening food deficit. There is,
therefore, a gigantic need for increased food production, both immediately
and on into the indefinite future, until populations shall have stabilized
within the resources available to sustain them. I reject, as being just
silly, arguments to the effect that the world has a distribution rather
than a food crisis; that the rich temperate countries need only give up
feeding grain to animals and ensure equitable distribution for all to be
well. Even temperate agriculture could not sustain an indefinite process
of doublings, and imagination boggles at the socio-politico-economic
obstacles to distribution. Food aid in various forms against local crises
will no doubt be with us for decades but the fundamental need is for the
enhancement of indigenous tropical food production. This can be achieved,
in effect, only by intensification of production on existing arable land
because, on the world scale, the potential for cultivating new lands is
small and for irrigation of dry areas probably rather limited.

Food production in this context means essentially vegetable food

with no particular emphasis on one kind at the expense of another.
Cereals, pulses, oilseeds, tubers, vegetables, and fruits all have their
places; so do animals, mostly as scavenging accessories to crop
agriculture, but sometimes as a means of getting something out of
environments too poor for crop production. The mythology of "the protein
gap" (Payne, 1978) is now dead, and the related mythology as to the need
for more of specific nutrients is nearly dead too. Nutritionally,
reasonably diverse vegetable diets, supplemented with odd animal products,
are a rational objective for agricultural research.

The crucial need - for enhanced yields in existing cultivations

- was perceived by the Rockefeller Foundation in its Mexican program in
the 1940s. The wheat production package followed and was exactly
paralleled some years later by the rice package developed by the
International Rice Research Institute established in 1960 jointly by the
Rockefeller and Ford Foundations. The story of the ensuing Green
Revolution has been often told but not always well understood. The
essence of it was the large-scale exploitation of positive genotype-
environment (GE) interactions (cf. Chapter 2): semi-dwarf, quick-maturing
varieties of both crops plus enhanced environmental inputs (water,
fertilizers, other chemicals) produced far greater yields than could be
produced either by the new varieties alone or by the inputs alone.
Indeed, the new varieties under low-input conditions and the old

-1I-
varieties under high inputs were sometimes actually worse than the old-low
combination. Furthermore, day-length insensitivity and quick maturity
meant that, sometimes at least, the way was open to multiple cropping,
with two or three crops a year instead of only one.

This exploitation of GE effects, by way of semi-dwarf varieties

of small-grain cereals responsive to high-input environments, was not, of
course, new. It was the same as what had been happening for decades in
temperate agriculture, especially in Western Europe and Japan, where
cereal plants had long been getting smaller and inputs higher. But the
Green Revolution did quickly and in one dramatic step what had taken many
years elsewhere.

The Green Revolution was, with qualifications, a triumphant

success. Wheat and rice yields went up (Dalrymple, 1974), consumer prices
came down and adopting farmers profited. The qualifications were that the
technology was only possible where there was adequate water, where farmers
were economically able and willing to provide the inputs, and where the
economic infrastructure of supplies and marketing allowed. Criticisms
(often apparently by politically motivated writers) to the effect that the
process favored large farmers as against small ones and tended to cause
unemployment have been shown to be without foundation; benefits are
virtually scale-neutral (Ruttan, 1977). Other critics have urged that the
use of chemicals in agriculture is environmentally hazardous, but often
without also observing that starvation, to the victims, is also hazardous.

From the widespread (though not universal) euphoria generated by

the Green Revolution there arose the notion that large-scale agricultural
research on tropical food crops could initiate a whole series of new Green
Revolutions, crop-by-crop, new responsive varieties matched to suitably
enhanced inputs. This explains the emergence of the CGIAR system of IARCs
in 1971 and its development to a powerful group of 13 constituent bodies
with a current expenditure of about US$168 million. It was also behind
the marked strengthening of NARSs in the past decade and the emergence of
a body within the CG system (ISNAR) devoted to developing national
systems.

From the start, the CGIAR and the IARCs were committed to the
interests of the small farmer in respect of annual food crops (cereals,
pulses, tubers), with some effort also on animals (especially in semi-arid
Africa). Perennial crops were not included (a curious omission in view of
their social importance) and cash crops (at least export cash crops) were
deliberately eschewed. The IARCs have done much good work and, either
directly or indirectly (through NARS), have had diverse local impacts on
agricultural production. But there have been no more Green Revolutions;
in retrospect none was to have been expected. The Green Revolution
succeeded where wheat and rice were already grown under irrigation and
were susceptible to application of the new technology. Elsewhere, the
same crops in rainfed lands have hardly been touched and it is doubtful
whether upland rice, on average, yields any more now than it did 20 years
ago. Furthermore, irrigated wheat and rice farmers, though small, were
evidently prosperous enough to take up the new technology quickly, even
enthusiastically, and lived in places where it could readily be made
available. Most tropical small farmers, however, have no such advantages;
for them, money is scarce, advice hard to come by, markets uncertain,
supplies and communications poor. Can one really imagine a Green
Revolution in cassava production, whatever the varieties and technology
that might be available? I think not.

In short, I argue (following Wortman and Cummings, 1978) that

the Green Revolution proper achieved the big, quick, spectacular success
that was necessary to get tropical food crop research on the road (which
it did) but that we are now faced with a much slower and more laborious
process: of pushing up yields, step by step, a little at a time; of crops
grown in socio-economic circumstances that simply are not susceptible to
revolutionary change, or are susceptible to it only under really massive
governmental intervention. I shall argue later in this report that the
last will sometimes be necessary. Here I note that none of the very many
deeply informed people that I have talked to in the past few months seemed
to believe in revolutions. All thought and spoke in terms of evolutionary
change, of research matched to the socio-economic circumstances of farmers
and thoughtfully applied and diffused, but in time scales of decades
rather than of years.

We now reach (somewhat circuitously, I admit) the context of

Farming Systems Research (FSR). FSR developed, I think, from the
(intuitive as much as explicit) realizations that: the Green Revolution
was a "one-off job" that hit the socio-economic centers of interest of two
important groups of farmers, the irrigated wheat and rice growers; that
no more revolutions were in prospect; and that other farmers would adopt
new technology only when they themselves (like the wheat and rice growers)
perceived it to be in their own socio-economic interests and capacities to
do so. The last phrase, of course, raises the question of how to identify
those same socio-economic interests and capacities. This is what a
substantial part of FSR is about.

Tropical Agricultural Systems

The standard work on the subject is that of Ruthenberg (1980)

and I have used it much. Reference may also be made to Duckham and
Masefield (1970) and Grigg (1974) (though not limited to the tropics);
also to Webster and Wilson (1980), the standard general text on tropical
agriculture. The third (1980) edition of Ruthenberg contains appendices
by Collinson and by Zandstra summarizing FSR ideas. Johnston (1958) gives
a useful systematic treatment of West African systems.

No simple but satisfying classification of tropical agricultural

systems is possible. Any general classification would have to be
multi-dimensional, taking in rainfall, altitude, crop/stock relations,
annual/perennial crop components, and irrigation at the very least. Two
dimensional classifications (Figures 1.1 and 1.2) are always deficient.
The figures therefore merely serve to set the present study in context.
They are based on Ruthenberg and they identify only macrosystems. On the
ground, there is virtually infinite variation in detail and, at a
practical level, any useful definition must be geographically restricted.
 -4-

Rgure 1.4: TropicalAgdrculuralSytems (1)

Perennb -ia

I ShlttIngCultieation =

WET 2 BushFaliow I…--

BushIi'

hN~~~~~~~~~~~~~~~~~~~~~~~~~
I I 1Ii

- Nomadic | J
Ranching - - - - - - - -
Grass

DeW Amuble|
H t_ ____ _

Permannent
Upland

Examples(mostlyreferredto In this report): (1) (2) IITAprogram In wet WestAtrica, (3) ICRISATworkIn India,CIMMYrprogramsIn SAT
Africa, ILCA'sHighlandProgram,CIAT'sbean program In the Andes: (4) IRRI's ACSN(now AFSN)In southeastAsia,(5) ThePhilippine
coconut program;(6) ILCA(1980, 1983);(7) ClAT'spasture program In the llanos,EMBRAPA In the cerrodosof Brazil.
 -5-

Figure1.2: TropicalAgricultural Systems(2)

Nonmadic
Herding&
Ranching

Dryland Dryland
Mlxed Seasonal
Farming Cropping

Shifting
Ranching 9atk Cultivatlon
Static
Mixed
Farming
Static
Cultivation
wet

LOw CROP/STOCK
RATIO High
 - 6 -

At the limit, no two farms are identical, so all FSR activities work at a
micro-level of classification, a level at which farms may be judged
sufficiently alike (for the purpose in hand) to be deemed members of a
local farming system.

Formally, the problem is common to nearly all biological classi-

fication, whether of plants, animals, soils, or farming systems. The
question is: how to partition variation so that it is maximized between
taxa and minimized within taxa? Even when useful discontinuities occur
(and they are few in farming systems) the problem remains of coping with
the residual continuous variation. The practical answer that always
emerges is that a taxon (an entity of classification) is that which is
defined by the competent taxonomist with a particular purpose in mind.
Thus the plant taxonomist might be content with Solanum tuberosum for the
cultivated potatoes but the potato botanist would need to recognize
cultivar groups (based on cytogenetic criteria), clones, maybe even
sub-clones. Similarly, the coarse classifications of Figures 1.1 and 1.2
may serve for some purposes, but are quite inadequate for any kind of
FSR. Thus Permanent Upland Systems are very different things in India,
Ethiopia, Kenya, the sub-Sahel of West Africa, and in the Andes. And,
within these regions, numerous local systems can (and indeed must) be
recognized.

I shall dispose here of another point which is sometimes a

source of confusion. Mixed cropping is very common in'tropical
agriculture and one encounters the notion that the study of it is, in some
sense, FSR. It is not. Mixed cropping is rare and unfamiliar in
temperate agriculture but frequent and important in tropical agriculture
because it contributes both to gross yield and to stability of overall
performance. That mixed cropping has favorable features has long been
recognized (in India back into the 19th century at least) and there is
certainly room for more experimentation on and better understanding of the
subject. I know of no comprehensive review but a partial one has been
presented by Willey (1979) (see also ICRISAT, 1979). One would like to
see a full treatment. Meanwhile, I regard it as simply a phase of
agronomic-physiological research that must have something to contribute to
tropical agricultural development. The nomenclature is sometimes
confused. Figure 1.3 is based on Ruthenberg. There will be no more about
mixed cropping per se in this report though researchers, of course,
frequently encounter it in the field and the subject is present in the
programs of several IARCs (notably, CIAT, CIMMYT, IITA, ICRISAT, IRRI) and
NAR systems (as in India, the Philippines, and Indonesia, to name but a
few) (Chapter 5).

Figure 1.3 relates to mixed cropping with annuals, the aspect

that has had most study. Perennial crops are no less often grown mixed,
with each other and with annuals. Watson (1980, 1983) gives valuable
systematic treatments of such mixtures, of which there must be hundreds of
variants. Again, perennial crop mixtures are not per se farming systems,
though they may be very important components of them.
 -7

Rgure1.3:AnnualPlantCroppingand IntercroppingSystems

On a Substantial
Piece of GroundOv&
a Whole Season

L J: z [ ~~~~~~~~~~~~~~~~More
Than

One Cro at ore Than One

KAl
a Tlrne Crop at a Tlrne

l~~~~~~~~~~~~~Sqeta .roppnI lntI

| Monocropping
< ~ ~ ~ ~ ~ ~~Copin

Mixed

Synchronious
Planting Deleberately'
D-s-synchronous
Flanting

ShTip Row Mixed Relay

Cropping Cropping Cropping Crpplng
 Innovation and Change

It is often said that farming systems tend to be stable unless

perturbed. So they may be, but perturbations are rarely wanting for long.
Thus, while bush fallows and shifting cultivations have no doubt existed
in the wetter tropics for millenia, they must have changed greatly in
structure from time to time, as new crops and stock became available and
new opportunities appeared. Bananas entered Africa maybe 1,000-1,500
years ago and locally transformed the agriculture; the sweet potato did
the same in Oceania, and the coconut was not without effect in its immense
travels. Maize, cassava, and the aroid tubers have been in Africa only
for about 500 years and bananas and rice in America for a like time.
Cattle, sheep, goats, pigs, and horses are all Old World beasts introduced-
to the Americas about 500 years ago. All these (and many more)
introductions wrought revolutions, invented and carried through by
farmers, perhaps rather by rural societies as wholes.

So stability, if it ever existed, was always punctuated by

change. On a short, backward view, however, a quasi-stability is often
apparent: multitudes of small tropical farmers are probably now living
and working much as their grandfathers did. Any apparent stability is
now, however, plainly impermanent. Virtually all tropical agriculture has
been or is being destabilized by population pressures that demand more
food from limited (always finite, all too often diminishing) resources.
Change is therefore, willy-nilly, rapid, even on personal time scales. It
is the business of agricultural research to try to promote change in a
socially favorable sense and that phrase lies at the heart of farming
systems research, broadly understood. Futurology is no part of my remit,
but I observe in passing that agricultural research can defer acute
conflict between population and food supplies; it cannot, by any means
technically conceivable, assure food for indefinitely increasing
populations. In a serious sense, therefore, population control is a more
fundamental matter than food production. The point has been made often
enough but bears repetition.

There are several causes and agents of technological change in

agriculture. I recognize four of each in Figure 1.4 which could, perhaps,
for some purposes, be subdivided. Comments follow.

Case A, the more or less static socio-economic environment, must

nowadays be pretty much an abstraction if only because there can be few
places in which population and/or economic forces favoring "efficiency" do
not press hard upon agriculture. Case D has, in the past, been of great
importance; one thinks of the history of the great plantation crops and
of small-farmer export crops such as cotton and tobacco, also of new local
demands for such products as wheaten bread and beer; alas, such
opportunities seem to be becoming scarcer as the rich countries promote
sugarbeets and oilseeds or substitute synthetic materials for natural
fibers and drugs. In our context, that of farming systems, cases B and C
are surely the most important, the former reflecting the rising needs for
food of burgeoning populations, the latter the ever increasing pressure on
land or other limiting resources by those same populations. The preceding
is a short list of the external causes of change (or [A], non-change).
The lower part of Figure 1.4 recognizes four agents of change, remarks
upon which follow:
 -9-

Fgure 1A: Causesand Agents of Technological Change

Socio-Enonomic
Environnment

Static| Changing

Altered Mabrkets ShiltIng NewNMarket

for Estabuished
typpotunities, Resource
Products Pressures
wous. Extranm

Adoption of NewTechnologyby Famiea Due to:

(1) Famer' own prceptionsof economic
opportunltye
(2) Promotlonby public RD& Es,em. with or
without FSRcomponent,
(3) Promotionby govemmental/binancal/
commercial agenciesof actions designed to
generate changes(often maJorones),usually
with speclfic RD& Ecomponents.wlth or
wlthout FSR/OFR component,
(4) Promotlon bVcommercial bodies selling
products,often with a strongOFRcomponent.

Note: Thenomenclatureof this dclgram & the accompanying text antIcipates that set out In the next section.
 - 10 -

1. Examples are rarely documented but probably innumerable. Most

farming systems have evolved thus and continue to do so. It is a
substantial agent of change in temperate agriculture where enterprising
farmers often do something first and leave the research, development, and
extension system to tidy it up. But it is far from unknown among tropical
small farmers, for example: the recent evolution in the Antioquia area of
Colombia of potato-maize-bean cultures from more or less random mixture,
through row cropping to an orderly, manured rotational-relay-cropping
(with numerous local variants); in Costa Rica, evolution from coffee with
full Erythrina shade, to coffee with pollarded shade, to to pollarded
shade with a third-storey tree, Cordia; there are numerous examples of
farmers exploiting a good new variety quickly, even to the extent of
stealing it before release; and Dr. G. Gibbon has told me of Sudanese
farmers who adopted camels for ploughing because the advent of lorries for
load carrying had reduced the price of the beast.

2. The Green Revolution is a plain example (indeed the best one

extant) of a research-push innovation, lacking any substantial and formal
farming systems component, that went well because farmers rapidly
appreciated the benefits. The farming systems perspective (FSP, see
below) was, of course, there (though maybe intuitively) in the
imaginations of the parent foundations and researchers. All current
efforts by IARC/NARS groups, separately and jointly, are of this nature,
with a strong and generally increasing farming systems component. This
track is characteristic of all publicly funded agricultural research,
development, and extension efforts and it is, of course, mostly aimed at
meeting the pressures generated by causes B and C above.

3. The promotion of new crops (category D) is possible only

when markets and adequate supply are assured. Historically, all the great
plantation crops come here, typically estate cultures, but often with a
strong small landholder sector (as in sugar, rubber, bananas, coconuts,
tea, cacao, and coffee). Groundnuts in Senegal, cotton (very widely in
Africa and Asia), and tobacco in all the continents are examples of small
landholder developments that worked pretty well because the
official/commercial agencies responsible were able to impose the necessary
production disciplines. The current promotion of wheat in semi-arid West
Africa is a contemporary example. All such developments require a strong
technical/rural development/co-operative base provided by governments,
banks, or industries or combinations of them. Broader rural development
schemes (e.g., irrigation works, trans-migration or settlement schemes)
make similar organizational demands but are usually aimed at enhanced
production of existing crops (B and C) rather than promotion of new ones.
The farming systems component was traditionally small and the technn!-v -
transfer bit of new crops programs was pretty much a "top-down" pro -
(Chapter 2, "'Upstream' and 'Downstream"') as far as the smalL landhoLder
was concerned. Maybe this is changing and, nowadays, such schemes, if
they have a research component at all, recognize the need for a farming
systems element (see Chapters 4 and 5).
 - 11 -

Rgurei.5: TheThreeMain Categodesof FSRWork (sensu lato)

Univers of FS

ChosenFS
hairty FRsensu
snt
NarraowlV
Deflned

Sewshnge,
Enharcernentof OFRFSP
Existlng FS\

NewFS _NS

for Old-- v
 - 12 -

4. Fertilizers, herbicides, fungicides, drugs, insecticides

(indeed, agrichemicals generally), and machinery come to mind. Plant
breeding companies are already strong (and highly beneficial) in temperate
countries and must be expected to develop, too, in the tropics
(especially, perhaps, for hybrid maize, in the knowledge that it has done
very well in Kenya). Such businesses have a long history of on-farm work,
at least in temperate countries, perhaps less so in the tropics. The
objectives of on-farm work by commercial companies are not only to test
innovations (chemicals, varieties, machinery) in farm practice but also,
of course, to advertise the product.

5. I shall now try to place the foregoing discussion in the

context of the present report. Of the four main agents of technological
change identified above, it has been the growing interest of publicly
supported research organizations (agent 2 in the list of Figure 1.4) that
has been mainly responsible for the promotion of explicit farming systems
ideas in the research process. The ideas have developed rapidly and not
very coherently. Hence this report is mostly concerned with the work of
IARC/NARES bodies, and my travels were virtually wholly concentrated upon
a sample of them. The discussion above, however, will have made it clear
that farming systems ideas are relevant to all four agents of change, in
tropical and temperate agriculture. I shall return briefly to the theme
later but the bulk of what follows is devoted to the CGIAR/NARES context.

Anticipation of Argument to Follow

On the excellent advice to a writer to say what he is going to

say, say it, and then say he has said it, I interject here a short
statement of what I understand by FSR in the broad sense. I have come to
recognize three (fairly) distinct elements and use the nomenclature
summarized in Figure 1.5. Justification of the terminology and the
classification will appear later. Referring to the figure, I define the
three elements as follows:

1. FSR sensu stricto is research on farming systems as they exist,

their description, analysis, classification and understanding. It can go
to any depth but typically goes deeply into the agriculture, economics and
social context of the system studied. It is essentially an academic
activity good for generating Ph.Ds, but not much use to agricultural
research.

2. On-farm research with farming systems perspective (OFR/ FSP)

starts from the FSP bit which is just enough FSR sensu stricto for the job
in hand (but no more). It uses the FSP to help to define the on-farm
research (OFR) necessary for practical progress. It is a "style" of doing
agricultural research founded on the well-justified assumptions that
changes need to be adapted to the circumstances of their users and that
on-station experiments by no means always predict farm experience. The
OFR/FSP "style" broadly assumes that progress will be step-wise rather
than revolutionary and devotes itself to a cautious, empirical,
evolutionary process. The terminology is that of Byerlee et al. (1982).
 - 13 -

3. New farming systems development (NFSD) contrastswith the

preceding in seeking to generate revolutionrather than evolution,to
build radicallynew systems ab initio. It is essentiallypractically
orientatedagriculturalresearch and must (obviously)be founded on at
least some FSP (but maybe not much). It differs from OFR/FSP in degree
rather than in nature (both seek to generate beneficialchange) but it
necessarilyhas a lesser OFR component (indeed it may have almost none, as
we shall see later). The terminologyis mine, invented because,so
far as I can ascertain,no one has invented a phrase that clearly
distinguishesthis sort of activity from other aspects of FSR sensu lato.
From here on I shall use this terminologyand, if I need to refer to FSR
in the rag-bag sense unhappily prevalent in the literature,shall use the
term FSR sensu lato.

Finally, to close this chapter and to anticipatea little

further what is to follow, I shall argue that; (1) the practice of
OFR/FSP is just becoming (and quite reasonablyso) a standard component of
the large body of agriculturalresearch that seeks to generate step-wise
changes in farming practice; it has, I think, proved its value and has
come to stay; (2) NFSD, though more ambitious and much more uncertain of
outcome than OFR/FSP, is no less necessary; in the real world not all
changes should or can be step-wise; many farmers' circumstancescry out
for radical alterations,however difficult they may be to achieve in
practice; and there must surely sometimes be room in agricultural
research for somethingwider and more imaginativethan the step-wise
process, even if the last, is, in real life, the norm.
 - 14 -

CHAPTER 2. FARMING SYSTEMS RESEARCH IN GENERAL(FSR sensu lato)

History

Knowledge of, and interest in, tropical farming systems is not

new. Many government agricultural officers in the colonial days had
profound knowledge of the systems with which they lived, often for
decades. That knowledge usually died with them or was lost in the files.
Publications such as those of Tothill (1948), Jameson (1970), and Arnold
(1976) summarize a little of the great knowledge that was available, and I
have no doubt that similar works written in French and Dutch could also be
identified. Bunting and Watts-Padwick (1983), in a recent perceptive
article on tropical agricultural research, remind us that a quite explicit
statement as to the importance of understanding what farmers do (i.e.,
FSP) was made as long ago as 1889 by Voelcker, writing of Indian
agriculture.

The important ideas that have emerged to change the scene in the
past 20 years or so have been, I think: (1) that the methods of farm
management economics, well established in Europe and North America for
decades, could be adapted to the circumstances of tropical small farmers;
(2) that tropical small farmers were economically rational (though not
necessarily profit-maximizing); (3) that risk, uncertainty, and therefore
caution were dominant features of their existence; (4) that statistical
and operational research methods, with or without computers, had provided
means of thinking about uncertainty and the working of complex systems;
and (5) that innovations proposed by agricultural researchers for
extension were, rather often, simply not adopted or were adopted only in a
partial or modified form.

The notion that small farmers in general present problems

unfamiliar to conventional economics goes back to Chayanov in 1925
(Thorner et al., 1966; Levi and Havinden, 1982), though Chayanov's view,
that the peasant economy really demanded a different "system of
economics, would not now be generally accepted. The adaptation of
farm-managementeconomic ideas to the circumstances of tropical small
farmers was pioneered in India in the 1950s (Mellor, 1966) and developed
in Africa by D.W. Norman and M.P. Collinson during the past decade
(Norman, 1974-83, Norman et al., 1981, 1982; Collinson, 1982, 1983); this
latter initiative developed into the OFR/FSP now widely practiced. Other
useful general references are: Andrew and Hildebrand (1982), Avila et al.
(1982), Byerlee and Collinson (1980), Byerlee et al. (1982), CIMMYT
(19 81a, b) Dillon et al. (1978), Eicher and Baker (1982), Galt et al.,
(1982), Gilbert et al. (1980), Harwood (1979), Ngambeki and Wilson (1983),
Perrin et al. (1976), Shaner et al. (1981, 1982), Sheppard (1982), and
Zandstra et al. (1981).

Formal "systems" ideas have proved, I think, far less pervasive.

Full understanding of a system implies complete numerical specification,
and this turns out, in the context of tropical agricultural research, to
be rarely necessary and sometimes just a waste of time (as I shall argue
further below). At all events, the important thing has turned out to be
the systems perspective (i.e., FSP) rather than the formal models. To
 - 15 -

say this is not to deny that models have their uses, even in an OFR/FSP
context (see examples in Chapter 3) and they have certainly found a niche
in rich temperate agricultures, as the pages of the journal Agricultural
Systems and such useful animals as the "Edinburgh Model Pig" testify.

FSR sensu lato entered CGIAR thinking at an early stage and it

figures in the mandates of most of the centers. This stemmed, I suppose,
from the realizations that more Green Revolutions were not in prospect,
that point 5 above was true, that agricultural economists had their uses,
and that tropical food crop research had somehow to be adapted to farmers'
circumstances. These were all, I think, accurate perceptions but, in the
outcome, FSR was interpreted by the centers in very diverse ways, as the
important TAC Stripe Review (Dillon et al., 1978) made clear. The
diversity of views is hardly any less now, and they range from a little
more or less pure FSR sensu stricto, through much OFR/FSP to some NFSD
(or, at least, component research pointed towards NFSD).

Small-Farmer Characteristics

The socio-economic characteristics of small farmers have been

stated many times (e.g., Collinson, 1984) and may be summarized as
follows:

1. they are poor and have little ready cash;

2. loans to them are usually unavailable or expensive;

3. they are conscious of an uncertain environment, of cash

shortage, and of family responsibilities and therefore;

4. they are risk-averse;

5. they often suffer cyclical labor shortage and

under-employment;

6. they may have opportunities for competing off-farm

employment;

7. they are economically rational but not necessarily

profit-maximizing because;

8. they (like the rest of us) have their own scales of

utility;

9. they live in countries in which the social infrastructure

of markets, supplies, and communications is often weak and
not to be relied upon;

10. they live in societies which normally have fairly clear

codes as to what is socially acceptable and what is not.
 - 16 -

This is a formidable list. It differs in nearly every point

from any comparable list that could be written for rich farmers in
temperate countries (or, indeed, locally in tropical ones). Given such
constraints, and they do indeed seem to be universal, it is not surprising
that small farmers generally approach innovation cautiously because it may
be both costly and risky, however profitable.

Ideas on farmers' attitudes to innovation are summarized in

Figure 2.1 and comments follows. The scheme of Figure 2.1 is quite
general and must apply whatever the source of innovation and whatever the
scale of operation of the farmer. Sources of innovation may be:
endogenous to the farming community or exogenous (public R & D, official
promotion, commercial) (see Figure 1.4). We concentrate here on the small
farmer in the FSR context.

Costless innovations

The leading example is of new crop varieties, which will

generally be accepted and quickly adopted if similar to the established
one(s), but plainly better in some significant respect such as yield or
maturity (usually earliness); more subtly, the more stably performing
(reliable) variety will be preferred, though it may take several years of
experience for a body of farmers to decide. New varieties will be
rejected, as many (probably most) have been, if they do not meet farmers'
needs in respect of yield on the farm (not the experiment station). They
will also be rejected if they do not meet farmer requirements in some
secondary field characteristics such as yield of straw for stockfeed,
recovery from grazing in barley, or strength of stalk in maize destined to
support climbing beans. Again, they will be rejected, or at least
discounted, if they fail to meet local quality requirements such
as seed size in pigeon peas and chickpeas, grain color in sorghum (white
or red locally preferred), grain color and texture in maize (usually
strong local preferences), seed color in beans (only a blotchy pink will
do in the Antioquia area of Colombia). But there are trade-offs, and an
exceptionally good variety may be taken up even if quality is less than
ideal; thus I heard of a purple-seeded bean that was going well in
Colombia because it was outstandingly drought-resistant, and one
recalls that IR8 rice succeeded despite a "chalky" grain character.
Market preferences are not immutable if the price is right; and tastes
can change.

Cash inputs needed

The obvious examples concern chemical inputs such as

fertilizers, herbicides, fungicides, and insecticides, all relatively
small recurrent expenditures; and capital expenditures on such items as
new stock, fencing, field equipment, spraying equipment, grain storage
facilities. The former have shorter pay-back times (say, a crop season)
but all are subject to the exigencies of the farmer's cash supply and
discount rate. It will be noted that fertilizers are divisible,
so that low-level partial adoption is feasible, whereas the other items
listed are not; furthermore, capital expenditure normally incurs some
prospect of future demands for maintenance costs.
 - 17 -

Flgure2.4: Farmers'Decisionson Innovation

Instfutknal
support
Usually
| Inrnovtlon l Essential

Costle | Costs | Costs

_ - = No <ttroctivef~~~N

Cash Eff~~~ort Paty

Modifiab,9a_ 7|

o Y
~~~~~~~~~~~~~N

l ~~~~~~~~~~~~~hl
X~~~~~~~~~~o
I e
 - 18 -

Innovations involving cash expenditure are notoriously likely to

be rejected. First, the inputs may simply not be available, either
generally or at the local village level due to failure of distribution
mechanisms. Second, even if demonstrably attractive technically, the
farmer may be unable to find the cash or unwilling to take the risk of
borrowing. If the farmer does adopt, say, the use of fertilizer, then he
is likely to do so at a lower level than the recommendation and later
maybe adjust upwards to approach his preferred marginal rate of return
(Figure 3.4).

Extra effort needed

The growing of two short-cycle crops instead of one longer-cycle

crop might be virtually costless in cash terms, but make considerable
labor demands of the farmer, of his family, and of his work animals.
Examples are the undercropping of coconuts in the Philippines and the
addition of ratoon rice and a quick pulse crop after maincrop rice in
Indonesia (see Chapter 5). Numerous conflicts between such demands and
those of other farm enterprises or off-farm earnings occur. Economists
examining such problems by partial budgeting techniques will, of course,
translate them into cash flows (see beginning of Chapter 3). And one
should recall that an already hard-working farmer (who is not necessarily
a profit-maximizer) also puts a value on hard-earned leisure.

Compound innovation

All the more radical innovations come here. Thus the ICRISAT
proposals for the black cotton soils of India, and the Asian Cropping
Systems Network proposals for irrigated rice, demand: new varieties
(generally of quick maturity), cash inputs for chemicals and tools and
more labor, all with complex biological and social interactions. ILCA
proposals for upland Ethiopia involve a fairly modest change in cropping
pattern but a profound change in kind of cattle, affecting working methods
and equipment and milk production. There is no sharp discontinuity
between such proposals (all already in some degree of adoption) and the
still more radical proposals for virtually new systems (elsewhere in this
report referred to as NFSD) that are now being explored; for example, the
studies by IITA in Nigeria that seek to develop stable farming systems on
poor, acid erodible soils in place of shifting cultivation (see Chapter 5
for all examples). All such compound innovations, from the simpler to the
most adventurous, will need strong institutional (ultimately governmental)
support to provide the essential communications, material supplies,
marketing, research and extension services, credit, seeds, stock,
and so forth.

To conclude this section, the heart of the matter is surely

this: that, if innovations are to be effectively promoted, they must
either fit the farmers' economic circumstances or those circumstances must
be changed to make the innovation work. OFR/FSP seeks the former, in
shape of relatively simple, even unit, innovations which fit the
circumstances; complex changes, including NFSD, must generally seek to
change the economic circumstances to fit the change.
 - 19 -

A final point is worth making. In the new-found belief that

small farmers are economically rational, we ought, perhaps, to be wary of
the assumption that they are always right, have always optimized within
their own constraints. Even rich, well-educated farmers can be
collectively wrong under persuasive advocacy from articulate neighbors or
from fertilizer salesmen or simply from conservative retention of a
judgment that was correct in the past. I can see no reason to suppose
that small farmers are immune to error from similar causes. Sometimes,
perhaps, the outstanding farmer (even the scientist) may point the way
against the general trend of opinion?

Systems

The ideas of systems grew out of wartime operational research

applied later to industrial problems. Numerical application became widely
possible when computers became available. Thus complex, interacting flows
over time and space, far too complex for algebraic or analytical treatment
and far too bulky for pencil-and-paper numerical methods, are now
generally accessible to fairly exact specification (provided the data are
good enough). Thus, no self-respecting treasury is now without
macro-models of the economy, weather forecasts are at least somewhat
better than they used to be, road traffic, tide and river models are much
used by civil engineers, and any substantial manufacturing company runs
optimizing models of plant throughput, cash flow, and distribution
system. Even world systems have been attempted.

Systems ideas have also entered technology-based agriculture in

the rich countries where enterprise, whole-farm, and production-unit
models, usually with a maximizing/optimizing object in view, are ever more
widely used; when the physical, biological and economic data are good (as
they commonly are), such models are generally agreed to be potent aids to
efficiency. In agricultural research, too, systems models have their
place, for example, of crop and animal growth, of epidemics, and of
production systems.

There is a vast literature of systems in general and a

substantial one on agricultural and biological applications on which the
following are standard general texts: Dalton (1975); Dent and Anderson
(1971); Spedding (1979).

Any systems understanding starts from qualitative enumeration of

components and their interactions, goes on to quantitative description of
states and flows, and, only when the latter are tolerably well defined,
attempts modeling or synthesis. In our context, that of FSR,
understanding rarely gets far beyond the initial, qualitative stage and is
usually confined to a bit of the whole, or a subsystem (Figure 2.2).

In Chapter 1 I made the point that the macro-systems such as

permanent upland systems, irrigation systems, and so on are far too wide
and heterogeneous to be much use for the FSR worker; at the other extreme
the single farm is far too narrow. The choice of what to call a system
for practical purposes is always arbitrarily determined by the inquirer's
interests. I return to the point later in discussing what constitutes a
 - 20 -

FRgure
2.2: An AgriculturalSystem

/ SuibsVtems \Suibsems \\

\ ~~~~~~~~Animal/

\~~~~~~Sbvtm
<SZ
 - 21 -

"recommendationdomain" (Chapter3) and note here that, for Figure 2.2, we

imagine that a set of farms sufficientlyhomogeneousfor the purpose has
been identified.

The grandly "holistic"approach of FSR sensu strictowould be

satisfied by no less than a descriptionof the entire system, Z. More
realistically,and certainly for OFR/FSP purposes,a subsystem is selected
(e.g., A or A', depending on the interests of the investigator)in
awareness of the interactionswith other subsystemsimplied by the
intersectionsmarked with dots in the diagram. Sub-subsystemC' is a
crop-non-interactive one (should such exist). Though diagrams such as
this are helpful in illustratingideas, they would have to be
multi-dimensionalto be accurate and can give no quantitativeinformation
about componentsand interactions. At a low, far from truly holistic
level, flow charts can convey a good deal of quantitativeinformation
(examplesin Hart, 1975; Gibbon, 1980; Dillon and Hardaker, 1980) but
even they quickly become too complex to be easily read.

It would not be difficult,but might not be very profitable,to

invent diagrams of the type of Figure 2.2 for the examplesdiscussed in
Chapter 5. Thus for the CIAT bean program subsystemsB and C would be,
for all practical purposes, absent and A would be representedby a set of
strongly interactiveannual crops; ICRISAT might be thought of as
concentratingon A', cropping of the black cotton soils, interactivewith
the draft animal component; ILCA might be regardedas concentratingon C,
the cattle, interactivewith A, but in the absence of B; the Philippines
coconut programwould contain all three subsystemsbut with concentration
on the intersectionof A and B, undercroppingwith annuals.

In real life, therefore,systems isolatedfor study are always

subsystemsarbitrarilydefined for the purpose in view. They are never
holistic in any serious sense of that rather over-usedword. In practice,
what is wanted is sufficientunderstandingto attain the necessary level
of FSP and no more. I wish the words holism and holisticwere avoided in
FSR contextsexcept when a really deep analysis of a whole-farm system is
being attempted,that is, FSR sensu stricto (approached,e.g., by Norman
et al., 1982) for the Nigerian Savanna area). For OFR/FSP, a partial,
non-holistic,subsystemknowledge will suffice or, anyway, has to suffice
in practice.

FSR sensu stricto and OFR/FSP

The essential distinctionsbetween FSR sensu stricto,OFR/FSP,

and NFSD have already been drawn in Chapter 1 and in the earlier parts of
this chapter. I elaborate them in this section,but defer consideration
of operationaldetails of OFR/FSP until Chapter 3. Figure 2.3 extends the
very brief statement of Figure 1.5. It makes clear that, a specificFS
having been chosen for attention,deep analysis of it leads to FSR sensu
stricto, but superficialanalysis (FSP) sufficesfor OFR/FSP. Some
readers might object to the implicationthat superficialanalysiswould
suffice to decide that profound change and hence NFSD were necessary,but
I think the implicationis correct. It requires, for example,no deep
socio-economicanalysis of numerous shiftingsystems in the humid African
 - 22 -

Fgure 2.3: FSRsensustticto, OFR/FSP

and NFSD

at Large

kintcto FS Modiflale
d Neeped
FS l

of ~4
Chse NFSD.
 - 23 -

tropics to conclude that they simply cannot sustain ever increasing

populations,effective though they may be to sustain small, stable
populations. No amount of FSR can negate the brute technical facts of
old, erodible,infertile soils under heavy rainfall. The diagram does,
however, make it clear that understandingat the FSP level will (or at
least should) contributeto any NFSD that might be invented.

FSR sensu stricto and OFR/FSP, though quite different in

intention and outcome,have, at least, initial stages in common as Figure
2.3 shows. I shall now examine the overlap in a little more detail. The
listing below distinguishesbetween the broad survey, or background,work
needed to identify the chosen FS and the analysis of it, once chosen. The
listing follows.

A. BACKGROUND

1. Objective:

To understandthe agro-ecosystemof an area well enough to

identify specificFS worthy of more detailedanalysis.

2. Activities (in descendingorder of scale):

a. Understandingof climate, soils, topography,

vegetation,biotic factors (such as erosion and
fire) of a substantialarea of land; by collation of
existingknowledge and some survey if necessary.

b. Understandingand mapping of land use, including

forestry, of the area.

c. Understandingof agriculturalcomponentsin broad

terms, including cropping/stockingpatterns,
perennial crop usage, forest/timberrelations,land
tenures.

d. Identificationof "target"FS for detailed study on

basis of special interestor social importance;
possible designationof characteristic("benchmark")
sites/areas.

3. Observations: broad patterns defined above will usually be at

least moderatelywell known but some filling in may
be necessary,especiallyas to item d.

B. ANALYSIS

1. Objective

Analysis of chosen FS with aim simply of understandingthe

system (FSR sensu stricto) or of identifyingpotentially
favorable innovations(OFR/FSP).
 - 24 -

2. Activities (in ascending order of detail):

a. Analysis at verbal/descriptive level; identification of

crops, annual and perennial, and of animal components
(including fish and fowl); cropping patterns in space and
time, rotations; crop/stock integration; farming-woodland
relations, sources of fuel and timber; influence of land
tenure; technology used; labor; within and between
family relations; farmers' objectives; cash and energy
flow patterns over time; utilization of farm products,
marketing; more or less intuitive identification of
limiting factors.

b. Analysis at semi-quantitative level; elements

listed under (a) quantified by appropriate surveys
in terms of cash/energy/labor flows; development of
descriptive diagrams; some explicit economic
analysis (e.g., partial budgets); identification of
limiting factors.

C. Analysis at system-synthesis level; extension of

(b) to numerical computer models with object of
specifying the system fully, identifying critical
interactions, predicting the effects of specified
changes or perturbations, investigating robustness
of data and assumptions.

3. Observations

FSR sensu stricto will tackle 1 and 2 in depth and may

proceed to 3. OFR/FSP will do just as much 1 and 2 as is
deemed necessary and will either eschew 3 as being
unnecessary or use computer models rarely, cautiously, and for
specific purposes.

The listing is obviously highly generalized and would rarely,

perhaps never, be fully followed in practice. Thus ICRISAT, with rather a
wide mandate for crops in the semi-arid tropics, has done extensive survey
and agro-climatological background work under A(2) in India and Africa but
has, naturally, concentrated on cropping pattern improvements in
Peninsular India; while ILCA has naturally (and by mandate) concentrated
upon animal production systems in Africa, especially in upland Ethiopia.
The crop-oriented institutes such as CIAT, CIMMYT, and IRRI, under their
mandates, can perfectly well by-pass most of A(2) and look hard at their
chosen crops (while keeping an eye on others and on stock). Thus the
notion that a sequence of ever finer analyses of agro-ecosystems will
finally reveal those farming systems worthy of detailed study is a bit of
a fiction. The FS chosen are, in practice, chosen pretty much on the
basis of the mandates of the researchers, of their skills and
understanding, and of sheer practical convenience.
 - 25 -

Under B. ANALYSIS, we need only observe that, as remarked

above, FSR sensu stricto will go deep (certainlyin activities1 and 2,
maybe also in 3, take a long time about it, and will choose systems to
study on the basis of its own criteria of importance,social, political,
historical, scholarly,or whatever. OFR/FSP will analyze the chosen
system superficially(in the literal,not denigratory,sense), as quickly
and cheaply as possible, with the object of finding out where and how
research may have a useful economic impact. The work will lie largely in
activities (a) and (b) but will be highly selective in those areas and
will only very rarely touch activity (c). We explore this further in
Chapter 3.

"Upstream"and "downstream"

The words "upstream"and "downstream"recur repeatedlyin the

literatureof FSR. I do not propose to use them often but they must be
explained and interpreted. I remarked above that one of the several
reasons for the emergence of FSR as a component of the researchprocess
was that, too often, ideas proposed for extension failed: farmers did not
adopt them or adopted only partially or in modified form. Such failures
have certainly been numerous: of varieties that looked good in the
breeders' plots but were rejected on the farm; of tillage,fertilizer,or
plant protectionrecommendationsthat were technicallysound on the
station but unadopted on the farm; of grazing managementor stock-feeding
proposalswhich came to naught; and so on. It is generallyheld that
such failureshave been due to the researchprocess having been of an
upstream, top-down,or research-pushnature, innovationstechnicallysound
in themselvesbeing ill adapted to farmers' circumstances. The upstream
process is to be contrastedwith the downstream,bottom-up,or
"farmer-pull"process characteristicof research guided by OFR/FSP (Figure
2.4). The differencelies in two features: first, that the traditional
upstream approach started from such intuitiveFSP as the researcher
already had rather than from explicit analysis; and, second, that results
were translateddirectly from experimentstation to extension,without an
interveningOFR stage. The two routes are indicatedin Figure 2.4,
respectivelyby the numbers 1 to 10 for downstream(with provision for an
iterative cycle at 7) and the letters A to G for upstream.

There is now, I believe, little doubt that the downstream

approach,with its OFR/FSP emphasis, is the proper and effective one in
the small-farmerfood-croppingcontext, and this belief (now nearly dogma,
perhaps) is widely shared by IARCs and NARSs alike (but not by Arnon,
1981, who advocates a very top-down approach). It is not true,
however, in the context of technology-basedagricultures,where what is at
least a nominallyupstream approach works very effectively(though no one
would claim perfectly). The differencesare that: first, growers are
generally well educated, technicallyand economicallyaware, interestedin
and often eager for innovation,and rich enough to be adventurous;
second, the researchersare often in close touch with growers, so the FSP
is there all along; third, though there is rarely a formal OFR phase
(sometimesa deficiency,I think), there is no lack of informalOFR
activity because there is nearly always a core of growers who will try any
idea even before the researcherthinks it is ready; and, fourth,
 - 26 -

Flgure2A: Upstreomand Downstream

| b ^ 9~~~~( FolnV SvbWm

Research
Dxen De Dred
'StaOrlo _ _ Servilce t ~ ~w
S Kr,ow$edge. ~ ~
F nrm* ~ ~ AOtcid t
Understanding of~Unomo

& Tsted
IE
on|
IC()E
_ _Systern
(D syt
kIeas tor Urndertondingotf l
Reoarh ResearchNeeds

0[]
~ @ i~ ~ ~ _. ~ ede It
ters~~~~~~~~~~~~Wtion __
Adoptlon
. ~~~~~~~~~~~NecessaryI Voryiee
ResearchDone
& Testedon

Resecarch On-Farm Researched

Proposedfor Research, Exended
Extension OFR on Forms
 - 27 -

communications, by press, radio, meetings, and, not least, by personal

contact, tend to be good, even excellent. The above is as true for
privately run tropical estates as it is for prosperous farms in temperate
countries. In historical retrospect, therefore, it was the unquestioning
transfer of an upstream model, from social circumstances in which it
worked (and continues to work) very well, to a socio-economic milieu in
which it did not, that caused the difficulty. It took a long time to see
the point but it is now well taken.

New Farming Systems Development (NFSD)

I now try to draw together points from the foregoing discussion

and list what I take to be the main features of NFSD, as follows:

1. any NFSD consists of many and complex changes which must,

of their nature, be made more or less simultaneously;

2. it is inaccessible to the stepwise change characteristic

of OFR/FSP, at least on any reasonable time-scale;

3. of its nature, NFSD is a top-down or upstream process

which must owe something to a FSP of what is already there
on the ground but cannot be other than an invention by
imaginative researchers with some conception of what is
technically and economically possible;

4. while OFR/FSP seeks to adapt new technology to the socio-

economic circumstances of small farmers, NFSD has to do
the opposite, namely, adapt the economics to the techno-
logy; government intervention in one form or another is
implied;

5. there are no rules for how to invent a NFS because every

case must be sui generis;

6. nor are there, for want of experience, rules for how to

develop a NFS into practice; I can only suggest
(tentatively, because there is probably much literature
that I have not seen) that an approach through spreading
clusters of model/unit farms might be appropriate (cf
Chapter 5); a somewhat authoritarian element here seems
to be inescapable;

7. no NFSD would be at all likely to be "correct" from the

start; so there would have to be an adaptive phase, which
might well take the form (if point 6 is right) of an
OFR/FSP approach to the problems of the hypothetical
model/unit farms.
 - 28 -

Historically, there have been many examples of land-settlement,

irrigation, rural development schemes (by whatever name) which were, in
effect, attempts at NFSD, or were built around NFSD ideas. One's
impression is that a great many of them failed, though some (the Gezira
scheme in the Sudan comes to mind - Tothill, 1948) worked well for
decades. A critical survey would be worthwhile because analysis should
yield clues as to how and how not to try to implement NFSD schemes in the
future.

As a historical aside, one wonders what might have emerged from

the Namulonge farm (Arnold, 1976), as the center of a network of
unit/model farms, had history been otherwise. Namulonge, I suggest, bears
thinking about as an example of a possible starting point for a NFSD.

If points 4 and 6 above be accepted, it is clear that the

development phase of a NFSD would demand a high level of management and
would be expensive. There is a possibility that the needs for expertise
and money would just be too high to be acceptable and Collinson (1983, and
personal communication) inclines to this view. This is discouraging but I
do not believe that the possibility absolves us from the responsibility of
thinking and trying.

All the above is written in the context of tropical small-farmer

food crop agriculture. The market pull of an important cash-export crop,
of course, changes the situation fundamentally. The great banana,
sugarcane, coconut, rubber, oil palm, tea and other cultures were all
outstandingly successful NFSDs that carried along with them important
small landholder developments (Chapter 1; Figure 1.4; Chapter 4). Perhaps
one can discern here a clue that future, deliberately contrived, NFSD need
a cash-pull element and a degree of autocracy in their management?

Genotype-environment Interactions

I stated in Chapter 1 that a genotype-environment (GE)

interaction was an important element of the Green Revolution: the
matching of the responsive semi-dwarf varieties to environments enhanced
by irrigation and agri-chemicals. Plant breeding is a substantial part
(about 50 percent - Coulter, 1979) of the crops research effort of the
IARCs and is probably a comparable proportion of most NARS programs.
Temperate crop agricultures have achieved very large yield advances in the
past four or five decades, generally of the order of a doubling; such
calculations as have been done (surprisingly few, in fact) tend to show
that plant breeding and improved husbandry have been roughly equal in
effect but with a large GE component as well. Improving crop environments
have, therefore, generally (not always) been matched by responsive
varieties (Simmonds, 1981). Such varieties, when compared with their
predecessors, would typically show regressions of the VH type of Figure
2.5, superior in a "high environment" (EH), little better, even slightly
inferior, in EL. Though selection for this kind of "responsiveness" seems
to have been the general pattern as crop environments improved, there are
varieties (e.g. of potatoes and sugarcane) that show, at the extreme, a
contrasted pattern of response (VL in Figure 2.5): excellent performance
in EL, unremarkable performance in ER and low regression slopes.
 - 29 -

Interactlon
Rgure2.5: Genotype-Envronment

Yield of New /
Vadeties /
VL& VH/

!s~~~~~~~~~~~~~~~~~~~~~~~I
/ / / ~~~~~NUmerCdl

1,~~~~~~
/S VO 10 26

.,00 VH/~~~~~~~~~~-
10 32 n
}^ ~~~~~~VL
14 26

~~~~~~~~~~~~~~~Yleld of Old Vadiety,VO

EL E' Environn-ont
 - 30 -

GE interactions are omnipresent and pose endless problems for

plant breeders (review in Simmonds, 1979); animals show them, too (cattle
example in Chapter 5), but temperate animals are usually kept in such
uniform environments that breeders there hardly have to reckon with GE
effects. Their importance for improving tropical food crop yields is
great but not yet, I think, sufficiently recognized.

Consider the following points:

1. Selection in EH (e.g., on the experiment station) automatically

tends to favor the responsive VH type of variety with a high regression
slope and tendency to do poorly when grown at EL, in rough farm
conditions.

2. Plant breeders are brought up to take pride in their plots and

to be pleased by high-yielding trials. Very few breedersindeed (though
the number is increasing) do the obvious thing and select for VL in EL.
Strict on-farm selection would rarely be feasible but deliberate
simulation of EL on-station would be open to trial; CIMMYT already
practices some low-input selection in maize, and CIAT simulates the
climbing bean's small-farm maize environment (Chapter 5). There is room
for much development along these lines, I think.

3. Given an array of potentially good, new varieties widely spread

for local trial, selection, and use, how are the best ones for local small
farmers (generally operating in EL under low or moderate inputs) to be
identified? More trials at E. (international nurseries or whatever) must
simply tend to pick the responsive VH types again, not necessarily the
best for small farmers. A partial answer must surely lie in variety
testing on-farm in OFR/FSP programs and this has of course been, and is
being, widely practiced; but the samples so tested have already been
biased by at least one and often two cycles of trials selection in EH,
with unavoidable contra-selection against the survival of such of the VL
type of variety as might have survived the original selection process.

4. Identifiable environmental stresses such as peculiar soils,

drought, pests and diseases are coped with by the plant breeder by
selecting for resistance in the presence of the stress. To select
resistance is deliberately to try to exploit a GE effect (though not
normally thought of in these terms). Conversely, not to select for
resistance to a particular stress (because it is absent from the breeder's
environment, for example) is to court disaster when the varieties that
emerge are grown more widely. There are plenty of examples of unforeseen
disease susceptibility; neither disease resistance nor tolerance of
highly adverse soils is likely to arise by accident.

From these points, I conclude that GE effects, both macro and

micro, are omnipresent, that plant breeding is therefore essentially a
local activity, that over-centralization tends to lead to local
disadaptation, and that, insofar as small farmers working at low yield
levels are the ultimate customers of a program, FSP at the least (and
maybe more OFR as well) ought to have a substantial place in framing
breeding plans.
 - 31 -

Some readers might think that, in arguing thus, I am merely

over-emphasizingprofessionalinterests at the expense of the proper
subject of the report. I do not believe that I am. If plant breeding is
roughly half of crops research and representsroughly a half of the
potential for progress, and GE effects are as importantas I contend, then
the matter is economicallyweighty and there must be a substantialarea of
intersectionof interests between plant breeding and OFR/FSP yet to be
well exploited. It cannot be good enough, these days, simply to hope that
widely adapted "universal"varieties will turn up so long as one tries
hard enough.

I note, finally, that researchersin India have come to very

similar conclusions. The reliabilityof conventional,high-inputtrials
as predictorsof farm performancehas been questionedfor some years, and
IARI is thereforemoving towards a policy of on-farm testing and the
release of numerousvarieties rather than few (Jain and Banerjee, 1982).
 - 32 -

CHAPTER3. THENATUREOF ON-FARMRESEARCH WITHA FARMING

SYSTEMS
PERSPECTIVE(OFR/FSP)

OFR Procedures

Introduction

A distinctionhas been drawn in preceding chaptersbetween FSR

sensu stricto and OFR/FSP. In this chapter I concentrateupon the latter
with the object of outlining the phases recognizedand the procedures
recommendedby its practitioners. The leading referencesare: Byerlee
and Collinson (1980),Perrin et al. (1976), Zandstraet al. (1981) and
Shaner et al. (1982). Wide variation in terminologytends to conceal the
essential similarityof all the approachesand the fact that the
"methodology"(an over-usedword) is, in essence, straightforward. This
report is not a handbook of OFR and my only concern is to try to reveal
the leading principles. Those desiring specific practicalguidance could
best refer to the referencesabove, in the knowledge that they are by the
leading and most experiencedpractitionersin the field and represent the
best opinionswe have.

Figure 3.1 makes the main points and, of necessity,reiterates

the FSR sensu stricto - OFR/FSP distinction. The diagram is set in
the context in which, historically,systematicOFR/FSP has developed,
namely, that of an IARC (specificallyhere CIMMYT and IRRI) working in
collaborationwith NARESs. The main featuresare as follows;

1. There is a team (multi-disciplinary)

in the IARC which

2. identifiesthe "target"FS or "recommendationdomain" and

3. analyzes its technical/economicstructurejust deeply

enough for the purpose in hand;
4. the team identifieseconomicallysensible innovationsand

5. does researchon them, on the experimentstation, on few

selected farms, on more numerous farms, according to
circumstances;

6. under points 2-5 the team solicits the interest and

collaborationof the NARES and

7. the experimentationis iterated as necessary;

8. the experimentsare monitored/analyzed economicallyand,

if successful,lead to recommendationsfor the national
agriculturalextension system (NAES) to exploit;
 - 33 -

FSRand OFR/FSP
Figure3.1:RelationBetween

Tralnlng

IARC NP

FSR ~~~~~~~Agrc.
Reserc
TFom
Team ~~~~~Workers
NAES

bientiy Unclertornding
'Target FS ti Guldance

AnalyzeFS
Superficially

T~
Deeply

l l~~S

identity - . -
Potentialty
Useful
innovations,
Do Research. ;

Try Out Innovation

(a) On Exp.Station
(b) On Selected Farns Monitor,
Analyze,
(c) On NumnerousForms
* ~~~~~~Recornmend.

l~ ~~~~~~~~ Exli-
~
L~ /____ Sucssu
___________J~~~~~~~~~~~oclF
 - 34 -

9. there is a continual feed-back of informationto the

IARC/NARES consortiumas experienceand understanding
widen;

10. given that there is agreement that the OFR/FSP "style" of

researchis valuable, the IARC must assume a
responsibilityfor trainingand network development(see
"Trainingand Networks" section below).

The above list is stated in the existing context of IARC-NARES

collaborationin the enterprise. As time goes on and experience
accumulates,the responsibilityfor OFR/FSP must move towards the NARES
(Chapter4). With this qualification,the list is, I think, perfectly
general and is likely to remain a satisfactorylogical strupture,whatever
the vagaries of nomenclature.

The "recommendationdomain"

The "recommendationdomain" (or "target"farming system) is the

farming system narrowly enough defined that any recommendationsthat arise
from OFR work may reasonablybe expected to apply, in large part at least,
to all the constituentfarms of the domain. I remarkedabove that any
classificationof FS is arbitrary and will go just as deeply as is
requiredfor the purpose in view and no further. The depth required here
will be a matter of judgment for the OFR team asking the question: "Is
this assemblageof farms sufficientlyalike for the purpose?" In practice
(Chapter2) the questionwill always be asked of a subsystem (a crop, a
group of crops, an animal) in a limited geographicalarea but in awareness
(FSP) of possible interactionswith other subsystems. The factors of
which the researcherswill have to take account are usually classified
thus:

1. Technical: a. physical

b. biological

2. Human: a. Endogenous: farmer resources,goals,

attitudes,and constraints

b. Exogenous : community relationships,

institutionalarrangements
(communications,markets,
governmentintervention,
etc.)

In real life, obviously,no great depth of understandingwill be

possiblewhen numerouspossible recommendationdomains present
themselves. Generally, that one will be chosen which representsthe
greatest number of farmers, the greatest area of land or some such
criterion,tempered,no doubt, by some a priori sense of what is
technicallyfeasible. The choice, however arrived at, will not be
independentof other possible choices, because there will generally be
 - 35 -

overlaps/intersections(Figure 3.2) with them such that decisions/recom-

mendationsfor one domain will be transferable,in part, to others.
Within reasonablelimits, the wider the domain the better.

Survey or diagnosticphase

A recommendationdomain having been defined,there usually

follows a farm survey to define OFR activities. If local knowledge is
already good, of course, a formal survey may be unnecessary. Experienced
practitionersare agreed that short, simple surveys suffice (Table 3.1).
A few man-weekswill normally be enough, encompassingan exploratory
survey to decide what specificquestions to ask and a formal (but still
simple) numericalsurvey. As to the former, experienceis all-important.
As to the latter, there is an abundant statisticaltheory of surveyswhich
will, no doubt, be kept in mind but rarely exactly observedas to
precepts: if defects of public data are such that farmers can not be
accuratelyenumerated,let alone describedas to economic
characteristics,strictlyrandom or stratifiedrandom procedureshardly
apply. Accessibility,co-operativeness,the opinionsof experienced
extension agents, and so forth will probably be more weighty factors in
choice than the statisticalniceties. Even if statisticalprecisionmust,
perforce, often be foregone,results should, in experiencedhands, be
reasonablyrepresentativeand should yield useful measures of uncertainty.

In connectionwith the design of the survey questionnaire,a

good working principleis: as short and simple as possiblefor the
purpose in hand. Long questionnairesare self-defeating,and I have seen
a sad example of one of 50-odd pages, in two languages,rarely properly
completedand ill adapted to the local computer; the local OFR workers
had already waited months for non-existentanalyses.

The data taken in a well-designedsurvey will take the form of

time-tabledflows of labor, material inputs, other costs, yields, prices,
and other returns. Collectivelythey will allow the economiston the OFR
team to construct cash flows (imputingcosts and prices where necessary),
to get at least some idea of the farmers'situationwith regard to risk
and uncertainty,to detect potential conflictsof demand for labor and
cash, and at least start to identify criticalor limitingprocesses. The
methods are essentiallythose of conventionalfarm economics,but with
some necessarytranslationof non-cash items into cash equivalentsfor the
purpose of calculation. Adherents of Chayanov might be unhappy with these
proceduresbecause they treat the small farmers as a more or less
conventional"firm" that employs family labor at imaginarywages and sells
goods to itself at imaginaryprices. Chayanov argued that the
conventionaleconomictheory of wages-interest-rent-profits was simply
inapplicableto peasantswho paid themselvesno wages and sold few goods.
They were, he thought,simply outside the market economy. The test of the
OFR economists'approximationsmust be empirical: so far they seem to
work pretty well. Perhaps a re-readingof Chayanovwould suggestuseful
refinements?
 - 36 -

Figure3.2: The"Target" FarmingSystemor "Recommendation Domain"

Isolated System: Might ||Agroecological|

be a Patch of Coconuts stem
AStudg Irrigsted Riceor
• Pbtch ot RiceIn
• aDlnd/
Grozlhg //Non-Farm ng
Area Cornponents

\~~~~~~F

FS3Chosen for Detailed

Study: Intersections Indlcate
That Resultswill have
Impllcatbons for FSV,FS2'
FS4,but not FS5.
 - 37 -

Table 3.1. SEQUENCEOF OFR/FSP ACTIVITIES

Season Time
(year) Activity (weeks)

First 1. ExploratorySurvey)
)
) In
2. Formal survey ) crop
Prepare ) 2
Execute ) 3
Analyze ) 4

3. OFR work
Plan 4
Prepare 8

Second Execute (crop season) 20-25

Evaluate 6
Report and Plan 8

Source: Byerlee and Collinson (1980)

 - 38 -

Survey questionnairesare notoriouslydifficult to design really

well to give accurateand unbiased answers without excessive length. The
difficultiesin the OFR context are often compoundedby language problems,
by socio-economicallysensitive issues, even by an amiable desire to tell
the interviewerwhat he wants to hear. These difficultiesare usefully
listed by Matlon (1983). The only remedy is experience,cross-checking
when possible, and cultivationof a cautiousskepticism.

OFR work proper

OFR operationsin the field start with planning and preparing by

the team on the basis of the findings of the preceding survey (Table 3.1)
and of prior scientificknowledge of the system (or of related systems).
Beyond stating the obvious, that simple unit changes or small packages of
changes will be preferred,no simple rules can be given: all depends upon
the circumstancesand the knowledge,experience,and imaginationof the
members of the team. Nor can simple rules be given as to what follows:
Table 3.1 supposes that it has been possible to take appropriatetechno-
logy off the shelf, so to speak, and go straight to the farm in the next
season. But, often, preliminaryon-stationexperimentationwill be needed
(Figure 3.3) and entry on-farm deferred. Similarly,the number of cycles
of OFR needed is generally unpredictable; sometimesa single cycle of a
couple of years will suffice; more often, perhaps, two or more cycles,
narrowing down the alternatives,will be necessary. If the last is true,
then the first cycle will probably consist of researcher-managed trials
(statisticallystructured,often factorial)and the second of farmer-
managed trials of, at most, two or three treatmentsagainst traditional
practice as the control. In the latter, replicationwill be of substan-
tial plots across farms so that, formally, farms are treated as blocks.

All on-farm trials are economicallymonitored,and there will be

regular feed-backof scientificand economicunderstandingto the OFR team
to guide subsequentexperimentationor to reach a decisioneither to
abandon or recommendadoption (Figure 3.3).

All experiencedOFR workers are agreed that the NARES must be

involved from an early stage and throughout(Figure 3.3). The extension
worker'sFSP is agreed to be a potent complementboth to the formal
economic survey and to the actual conduct of on-farm experiments;
furthermore,there can be few things more likely to conduce to successful
extension than officialNAES involvementand interestab initio.

The compositionof an OFR team (whateverthe institutional

arrangements)has to be, in practice,decided ad hoc. There will
generally be an agronomistand economist,at least, with technical
assistance,and contributionsfrom specialistssuch as plant breeders,
pathologists,entomologists,and soil scientistsaccording to need.
Practice is beginningto suggest that some four to six workers will be
fairly deeply involvedin any one operation. Institutionalstructure is
referredto later in this chapter.
 - 39 -

Rgure 3.3: Formal Structureof 0FR/FSPWork

CAoroecdogyofn
WhoerArea

Gernrarl
r--- Uroderstan
ding

Describe _
Compronraent
FS

s | [ 1 l l T Frl~~~~~~~~~~~~~~Por
Sclentffic

.| FS1 FS2 FS3 | |Knowiedoe of FS3|

Jb Economic SelectFS3 <

Knowlecdge

DesignI
_ ~~Expermrwnts_|

Ecnomicta Do Expts, rnantcis

Knowleget OS&/or OF. Banks,etc.

Evalut Instftulboril |
SelectNeaw Backoround
A ~~~~~Teachnology
Abandon,
1 | ~~~~~~Modity. J

FurtherEnhanced | |WdrO etngi

|Ecoromnk < of Fewer NARES
|Knowledge AJtemntie

L Systernm
 - 40 -

Practical points for OFR workers have been usefully discussed by

Tripp (1982), and a condensed summary of his list follows:

1. Site-collaborator selection

a. Domain: See above.

b. Sites: Chosen within farms, suitable as to

rotation and access.

c. Logistics: Sites and farmers easily visitable.

d. Diversity: Rotate sites between farmers over

years; do not stick to a few chosen
collaborators.

e. Extension: Use extension officers' experience.

f. Biases: Adjust choices of sites over years.

2. Communication

a. Details: Agree formally with farmer as to sites,

treatments, cultural practices.
b. Visits: Make them in presence of farmer
whenever possible.

c. Address: Adopt courteous forms of address.

d. Local
knowledge: Learn about farmer's circumstances.

e. Opinions: Encourage farmer to express his own

opinions.
3. Data

a. Fieldbook: Should be pre-designed in field format.

b. Records: Made immediately but preferably not in

farmer's presence.

c. Opinion: Record farmer's opinions

d. Visits: Record each one.

e. Other fields: Make notes on.

f. Posterity: Aim for complete records, usable by

successors.

g. Approach: Keep it seemingly casual.

 - 41 -

4. Miscellanea

a. Other farmers:Visit and learn from them.

b. Markets: Learn about from local merchants and

traders.

c. Research: Inquire whether there is other relevant

work in the same scientific/
geographicalarea.

d. Experience: Use that of former collaborators.

e. Contacts: Maintain close touch with extension

personnel.

f. Questions: Develop lists of questions arising.

The question of compensationto farmer-collaborators is a tricky

one. I think a fair summary of rather diverse views would be as follows.
In researcher-managed trials, the inputs are provided and the farmer is
guaranteedthat he will not be worse off at the end of the season than he
would have been had he used the land himself; he might, for example,
receive the produce (rather than any cash payment). In farmer-managed
trials, the less free inputs the better, though some may be unavoidable;
the aim, at the end of the cycle of experimentation,anyway, should be
full economicresponsibilitylying with the farmer, because only thus can
the researchersbe fully confident of the economicviability of the
innovation.

Economics

As I said above, this is not a how-to-do-ithandbook, so

detailed economicswould be out of place as well as beyond my competence.
The referencesat the start of this section,togetherwith Collinson
(1983),Dillon and Hardaker (1980) Valdez et al. (1979) will provide
details for those who need them. Levi and Havinden (1982) give a useful
general introductionto African agriculturaleconomics. Here I shall only
refer to a few general principles and give some examples.

There are four main points, as follows. First, the methods are
essentiallythose of conventionaltemperate-countryfarm management
economics,but using imputed costs and prices where necessary (see above
in this section). Second, small farmers are poor and risk-averseand this
factor is taken into account in two ways: by doing specific uncertainty
analyses (could the farmer stand a 10 percent chance of failure?) and by
setting the discount rate at some arbitrarilyhigh level (maybe 40%) so
that marginal returns must be high. Third, it is assumed that the farmer
is nearly enough profit-maximizingfor practicalpurposes, though the
balance to be struck between costs, profits, and rates of return must be a
matter of economic judgment. Fourth, some innovationsare divisible
(e.g., fertilizers- Figure 3.4), some indivisible(e.g., packages of
agriculturalpractices- Tables 3.2 and 3.3); the former can be adopted
 - 42 -

Rgure3A: EconomicResponsesof Farnersto a DMsibleInput

NmaX# R =0

B Grs
Benefit

Large Farmer
/ R~~~~~>O

SrnallFarner
R>>O
C cost

T Levelof Treatment
 - 43 -

Table 3.2. ON-FARM ECONOMIC ASSESSMENT OF RICE-CROPPING PATTERNS IN THE

PHILIPPINES

(arbitrary money units)

Net Marginal
Patterns V'ble costs Returns Returns Rates
(a)* (b) (a) (b) (a) (b)

Traditional Pattern C C' B B' N N' B/C' N/C'

Rice-fallow 8 -- 20 -- 12 -- -- --

Alternatives

Rice-mungbean** 11 3 25 5 14 2 1.7 0.7

Rice-ratoon 10 2 27 7 17 5 3.5 2.5

Rice-rice 22 14 41 21 19 7 1.5 0.5

* Of the columns headed (a) (b), the former are the actual
figures, the latter the marginal differences over the
traditional base figures.
** Rice-mungbean was already partly adopted technology known
the farmers.

Source: After Zandstra et al. (1981)

 - 44 -

Table 3.3. RETURNS FROM CROPPING SYSTEMS INTRODUCED IN INDONESIA

(US$/ha.)

C B N = B-C (B-C)/C
System Costs Benefits Net Benefits Rate of Return

Farmer practice 170 360 190 1.12

Introduced system 464 1,070 606 1.31

Differences
(Marginal C, B, [B-C]) 294 710 416 1.42

Notes: (1) See Figure 5.7, Section 2, for general patterns of systems
tested (rainfed multiple cropping).

(2) The marginal rate is 142 percent.

Sources: Means of three bodies of data from various sites in Sumatra

and Java, 1976-82 (AARD, 1981; Inu, 1983).
 - 45 -

at a low (sub-optimal)level, thus leaving open the option of subsequent

upward adjustmentif/when the farmer feels inclined to greater
expenditure; the latter incur a fixed cost per unit but need only be
adopted over part of the farm, so they, too, are functionallydivisible.

I now give three examples,with brief discussions. Figure 3.4 is

based on the maize fertilizerdata of Perrin et al. (1976). Given a fixed
dose of phosphate,the nitrogen response curve is of the form shown and
can be describedby a simple quadratic. The net benefit (N = B-C) curve
has a maximum when R, the marginal rate of return, is zero, a point which
few farmers would choose. The rich, non-risk-aversefarmer, with a secure
low discount rate, would operate near the maximum; the small, poor farmer
could reasonablybe advised to use nitrogen at a rate correspondingto
R = 0.4, say (40 percent), but it might be prudent to suggest a somewhat
lower rate, in the knowledge that adopting farmers are perfectly capable
of using collectiveexperienceto adjust rates upwards over time. (I
should add that this is a simplificationof the example, adopted for
heuristic reasons; Perrin et al. treat the data as discontinuous,an
approach that has computationaladvantagesbut is less easy to expound).
The number R=0.4 just cited represents simply an order of magnitude for
the discount rate suggested by the empirical experienceof the CIMfYT
workers. It is not a universal constant.

Examples of indivisible "packages"are given in Tables 3.2

and 3.3. In the former, three alternativesare comparedwith the
traditionalrice-fallowsystem. A second full crop of rice looks less
attractive than the other two: it is, in one sense, the most profitable
but it incurs high costs and offers a low marginal rate of return. No
single, simple criterionfor decision is apparent. The Indonesianexample
of substitutingrelayed row cropping of maize, rice, peanut, cowpea and
cassava for the traditionalmaize-ricemixture relayed with cassava looks
very attractivein terms of marginal return but demands a heavy extra
expenditure(costs are more than doubled). Could the farmer find that
amount of extra money? Would he want to do so? Is there a market for
extra produce? These are the sorts of questions the OFR economistwould
have to try to answer.

Modeling

I talked to many people about the use of numerical system

modeling and found many and diverse views about the subject. Three uses
of models need not detain us here: (1) whole farm models as outcomesof
FSR studies sensu stricto (which are academic in nature rather than
practical); (2) plant or animal growth models as aspects of physiological
research; (3) macro-economicmodels as componentsof economic studies
wider in scope than OFR/FSP (ICRISAT analysis of the "common property"
problem would come in this category, I think).

For the present purpose the question is: what place do models
have in practicalOFR/FSP work? Most practitionerssay "none," that the
questions asked and prospective changes in farm practice are so restricted
in scope that the experienced economist,using conventionalfarm-
managementprocedures,can discern what is workable and what is not.
 - 46 -

Further, time usually presses (Table 3.1), and the need is for quick,
sensible decisions in a feasible economic area rather than precise,
optimal decision. The proponents of this view would add that both data
and models are rarely good enough to allow optimal prescription anyway,
even if it were wanted; and they sometimes add, as well, that the
delights of modeling per se are liable to divert the modeler from his
proper function, namely, practical economic advice. A variant of these
views (not widely encountered) is that models have their uses in defining
uncertainties and feasible areas of decision (even if optimization is
rejected as the object).

The arguments against modeling in OFR/FSP just outlined come

from practitioners experienced in work with annual crops and, in that
context, I find them persuasive. Our views of OFR/FSP procedures have, so
far, been dominated by students of small changes in annual cropping
systems and for these, as I have just argued, modeling seems to have
little or no useful place. One notes that successive years can be
regarded as essentially independent of each other and that the changes
sought are small enough not to perturb the whole system. When there are
effects spread over time (year-to-year correlations) or when larger
changes are sought, the situation may well be different and modeling may
have a more important place. Thus the CIAT pasture-beef workers argue
(convincingly, I think) that both pastures and herd structures change over
time, results in each year being largely determined by what preceded them,
and that data simply can not be handled by pencil-and-paper methods.
Numerical models, they contend, are the only available way of thinking
about possible consequences, several years ahead, of relatively small
changes made now. Without such models, indeed, it would be hard, perhaps
impossible, even to predict the character and timing of equilibria let
alone their responses to changed inputs.

As I say, I find this argument convincing and believe that crops

OFR work, if and when it moves to more complex changes, may find
increasing need for models. Likewise, NFSD, involving very complex
systems with interactions beyond what the unaided imagination or mere
diagrams could cope with, is, I think, likely to need models to help to
identify the impossible, if not to define the best, solutions.

Finally, a few words of caution are necessary. First, I have no

personal experience of numerical modeling and the comments above may be
regarded by the experts as naive. Second, models are as good or as bad as
the skill and the data that go into the making of them; they can be very
wrong without the fact being apparent or testable. Third, the bigger they
are, the more complex the systems that they describe, the more assumptions
and guesstimates that they embody, the more likely they are to be
inaccurate, but the more useful, in principle, they ought to be. And,
fourth, modeling is evidently a compulsively interesting activity for
those so inclined (as well as being a fashionable one); the dangers of
doing too much and of believing the results too readily are evident.
 - 47 -

Costs and benefits of OFR

To get a rough idea of costs of OFR I examined recent annual

reports of several centers and counted economists (and a few
anthropologists)as a proportionof total senior scientificstaff
(administration, works, libraries,etc., excluded). For six centers
(CIAT, CIMMYT, IITA, ILCA, ICRISAT, IRRI) I found a grand mean of 8.4
percent economists,rather a wide range (3.1 percent IITA, 6.9 percent
ICRISAT, to 19.6 percent ILCA) and a tendency for those places most
committed to crops OFR to lie around 9 percent (CIAT 8.4, IRRI 9.4, CIMMYT
10.8). Assuming that some of the economics can not be assigned to OFR
(which is certainly true, e.g., at ICRISAT), it looks as though about 5
percent of the centers' budgets might be attributable. OFR work, of
course, incurs other costs over and above on-stationresearch(Chapter5)
so it looks as though total costs are, at most, under 10 percent of
budgets. If anything, commitmentis tending to rise but I can think of no
way of calculatingan objective "best" or "optimal"number.

The uncertaintyattaching to the above crude estimateof "under

10 percent" draws attention,I think, to a general problem. Accounts of
the IARCs are not reported on a very informativebasis so it is next to
impossibleto derive good estimates of expenditureon anything; to do
this, a several-dimensionalmatrix of expenditurewould be needed. In the
FSR context,some institutesidentify expenditureunder this head but
include much that should really be attributed to componentresearch
(soils, agronomy, machinery,etc.) while others do FSR sensu lato but do
not call it that.

As to benefits of OFR, we have the one availableC/B analysis

referred to later (Chapter 5) to show that, in that one case, B fairly
certainly much exceededC on a DCF basis. In general, if OFR work can
indeed, as is claimed for it, point research effectivelyat realistic
practical objectives,deter the adoption of unrealisticobjectives,and
expedite uptake on the farm, then it must be generally beneficial. I
think these claims are reasonable,which is not to predict universal
success, for there must be some failures; nor does it help in calculating
an optimal level of OFR activitywhich must, for the present at least,
remain a matter of informed judgment.

I have no cost data for national programs and can only note that
economistsare already quite widely employed. As to benefits,a
widespread commitmentto OFR methods in the tropics of all three
continents(cf. Chapter 5) implies local acceptanceby NARESs of the
likelihoodof substantialgains.

Training and Networks

Traininghas been an important element in the work of virtually

all the IARCs since the early days of the CGIAR, it having been recognized
that one of the their prime functionswas to help the NARESs to develop
their own local skills and organizations. While much of the traininghas
been technological,there is already a strong OFR trainingelement at
several centers (CIMMYT,IRRI, ICRISAT) and prospectivedevelopments
elsewhere (CIAT).
 - 48 -

Training and networkingnaturally go togetherand, jointly, they

surely representthe most economicalway available of diffusing the skills
and knowledge of the Centers where they will do most good, that is, on
the ground. To be only a little frivolous,the "old-boynetwork"
(politelycalled an "invisiblecollege")is probably the most potent agent
for influenceand informationever invented (librariesand computersnot
excepted). Its value in drawing togethera dispersed body of people with
common interests is incalculable.

The structure of IRRI's ACSN is summarizedin Figure 3.5. It

grew naturally out of a trainingprogram in the technologyof rice
agriculture,the outpostingof IRRI staff to Asian national centers, and a
sharply focused program on intensivewetland rice production. Now there
are about 120 sites in 11 countries,each site with a local NARES-OFR
team. As the results for wetland rice become assimilatedinto local
practice (which has widely though not universallyhappened),so the
program is diversifying,by a natural progression,into the wider, but
still rice-related,contexts indicated at the bottom of the figure. Hence
the emergenceof the AFSN (Chapter 5).

The operation of the network, beyond the fact that it is a

centralizedbut two-way system, is not detailedin Figure 3.5. In Figure
3.6, I outline the structure of the CIMMYT East African program which has,
since 1975, evoked local interest in member countries,trained numerous
local staff, done short survey/diagnosticjobs, and got the network
going. The main activitiesof the network are: a newsletter,to sustain
interest and attention among workers on the ground; occasionalmeetings
of NARES administrators(e.g., CIMMYT, 1983); technicalworkshops; and
visits and travels for specificjobs. All this is designed to provoke
both the necessaryadministrativesupport for the OFR approach (sometimes
not readily obtained, since the ideas are fairly new) and, no less
important,the diffusionprocess (the old-boy network under a different
name). As the network develops, so there is a natural tendency to shift
training from in-centerto in-country because: it is cheaper to do so;
national interest is sustained; and local needs can be better met.

Economists,Anthropologistsand Institutions

I remarked in Chapter 2 that economistsentered the IARCs at an

early stage and that their use seems to have been an article of CG
policy. It is to the economists,I think, at the fairly mundane level of
farm-managementeconomics,that we owe the developmentof OFR/FSP and the
decline of the traditionalupstream approach in tropicalfood-crop
research. That economists,in this role, have come to stay, and that they
perform an extremelyuseful function is not, I think, now in doubt.

Some problems remain, though. Besides institutionalquestions

(see this section, below), the relation scientist-economistis not always
an easy one if neither fully understandswhat the other is doing and if
each tends to defend his professionalinterests. On the one hand, I have
heard it argued that the essential economicsis fairly simple (which it
is) and that all that is required is a bit of economic education for the
scientist; on the other, it is argued that the necessary economic
 - 49 -

Rgure 3.5: TheAslanCropping (Farming) SystemsNetwork

IRRI Country1
QStteswith
OfR
Reseorch = Troining AQronomist
Ag Crop
Protecflonlst&

FIn
Economistat Each;

r- L~~~~Repnsli
Tf
~ ~ ~ ~
r 1
NARS.
Nonal
wkgGroup|

Country2

I
NeNtork_=

IrE

Long-Term e StudsCountrle &

hIrrigtedRiceProgram Ab\ /
IRQINow Mavlngto:\/
Qalnfed& Upland
of Rice
Animals0 its
/ \Mthiuces
F _ ~~~Companlon Crops(e.g.
mopo,Weat)
Long-TermFerhityStudie
\ of Animals
IAditon /
 - 50 -

Fgure3.6: Training and Networking In OFR/FSP,

Based Upon the CIMMyr EastAfrican Prgram

1. ShonDKonnosni
Surw Dore
Ethiopia.Uganda,M.2, Natll Intere

Training
pa|l 7_ 3, Cntraplad
Trainingalnlng Started
InMatlang Country Country Country
Bodv/Cete 1 2 3

D|;;n RoeA Network: Foundatbons:

NetworkBecormeF 1, r sletterl> 'N1. SorneFSP
SeAdmin.
nWk2.Meetinss s 2 Sore Interest

[ ~~~~~4.
\Asits/Travels _

Continued Tralning
Tendlngto be In-
Country,RaitherThan
In-Center.

Self-Sustaining 1, Newsletter~~Neslete, Metigs

1975 1981

Surveys/Dbognostics
in Kenya,Tanzania,Zambla.,
Ethiopia,Uoganda,Malawl,Zimbabwe. 5

> NARSCommitment Developed, MavlngTowards

4 pariTralning
passu. ~~~Tralnlng
, ~~Netwrk
Development. \
5, Newsletter,Meetlngs,
' << ~~~~Wo,*shops,
\lsits.
 - 51 -

judgment, the ability to handle farm management data critically, comes

only from experience and that the scientist with a smattering of economics
is unlikely to arrive at that judgment. I incline to the latter view but
add that some process of mutual education is needed; both scientist and
economist must know enough of each other's fields to be sensible.
Sometimes, a sort of osmosis by personal contact will suffice; sometimes,
an explicit educational effort might be desirable (see below). In
general, scientists, brought up in the conventional B.Sc-M.Sc.-Ph.D mill,
will be more in need of economic understanding than agricultural
economists (who can hardly avoid knowing something about agriculture) will
be in need of scientific education.

So economics has come to stay. The position of anthropology is

less clear. Anthropologists are inclined to argue (e.g. IRRI, 1982) that
their skills in social understanding transcend those of economists and
that they could therefore get under the crude questions of cash flow and
marginal rates of return to the deeper social realities. Maybe, but one
recalls the not altogether unfair stereotype of an anthropologist living
in a village for years and emerging at the end with the view that the
villagers are all splendid chaps who ought to be allowed to get on with
agriculture in their own way regardless of the fact that the world around
them will not allow them to do so. Thus, if there is a place for
anthropology at all, I incline to the view that is for short, highly
specific inquiries closely co-ordinated with agricultural research
programs and directed at answering important questions beyond the reach of
economics. The economic anthropologist rather than the strictly social
kind would probably be the most useful in this context: there might be
little to distinguish him from the economist with well-developed social
perceptions. But any generalized adoption of social anthropology would
be, I believe, merely an expensive way of avoiding a few, not very costly,
mistakes by OFR/FSP teams.

To conclude this section, I refer to institutional arrangements

for OFR/FSP work. As preliminary points, I note that: (1) OFR/FSP is
hardly a decade old and the term itself was only coined a year ago; (2)
it developed more or less independently in several different institutions
and had to be fitted into structures and personalities that were already
in place; (3) the predominant discipline and/or commodity orientations of
most institutions (IARC and NARES alike) meant that OFR, essentially a
non-traditional and multi-disciplinary activity, did not always fit easily
into existing structures; (4) ad hoc arrangements have therefore
dominated the scene.

So no consensus as to the "best" institutional arrangement has

emerged and it is perhaps doubtful whether one exists. In practice,
structures range widely from those (e.g., CIMMYT) in which an economics
program runs the OFR work, drawing on specialist colleagues' knowledge as
necessary, through those (e.g., ICRISAT) in which a multi-disciplinary
team is drawn from several departments (including economics), to those
(e.g.,CIAT, ILCA) in which economists are individually committed to
membership of OFR commodity teams. I confess to a personal fancy for the
last model, the commodity-team-cum-economist, because I believe that it
should best promote the scientist-economist interchange that is so
 - 52 -

necessary. But it may not always be applicable if the work is not focused
on a commodity (as ICRISAT's is not). Again, what is one to say of a NFSD
program such as that of IITA? It can not be commodity-orientated but must
clearly be multi-disciplinary and include economics alongside the science
(which it does).

I conclude that existing structures are diverse and likely to

remain so. The only guiding principle evident is the need for integration
of diverse disciplines - not a very surprising conclusion!

The need for integration and cross-discipline understanding is

widely recognized but perhaps less widely achieved. I have certainly met
specialist scientists who wanted to stay that way and had rarely been on a
farm; also economists more interested in linear programming than in FSP.
The problem is a real one and will not go away; blame, if any, attaches
to the excessively specialized educational schemes adopted by universities
in rich countries and copied elsewhere. The Ph.D. system, with its
tendency to generate a series of near-replicas of the teacher in terms of
skills and interests, is perhaps the most damaging part. There is
certainly a case for debunking the Ph.D. as a general license to practice
science though this, alas, is unlikely to happen. For the time being,
anyway, the example of senior colleagues at working level is likely to be
more potent than exhortation. Something could surely be done, by IARCs
and NARESs alike, to ensure that young scientists see enough of farming
systems on the ground to generate at least a modicum of FSP? If my
earlier points about GE interactions (Chapter 2) be accepted, it will be
evident that plant breeders are no less in need of FSP education than
other scientists (which is not to deny that many are already well
informed).

Relation to Extension

The importance of close links between OFR/FSP work and the local
extension system has been emphasized by many writers and is referred to
above. Given that there is an extension system this must be correct, but
there are, I think, grounds for believing that the whole structure may be
open to change.

First, OFR/FSP has itself a rather potent extension effect, as

we are now just beginning to understand. Small farmers are far readier to
take up and diffuse innovations than was realized as little as a decade or
so ago. If OFR/FSP is practiced on a wide scale with numerous farmers,
then useful changes go far and fast without any formal intervention by
extension agencies; or so accumulating experience seems to say.

Second, extension agents on the ground tend to be young,

inexperienced, and less knowledgeable than the farmers they are supposed
to guide. This was, of course, the basis of the "training" part of the
training-and-visit system of extension which is now more or less
conventional wisdom. So farmers often do not take extension agents very
seriously, but they do listen to researchers and senior assistants who
clearly know what they are talking about; hence the evident efficacy of
an OFR/FSP operation that really has something to offer.
 - 53 -

Third, if the two foregoingpoints be accepted, there is clearly

room for speculationas to the future place of extension. A somewhat
cynical view is that, if there is nothing to extend, extensionis helpless
and that, if OFR/FSP is effective,then farmers will quickly know about it
and adopt anyway. If this is even nearly true, then the future may lie
with smaller but better-trainedextension servicesworking in the OFR/FSP
area in close collaborationwith researchers,leaving diffusion rather
than exhortationto do the rest. Only time and experiencecan tell, I
think, but this trend appears to be at least a possibilityto be taken
seriously. It amounts to saying that OFR/FSP shows signs of bridgingthe
gap between researcherand farmer that it was the traditionalfunctionof
the extensionservice to cross.
 - 54 -

CHAPTER 4. WIDER QUESTIONS

TechnologicalChange and the CGIAR/NARESContext

In Chapter 1 I enumeratedthe main causes and agents of

technologicalchange (Figure 1.4). Of the four agents recognized,
farmer-generatedchange and the activitiesof commercialbodies selling
productsdo not concern us here. There remain the rather wide body of
changes deliberatelygenerated by governmental-industrial agencies (Figure
4.1). Consider, first, the export-cash crops at the left-hand side of the
diagram to which the following points are, I think, relevant.

1. They have a long history of industrial/governmentalsupport

by way of highly efficient research, development,and
extension.

2. Some such programs have moved entirely into the hands of the
industries themselves (sugar in the West Indies, rubber and
oil palm in Malaysia) while others retain joint
governmental/industrial interests (diversecrops in Brazil,
India, the Philippines,and Indonesia,for example): thus,
while NARESs, world-wide,have considerablecommitmentto
industrialcrops, the CG system has ignored them.

3. I know of no general historicalassessmentof the

socio-economicimportanceof these crops but believe it must
have been immense, in several ways: in drawing in external
developmentcapital; in generatinglocal cash; in creating
communicationsinfrastructuresof shipping,roads,and
railways;andin demonstratingnew agriculturaltechnology
transferableto other crops.

4. Plantationagricultureled the way and has been widely

criticized(I believe over-criticized)for its exploitative
features; but all the crops listed in Figure 4.1 have small
landholdersectors, sometimesstrong and long-established
ones, of substantiallocal economicimportance.

5. The technologiesof growing such crops efficientlyare well

established,and transmissionto growers must, of its
nature, be rather a top-down process.

6. In the plantationsector, OFR/FSP is hardly relevantbecause

the researcherslive with the crop, FSP is not lacking,and
competentestate managers practiceOFR anyway (without
calling it that); for small holders, industriescommonly
maintain specializedextension/advisory services but it is
common experiencethat yields and quality are below what
might reasonablybe expected; maybe there is a place here
for more OFR/FSP input?
 - 55 -

Figure 41: TheWider Context of Tropical Agricultural Research

Older Agr. Res.Systems

ECort FoodCros
crops

Inuns ental Colonial National More Recent Developments

Govemments
Dev. Banks
Foundations

For Example:
Sugar _ Internatlonal Funds
Rubber for Agr. Res
Bananas
Oil Palm
Coconut
Coffee

Tea
Cacao b_ _

Cottan

l~~~~~~~ _ I~~~~eeac
Contemporary CG Syoem
WAES Of iARCs

Ca s Pull* Tr aining.

of Expfflr.t

l ~~~~~~~~~~~~~~~~~~~

Area of Cuffent Concern

Impacts on Agr.Practices, with OFR& NFSD
TechncogoicalCharge.

Small
Faffm
Sector

Other Impacts
Autonomous or
Commercial
 - 56 -

7. An efficient industrial-advisorybase for the technologyof

the chosen crop can, in principle (and sometimes does in
practice),have three beneficialside effects: it maintains
a high level of farmer/advisercontact; it obviateswhat
may be an inefficientbureaucracyin the supply of cash and
physical inputs; and it makes possible some support of
crops other than the leading one (as the Colombian Coffee
Federation,I was told, is very effectivelypromoting
cassava).

8. All-in-all, a good export-cropindustry is an excellent

focus for agriculturaldevelopment.

9. The difficultiesof finding reliablemarkets for export

products are well known and are more political than
technicalin nature; but point 8 remains good in principle.

Turning now to the rest of the diagram (Figure 4.1), the food-
crop sector, contemporaryNARESs are often direct historicaldescendants
of older colonial or national systems, though generallygreatly expanded
in the past 30 years or so. The CG system is both recent in origin and
quite small in size, with expenditureof the order of 10 percent the
amount spent on NARSs (Coulter,1983). In my opinion,history will praise
the CG system for having done four things, namely:

1. it recognizedthe need for centers of excellencein various

areas of tropical agriculturalresearch at the "strategic"
level (not "basic" - no agriculturalresearch is basic in
any useful sense of that often misused word);

2. it recognizedfrom the start that it would have to work

through and help to build up NARSs (hence the invention of
ISNAR);

3. it recognized,following point 2, that training and the

developmentof networks were activitiesof central
importance;and

4. it recognizedthat the traditionaltop-down approach to

small farmer extensionwork was not good enough, so
institutedthe use of economistsand the study of FSR ideas
which eventuatedin the subject matter of this report.

These are all excellent achievementsthough the praise should

perhaps be a little qualified: thus IARC/NARESrelations,though
sometimes excellent,are not always so and FSP, though widespread and
growing,is yet far from universal.

What then of the future of the CG system especiallyin relation

to FSR sensu lato? The followingpoints are, I think, relevant:

1. the NARSs are growing, and it must remain a major function

of the IARCs to help them to do so effectivelyby scientific
exchange, by training, and by networking;
 - 57 -

2. as the NARSs grow, they must, for obvious logistical

reasons, take over most of the large body of OFR/FSP work
that will be necessary;

3. given point 2 and assuming that the principlesof OFR/FSP

are well enough established (as I believe they are), I can
foresee no need for any substantialexpansionof OFR/FSP
work in the IARCs, indeed perhaps the reverse; it will be
for the NARSs of adopt the OFR "style" (as many already do)
and for the IARCs to back up the effort by their networking
activities;

4. if my earlier analysis is correct (Chapter2), FSR sensu

strictowill have little or no place in the scheme of things
and OFR/FSP will be mostly for the NARSs (point 3), which
leaves NFSD for consideration; I believe this is an area
that cannot be ignored and that it must lie in (but will
not, of course, be confinedto) the domain of the IARCs.

I think the last point is important,though I can offer no recipes as to

how to do it. Certainly, much tropical agriculture,willy-nilly,faces
profound change in the next few decades,and it were better, surely, that
it were purposefulrather than haphazard. Workers in the area will have
to forsake the comfortingcertaintiesof componentresearch on annual
crops and think of unlikely combinationsof annuals,perennials,and
animals; they may have to resurrect the unit/modelfarm idea and, in the
end, try to persuade foundations,banks, and governmentsthat they really
have somethingworth trying. OFR/FSP, I think, excellent in itself as an
aid to sensiblestep-wise change, does not lead in the right direction.
Imaginativejumps are needed and, so far, no one has been willing to make
them. They will be made only by workers able to transcendthe customary
discipline/commodity boundaries;in this I suspect that multi-disciplinary
thinkingmay be more important than multi-disciplinaryteams.

Agro-forestryand the Neglect of Perennials

Perennial plants, mostly- but not all, woody, are of profound

socio-economicimportancein tropical agricultureand, broadly, the wetter
the place, the more important they are. Under populationpressure, trees
have almost disappearedfrom huge areas that once grew them and remaining
woodlands are fast vanishing. Locally, firewoodgatheringmakes labor
demands comparablein magnitude with those of agriculture. Elsewhere,as
in the backyards of SoutheastAsia and Central America and the banana
gardens of Central Africa, perennial plants make major direct
contributionsto human nutrition.

With the partial exception of IITA (which has some work on

bananas) the CG institutes do not work on perennials. I understand that
work on tree crops has been consideredbut, so far, rejected. Perhaps the
time has come for a review of the question? Trees, however, have not been
universallyneglected. Thus CATIE has a long history of taking perennial
plants seriously and of integratingthem into Central American farming
systems (Combe et al., 1981; Sales, 1979); and the NARSs of India, the
Philippines,and Indonesia are hardly less aware of their importance; all
 - 58 -

devote substantialshares of their programs to them. Further, the IDRC

inspired a prescient report (Bene et al., 1976) which eventuatedin the
foundation(1977) of the InternationalCouncil for Research in
Agroforestry(ICRAF). ICRAF has only very recently (say 1981) really got
going, and the intention seems to be that it shall operate world-widein
collaborationwith local institutions,both IARCs and NARSs (ICRAF,
1983; MacDonald, 1982; Lundgrenand Raintree in Nestel, 1983; Raintree,
1983). There will be institutional,even psychological,difficulties,of
course. Many agriculturiststhink of trees as things which should be
confined to forests or, perhaps, hedgerows,while many foresters regard
agriculture as a competitor,a despoiler of woodland or, at best, as a
transientphase in the re-establishmentof forest (taungya). These views
are, happily, by no means universal and there are welcome signs of
convergenceof interests in the general area of FSR and forestry,
evidenced,for example, by the extremely interestingreview by Burley and
Spears (1983).

From the FSR point of view, there are three points to make, as
follows:

1. OFR/FSP, concentratingupon improving the yield of maize

or cowpea or whatever, could quite well note that trees
were there (or not there) as a matter of FSP, but
thereafterignore them, unless they were in some sense
critical for the performanceof the chosen crop (maybe
yam poles are required or sheep have to be cheaply
confined);

2. at the other end of the scale, FSR sensu stricto would

include the trees (or lack of them) in its descriptionbut
would provide no practical prescriptionas to what to do
next;

3. the agro-foresterwould be inclined, I think, to perceive

critical local needs for trees for various purposes (fruit
here, firewood there, fence posts and browse in the next
place); but he would be constrainedby mandate to trees in
agriculture"- he is not in the NFSD business.

I conclude that agro-forestryis likely to make valuable contributionsto

tropicalagriculture,provided that its activitiescan be operationally
integratedwith those of agriculturalresearch; but that it is no more
likely than conventionalagriculturalresearch itself to arrive at what is
surely needed, namely, practical and testable visions of NFSDs for the wet
tropics on poor soils.

To pursue this vision, I note that most of the items in the box
of Figure 4.2 might well be involved: a patch of rubber, cacao, or citrus
for sale; a mixed patch of breadfruit,avocado, citrus, mango, pechibaye,
coconut, other fruits for home consumption; bananas (which are "trees"
for this purpose)everywhere; woven legume fences to contain the animals;
hedged contour beds; mixed leguminousalleys pruned for mulch, browse, and
 - 59 -

Rgure4.2: The Place of Trees

NaturalForest

_ (Semi-) Natural _ .t
_ Forest P'ermanently limber
_ Eixploited As _ _ etc.
_auno
etc. PManaged

Woodlands

ShiftingBt Friree& 1
Cultivation_EgS CiCtitiions N) sMurhs

Forms Domain~L
Other tbl
Use; Myqclue
Seodad o_Tesi__ __ Landans
~~~~~~~~~~~~~~~
CultWlons
Culotion ~ Patue

| \ SseciAcutel

AnnL…al Pereneaal
I nbrak I

I~~~~~~~~Fi
|
Mus Tak Ths
~~~~~~~~~~~~~~~~~ruit
iDomestic
Trees
nocut
l~~~~~~~~~~xoi

I ~~~~~~~Woodiots
forFRrewood. P'oles & Timbver,
I ~~~~~~~~Shojde
& Live iFences.I
Erosion Conttol or) Slopes.
Sdll Ameiioratio (Mulch + N).
Wl~~~~~&ndbreaks
i3rows & Feed for Stock.
,1 ~~~~~~~Interoctive
wlth Eoch Other &kwith
Other Faoms of Larnd Use; MnySeous
g P~~~~~~~SR
MustTakeTheseIntoAccount.
 - 60 -

feed; annual crops in the alleys; poles and firewood from the alleys and
hedges, and maybe also from a special Eucalyptusor Gmelina woodlot; the
soil as nearly as possible invisible,almost untouched but probably a
little fertilized. There may be a few small farms like this here and
there (in the Philippines or Indonesia perhaps?) but they cannot be many.

The potential diversity of patterns of this kind is virtually

indefiniteand only extensive on-farm experience could tell which, if any,
were workable. Some of the elements of the technologyare available,of
course. IITA and ILCA (see Chapter 5) have the beginningsof
understandingof leguminousalleys; the soil-conservativeeffects of tree
cover and mulch in the wet tropics is establishedbeyond doubt, as well as
the nitrogen-supplyingcapacity of leguminoustrees or bushes. But we are
essentiallyignorant, I think, of the productive capacityand economics of
mixed stands of perennial food-fruitplants. As Watson (1980) points out
in a valuable compilation,perennial crops are there all the year round
and must have high biomass potential,but we know very little of partition
ratios; we do know, however, that breadfruitand bananas are virtually
non-seasonaland (I should suppose) as productiveas tuber crops but less
labor-demanding. There is a long way to go, not only in understandingthe
food potential of the perennialsbut even in introducingthe species
around the world and getting them known; many (maybe 100?) species, each
representedby diverse genotypes,should be very widely spread round the
wetter tropics as a preliminary to local selection,followed by purposeful
breedingas local needs emerge to view. There is here an immense and
almost wholly neglected task.

I believe that any serious developmentof the NFSD, so sorely

needed in the wet tropics,must be built around a far better understanding
of the perennial crops (and other trees) than we yet have; they have been
terriblyneglected. This view is growing (e.g., Harwood, 1979) but is
hardly yet widespread. I conclude this section with an apposite quotation
from a masterly recent review by Watson (1983), which should be required
reading for anyone concernedwith tropicalagriculturalresearch:

Under these circumstances,prospects for the rural poor are grim

and there is a great deal to be said for the suggestionthat the
developmentof tree crop farming systems is one area where aid
programs should be directed, rather than continuingto channel
capital and expertise (togetherwith subsidisedfood imports)
into the urban-industrialcomplex....Treecrops are needed: for
the fruit, nuts, oil, rubber, cocoa, coffee, pulpwood and timber
they provide, for the rural industriesthey support and for the
protectionof water catchmentareas and regenerationof
abandoned land. Not least, they offer shade and protection
against the tropical climate,and can help to bridge the gap
between rural subsistenceand urban affluence. They protect the
environment,offer a diverse and steady income to the
subsistencefarmer and a source of amenity and recreationfor
the city dweller. If the social value of mixed tree crop
systems were added to their more easily-quantified direct
returns, they would undoubtedly be seen as deserving a greater
allocationof the developmentresources now available.
 - 61 -

The Interests of the World Bank

The Bank spends money on tropical agricultural research in two

ways: in its support of the CG system to which it is a major donor of
funds (12 percent in 1983); and in its agricultural/rural development
projects in various countries. In this section I make some observations
on how the Bank's spending might be influenced by FSR ideas. First, I
state three premises which have emerged from preceding sections of this
report:

1. FSR sensu stricto is irrelevant (except to the extent that a

little of it provides the necessary FSP);

2. OFR/FSP is now a well-established and effective adjunct to

research that aims to make modest step-wise changes in
established farming systems; methodology is secure enough
and the "style" has come to stay;

3. the longer term need for NFSD is implied in many systems

(Chapter 5) and there is acute need of it in the lowland wet
tropics, where any sensible development is likely to be
based upon the perennials, of which our understanding is yet
very deficient (see above).

Following these premises, I think the Bank should, in relation

to its CG responsibilities, help to encourage the CG institutes:

1. systematically to develop FSP among its researchers, use

OFR/FSP methods themselves when they are relevant, and
enhance training and networking for their NARS colleagues;

2. to take perennial food crops and other trees, especially in

the wet tropics, really seriously, in a deliberate attempt
to develop the NFSD which are so sorely needed.

In relation to its own agricultural development projects, the

Bank's interests can not, I think, be quite so concisely defined. I have
the impression that agricultural development projects at large (not only
the Bank's) have not, in general, been spectacularly successful. Broadly,
such projects have usually had three components: infra-structural (roads,
land clearing), supplies (fertilizers, planting materials) and extension
(with recent emphasis on training-and-visit methods). The approach has
usually been of a package of practices delivered top-down and time scales
usually short (say five years). I am in no position to analyze the causes
of such failures as have occurred (cf. Chapter 2) but have encountered a
growing belief among experienced people that top-down "packages" ill
adapted to local circumstances have been a significant element. This
seems to me to be at least likely. I note that the three current
agricultural development projects in the savanna zone of Nigeria have all
introduced research elements with explicit OFR/FSP components into their
programs and are convinced of the good sense of doing so (cf. Chapter 5).
I understand (non-authoritatively) that Bank thinking is tending to
replace the traditional five-year project, founded on a pre-determined
 - 62 -

package, by longer, more cautious, flexible, phased schemes that embody a

research component with OFR/FSP in it. I think we now know enough of
OFR/FSP to assert that this seems entirely sensible in areas where step-
wise enhancement of the agriculture is feasible. Collinson (1984), it
seems, would take the idea still further and would make the formulation of
an agricultural development project conditional upon a successful
preliminary OFR/FSP study, and he points out that the OFR economist would
then be singularly well placed to conduct any subsequent
monitoring-and-evaluation operation.

It is hard, however, to see such ideas working very well in

shifting/bush fallow areas in the lowland wet tropics where the
opportunities for step-wise improvement are severely constrained and
massive land clearing leads merely to disaster. Here, the Bank can hardly
evade confrontation with the problems of NFSD. But, due to our lamentable
lack of understanding of perennial food crops and the use of trees
(see above), working NFSDs can not yet be defined and it may be decades
before the knowledge is available to do so really sensibly. In these
circumstances I believe the Bank should be adventurous, try to
short-circuit the missing research, put some groups of unit farms (on a
best-guess basis) on the ground, and expect to lose money. Even if they
more or less failed, the FSP feedback to the researchers could be of great
value. It could be argued that deliberate financial imprudence should
have no place in a respectable Bank's activities, but the cash at risk
would be small in relation to total spending and the gains potentially
great if the technical and economic issues were clarified.

I conclude that, in relation to its agriculture development

projects, the Bank should:

1. continue to introduce a research element, with OFR/FSP,

into schemes in which step-wise change is intended and
phase those schemes in such a way that objectives can be
adapted to improving knowledge;

2. make some bold, if financially imprudent, efforts to

generate NFSD in the lowland wet tropics, on a best-guess
basis, in the hope of at least partial practical success,
of short-circuiting the research process, and of
generating FSP of the problems involved.
 - 63 -

CHAPTER5. EXAMPLES

Beans in Highland Colombia

Phaseolus beans, a major element in the CIAT research program,

are a very important food crop in upland tropical America. Some 50-70
percent of them are climbing beans grown on maize stalks, so inter-
cropping is the rule. They are an important (often the leading) cash crop
for small farmers. Though several main cropping systems can be recognized
(e.g., four in Colombia: Table 5.1), there are many local variations
(e.g., climbing beans on poles and wires and bush beans for green
vegetables locally in the eastern Antioquia area).

I saw something of the OFR work by CIAT/ICA in the Antioquia

province, near Medellin. OFR/FSP methods are essentially identical with
those of CIMMYT. Antioquia is high and only one crop a year is possible.
The beans are rotated with potatoes (or vegetables locally) and are
relay-planted into maize. The maize is broken over at maturity to form a
fence upon which the beans climb and the cobs usually dry off in the
field, to be picked as needed. The area presents an interesting example
of farmer innovation, from fairly crude potato-maize-bean mixtures 50
years ago to an orderly rotation-relay cropping about 1970, though at low
yield; bean yields are estimated to have nearly trebled in the past
decade as a result largely of fungicide use and some fertilizer applica-
tion. The beans are essentially a cash crop and only a mottled pink bean,
like the current land-race variety Cargamanto, is acceptable to the market
(though trade-offs are conceivable: cf. observations on Figure 2.1).

The ICA/CIAT OFR team concentrates on step wise improvements of

agronomy (which tend to be interactive with variety). The breeders have
had the oft-repeated experience of breeding varieties that were excellent
in station trials but unimpressive on-farm. Stability of yield is clearly
(and explicitly) very important to the farmer, and there is an interesting
suggestion that the genetic heterogeneity of the local land race
(Cargamanto) somehow contributes to it; at least, pure lines isolated
from Cargamanto seem less reliable over sites and seasons. So the CIAT
breeders are considering the possibility of deliberately retaining
heterogeneity (which seems to me to be an excellent idea: purity is often
over-valued). In their station plots, the breeders grow and select their
beans on maize under diverse cropping regimes. To the farmers, the maize
is relatively unimportant so long as it yields tolerably and holds the
beans up. A tall local variety at wide-spacing is used and stalk strength
is crucial. Shorter, modern maizes at high density, with thin stalks, are
unsuitable (especially if susceptible to Helminthosporium). Perhaps maize
breeders somewhere should do the complementary selection and pick
varieties tolerant of beans hanging on them?

Local uptake of new methods seems to be good, and there is no

doubt that the farmers will be receptive to good new varieties when they
emerge (CIAT beans are being adopted elsewhere). There is an incipient
training-network scheme in hand (Woolley and Pachico, 1983). Sanders and
Lynam (1982) comment upon the CIAT bean and cassava OFR programs.
 Table 5.1. BEAN-CROPPING SYSTEMS IN COLOMBIA

(1) Northern (2) Central (3) Eastern (4) Southern

Element Narino Narino Antioquia Narino

Alt. (km) 0.9-1.5 1.8-2.3 2.0-2.3 2.4-2.9

Period
(months) 3-4 4-5 5-6 8-9

Crops per
year 2 2 1 1

Rotation none peas potatoes or potatoes or

vegetables barley

Plant bush or semi- bush climbing climbing

habit climbing

Association row intercrop monoculture relay intercrop associated with

with maize with maize Vicia, and urbits

Technical
Index* 23 64 226 145

* A round-figure measure of the technical inputs already used by farmers.

Source: Woolley and Pachico, 1983.

 - 65 -

Maize ProductionMethods in Panama

Several examples of CI1MYT'sOFR/FSP work in diverse countries

in Latin America and Africa are outlinedin CIMMYT (1981). In this
example I briefly describe the IDIAP (Panama)program in the Caisan area,
1978-82 (Martinezand Arauz, 1983; Martinezand Sain, 1983 [economics]).
Survey revealedtwo recommendationdomains of which one was chosen.
Spacing,weed control and fertilizingwere identifiedas probable limiting
factors to maize production.

The first season of OFR work concentratedon factorial

experimentson these factors and results suggestedthat higher planting
density, better weed control (by herbicide),and less fertilizerwould be
favorable. The second season thereforerestrictedthe treatmentssharply
but added the idea of zero tillage (reasonablein the context of an
erodible soil, labor shortage, and herbicideusage). The third season was
devoted to verificationtrials of the package: no fertilizer,zero
tillage,higher density,herbicide. Farmers proved keen and made highly
practical suggestionsfor enhancement. Uptake was excellentand, by the
end of the study, 60-80 percent of some 300 farmershad adopted (in rather
varying degree). Fortifiedby the experience,the NARS (IDIAP) has
greatly expanded its OFR/FSP activities.

Martinezand Sain (1983) provide the only example yet available

of an attempt to examine the economic consequencesof OFR work by
cost benefit analysis. Their analysisshows pretty convincingly,I think,
that B/C>1. Costs for 1978-82 attributableto the OFR work
amountedto about 78 k$. They used Griliches'smethods of estimating
consumersurplus (thoughmethodologicaldoubts attach, as to all
cost-benefitanalysis)and comparedthe cost flows with the flows of
returns estimatedin the presence of OFR with hypotheticalreturns at
lower rates of farmer uptake had OFR not been availableto speed the
process. The last is the most doubtful bit (Figure 5.1) but a range of
reasonableassumptionsall showed substantialnet benefits,with the most
likely B/C ratios> 10 and (the Griliches)rates of return around 200
percent. The authorsused a discountrate of 15 percent and assumed
(reasonably,I think) that CIMMYT experiencewas an external "free good."
I hope that no one will now say that OFR must be a good thing because it
yields a social rate of return of 200 percentupon investment.

NFSD and OFR/FSP in Nigeria

NFSD work at IITA

The problemsof replacing shiftingcultivationsor bush fallows

by somethingmore stable and more productivehave been recognizedfor
decades,even though they are workable (even more or less stable) systems
at low populationdensity. Very similar agro-climaticproblemsare, of
course, faced in the wetter parts of lowland tropicalAmerica and
southeastAsia (see below).
 - 66 -

Flgure5.1: Cost-BenefitStudyof Cimmyt/IDIAPOFR/FSP

Workon Maize In Panama

Percent
Adopflon

lnme
-

Note& Socialretums(as economic surplus)are estimabletor each curve. Curve(1) relatesto OFRas it was
appiled. Curves(2)-(5) representIdeasas to what might have happened in the absenceof OFR.
Somethinglike(2) seemsthe mostplausible. Benefitsattributable to OFRare estimated tromareas
between curves. Curve(5), the abscissa,representsthe extremecase of no adoptlon (not considered
by Martinez& Sain,1983).
 - 67 -

FSR sensu lato has been an element of the IITA mandate from the
start. It has been interpretedas, essentially,a NFSD problem,which it
surely is. Accordingly,the Institutehas devoted itself primarilyto
componentresearch,recentlyauthoritativelysummarizedby ter Kuile
(1983) as: (1) land-development methods and forest-clearingtechniques;
(2) minimum or zero tillage,mulching,and fertilizers; (3) varietal
improvement,especiallyin disease and insect resistance; (4) mixed and
relay croppingmethods; (5) agro-forestrysystems,especiallyalley
cropping; (6) integratedpest and weed management; (7) appropriate
mechanizationand farm energy development; (8) integrationof livestock
into farming systems.

Present practicesare generally agreed to be under acute

pressure and of decliningproductivityas cycles shorten and soils are
"mined" (cf. Ruthenberg,1980, Figure 3.5; Greenlandand Lal, 1977; Lal
and Greenland,1979). They are summarizedhere in Figure 5.2; Figure 5.3
presentsa fairly imaginativeinterpretationof what an emergentNFSD
might look like.

IITA is certainlythinkingof NFSD. Apart from some survey

work, the Institutehas hardly been on-farm in the CIMMYT/IRRIsense but
more OFR is planned for the near future (Ngambekiand Wilson, 1983) and
training-networking activitiesare in preparation. IITA workers are
inclinedto think (and I agree with them) that, of the available component
technology,alley croppingis probablythe most promisingstartingpoint
(Kang et al., 1981). The ILCA group at IITA alreadyhave substantial(and
seeminglyfairly encouraging)experienceof alley croppingon-farm in the
area, primarilywith a view to feeding small animals (Sumbergand Okali,
1983). No doubt the two groups will have much to learn from each other.

Even if alley croppinggoes well on-farm,it can hardly suffice

in itself. Any effectiveNFSD for the low, wet tropics must surely be
more complicated. If such is to be developedinto practice,it is hard to
see how it could come about except with suitablegovernmentintervention
by way of model (unit) farms, credit, and infra-structural development.
Unit farms have indeed been tried by IITA (in the context of wet
bottomlands: Menz, 1980). A key element may well turn out to be land
tenure: communalland tenure is not conducive to improvement. Herein
probably lies the differencebetween much of Africa and Asia. In
SoutheastAsia, in similar physicalenvironmentsbut with private tenure,
there is already much static (as against shifting)agricultureand some
prospect of step-wiseadvance (see below). Nor are examplesof static
agriculturein humid Africa wanting: the great banana-basedcultivations
of East Africa (Jameson,1970; Arnold, 1976) may owe somethingto soils
and altitudeand are probablynot a general model for developmenton low,
moist areas of degraded land; but they suggest bananas as a basis for
developmentand serve to remind us, yet again, of the extraordinary
neglect of perennials.

The key questionsbefore IITA seem to me to be: (1) the extent

to which step-wiseadvances are possibleand the use of OFR methods in
generatingthem; and (2) whether one or more socio-economically feasible
NFSDs can be inventedand proved in practice. The bigger question
remains: whether African governmentswill have the power and will to use
appropriateNFSDs to counter what is generallyagreed to be already a
nearly criticalfood-supplysituation.
 - 68 -

Figure5.2: Generalized Shifing Systemof Humid and Subhumid Mfica

~
Hec2W ~ Weed&
Labor Pathogen
Demands Problems

L__ BushUncler Selectively Hoe CultKiatlon

? Ranching Communal f Clearedby of Annwls
r _ Tenure Hand Residual Tree

DegrodeIdTreeCro
Gross
_ Planting

== | | ~~~~~~Eroslon.
Land Abandorecd Leaching, Small Vegetable
or Fallowed 4 Dehumiflcatlon, Animals Products
_ ~~~~Mineral
Offtke.

Cash &
Subsistence

/Losses \\/ \ / \
/ Reduced by: \/ Useof IgnoVrediy
Contourlng. , Bovines \/ 0oe
Mulchlng. Vrtually
\A l by W?ZCs
DrainageControl Forbidden || ThoughNot
 - 69 -

Figure 5.3: A View of Possible Stable Forming Systems in Humid/Subhumld Africa

, .W I XA FB
Bush Machine-
(PRt'ate * Assdsted _ 0 MoreVulrerabb Areas
Tenure) Ceadrln UnderPerenniols
(Bananas,TreeCrops).
j K ~~~~~~Maybe
Undercropped.
1 Zero-Tiled,
_~~~~~~~~~~~~~-O

Inputs:
AdloptedVarSb
PothoWe Conro BC
Herbkicids _ _tr
SmallMochines
or Ox-Work _Less VulnerableAreas
Feltillzr Mlxed Cropped with
_ _ ~~~~~~~~~~~~~~Annuals,
TreessJShrubs
Present.Contoured,
~~~Mulched, Minirnally
_ Tilled, Dralnoge

B Work Dung Controllbd.

Stocke Productstor

t_1 + ;~~~~~~~Doresi Subslsence_

|
L Wastes

ObseNatlons:

A] Sxoll CharVe in LandTenure,

I Cash Avollobbe.MarketsSecure,

M IITAWorkRelevnt

FC]ICRAF WorkReW.vant

DI ILCAWorkReSvant
 - 70 -

OFR/FSP in the Nigerian Savanna

If the problems before IITA in coping with NFSD for the lowland
wet tropics appear to be peculiarlyintractable,those of the drier
savanna zone of Nigeria appear, to many observers,to be somewhat less
so. At least, there seems to be some prospect of immediate step-wise
improvementby OFR/FSP methods. The context is, not of IARC/NARS
research, but rather of the pressing practical problems of agricultural
developmentprojects already fairly far advanced. Consortia consistingof
the Nigerian Government,state governments,and the World Bank have three
major agriculturaldevelopmentprojects in the center and north of
Nigeria, in Kwara, Niger, and Onde States.Activities are co-ordinatedby
the Federal AgriculturalCo-ordinatingUnit (FACU) in Ibadan. All three
have experimentstations and all run OFR/FSP activities,using their own
agriculturalresearch staff working in collaborationwith the NARS and
with IITA economists. The work is on a very considerablescale (the Kano
project has 800 collaboratingfarmers)and is primarily,though not
wholly, directed towards small farmers by way of infra-structure,
supplies, and extension.

A substantialpart of the work is directed towards the

management of valley bottoms, or fadamas. These are, in total, very
extensive in the savanna zone and offer patches of land which, given some
modest water control, are potentiallyrather productive; at least they
are better watered and more fertile than the dry sandy areas around them.
Large irrigationschemes are not in prospectbut much can be done, it
seems, with quite modest resources. There are problems,of course, for
example: locally high concentrationsof ferrous iron, indicatingthe need
for accurate drainageor plant (rice) breeding for toleranceof iron
toxicity,or maybe both. I did not see anything of the OF work in the
field in the fadamas or in the dry part of the Kano project area but
understoodthat the philosophyadopted was to minimize the formal
researcher-runexperimentationand to go as quickly as possible to
numerous, simple, widely dispersed farmer-managedtrials.

I saw something of the OFR work in the Ilorin area of the Ilorin
agriculturaldevelopmentproject. The project is aimed primarily at
enhancing the agricultureof about 60,000 small farmers in about one-third
of Kwara State. The OFR work is backed up by a researchstation used for
trials and seed production. The local farming system is a bush fallow one
in savanna woodland and is based on a two-year cycle of croppingrepeated
two or three times, followed by about eight years of fallow. A common
croppingpattern is early maize, relayed with sorghum,relayed with yams
grown on a trellis of sorghum stalks. The terminal yam crop is often
replaced by cassava. Cowpeas used to be importantin the area and are
much favored for food and cash but have declined due, it is thought, to a
combinationof recurrent drought and insect pests (the farmers agree).
The OFR work concentrateson new varieties of maize, sorghum, sweet
potato, cassava, cowpea, soybean, and okra and on sprayingcowpeas with
insecticide. Ultra-low volume spraying looks promising (water is scarce)
and is likely to be just within the economic reach of many farmers, either
on their own or by contract. They took very well the point that, sprayed
early, cowpeas promised to restore a crop that they valued highly. They
 - 71 -

liked the IITA cassavaswith mosaic and bacterialblight resistance(their

own varietieslooked terribleand the IITA materialwas being
enthusiastically passed on to neighbors),insistedthat any new sorghum
had to be tall in order to hold the yam vines and liked the idea of
streak-resistant maize (it had been a very bad year for the disease
locally). In general,the farmerswe saw were criticaland receptive,and
I do not think there can be any doubt that the Ilorin OFR work is
contributingmateriallyto the progressof the project. As I understood
the situation,the Ilorin project is somethingof a pioneer in this area
of work and was the first deliberatelyto seek (from IITA in 1981)
collaboratingeconomiststo help to design the program and monitor
progress (cf. Chapter 4).

Rainfed Farming in the Mediterranean

The ICARDA mandate containsspecific referenceto the adoption

of an FSR approach..Studies in the early years of the Institute(1977-80)
of Mediterraneanfarming systemshave been summarizedby Gibbon (1980,
1981) and by Martin (1981). These authors present fairly detailed studies
of samples of rainfed arable farming in the area, concentratingmostly on
specificvillages. The general approach might be describedas FSR sensu
stricto but with the explicitobjective that others should identify
researchneeds and develop OFR work from them.

Rather broadly,four main kinds of agricultureare recognized:

(1) nomadic or semi-nomadicherding of sheep and goats on steppe
grassland; (2) arable rainfed farming on deeper soils in moister areas
(fallow-wheator fallow-wheat-legume rotationswith orchards- (olives,
vines, figs) - as the main cash source); (3) arable rainfed farming on
shallow soils in drier areas (fallow-barley,with sheep, often grazed on
the immaturebarley, as a main cash source); (4) irrigatedfarming,in
which, of course,very diverse croppingis possible. There are important
interactionsbetween (1) and (3) whereby sheep move seasonallybetween
steppe and arable,with increasingrelianceon stock the drier the local
climate. Rainfallunreliabilityis a major source of economicrisk to
farmers. There has been considerablemechanizationof arable crops and
consequentrural unemployment.

OFR work is in hand at ICARDA (some, at least, has been going

since 1977 and is being developed- ICARDA, 1982) but I find it hard to
judge from the literaturethe degree of commitmentof the Instituteand
its associatedNARSs to the general OFR approach(which may not have much
to offer the rich irrigatedfarmer but surely must be relevantto the
small man on the dry land).

Mixed Upland Farming in Ethiopia

Ethiopia is a large countrywith diverseagricultures

(systematicallydescribedby Amare Getahun, 1978, 1980). ILCA., with a
farming-systemscommitmentin its mandate, has done a good deal of survey
work on Africa, pastoralsystems in general (ILCA 1980, 1983) and, in its
highlandprogram in Ethiopia,has concentratedon the dominantmixed
upland systems of the plateau. Like the Deccan of India, the soils are a
 - 72 -

complex mosaic of less fertile sandy soils on the (often) gentle slopes,
with more fertile but ill-drainedblack clays in the bottoms. In
Ethiopia, the sandy soils are arable and the clays provide seasonal
pasture for milk cows, work oxen, and miscellaneoussmall animals.
(Diseases are little problem in upland Ethiopia, unlike most of lowland
Africa.) One form of the system (there are several, of course) is
outlined in Figure 5.4 which applies at Debre Zeit, an ILCA station and
on-farm site near Addis Ababa. The farmers are small (about 2 ha each)
and operate a complex land tenure system, with common pasture in the
bottom and cropland rotating among members of an "association." Tef
(Eragrostis)is the preferred cereal here (wheat elsewhere),despite its
low yield and difficultharvest: its retention is determinedjointly by a
strong market preference,by its toleranceof waterlogging,and by the
facts that its seeds store well and that its straw is good for fodder.
The pulses are peas, horse-beans,and chick peas, and it is noteworthy
that it is these that the farmers chose to sacrifice to grow fodder under
the ILCA scheme. The ILCA program startedwith surveys and proceeded to
OFR work using the ideas sketched in Figure 5.4. The key notion is to
replace numerous small work oxen and cows by fewer, larger, hybrid beasts
and feed them better. An extension of the idea (yet to be developed far
in practice) is to abolish oxen and use (well-fed)milk cows for draught.
There is certainly a local market for extra milk and the local
farmer-adoptersaround Debre Zeit seem pleased.

The farmers are good stockmen and take well to the Boran x
Friesian crosses that have been tried. Dr. G. Gryseels tells me that
average milk yields (liters per year) are about:

Local Cross-bred
(Boran) (B. x Friesian)

On-farm 300 1,500

On-station 600 2,800

so a spectacularincrease in milk yield and a clear GE effect are evident.

The ILCA scheme looks attractivebut implicationsfor the NARS

are daunting. First, there are 30 million cattle in Ethiopia so, even if
the numbers declined, there would be a gigantic breedingdemand. Second,
the best breeding plan has yet to be determined (50:50 Boran/Europeanor
perhaps a rotationalSahiwal/Europeancrossing program?); whatever is
chosen could be costly and would have long-termimplications. Third,
though there is now a market for more milk, greatly increasedsupply would
probably imply a need for uptake in other than liquid form and, again,
government intervention. Though establishedby OFR methods, this is not,
I think, an innovationthat the farmers,unaided, can themselvesexploit.

Timber is generally scarce in the plateau but trees grow well.

This is an obvious area for agro-forestry,with the attractionsof
supplying local timber, fuel, and fodder and of saving dung for
fertilizer.
 - 73-

Figure 5A: ILCAProposalsfor Mixed Upland Foring in Ethiopia

1 1.~~~~~~Drf

Plough Lanci Stock,Say:

Rotation:~~~~~~~~~ 2 Oxen __
Cereal __Cmo
Cereal _ 1 Cow _ _ ~~~~~~~~~~~~~~Pasture

Residues ;G(rsaozngl

Grain Milk MetDung

| Subsistence | ah | Fuel

) I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1- 1 \ / ~~~~~~~~~~~~~~~~~~~~~~~~Cattle
for Small Borain.\
t~ ~~ ~~~b OatusV4tcHutFr 2. Peduce Oxen to One &\
~ ~~{
_~ ~~~r
b_H foatsVthy Modify Plough & Yoke.
\ MlxtureforHay
/ 1 ~~~~~~~~~3.
EvenKeep No Oxen &
\ / \ ~~~~~~~~~~~~~~~~~~~~Use
Cow(s) for Draft./
\ / \ ~~~~~~~~~~~~~~~~4,/
Meet Increased Feed Demn
/ t X ~~~~~~~~~~~fromOats-Veth.

Ehance Mlk&< _E Ntoo ProblernsHfLarge\

\ ] ~~~~~~~~~~~~~~~~~~Socile
Aoption Is to be \ /
\ Netincome
/ ~~~~~~~~~~~~Achieved:
Chlce of Parent |
/ t ~~~~~~~~~~~~Storcks
& Breeding Pattern:
\ /~~~~~~~~~~~ _ \~~~~~~Maintencince
of Nuclear|
< / ~~~~~~~~~~~~~~~~~~~Stocks, Marketlng of/
 - 74 -

It is interestingto reflect that it was ILCA's mandated

preoccupationwith animals that led them in the direction they have
taken. I suspect that if ICRISAT had looked at the plateau of Ethiopia
they would have seized upon the relativelyfertile black soils in the
bottoms, drained them for cropping,and banishedthe stock to the sandy
slopes. This thought is in ILCA's mind, and it has been found that
drained beds on the black clays do indeed seem to be (at least locally)
workable and productive. But ILCA is presumablydebarred by its mandate
from proceedingfar with the idea. To invert the system thus, so to
speak, would certainly be to propose a NFSD.

Cropping Systems in the Deccan of India

ICRISAT, from the start, was committed to FSR sensu lato and it
conceived this in terms of NFSD; in a recent authoritativesummary,
Dillon and Virmani (1983) use the phrases "system replacementin India"
and "virtuallya new farming system" (see also ICRISAT, 1983). The
outcome to date has been rather less radical; it is, in effect, a set of
proposals for intensifiedcropping of the black cotton soils.

In Andhra Pradesh (and characteristicalso of a good deal of the

Deccan) the agriculturalscene is dominated by a gently undulating relief
with a mosaic of black, cracking clays in the hollows and red sandy soils
on the slopes, interspersedwith knobbly granitic outcrops. The black
clays (vertisols,black cotton soils) are sticky when wet but retentive,
hard to work when dry, excellentlystructured,and moderatelyfertile;
the red sandy soils (alfisols)are infertile and easily worked but very
sharply drained. Ploughing (with a pointed, not a mold-board, plough) is
done by oxen; their power is generally insufficientto till the black
cotton soils in the dry season so they are ploughed in the rains and,
typically,one crop per year is taken on residual soil moisture (Figure
5.5). Farmers, however,know very well that some patches can be tilled
dry so that end-of-dry-seasonsowing is possible and hence, locally,
double-cropping(kharif and rabi). The ICRISAT scheme simply generalizes
this idea by exchanging the traditionalplough for a wheeled tool-bar
which the oxen can work in the dry season and which can be variously
adapted to forming graded beds-with-furrows. Fairly reliable double
cropping becomes possible (Figure 5.5); cultivationis faster and more
timely, farmers like to sit on a tool bar rather than walk behind a
plough, and ICRISAT calculationssuggest that the cost of the equipment
will be within reach of many farmers.

ICRISAT has had farm survey work going back a decade or more but
the black-soil-toolbartechnology (Thierstein,1983) outlined above was
developed on-stationand only taken to the OFR phase fairly late (1981
on). Now, a collaborativeOFR program with the All-India program
(AICRPDA;ICAR, 1982), state governments,and universitiesis being
developed. No doubt, generalizedOFR economic assessmentswill soon be
available (they look very attractiveon-station). The farmers I talked to
in Andhra Pradesh were enthusiasticbut the overall economics is not yet
clear.
 - 75 -

Figure5.5: ICRISAT
Proposalsfor the Cotton Soilsof the Deccan

March

ICRISAT
Cropping

r-------- IWatershed:
I : Surveyed.
Dralned.
I l ~ ~~~~~~
Contoured.

I Till& sow I

Rain
June I

I_LP Equipment

I Essentlal;Oxen
Trod'I WeaklWorkIn
Cropping _ HottestWeather

& ,
-nTil "i
Sept I So I

I I _ Competitivewith
I i} LaborDemands

Dec _
Rabi
r Residuai\ Cereal
t Moisture1 | Innumerbie \
Different
LPy _ 2 _ Cropping
Sequences

> _ v~~~~~~~~~~~~~~~~~~Psii
 - 76 -

The essential outcome of the ICRISAT program so far has thus

been a very promising bit of technologybut only one component of a much
wider scheme. The Institute conceived of "watershedmanagement"(very
small watersheds)with contoured cultivation,controlleddrainage,grassed
drains, integratedcropping of black and red soil patches, perhaps water
storage in small tanks fed by drains (or improvedwatering of the rice
paddies). In practice, integratedwatershed managementwould be a huge
task and would imply extensive consolidationof dispersed landholdings
(somethingthat the local farmers evidently do not fancy). As to the red
soils, ICRISAT grows excellent crops on them on-stationwith high inputs;
what farmers might do is likely to be revealedonly by OFR/FSP methods, I
think; leguminousmulching/incorporation is one interestingpossibility.

There are wider questions. I remarked earlier in the chapter

that, if ICRISAT looked at upland Ethiopia, it would probablyhave gone to
the black soils rather than, as ILCA did, to the cattle. Indeed, both
instituteswere effectivelyconstrainedby their mandates to do what they
did. ILCA arguments for fewer and more productive,better-fed cattle
apply also in India, where cows mother oxen and buffalo produce milk. A
social revolutionwould, of course, be implied but one can at least
imagine a NFSD for the Deccan, even wider in scope than that conceived by
ICRISAT. Managed catchments,small machinery,double-croppedblack cotton
soils, mulched red soils, agro-forestryfor timber, fuel, fodder, and
mulch, fewer but better-fed cattle for both work and milk, and so on. In
short, ICRISAT started with a NFSD concept but has arrived at OFR/FSP in
collaborationwith local agencies. The Institutehas a strong training
program with FSR ideas as a component of it.

The NARS (that is, AICRPDA in the context of dryland farming -

see ICAR, 1982) and their State counterpartswhom I met in Andhra Pradesh
and Karnataka seem collectivelyto have, broadly,a commodity/discipline
orientationbut with explicit commitmentto OFR/FSP methods (usuallyunder
the title "operationalresearch"- see, e.g., Sanghi, 1982).

Rice-based Systems in Asia

The outstandingsuccess of the Asian Cropping Systems Network in

raising irrigated rice yields in Asia is well known. The ACSN arose
directly out of IRRI's perception that the quick maturity and day-length
insensitivityof some of the newly available dwarf rices would permit
sequentialmultiple cropping to two or even three rice crops per year. So
IRRI concentratedfirst on areas in which good water control was available
(or could be developed)and, recognizingthat collaborationwith local
NARSs was of the essence, developed the ACSN, with strong emphasis on
training, informationflow, and OFR/FSP operations (Figure 3.5). The last
was not so entitled, but OFR/FSP was in fact how the ACSN operated (Zand-
stra, 1979; Zandstraet al., 1981; Quisumbing, 1982.) The outcome was
not just sequentialrice-croppingsystems but, sometimes,such sequences
as rice-rice-cowpeas(Figure 5.7). Now that the potential for developing
fully irrigatedrice systems is nearly exhausted,and the rainfed crop has
been intensivelydeveloped too, attention is shifting to rice under less
than perfect irrigation,to upland rice, to associatedcrops (e.g.,
cowpea), even to animals; hence the change in title to AFSN noted in
 - 77 -

Figure 3.5. The Network's attention to fully irrigated rice will not,
presumably,be allowed to lapse completelybecause yields can yet go
higher and second-generationproblems are emerging, for example: possible
minor nutrientproblems (a consequenceof past success)and effects of
agri-chemicalson fish and other wildlife.

The outstandingsuccess of the ACSN was due, I think, to the

fact that the objectivewas never in doubt (raise wetland rice yields per
ha/yr), that it was shared by all participants,and that the potency of
the OFR/FSP approachin the contextwas fully appreciated. And it may be
that, in retrospect,it will appear that not the least achievementof the
ACSN is to have stimulated,by example and training, the participantNARSs
to develop their own research systems and local OFR/FSP. Certainly, the
Philippinesand Indonesia seem strongly committedto developinglocal
agriculturalresearchwith OFR components(cf. next two sections).

Small-farmerCoconuts in the Philippines

Some 95 percent of the 9.6 million hectares of coconuts are in

SoutheastAsia and the Pacific and the small farmer is far larger than the
estate sector. The Philippine islands are the leading copra exporter and
the small farmers there mostly grow for cash (elsewhere,coconuts are more
of a subsistencecrop). The system is broadly summarizedin Figure 5.6.

The PhilippineGovernment, through the agencies of PCARRD, the

Ministry of Agriculture,the University of the Philippinesat Los Banos,
and local Colleges of Agriculture (e.g., VISCA, the Visaya College in
Leyte) are committed to an OFR/FSP approach to enhancing diverse
agriculturalsystems, including the coconut one (PCARRD, 1982). I saw
somethingof the work in the field near Jaro, in the TaclobanCity area of
Leyte. The Eastern Visayas are Region 8 of the 12 Regions of the
Philippineislands; six areas have been identifiedin the Region, the
islands of Leyte and Samar; in each area, several on-farm sites
representativeof "recommendationsdomains"have been identifiedand
preliminarysurveys carried out. OFR teams, including economists,live at
each site.

Farms are small and, as noted in Figure 5.6, the coconutsare

tall, old, gappy, and running out, since most are residual from a surge of
planting in the early years of this century. Since nothing dramatic can
be done about the coconuts in the short term, the OFR work at Jaro
concentrateson seeking to undercrop the coconuts much more widely than
they now are. Diverse food crops such as dryland rice, bananas, and
cassava and cash crops such as pineapplesand cocoa are being tried;
small animals are another possibility (though the goats I saw were not a
spectacularsuccess: they looked unhappy, needed shelter and ate the
neighbor'svegetables). The farmers were clearly responsivebut no
economic analysesof the outcomeswere available.

The OFR work along these lines may well prove beneficialto the
farmers because the soils are quite good, rainfalladequate, and the
potential for enhanced productionis clearly there. A wider question
remains,however, as indicated at the bottom of Figure 5.6. The coconuts
 - 78 -

Rgure 5.6: Smaliarmrer Coconut SystemsIn the PhDlppne

r ------------- mmm mmmm l

Coconuts

_ Estate SroloO"he Enterprbes

Sector Farunnis FoodGaWdens
ne.g
Asa _ Pacific:Rce Small tches, S
m mLlm
evestock,
mm mm mm mm Than
m Estatem shing.

Copra: Domestic:
Oil & Food,t)rnk. ul
8y4kducts Thatch,Wood,\ S
HandkraffsCcrt d al
IL ~ Cash _Weedv. UnpouHe
RunnlngOut,Wide
Op sLocal VarHon In
Exhance RepaCSubsistence
Ei ExisthgCoconut FS f Cnpen
95%of9.6MhaSsniSE \
Asia & Pacific.Smaull |
I FamferForLargerw
| ThanEstateSector.

Exsig Rpatfr Coonts

Existing Replanttor ~~~~CopFO-Cash Invenuts

Systemns Copra-Ccsh & Subsistenoe NewFS.
 - 79 -

everywhereare old and the PhilippineCoconut IndustryBoard has been

embarked, for some years, on a policy of replantingwith hybrid
semi-dwarfs. Though some doubt apparentlyattaches to the choice of the
best hybrid combination,it can not be doubted that semi-dwarfsare the
correct combinationfor the estate sector, if coconutsare to be replanted
at all. However, competitionfrom other oils such as that of oil palm
makes the future for coconut oil less certain. So what is the future for
small-farmercoconuts? Should they be replantedwith whatever hybrids are
chosen for copra-cash (the best choices are far from clear anywhere),at
high density,and with little opportunityfor under-cropping? Should the
same hybrids be used but at low density, deliberatelyto leave space for
intercropping? Or should coconuts be abandonedas the old talls die out
and some NFSD be invented to take their place? These are hard questions
to which the Philippine authoritiesare no doubt applying themselves.
Meanwhile all that the OFR/FSP workers can do, I think, is to try to
devise ways by which the farmers can get more out of the old coconut
system while it is still there. Thus, what is now plainly an OFR/FSP
problem may become transformedin time into an NFSD one.

Small-farmerFood Cropping in Indonesia

Indonesiawas very active in IRRI's ACSN (Figure 3.5) from the

start and, fortified by the success of that program (Bernsteinet al.,
1982) for irrigated rice, adopted the OFR/FSP approachgenerally for
subsequentsmall-farmdevelopments. The basic commodity orientationof
the NARS (AARD, 1981) is not thought to be an obstacle to the OFR
approach, because specialistsare drawn into OFR teams as necessary. The
government (AARD) officerswork with collaborativesupport from outposted
IRRI staff and internationalagencies and, in total, the effort is a very
substantialone, with several sites in all the main islands. The
practical problems have not been eased by the fact that much of the work
has had to be done in "transmigrationareas" of Sumatra and Kalimantan
where migrants from the more fertile but over-populatedlands of Java have
been settled. Even with migration,however, Java remains over-populated
and there is severe pressure on the erodiblehilly lands (unhappily,often
referred to as "watersheds"). The general problem therefore is to enhance
and stabilize food production in rather diverse soils, altitudes,and
rainfalls (but exclusive of irrigated rice on lower ground).

The main points of the approach are summarizedin Figure 5.7,

and further details of OFR methods (essentiallythose of the ACSN) will
be found in AARD (1981), Inu (1983),McIntosh and SuryatnaEffendi (1981).
Some economic results are cited in Table 3.3. It will be seen from the
figure that three levels of difficulty can be identified. With moderately
good irrigation,a dry-seededrice/irrigatedrice/cowpeasequence is
possible and being adopted; it lies in the area now being developed by
IRRI under its AFSN (see above). This part seems fairly secure and, to
the extent that irrigationcan be further improved, could be transitional
to really intensive rice productionon the IRRI model. Rainfed
agricultureon moderate slopes is a major concern in the trans-migration
areas where multiple cropping in space and time with moderate fertilizer
inputs are evidently attractive (Figure 5.7 [2], Table 3.3). The steeper
slopes present the hardest problem and, though the general form of the
technologyneeded is clear enough (Figure 5.7 [3]), economic feasibility
of possible solutionshas not yet, I think, been established.
 - 80 -

Rgure 5.7: OFRand NFSDin Indonesia

Oct. Jon. Apr._ Jul.

1. Irrigated

II Irrb~~~~~lflotlon .

Rice Tradltlonal

I | , ! |~~~~~~~~~~~
New
[Z17 Rlce Rice |iCowpea
I I

2. Uninigated,
Moderate ±

I | | Troditonal

Malze& Rice
Mix Cassava
Relayed

Maize (Rows)Ne

Rlce(ROWS) Peanut cwe

Cassawva
(RekbVedRows)

3. UnlTlgated,SteeperSlopes

a. Terraced,
withContourLegumes,
MlxedCroppingAsIn2.
b. TreeCropstor Food& Cash.
c. FoddertorStock:Poultry& Small
AnimalsThere & Important,
AJso
Wok Butfflo:?DairyCattleDevelopment.
 - 81 -

Technically,the problems of rainfed agriculturehere look much

like those of humid Africa (sectionon Nigeria,above). However, the
soils of Indonesia are probably rather better than most of those of
Africa, people are accustomedto settled agriculture,land is privately
owned, there is a long tradition of more or less integratedricetdryland,
food/mixedtree cropping, terracinghas been understoodand practiced for
millenia, and stock are not particularlythreatenedby diseases. So the
ingredientsare there and it seems to me likely that a systematicOFR
effort, coupled with good extension,could lead, step-wise,to stable and
productivesystems (howeverdifficult it is to conceiveof such a change
in the wetter parts of Africa now dominated by shifting systems). So the
problem before Indonesia,even on the steepest ground, has not necessarily
the characterof a NFSD; step-wiseprogress seems conceivable.

Another area of activity in Indonesia,currently assistedby the

World Bank, lies in the tidal swamps. These are very extensiveareas of
difficult,acid soils and peats which have, nevertheless,considerable
potentialwhen drained. I do not know enough to try to review the
problems but simply note that: (1) local farmers have already shown
considerableingenuity in growing annuals, perennials,fish and fowl and
that there must be much to be learned from them; hence (2) multiple land
use will almost certainly be appropriate; (3) given wetness and very acid
soils, fairly massive governmentinterventionin respect of drainage and
fertilizerswill be necessary; (4) whatever ingenuity the farmers show in
adaptingdetails, a degree of centralizedcontrol and something
approximatingto NFSD will be necessary.
 - 82 -

CHAPTER6. SUMMARY

Introduction

Diverse activitiescollectivelydescribedas Farming Systems

Research (FSR) have become a prominent feature of the work of the
InternationalAgriculturalResearchCentres (IARCs) and of many National
AgriculturalResearch Systems (NARSs),too, in the past 10-15 years. The
Bank's desire to sort out a confused terminologyand assess the usefulness
of FSR studies in agriculturalresearch prompted it to commissionthe
study reportedhere.

The broad context of the study is that populationsin most

tropical countriesare growing rapidly and that there is an ever more
acute need to produce more food from finite (sometimesalready limiting)
patches of land. The Green Revolutionvery effectivelyincreased yields
of wheat and rice over large areas by exploitationof genotype-environment
interactions(i.e., dwarf, responsivevarieties plus enhanced husbandry).
Green Revolutionsin other crops did not follow and it became apparent,a
decade or so ago, that many innovationsproposed by agriculturalresearch,
whether in unit or "package"form, were simply not being adopted by
farmers.

It became apparent that the reason for non-adoptionwas

generally that the innovationswere unsuitablefor the socio-economic
circumstancesof the farmers. The adaptationof the ideas of temperate
farm-managementeconomics to tropical small farmers revealed that they,
the small farmers,were economicallyrational but risk-averseand sharply
constrainedby uncertain environmentsand shortage of cash; they were
ready enough, however, to adopt innovationsthat they themselvesperceived
to be economicallyattractive. Accordingly,the doctrinegrew that
research should be determinedby explicit farmers' needs rather than by
the preconceptionsof researchers.

The ConsultativeGroup on InternationalAgriculturalResearch

(CGIAR) thereforeintroducedeconomists into all the institutesat an
early stage and explicitlyespoused FSR ideas with the object of linking
researchwith perceived farmers' needs.

Tropical agriculturalsystems are almost infinitelyvarious, and

the relativelycoarse classificationsnecessarilyadopted in the textbooks
are little use to the FSR worker in the field. Various kinds of mixed
cropping are characteristicof many tropical agriculturesand are
important in contributingto yield and stability of yield; FSR workers
very often meet them, not as farming systems per se, but as componentsof
cropping systems. Mixed croppingis interestingand important but not
central to FSR ideas.

Despite a frequent, seemingly stable situationin the

short term, most tropical farming systems are in a state of change. Four
agents of change may be recognized: the farmers' own perceptionsand
actions, the stimulus of publicly supported research,the economic pull of
 - 83 -

a cash crop, the commercialpush of a product sold to farmers by a firm.

The second of these is the main context of FSR and the question is: how
to do the right research, the kind that farmers really want, and transmit
it to them?

Farming Systems Research in General

The diverse activitiessubsumed by FSR in the broad sense (FSR

sensu lato) fall into three categories:

1. FSR sensu stricto, the study of farming systems per se, as

they exist; typically,the analysis goes deep(technically
and socio-economically)and the object is academic or
scholarly rather than practical; the view taken is
nominally "holistic"and numerical system modeling is a
fairly natural outcome if a holistic approach is claimed.

2. On-farm researchwith farming systems perspective (OFR/FSP)

is a practical adjunct to agriculturalresearchwhich starts
from the precept that only "farmerexperience"can reveal to
the researcherwhat farmers really need; typically,the
OFR/FSP process isolates a subsystemof the whole farm,
studies it in just sufficientdepth (no more) to gain the
necessaryFSP and proceeds as quickly as possible to
experimentson-farm, with farmers'collaboration; there is
an implicitassumption that step-wise change in an
economicallyfavorabledirection is possible and worth
seeking.

3. New farming systems development (NFSD) takes as its starting

point the view that many tropical farming systems are
already so stressed that radical restructuringrather than
step-wise change is necessary; the invention,testing and
exploitationof new systems is therefore the object; while
OFR/FSP seeks to adopt the technologyto the farmers'
economics,NFSD must usually imply governmentintervention
and the adaptationof economics to technology.

After a decade of work in the field, the economic characteris-

tics of small farmers are well enough understood for practical purposes,
though they present endless subtletiesof detail. Small farmers are poor,
economicallyrational (thoughnot necessarilyprofit-maximizing),
risk-averse,and subject to high interestrates. The methods of
farm-managementeconomics can be well enough adapted to their
circumstancesfor the practical purposes of OFR/FSP but experienceand
caution on the part of the economist are of the essence.

Several kinds of innnovation can be recognized: the costless

one (a new variety), the one that costs cash (chemicalinputs), the one
that costs effort (an extra crop in the sequence),and the complex
innovation ("package")that demands several inputs. Generally,the
simpler and cheaper,the readier the adoption; OFR/FSP is mostly
concernedwith unit changes or simple packageswhile NFSD has to face
the problems of multiple changes in complex packages.
 - 84 -

"Systems"have been made much of and may appear formally in FSR

sensu stricto. For OFR/FSP, the preoccupationis always with a sub-
system, often a small one, but in awareness of the interactionsthat are
perforce ignored; FSP rightly takes a commonsensicalrather than a formal
view of systems and only exceptionallyneeds to make numerical models. At
whatever level of analysis is relevant, the systems consideredby FSR
sensu stricto and OFR/FSP are far narrower than the systems recognizedat
textbook level (see above); in practice the only definitionof a system
useful for these purposes refers to an assemblageof farms (in a limited
area) which are like enough to each other, technicallyand economically,
for the purpose in view. No useful objective typology/classification is
possible at this level. A system is what an experiencedworker says it
is.

If FSR sensu stricto, however interestingper se, is largely

irrelevantto agriculturalresearch (above)and OFR/FSP is becomingwell
establishedas an adjunct "style" of research (also above), the position
of NFSD is far less clear. That there is local need of it is not in doubt
and there is much component technologyin several IARCs which has yet to
be synthesized. The need for NFSDs in the lowland wet tropics on fragile
soils is agreed to be acute yet there is little or no generalizedvision
as to how they might be accomplished. Somewhere,somehow, experienced
researchersmust step outside the component technologyand make
imaginativeguesses; and developmentagencies must be persuaded to try
those guesses in practice, even at risk of making some expensive
mistakes. I do not believe that OFR/FSP, useful as it is in its own
context, can make the kind of imaginativejumps which are needed. I
return to the subject later in relation to the general neglect of
perennial crops and to the activitiesof the Bank.

Genotype-environment(GE) interactionsoccur when different

varieties respond differentlyto different environments. They are a
significantfeature of all plant breeding; the more diverse the
environments,the more likely they are to be economicallyimportant.
Their existence implies that experiment-stationselectionmay not produce
farm-adaptedgenotypes and that breeding tends to be rather
site-specific. Decentralizationof breeding responsibilityand selection
are indicated. The matter is economicallyimportantbecause a very
substantialpart (probablyapproachingone half) of tropical agricultural
research is devoted to plant breeding and a like proportionapplies to our
expectationof plant breeding as a component of progress in crop
production. From the OFR/FSP point of view, variety adaptationon-farm,
as distinctfrom on-station,is the crucial point.

On-farm Researchwith Farming Systems Perspective

OFR/FSP (the term was invented by CIMMYT workers) has largely

been developed into practice by CIMMYT and IRRI working in collaboration
with local NARS in Central America and eastern Africa and in Southeast
Asia, respectively. Their methodologieshave been fully described in the
literatureand, despite differencesin terminology,are effectively
identical in structure. Essentially (cf. Figure 3.3):
 - 85 -

1. There is a team (multi-disciplinary)

in the IARC which

2. identifiesthe "target"FS or "recommendationdomain" and

3. analyzesits technical/economic structurejust deeply

enough for the purpose in hand;

4. the team identifieseconomicallysensibleinnovationsand

5. does research on them, on the experimentstation,on few

selectedfarms, on more numerousfarms, accordingto
circumstances;

6. under 2-5 the team solicitsthe interestand

collaborationof the NARES and

7. the experimentationis iteratedas necessary;

8. the experimentsare monitored/analyzedeconomicallyand,

if successful,lead to recommendationsfor the extension
system (NAES) to exploit;

9. there is a continualfeed-backof informationto the

IARC/NARESconsortiumas experienceand understanding
widen;

10. given that there is agreementthat the OFR/FSP "style" of

researchis valuable,the IARC must assume a
responsibilityfor trainingand network development.

The "recommendationdomain" in the precedingis the farming

(sub-) system chosen on the basis that it is broad enough to be
operationallysensible,but narrow enough that there is reason to think
that any recommendationsthat emerge would be applicablethroughout.
Choice is based on survey (as short and simple as possible) and economic
analysis. There can be no fixed schedule for the OFR part of the work but
close collaborationwith and feedbackfrom farmers is agreed to be
essential; generally,researcher-run,factorial-typeexperimentation(if
necessary)precedesfarmer-run,few-factorcomparisonsof selected
treatmentsor packages comparedwith traditionalhusbandry. But practical
empiricismmust rule; sometimes,the process may be shortenedand,
sometimes,iterationwill be necessary.

The involvementof the NARS and extensionservice throughoutthe

OFR/FSP process is agreed to be necessaryso that both the methodologyand
technologymay be transferredto the users. In pursuit of NARES
involvement,the leadingIARCs in the area, notablyCIMMYT and IRRI, have
long recognizedthe importanceof trainingand network development
(Figures3.5, 3.6). In the longer run, most OFR/FSP work must pass to the
NARSs but prolongedIARC involvementin trainingand networkscan be
foreseen. The network is a centralizedbut two-way informationsystem
which depends a good deal on personal contact (not the least of the
benefits of a sustainedtrainingprogram).
 - 86 -

As a result of historical accidents, the institutional

provisions for OFR/FSP work in the IARCs are quite various and likely to
remain so. No "ideal" structure is apparent but there is an over-riding
need to generate the essential FSP among researchers. Multi-disciplinary
team work is usually thought to be essential but the idea does not always
work well, and multi-disciplinary thinking by individuals could well be
more productive - if it were achievable. Economists are already well
established as elements in OFR teams and will remain so. I can discern no
clear need for social anthropologists.

Wider Questions

Export cash crops have been, historically, an important agent in

tropical agricultural development. They have been avoided by the CG
system but maintain fair to excellent R & D systems of their own and
remain a substantial element of many NARS programs. Many such crops are
perennials; they constitute an excellent economic base for small
landholders in the wet tropics and a focus for agricultural development.
The technology of managing them is well known and its transfer tends, of
its nature, to be of a conventional "top-down" nature; maybe there is
room here for more OFR/FSP in the small landholder context?

For annual food crops, the CG system has done a highly

creditable job, over the past decade or so, of introducing economic ideas,
of developing OFR/FSP into a working technique, of training, and of the
building of networks of collaborating NARSs. The groundwork having been
done, one can begin to perceive training and the sustaining of networks as
key activities in the future, as constituent NARSs grow.

The need for NFSD, especially in the lowland wet tropics, is

widely recognized to be important, yet the synthetic thinking that goes
beyond component research is hardly anywhere to be found (see above).

Agro-forestry ideas (Figure 4.2) have been abroad for several

years but an effective institutional structure (ICRAF) has only recently
emerged. Widespread recognition, that trees (more generally, perennial
plants) even exist in tropical small-farmer agriculture has been long
delayed, extraordinarily so in the light of their socio-economic
importance, especially in wetter areas. Interest, however, is now
beginning to stir. We may be certain, I believe, that trees will play a
central role in any serious NFSD in the wet tropics, though our scientific
understanding of their potential for food production and even simple
introduction-trial work are still grossly deficient. The neglect of
perennials is, I suggest, a lamentable deficiency in contemporary tropical
agricultural research.

With regard to the interests of the World Bank, I suggest that

it has a dual role. As a major donor to the CG system, I think that it
should encourage the CG institutes: to develop FSP ideas among
researchers, to use OFR methods where relevant, to enhance training and
network operation, to think hard about NFSD, and to take perennial plants
seriously. As a funding agency for agricultural development projects, I
think that the Bank should: retain an R & D element (with OFR/FSP) in
 - 87 -

such phased developmentprojects as are aimed at step-wise advancementbut

should also be prepared to be bold and try to constructand exploit one or
more types of NFSD in the wet tropics. The latter might lose money but
could provide, nevertheless,valuable guidance on how to "do" NFSD and
short-circuitwhat may otherwise provide to be an intolerablyprolonged
researchprocess. There is a real need for bold thinking coupled with
resources and who better to provide them than the World Bank?
 - 88 -

BIBLIOGRAPHY

AARD (1981). Five years of agricultural research and development for

Indonesia 1976-80.
Ministry of Agriculture, Jakarta, Indonesia.

Amare Getahun (1978). Agricultural systems in Ethiopia.

Agric. Systems, 3, 281-93.

Amare Getahun (1980). Agroclimates and agricultural systems in Ethiopia.

Agric. Systems, 5, 39-50.

Andrew, C.0., and Hildebrand, P.E. (1982). Planning and conducting

applied agricultural research.
Westview Press, Boulder, Co. USA.

Arnold, M.H. (ed.) (1976). Agricultural research for development. The

Namulonge experience.
University Press, Cambridge.

Arnon, I. (1981). Modernisation of agriculture in developing countries.

Wiley, Chichester, U.K.

Avila, M., Navarro, L.A., and Lagemann, J. (1982). Improving the small
farm production systems in Central America.
18th International Conference Agric. Economists, Jakarta, Indonesia.

Bene, J.G., Beall, H.W., and Cote, A. (1976). Trees, food and people
land management in the tropics.
IDRC, Ottawa, Canada.

Bernstein, R.H., Siwi, H.H. and Beachell, H.M. (1982). The development
and diffusion of rice varieties in Indonesia.
IRRI Res. Paper Ser., 71, pp. 25.

Bunting, A.H., and Watts-Padwick, G. (1983). Fifty years of experimental

agriculture. People politics and policies.
Expl Agr., 19, 1-13.

Burley, G., and Spears, J. (1983). Forestry research in developing

countries - time for a reappraisal?
World Bank, Washington and FAO, Rome.

Byerlee, D., and Collinson, M.P. (1980). Planning technologies

appropriate to farmers: concepts and procedures.
CIMMYT, El Batan, Mexico

Byerlee, D., Harrington, L., and Winkelmann, D.L. (1982). Farming systems
research: issues in research strategy and technology design.
Amer. J. Agric. Econ., 64, 897-904.
 - 89 -

CIAT (1982). Bean program annual report, 1982.

CIAT, Cali, Colombia.

CIMMYT (1978). CIMMYT training

CIMMYT, El Batan, Mexico (CIMMYT Today, 9).

CIMMYT (1981a). Assessing farmers' needs in designing agricultural

technology.
IADS, New York (IADS OccasionalPaper).

CIMMYT (1981b). CIMMYT economics.

CIMMYT, El Batan, Mexico (CIMMYT Today, 13).

CIMMYT (1983). Report of a seminar for senior agriculturalresearch

administratorsfor eastern and southernAfrica.
CIMMYT East African Economics Program, NetworkingWorkshops, Rep. 1.

Collinson,M.P. (1982). Farming systems researchin Eastern Africa: the

experienceof CIMMYT and some national agriculturalresearch
services, 1976-81.
Michigan State Univ., Intl Dev. Paper, 3.

Collinson,M.P. (1983). Farm management in peasant agriculture.

WestviewPress, Boulder Co., USA, 2nd ed.

Collinson,M.P. (1984). Farming systems research: Diagnosingthe

problem.
Agric. Symp., World Bank, Washington, 1984.

Combe, J., Jimenez,H., and Monge, C. (1981). Bibliographyon tropical

agroforestry.
CATIE, Turrialba,Costa Rica.

Coulter,J.K. (1979). Crop improvement. In R.C. Staples and R.J. Kuhr

(eds), Linking research to crop improvement.
Plenum Press, New York and London, 35-49.

Coulter, J.K. (1983).Personal communication.

Dalrymple,D.G. (1974). Developmentand spread of high yielding varieties

of rice and wheat in less developed countries.
For. Agric. Econ. Rpt. USDA, 95.

Dalton, G.E. (ed.) (1975). Study of agriculturalsystems.

Applied Science Publs., London.

Dent, J.B., and Anderson, J.R. (1971). Systems analysis in agricultural

management.
Wiley, Sydney, Australia.
 - 90 -

Dillon, J., and Hardaker, J.B. (1980). Farm management research for small
farmer development.
FAO, Agricultural Services Bull., 41.

Dillon, J.L., Plucknett, D.L., and Vallaeys, G.J. (1978). Farming systems
research at the International Agricultural Research Centres.
TAC of CGIAR, Rome.

Dillon, J.L., and Virmani, S.M. (1983). The farming systems approach.
In Symposium on Agro-Research for Australia's Semi-Arid Tropics,
Darwin, CSIRO, Canberra, Australia.

Duckham, A.N., and Masefield, G.B. (1970). Farming systems of the world.
Chatto and Windus, London, U.K.

Eicher, C.K., and Baker, D.C. (1982). Research on agricultural

development in Sub-Saharan Africa: a critical-survey.
Mich. State Univ. Intl Dev. Paper. 1.

Fitzhugh, H.A., Hart, R.D., Moreno, R.A., Osuji, P.O., Ruiz, M.E., and
Singh, L. (eds) (1982). Research on crop-animal systems.
CATIE, Turrialba, Costa Rica.

Galt, D., Diaz, A., Contreras, M., Peairs, F., Posner, J.,and Rosales, F.
(1982). Farming systems resaarch (FSR) in Honduras 1977-81: a case
study.
Mich. State Univ. Intl Dev. Paper, 1.

Gibbon, D. (1980). Rainfed farming systems in the Mediterranean region.

Univ. E. Anglia Sch. Dev. Studies, Disc. Paper 70.

Gibbon, D. (1981). A systems approach to resource management and

development in the ECWA region: the ICARDA experience 1977-80.
Univ. E. Anglia Sch. Dev. Studies, Disc. Paper 104.

Gilbert, E.H., Norman, D.W., and Winch, F.E. (1980). Farming systems
research: a critical appraisal. Mich. State Univ. Rural Dev.
Paper, 6.

Greenland, D.J. and Lal, R. (eds) (1977). Soil conservation and

management in the humid tropics.
Wiley, Chichester, England.

Grigg, D.B. (1974). The agricultural systems of the world: an

evolutionary approach.
Cambridge University Press, London.

Gryseels, G. (1983). Livestock in farming systems research for

smallholder agriculture: experiences of ILCA's highlands program.
Seminar on Agricultural Research in Rwanda, Kigali, Rwanda.

Gryseels, G. and Anderson, F. (1983). Farming systems research in ILCA's

Highlands Program.
Univ. Fla, Fmg Systs Support Project Newsletter.
 - 91 -

Harrington, L. (1982). Exercises in the economic analysis of agronomic

data.
CIMMYT, El Batan, Mexico (Econ. Wkg Paper, 82/3)

Hart, R.D. (1975). A bean, corn and manioc polyculture cropping system.
Turrialba, 25, 294-301, 377-84.

Harwood, R.R. (1979). Small farm development. Understanding and

improving farming systems in the tropics.
Westview Press, Boulder, Co., USA.

ICAR (1982). A decade of dryland agricultural research in India 1971-80.

AICRPDA, Hyderabad, India.

ICARDA (1982). Farming systems program research report, 1982.

ICARDA, Aleppo, Syria.

ICRAF (1983). Draft guidelines for agroforestry diagnosis and design.

ICRAF, Nairobi, Kenya.

ICRISAT (1979). Proceedings of the International Workshop on

intercropping.
ICRISAT, Hyderabad, India.

ICRISAT (1983). ICRISAT farming systems research, a special report.

ICRISAT, Hyderabad, India.

ILCA (1980). ILCA, the first years.

ILCA, Addis Ababa, Ethiopia.

ILCA (1983). Pastoral systems research.

ILCA, Addis Ababa, Ethiopia.

Tnu G. Ismail (1983). Country report. Research and development of

rainfed agriculture in Indonesia.
Seminar/Workshop in Rainfed Agriculture Development, Iloilo, the
Philippines.

IRRI (1982). The role of anthropologists and other social scientists in

interdisciplinary teams developing improved food production
technology.
IRRI, Los Bainos,Philippines.

Jain, H.K., and Banerjee, S.K. (1982). Problems of seed production and
varietal release in India.
Seed Res., 10, 1-17.

Jameson, J.D. (1970). Agriculture in Uganda. Oxford University Press,

London 2nd ed. (following Tothill, J.D., 1940).

Johnston, B.F. (1958). The staple food economies of western tropical

Africa.
University Press, Stanford, California.
 - 92 -

Kang, B.T., Wilson, G.F., and Sipkens, L. (1981). Alley cropping maize
and leucaena in southernNigeria.
Plant and Soil, 63, 165-79.

Kuile, C.H.H. ter (1983). Technologicalpotentialsfor food productionin

the humid and sub-humid tropics of Africa.
Conferenceon AcceleratingAgriculturalGrowth in Sub-SaharanAfrica,
Victoria Falls, Zimbabwe.

Lal, R., and Greenland,D.J. (1979). Soil physicalproperties and crop

productionin the tropics.
Wiley, Chichester,England.

Levi, J., and Havinden,M. (1982). Economics of African agriculture.

Longman,London.

MacDonald, L.M. (ed.) (1982). Agroforestryin the African humid tropics.

Un. Nat. University,Tokyo, Japan (Proc. Workshop, IITA, Ibadan,
Nigeria, 1981).

McIntosh, S.L., and SuryatnaEffendi (1981). Indonesianfarming systems

research and development.
Agency for Agr. Res. and Dev.( AARD), Bogor, Indonesia.

Martin, A. (1981). Concepts and approaches to farm systems research in

Syria 1977-80.
Univ. E. Anglia Sch. Dev. Studies,Disc. Paper.

Martinez, J.C., and Arauz, J.R. (1983). Institutionalinnovationsin

national agriculturalresearch: on-farm researchwithin IDIAP,
Panama.
CIMMYT, El Batan, Mexico.

Martfnez, J.C., and Safn, G. (1983). The economic returns to

institutionalinnovationsin national agriculturalresearch research:
on-farm research in IDIAP, Panama.
CIMMYT, El Batan, Mexico (Econ. Wkg Paper, 04/83).

Matlon, P.J. (1983). Breakdownpoints in the data gatheringprocess.

Fmg Systs Newsletter,CIMMYT, Nairobi, 12, 8 - 18.

Mellor, J.W. (1966). The economics of agriculturaldevelopment.

Cornell UniversityPress, Ithaca, N.Y.

Menz, K.M. (1980). Unit farms and farming systems research: the IITA
experience.
Agric. Systems, 6, 45-51.

Moscardi, E., et al. (1983). Creating an on-farm research program in

Ecuador.
CIMMYT, El Batan, Mexico (CIMMYT Econ. Wkg Paper 01/83).

Nestel, B. (ed.) (1983). Agriculturalresearch for development:

potentialsand challengesin Asia.
ISNAR, The Hague, The Netherlands.
 - 93 -

Ngambeki, D.S., and Wilson, G.F. (eds) (1983). On-farm experimentationin

farming systems research.
Report of a Workshop, IITA, Ibadan, Nigeria.

Norman, D.W. (1974). Rationalisingmixed croppingunder indigenous

conditions: the example of Northern Nigeria.
J. Dev. Studies, 11, 3-21.

Norman, D.W. (1978). Farming systems research to improve the livelihood

of small farmers.
Amer. J. Agric. Econ., 60, 813-18.

Norman, D.W. (1980). The farming systems approach: relevancy for the
small farmer.
Mich. State Univ. Rural Dev. Paper, 5.

Norman, D.W. (1983). Some problems in the implementationof agricultural

research projectswith a farming systems perspective.
CIMMYT Symposium on the Introductionof on-farm researchwith a
farming systems perspective as a tool in agriculturalresearch,
Nairobi, April 1983.

Norman, D.W., Newman, M.D., and Ouedraogo, I. (1981). Farm and village
productionsystems in the semi-arid tropics of West Africa.
ICRISAT,Hyderabad, India (Res. Bull. 4).

Norman, D.W., Simmons, E.B., and Hays, H.M. (1982). Farming systems in
the Nigerian Savanna. Research'andstrategiesfor development.
Westview Press, Boulder, Co., USA.

Norman, M.J.T. (1980). Annual cropping systems in the tropics.

Universityof Florida, Gainesville,USA.

Payne, P.R. (1978). Human protein requirements.

In Norton, G.(ed). Plant proteins, Butterworth,London, 247-63.

PCARRD (1982). State of the art. Farming systems research.

PhilippineCouncil for Agricultureand ResourcesResearch and
Development,Los Ba-nos,PhilippineIslands (Farm resources and
systems series, 2).

Perrin, R.K., Winkelmann,D.L., Moscardi, E.R., and Anderson, J.R.

(1976). From agronomic data to farmer recommendations: an economics
trainingmanual.
CIMMYT, El Batan, Mexico (Inf. Bull. 27).

Quisumbing,E.C. (1982). Linking farming systems research and extension

to the farmer: the Philippineexperience.
Kansas State Univ., Manhattan,Kansas (Farmingsystems research
symposium).
 - 94 -

Raintree, J.B. (1983). A diagnostic approach to agroforestry design.

ICRAF, Nairobi, Kenya (Intl Symp. on Strategies and Designs for
Afforestation, Reforestation and Tree Planting, Wageningen, The
Netherlands).

Ruthenberg, H. (1980). Farming systems in the tropics.

University Press, Oxford, 3rd ed.

Ruttan, V.W. (1977). The green revolution: seven generalisations.

Intl Dev. Rev., 4, 15-23.

Salas, G. de las (ed.) (1979). Workshop. Agroforestry systems in Latin

America.
CATIE, Turrialba, Costa Rica.

Sanders, J.H. and Lynam, J.K. (1982). Evaluation of new technology on

farms: methodology and some results from two crop programs at CIAT.
Agric. Syst., 9, 97-112.

Sanghi, N.K. (1982). Operational research in dryland agriculture for

semi-arid red soils of hyderabad.
AICRPDA, Hyderabad, Project Bull. 3.

Shaner, W.W., Philipp, P.F. and Schmehl, W.R. (eds) (1981). Readings in
farming systems research and development.
USAID, New York, USA.

Shaner, W.W., Philipp, P.F. and Schmehl, W.R. (1982). Farming systems
research and development guidelines for developing countries.
Westview Press, Boulder, Co., USA.

Sheppard, W.J. (ed.) (1982). Proceedings of Kansas State University's

1981 farming systems research symposium -small farms in a changing
world: prospects for the eighties.
Kansas State Univ., Manhattan, Kansas, USA (Farming Systems Research
Paper 2).

Simmonds, N.W. (1979). Principles of crop improvement.

Longman, London.

Simmonds, N.W. (1981). Genotype (G), Environment (E) and GE components of

crop yields.
Exp. Agric., 17, 355-62.

Spedding, C.R.W. (1979). An introduction to agricultural systems.

Applied Science Publs, London.

Sumberg, J.E., and Okali, C. (1983). Linking crop and animal production:
a pilot development program for smallholders in southwest
Nigeria.
Rural Development in Nigeria (in the press).
 - 95 -

Thierstein,G.E. (1983). The animal-drawnwheeled tool carrier.

ICRISAT, Hyderabad,India.

Thorner, D., Kerblay, B., and Smith, R.E.F. (eds) (1966).

A.V. Chayanovon the theory of peasant economy.
Richard Irwin, Homewood, Ill., U.S.A.

Tothill, J.D. (ed.) (1948). Agriculture in the Sudan.

Oxford UniversityPress, London.

Tripp, R. (1982). Data collection,site selection and farmer

participationin on-farm experimentation.
CIMMYT, El Batan, Mexico (CIMMYT Wkg Paper 82/1).

Valdez, A., Scobie, G.M., and Dillon, J.L. (eds) (1979). Economicsand
the design of small-farmertechnology.
Iowa State Univ. Press, Ames, Iowa, USA.

Watson, G.A. (1980). A study of tree crop farming systems in the lowland
humid tropics.
World Bank, WashingtonDC, USA (Agr. Tech. Note 2).

Watson, G.A. (1983). Developmentof mixed tree and food crop systems in
the humid tropics: a response to populationpressure and
deforestation.
Exp Agric., 19, 311-32.

Webster, C.C., and Wilson, P.N. (1980). Agriculture in the tropics.

Longman,London, 2nd ed.

Willey, R.W. (1979). Intercropping- its importanceand research needs.

Field Crop Abstr., 32, 1-10, 73-85.

Woolley, J., and Pachico, D. (1983). A pilot project in trainingand

network establishmentfor on-farm research on bean cropping systems
in Latin America.
CIAT, Cali, Colombia, in preparation.

Wortman, S., and Cummings, R.W. (1978). To feed this world - the
challenge and the strategy.
Johns Hopkins, Baltimore and London.

Zandstra,H.G. (1979). Cropping systems researchfor the Asian rice

farmer.
Agric. Syst., 4, 135-53.

Zandstra, H.G., Price, E.C., Litsinger,J.A., and Morris, R.A. (1981).

A methodologyfor on-farm cropping systems research.
IRRI, Los Ba-nos,Philippines.
 - 96 -

ADDENDUM
TO THE BIBLIOGRAPHY

A number of older publicationsto which I should have referred

in writing this report have come to my attention only since the report was
submitted,and a very relevant book has been published in the same
interval. After each referenceI refer briefly to the contents in the
context of my report.

Zandstra,H., Swanberg, K., Zulberti,C. and Nestel, B. (1979).

Cagueza: Living Rural Development.
IDRC, Ottawa, Canada.

The Caqueza book bears on most of the main theme of my report.

The project (in Colombia, 1971-75)descended historicallyfrom the Puebla
project in Mexico. The socio-economicfeatures of small farmers are well
explained,with an enlighteningdiscussionof risk and uncertainty,even a
brave (if not very successful)attempt to quantify the risk factor.
OFR/FSP, though not under that title, was an integralpart of the
operation,which must have been one of the earlier rural development
projects deliberatelyto adopt a "bottom-up"approach, in recognitionof
the general failure of the obverse. There is an effort to quantify the
social cost-benefitaspects of the project which, like the work of
Martfnez and Safn, leaves it fairly clear that there were positive net
benefits of uncertain magnitude. The book contains a good deal of rather
dull administrativedetail which, however, does give a feel for the sorts
of practical reasons for which RDPs work or do not work, as the case may
be. The Caqueza project was surely one of the pioneeringworks in the
area, and I regret that I did not assimilateit into my report.

Okigbo, B.N., and Greenland, D.J. (1977).

Intercroppingsystems in tropicalAfrica.
In MultipleCropping, ed. R.I. Papendicket al.,
American Society of Agronomy, 63-101.

Okigbo and Greenland give a valuable compendiumof intercropping

systems in tropicalAfrica, with special reference to annual crops in
shifting/bushfallow systems. Some 50-80 percent of all crops are grown
mixed.

Greenland, D.J. (1977). Increasingfood productionfrom the lowland humid

tropics of Africa and Latin America.
Ann. New York Acad. Sci., 300, 112-120.

Greenland discusses the problems of maintenanceof soil

fertility in the humid tropics under production regimes more intensive
than those of traditionalsystems. Though he is clearly inclinedto think
of intensificationof what is already there, his arguments have profound
bearing upon thinking about NFSD. He recognizesthe importanceof trees,
less for their products and soil surface protectionthan for their
capacity to extract basic actions from the subsoil.
 - 97 -

Huxley P.A. (ed.) (1983). Plant research and agro-forestry.

ICRAF, Nairobi, Kenya.

Huxley's book is a multi-authorone with some useful reviews of

perennial-basedsystems in, for example, Indonesia,West Africa, Costa
Rica, and Brazil. There is much here that is highly relevant to my
advocacy of trees as the bases of many farming systems,especiallywhen
any kind of NFSD in the wet tropics is contemplated. Once again,
perennialsmust be at the heart of any imaginativethinkingor
experimentationupon NFSD. This is but one among several recent
publicationswhich, collectively,show that the traditionalboundaries
between agricultureand forestry are beginning to dissolve: the paradigms
are shifting, to adopt the fashionablephrase. The sooner we come to
think of perennialsand annuals as complementaryand interlockingelements
in tropical land use systems the better. It is surely time that tropical
agriculturalresearch at large got out of the annual crop rut in which
much (but not all) of it resides and really looked at what tropical
farmers already do with perennial plants. This book presents many useful
ideas on where and how to look.
Worid Banki The Agricultural Development
Experience of Algeria,
their productivity. Valuable to policy-
makers, project designers, rural sociol-
PUb lCAtkaS Morocco, and Tunisia: A ogists, extension workers, and other
w
*
Comparison of Strategies for agricultural researchers.
of PAftfwd Growis o1984. 176 pages.
Intetest Kevin M. Cleaver ISBN0-8213-0301-5.StockNo. BK 0301.
Compares agricultural experience of $1350.
Algeria, Morocco, and Tunisia. Pro-
vides insights into the importance of
food and agriculture for development,
and determinants of agricultural
Adoption of Agricultural growth. Agricultural Extension: The
Innovations in Developing Staff Working PaperNo. 552. 1983. 55 Training and Visit System
Countries: A Survey pages. Daniel Benor, James Q. Harrison,
Gershon Feder, Richard Just,and ISBN 0-8213-0120-9.Stock No. WP 0552. and Michael Baxter
David Silberman $3- Contains guidelines for reform of agri-
Staff Working PaperNo. 542. 1982. 65 cultural extension services along the
pages. lines of the training and visit system.
ISBN 0-8213-0103-9.Stock No. WP 0542. The central objective-making the
$3. The Agricultural Economy of most efficient use of resources avail-
Northeast Brazil able to governments and farmers-is
Gary P. Kutcher and Pasquale L. achieved through encouraging and fa-
Agrarian Reform as Unfinished Scandizzo cilitating feedback from farmers to re-
Business-the Selected Papers This study, based on an agricultural search workers through extension per-
of Wolf Ladejinsky survey of 8,000 farms, assesses the ex sonnel who visit and advise farmers
Louis J. Walinsky, editor tent and root causes of pervasive rural on a regular, fixed schedule, thus
Studies in agrarian policy and land re- poverty in northeast Brazil. The au- helping research to solve actual pro-
form spanning four decades, grouped thors review a number of policy and duction constraints faced by the
chronologically according to Ladejin- project options; they conclude that farmer.
sky's years in Washington, Tokyo, and courageous land reform is the only ef- Explains the complex relationships in
Vietnam and while at the Ford Foun- fective means of dealing with the training and visit extension and draws
dation and the World Bank.Oxford problem. attention to the range of considera-
University Press, 1977. 614 pages (in- The JohnsHopkins UniversityPress, 1982. tions that are important to implement-
cluding appendixes, index). 288 pages. ing the system.
LC 77-24254.ISBN 0-19-920095-5,Stock LC 8147615. ISBN 0-8018-2581-4,Stock 1984. 95 pages.
No. OX 920095, $32.50 hardcover;ISBN No. JH 2581. $25.00 hardcover. ISBN 0-8213-0140-3.Stock No. BK 0140.
0-19-920098-X,Stock No. OX 920098, $5.
$14.95 paperback.

Agrarian Reforms in Agricultural Land Settlement

Developing Rural Economies Theodore J. Goering, coordinating
Characterized by Interlinked author
Credit and Tenancy Markets Agricultural Extension by Examines selected issues related to the
Avishay Braverman and T.N. Training and Visit: The Asian World Bank's lending for land settle-
Srinivasan Experience ment and gives estimates of the global
SNEditedby Michael M. Cerea, rate of settlement andthe world's ulti-
pages (includingreferences). John K. Coulter, and John F.A. mate potentially arable land.
Russell A World Bank Issues Paper. 1978. 73
Stock No. WP-0433. $3. Captures nearly ten years of experi- pages(including4 annexes). English,
ence with the Training and Visit Ex- Frnch, and Spanish.
Agricultural Credit tension System. Addresses five issues: Stock Nos. BK 9054 (English),BK 9055
Outlines agricultural credit practices farmer parficipation, the research-ex- (French),BK 9056 (Spanish).$5 paper-
and problems, programs, and policies tension linkage, training, system man- back.
in developing countries and discusses agement, and monitoring and evalua-
their implications for World Bank op- tion. Within this framework, extension
erations. system managers and evaluators from Agricultural Price Management
A World Bank Paper.1975. 85 pages(in- six Asian countries and six discussants in Egypt
cluding 14 annex tables). Notes the World Bank's sarongcom- William Cuddihy
Stock No. BK 9039 (English),BK 9052 mitment to agriculturaldevelopment Staff Working PaperNo. 388. 1980. 174
(French),BK 9053 (Spanish).$5 paper- in its member countries and to helping pages(includingannex, bibliography).
back. least advantaged farmers to improve Stock No. WP-0388.$5.
Agricultural Price Policies and Stock Nos. BK 9074 (English),BK0160
the Developing Countries (French),BK 0161 (Spanish).$5 paper-
George Tolley, Vinod Thomas, and back.
Chung Ming Wong Agroindustrial Project Analysis
Appuisi.lg
PouIty Entepris5forPmfibbility
This book first considers price policies James E. Austin A Mnfl toPouatiol tn-roto,,

in Korea, Bangladesh, Thailand, and Provides and illustrates a framework

Venezuela, bringing out the conse- for analyzing and designing agro-in- I
quences for govemment cost and reve- dustrial projects. ' \
nue, farm income, and producer and EDI
consumer welfare. Other effects, in- EDI Series in Economic Development. / < \ ,J

cluding those on agricultural diversifi- The Johns Hopkins University Press,

cation, inflation, economic growth, 1981.2nd printing, 1983. 224 pages (in-
and the balance of payments are also cluding appendixes, bibliography, and N

discussed. The second part of the book index). , x x\

provides a methodology for estimating LC 80-550. ISBN 0-8018-2412-5,Stock II
these effects in any country. Opera- No. JH 2412, $16.50 hardcover;ISBN 0-
tional tools for measuring the effects 8018-2413-3,Stock No. JH 2413, $7.50 1 1
on producers, consumers, and govern- paperback.
ment are developed and applied. French:L'Analysedes projetsagroindus- L i
The Johns Hopkins UniversityPress. 1982. triels. Economica,1982. ISBN 1-7178-
256pages. 0480-3,StockNo. IB 0537.$7.50paper- Appraising Poultry Enterprises
LC 81-15585.ISBN 0-8018-2704-3,Stock back. for Profitability: A Manual for
No. JH 2704, $25 hardcover;NEW: ISBN Spanish:Analisis de proyectosagroindus- Potential Investors
0-8018-3124-5,Stock No. JH 3124, $9.95 triales. EditorialTecnos,1981. ISBN 84- International Finance Corp.
paperback. 309-0882-X, Stock No. IB 0520, $7.50 pa- Decisionmaking tool for entrepreneurs
perback. and project managers considering in-
vestments in integrated poultry pro-
jects. Use this guide to conduct on-site
Agricultural Prices in China investigation of proposed project. Fig-
Nicholas R. Lardy Alternative Agricultural Pricing ure production costs and determine
Analyzes recent adjustments to Policies in the Republic of fixed asset and working capital for
China's agricultural pricing systems Korea: Their Implications for broiler operations. Analyze market and
and its effects on urban consumers Government Deficits, Income accurately forecast market prices. This
and overall production patterns. De- Ditribut d B I comprehensive guide tells how to
fines price ratios from key inputs and DStriution, an alance of manage integrated broiler operations,
outputs and examines price/cost rela- ayments gives specifications for broiler and
tions in view of the institutional set- Avishay Braverman, Choong Yong breeder houses and summarizes pro-
ting for price policy. Ahn, Jeffrey S. Hammer duction costs.
Staff Working PaperNo. 606,1983.84 Develops a two-sector multimarket TechnicalPaperNo. 10. 1983. 110 pages.
pages. model to evaluate agricultural pricing ISBN 0-8213-0165-9.StockNo. BK 0165.
ISBN 0-8213-0216-7.Stock No. WP 0606. policies, replacing insufficient standard $5.
$3. operational methods. Measures the
Agricultural Research impact of alternative pricing policies The Book of CHAC:
Agricultural Research on production and consumptionof rice Programming Studies for
Points out that developing countries and barley, real income distribution, Mexican Agricultural Policy
must invest more in agricultural re- import levels of rice, self-sufficiency in Edited by Roger D. Norton
search if they are to meet the needs of rice, and public budget. Provides a
their growing populations. Notes that valuable synthesis of the work that has andLeopoldo Solis M.
studies in Brazil, India,Japan, Mexico, been done to date on agricultural The principal tool of analysis is the
and the United States show that agri- household models. Helps economists sector model CHAC, named after the
cultural research yields a rate of return evaluate the impact of alternative pric- Mayan rain god. This model can be
that is more than two to three times ing policies aimed at reducing deficits. used throughout the sector to cover
greater than retums from most alter- Based on the experience of the Grain short-cycle crops, their inputs, and
native investments and cites some of Management Fund and the Fertilizer their markets. It can also be broken
the successes of the high-yielding vari- Fund in Korea. down into submodels for particular lo-
eties of rice and wheat that were de- Staff WorkingPaperNo. 621. 1983. 174 calities if more detailed analysis is re-
veloped in the mid-1960s. Discusses pages. quired. The model helps planners
the World Bank's plans to expand its ISBN 0-8213-0275-2Stock No. WP 0621. weigh the costs among policy goals,
lending for agricultural research and S5. which can vary from region to region.
extension, particularly for the produc- This volume reports the experience of
tion of food and other commodities Argentina: Country Case Study using the CHAC model and also pre-
that are of importance to low-income of Agricultural Prices, Taxes, sents purely methodological material.
consumers, small farmers, and re- and Subsidies The JohnsHopkins University Press,1983.
source poor areas. Lucio G. Reca 624 pages (includingmaps, bibliographies,
Sector PolicyPaper.1981. 110 pages(in- Staff Working PaperNo.386. 1980. 72 index).
cluding annexes).English,French,and pages (including3 annexes). LC 80-29366.ISBN 0-8018-2585-7,Stock
Spanish. Stock No. WP-0386.$3. No. IH 2585. $35 hardcover.
Building National Capacity to The Design of Rural Economic Aspects and Policy
Develop Water Users' Development: Lessons from . Issues in Groundwater
Associations: Experience from Africa Development
the Philippines Uma Lele Ian Carruthers and Roy Stoner
Frances F. Korten Analyzes new ways of designing rural Staff WorkingPaperNo. 496. 1981. 110
Staff Working PaperNo. 528. 1982. v + development projects to reach large pages (includingannex, bibliography).
69 pages(includingreferences). numbers of low-income subsistence Stock No. WP-0496.$5.
ISBN 0-8213-0051-2.Stock No. WP 0528. populations. The third paperback Economic Retum to Investment
$3. contains a new chapter by the
~~~~~~~~printing EcnmcRtrtoIvs et
author updating her findings. in Irrigation in India
Bureaucratic Politics and TheJohnsHopkinsUniversityPress,1975; Leslie A. Abbie, James Q.
Incentives in the Management 3rd printing,1979.260 pages(including Harrison, and John W. Wall
of Rural Development in-
glossary,appendix, maps,bibliography, Staff Working PaperNo. 536. 1982. 52
Richard Heaver dex). pages.
ISBN 0-8018-1769-2,Stock No. JH 1769, ISBN 0-8213-0083-0.Stock No. WP 0536.
Analyzes management problems in $9.95 paperback. $3
implementing rural development from French:Le developpementrural:1'expe F
a bureaucratic political standpoint. ri- Farm Budgets: From Farm
Emphasizes the need to take account enceAfricatne.Economica,1977. ISBN 2- Income Analysis to
of informalinterests in managing pro- StockNo. IB 0545,$9.95grcultural
7178-0006-9, t lyi
grams. Suggests possible methods for paperback. Maxwell L. Brown
assessing incentives. Economic Analysis of Clarifies the relation between simple
Staff Working PaperNo. 537. 1983. 74 Agricultural Projects farm income analysis and the broader
pages. Second edition, completely revised field of agricultural project analysis
ISBN 0-8213-0084-9.Stock No. WP 0537. and expanded and emphasizes the more practical as-
$3. J. Price Gittinger pects of project preparation. Gives
Sets out a careful and practical meth- guidance to those responsible for plan-
odology for analyzing agricultural de- ning in agriculture.
velopment projects and for using these EDI Series in Economic Development.
The Common Agricultural analyses to compare proposed invest- The Johns Hopkins University Press,
Policy of the European ments. It covers what constitutes a 1980. 154 pages.
Community: A Blessing or a "project," what must be considered to LC 79-3704.ISBN 0-8018-2386-2,Stock
Curse for Developing identify possible agncultural projects, No. JH 2386, $15 hardcover;ISBN 0-
Curse for Developing the life cycle of a project, the strengths 8018-2387-0,Stock No. JH 2387, $6.50
Countries? and pitfalls of project analysis, and the paperback.
Ulrich Koester and Malcolm D. calculations required to obtain financial Spanish:Presupuestosde fincas. Editorial
Bale and economic project accounts. Tecnos,1982. ISBN 84-309-0886-2,Stock
Examines the importance of the Euro- The methodology reflects the best of No. IB 0522, $6.50 paperback.
pean Community (EC) in global agri- contemporary practice in government
cultural trade. Points out that the EC agencies and international develop- Fishery
is the leading importer of agricultural ment institutions concerned with in- Highlights the importance of fisheries
goods and is the dominant exporter of vesting in agriculture and is accessible to the economies of developing coun-
a number of agricultural products. Em- to a broad readership of agricultural tries and recommends that the World
phasizes that policymakers in develop- planners, engineers, and analysts. Bank provide assistance to those coun-
ing countries must understand the im- This revision adds a wealth of recent tries that have the fishery resources
plications of the EC's common project data; expanded treatment of and are willing to develop them fur-
Agricultural Policy. Spells out how this farm budgets and the efficiency prices ther.
policy operates and categorizes impor- to be used to calculate the effects of an SectorPolicy Paper.1982.
tant commodities. investment on national income; a glos- ISBN 0-8213-0138-1.StockNo. BK 0138,
Staff Working PaperNo. 630. 1984. 64 sary of technical terms; expanded ap- $5 paperback.
pages. pendixes on prepanng an agricultural
Stock No. WP 0630. $3. project report and using discounting Forestry
tables; and an expanded, completely Graham Donaldson, coordinating
The Design of Organizations annotated bibliography. author
for Rural Development EDI Series in EconomicDevelopment. Examines the significance of forests in
Projects: A Progress Report The JohnsHopkins University press. July economic development and concludes
William E. Smith, Francis J. 1982. 2nd printing, March 1984. 528 that the World Bank should greatly in-
Lethem, and Ben A. Thoolen pages (includingappendixesand glossaryl crease its role in forestry development,
Staff Working PaperNo. 375. 1980. 48 index). both as a lender and adviser to gov-
pages.English and French. LC 82-15262.ISBN 0-8018-2912-7,Stock ernments.
035 (Eglish, BK9241
StockNos.WP No. JH 2912, $37.50 hardcover;ISBN 0- SectorPolicyPaper.1978. 63 pages(in-
Stock Nos. Wi' 0375 (English),BK 9241 8018-2913-5,Stock No. JH 2913, $13.50 cluding 7 annexes). English,French,and
(French).$3. paperback. Spanish.
Spanish:Analisis economicode proyectos Stock Nos. BK 9063 (English),BK 9064
Prices subject to changewithout notice agricolas.EditorialTecnos,S.A. ISBN 84- (French),BKL 9065 (Spanish).$5 paper-
and may vary by country. 309-0991-5.$13.50. back.
 major policy options open to the pages(including3 annexes,appendix,
World Bank in this field. map).
Forestry Terms.-Terminologie A World Bank Paper.1975. 73 pages(in- Stock No. WP-0332.$5.
forestiere cluding 2 annexes). Monitoring and Evaluation of
English-French; Francais- Stock No. BK 9042. $5 paperback. Agriculture and Rural
Anglais. Land Tenure Systems and Development Projects
Presents terminology related to for- Social Implications of Forestry Dennis J. Casley and Denis A.
estry development and erosion control Development Programs Lury
inwaodoandp semsiaridlands. Since fuel- Michael M. Cernea This book provides a how-to tool for
wood problems and desertification McalMCenathe design and implementation of
have become serious, particularly in Staff WorkingPaperNo. 452. 1981. 35 monitoring and evaluation systems in
Western Africa, the World Bank has pages(includingreferences,bibliography). rural development projects. Because
become increasingly involved in wood- Stock No. WP-0452.$3. rural development projects are com-
based energy and erosion-control and plex, they seek to benefit large num-
in forest-management projects. Assists bers of people in remote rural areas,
translators and researchers who work and they involve a variety of invest-
in this field. ments. The need for monitoring and
A World Bank Glossary-Glossaire de M .. mg El.ph.., DprHen evaluating them during implementa-
la Banque mondiale i ','r ''"v tion has been accepted in principle,
1984. 48 pages. . but effective systems have not hereto-
ISBN 0-8213-0175-6.Stock No. BK 0175. fore been formulated. The concepts of
_5. monitoring and evaluation are differ-
entiated and issues that need to be
Improving Irrigated considered in designing systems to
monitor and evaluate specific projects
Agriculture: Institutional are outlined, emphasizing the timeli-
Reform and the Small Farmer ness of the monitoring functions for
Daniel W. Bromley effective management. Elaborates on
Staff WorkingPaper No. 531. 1982. 96 such technical issues as selection of.in-
pages. pages. . dicators,
data selection
analysis, of
andsurvey
~~~~~~~~~~~~~~~~~~~~ogy methodol-
presentation. It
ISBN 0-8213-0064-4.Stock No. WP 0531. Managing Elephant is directed primarily to those working
$3. Depredation in Agricultural with specific projects and will be use-
India: Demand and Supply and Forestry Projects ful to project appraisal teams, to de-
Prospects for Agriculture John Seidensticker signers of monitoring and evaluation
James Q. Harrison, Jon A. Outlines procedures for managing ele- systems, and to project staff who work
Hitchings, and John W. Wall phants in and around project areas as The tohns system s.
part of the project design. Helps proj- TeJhsHopkins University Press.1982.
Staff Working PaperNo. 500. 1981. 133 ect designers plan activities that will 145 pages.
pages (including5 appendixes,references, protect wildlife and prevent financial LC 82-7126.ISBN 0-8018-2910-0,Stock
annex). loss from damage by animals. Illus- No. IH 2910. $8.50 paperback.
Stock No. WP-0500. $5. trates methods used to investigate ele- Monitoring Rural Development
phant behavior and ecology. Notes
in ~~~that
careful scheduling of project activ- in East Asia
Irrigation Management in ities is required to ensure that ele- Guido Deboeck and Ronald Ng
China: A Review of the phants are not isolated in production Staff Working PaperNo. 439. 1980. 91
Literature areas. pages (including annexes).
James E. Nickum Technical Paper No. 16. 1984.50 pages. Stock No. WP-0439. $3.
Analyzes irrigation management in the ISBN 0-8213-0297-3.
People's Republic of China. Major top- Stock No. BK 0297. $3.
ics covered are the institutional envi-
ronment, the organizational structure, Managing Information for Monitoring Systems and
water fees and funding, and water al- Rural Development: Lessons Irrigation Management: An
location. The report is based on
Chinese-language materials published from Eastern Africa Experience from the
in China and now available in the Guido Deboeck and Bill Kinsey Philippines
United States. Staff Working Paper No. 379. 1980. vii + Agricultural economists, planners, and
Staff Working Paper No. 545. 1983. 106 70 pages (including 5 annexes, index). field workers will find this 1983 case
pages. Stock No. WP-0379.$3. study report a practical guide for de-
ISBN 0-8213-0110-1. Stock No. WP 0545. signing efficient monitoring and evalu-
$5. Measuring Project Impact: ation systems for irriation and similar
Monitoring and Evaluation in projects. It illustrates the practical ap-
Land Reform the PIDER Rural Development plication of the principles covered in
Examines the characteristics of land re- Project-Mexico Ealuation of Agricultureand Rural De-
form, its implications for the econom- Michael M. Cernea velopmentProjects.Highlights the prob-
ies of developing counties, and the Staff WorkingPaperNo. 332. 1979. 137 lems as well as the successes.
1983. 162 pages. World Bank Staff Working Paper No. 387. permitting a full social cost-benefit
ISBN 0-8213-0059-8.Stock No. BK 1980. 108 pages. analysis of the project.
0059.$5. Stock No. WP-0387.$5. The JohnsHopkins UniversityPress. 1982.
336 pages(includingmapsand index).
Project Evaluation in Regional LC 81-48173.ISBN 0-8018-2802-3,Stock
Perspective: A Study of an No. JH 2802, $30 hardcover.
Opportunities for Biological IrrigationProject in Northwest
Control of AgriculturalPests in Malaysia Rethinkig Artisanal Fisheries
Developing Countries Clive Bell, Peter Hazell, and Roger Development: Western
D. J. Greathead and J. K. Waage Slade Concepts, Asian Experiences
Describes how to use living organisms This innovative study develops quanti- Staff Working Paper No. 423. 1980.107
as pest control agents, either alone or tative methods for measuring the di- pages (including references).
as one component of pest manage- rect and indirect effects of agricultural Stock No. WP-0423. $5.
ment. Biological control offers hope of projects on their surrounding regional
long-term-permanent-results, causes and national economies. These meth- Rural Development
no pollution, poses no risk to human ods are then applied to a study of the Discusses strategy designed to extend
health and is often cheaper than Muda irrigation project in northwest the benefits of development to the ru-
chemical controls. Gives methods and Malaysia. A linear programming ral poor and outlines the World Bank's
costs. Specifies controls for specific model is used to analyze how a project plans for increasing its assistance in
crops found in developing countries. changes the farm economy, and a so- this sector.
TechnicalPaperNo. 11. 1983. 55 pages. cial
economy is thenmatrix
accounting of theThis
estimated. regio-al
pro-
SectorPolicy
cluding Paper,1975, 89 pages(in-
14 annexes).
ISBN 0-8213-0164-0. Stock No. BK 0164. vides the basis for a semi-input-output
$3. model, which is used to estimate the Stock No. BK 9036. $5 paperback.
indirect effects of the project on its re-
Prices, Taxes, and Subsidies in gion. Thereafter, a similar methodol-
Pakistan Agriculture,1960-1976 ogy is used to estimate the project's
Carl Gotsch and Gilbert Brown effects on key national variables, thus
RuralFinancial Markets in
Developing Countries
J. D. Von Pischke, DaleW.
Adams, and Gordon Donald
Rural Development
DwightaH. Perkiopmens China
a inShlnahidYs Selected readings
rural financial highlight
markets oftenfacets of
neglected
DwvightH. Perkins and Shahid Yusuf in discussionsof agriculturalcreditin
developing countries. Considers the
Looks at China's rural development China's rural development policy. performance of rural financial markets
experience as a whole since 1949. Ana- Helps clarify both the strengths and and ways to improve the quality and
lyzes China's agricultural performance weaknesses of a self-reliant strategy range of financial services for low-in-
and traces it back to the technology of rural development. come farmers. Also reflects new think-
and other sources that made that per- ing on the design, administration,
formance possible. Goes beyond the . . evaluation, and policy framework of
conventional sources of growth analy- .- rural finance and credit programs in
sis to examine the political and organi- - r - ... developing countries.
zational means that enabled the . TheJohns Hopkins UniversityPress.1983.
Chinese to mobilize so much labor for 430 pages.
development purposes. '~ \*ISBN
t<,, 0-8018-3074-5.Stock No. IH 3074.
Describes the successes and failures of ' $32.50 hardcover.
.'- < - - X Rural Poverty Unperceived:
Problems and Remedies
-;^^ t-~--.' .. ~ < -+*Robert Chambers
. j,,I,~ ,
..-- r,k' f 'V,' -~. A , -^Staff
WorkingPaperNo. 400. 1980. 51
,,,
- t,S%&*-
' .X>f + ta pages(includingreferences).
gz-iIU±
~¶-.x -'Z >' - .~ Stock No. WP-0400.$3.
Rural Projects through Urban
,>* .- -. -,, t? ' ^ Eyes: An Interpretation of the
World Bank's New-Style Rural
The Jons 'Development Projects
TheJohns Hopkins University Press. 1984. . Judith Tendler
232 pages. - Stock World Bank Staff WorkingPaperNo. 532.
LC 83-049366.ISBN 0-8018-3261-6.Stock 1982. 100 pages.
$25hardcover.
No.JH3261. .i;-> 4.> : 's ' -'. ISBN 0-8213-0028-8.StockNo. WP 0532.
-. a 0 .J'. -' "1$3.
 most important determinant of overall ISBN 0-8213-0168-3.Stock No. WP 0561
economic growth, has been sluggish in (English)$3.
Sheep and Goats in Sub-Saharan African countries during ISBN 0-8213-0269-8.Stock No. BK 0269
Countrieas:Ther the past two decades. This overview (French)$3.
Developing Countries: Their takes a three-pronged approach to un-
Present and Potential Role derstanding the problems of agricul-
Winrock Intemational Livestock tural production in the 47 countries
Research and Training Center that make up the region. It outlines
Sheep and goats are viewed as an in- domestic and global constraints; sum-
tegral component of complex agricul- marizes price, trade, and consumption Training and Visit Extension
tural systems. This comprehensive forecasts for major agricultural exports; Daniel Benor and Michael Baxter
analysis leads to recommendations on and project trends. Contains a comprehensive explanation
the need for a balanced production Staff Working PaperNo. 608. 1983. 172 of the organization and operation of
system approach for research, training, pages(includingmore than 75 tablesand the training and visit system of agri-
and development programs. Assesses charts). cultural extension. Emphasizes sim-
the role of sheep and goats in food ISBN 0-8213-0221-3.StockNo. WP 0608. plicity and decisiveness. Defines or-
production systems by examining ad- $5. ganization and mode of operation and
vantages and disadvantages, aid/donor allows continuous feedback from farm-
support, constraints on contributions, ers to extension and research workers.
and overcoming constraints. Empha- A System of Monitoring and This method has been adopted in
sizes the need for a combination of Evaluating Agricultural some 40 countries in Asia, Africa, Eu-
support activities and marketing and Extension Projects rope, and Central and South America.
pricing policies for small ruminants Mihe .Cre n ejmnUseful to extension staff at all levels,
and their products. r ongoing Michael M. Cernea and BenjTamin agricultural research personnel, train-
projects. J. epping ers, and staff of agricultural organiza-
TechnicalPaperNo. 15.1983.109 pages. Staff WorkingPaperNo. 272. 1977. 121 tions, as well as universities and train-
ISBN 0-8213-0272-8. pages(including 9 annexes,bibliography). ing institutions involved in agricultural
Stock No. BK 0272. $5. Stock No. WP-0272. $5. and rural development and public ad-
Stock No. BK 0272. $5. ministration.
Sociocultural Aspects 1984, 214 pages.
olfDeveloping Small-Scale Thailand: Case Study of ISBN 0-8213-0121-7.Stockno. BK 0121.
oFisheries Delivering Services Agricultural Input and Output $15.
to the Poor Pricing
Richard B. Pollnac Trent Bertrand
Staff WorkingPaperNo. 490. 1982. 64 Staff Working PaperNo. 385. 1980. 143 Women and the Subsistence
pages(including references). pages (including2 appendixes). Sector: Economic Participation
Stock No. WP-0490. $3. Stock No. WP-0385. $5.NEW and Household
Decisionmaking In Nepal
Some Aspects of Wheat and Traditional Land Tenure and Meena Acharya and Lynn Bennett
Rice Price Policy in India Land Use Systems in the Fascinating analysis of the complex
cial, demographic,and economic fac-so-
Raj Krishna and G.S. Design of Agricultural Projects tors that affect women's decisionmak-
Raychaudhuri Raymond Noronha and Francis J. ing role in the subsistence sector. Data
Staff Working PaperNo. 381. 1980. 62 Lethem collected from seven villages show
pages(including2 appendixes,6 tables, The feasibility of agricultural projects women play a major role in agricul-
bibliography). and their intended impact are often tural production, both as laborers and
Stock No. WP 0381. $3. determined by traditional patterns of managers. Bringing women into the
tenure and land use. This paper pro- market economy would make better
vides agricultural project designers use of local resources and improve
with an analytical basis and rationale their status and economic security in
for examining systems and suggests Nepal.
Sub-Saharan Agriculture: how to use such information in de- Staff WorkingPaperNo. 526.1983. 160
Synthesis and Trade Prospects signing projects. pages.
Shamsher Singh Staff Working PaperNo. 561. 1983. 54 ISBN 0-8213-0024-5.Stock No. WP 0526.
Agricultural production, the single pages. $5.
 The World Bank
Publications Order Form
SEND TO: YOUR LOCAL DISTRIBUTOR OR TO WORLD BANK PUBLICATIONS
(See the other side of ihis form.) P.O. BOX 37525
WASHINGTON, D.C. 20013 U.S.A.
Date

Name Ship to: (Enterif different from purchaser)

Title Name

Firm Title

Address Firm

City State_ Postal Code_ _ Address

Country 'Tlephone ( ) City State_ PostalCode_

Purchaser Reference No. Country 'llephone

Check your method of payment.

Enclosed is my L Check O International Money Order O Unesco Coupons O International Postal Coupon.
Make payable to World Bank Publications for U. S. dollars unless you are ordering from your local distributor.
Charge my O VISA O MasterCard O American Express Oi Choice. (Credit cards accepted only for orders addressed
to World Bank Publications.)

Credit Card Account Number Expiration Date Signature

LiInvoice me and please reference my Purchase Order No.

Please ship me the items listed below.
CustomerInteral
Stock
Number Author/Title RoutingCode Quantity UnitPrce Toal AmountS

All prices subject to change. Prices may vary by country. Allow 6-8 weeks for delivery.
Subtotal Cost $
'Ibtal copies Air mail surcharge if desired ($2.00each) $S_

Postage and handling for more than two complimentary items ($2.00 each) S
Ibtal S
113RD4M Thank you for your order.
 Distributors of World Bank Publications

ARGENTINA GERMANY. FEDEIAL

REnUUC KUWAIT SRILANKAANDTHEMALDIVES
arlcw
Hiua, SRL OF EMRB LakeHouseooshop
GaleaGuemes LSOVeda P. Box5465 P0 Box244
Ronda165.4thFloorOfc.45v465 D300BotnI Kuwait 100,SirCluttmpalans
A.Gadner
1333 BuenosAjes Sunmxoctsse23 MAYS owatha
AUSTRALIA PAPUA NEW GREECE University
of MalayaCooperative Colombo2
GUINEA. FpI.SOLOMON UEMRB Bookhop, Limited n
ISLANDS, ANDVANUATU 24.Ippodau Seet P.0. Box1127,ala Pantai
Bk SWEDEN
Info-Line Athens-11635 KualaLumpur Forsnrletits:
Oversea DocumentDelivery HOGKN,MCUMUOABCE ! mKni.Hvowe
Baoi0. GPO HONa ONG
Lt AcAUFOTCo
.ynv Fenptze
Kng Hov
Box andel
Austraha
Australia
SW2001Debvey 6F1146Pnnce Ewad27
Edward
Road,
W, ~Stockh
SanLoretzn
153-11
Col.delVa 103ne
-2ox27Stock Boxl
13m
Kowloon Deleg.
Betuto
Ir Fosc;anm
SAHRAIN Hong
Kong 03100
Mexico
CGt Weaergen-WWillians
AS
P0 ox no INDIA MOROCCO BOX3004
PU0MRBox
22103 UV Publishe'Disthbutors
Ltd. MEORB 5-104
Man,maTown317 PostBox7013 25StocholM
2 RueMobere
RianWT AND
BANGLADESH NewDelhil1tOO2 Casablanca Lib e Pivot
MicronuhnesDevelopment 10Firt .ManRoad THENETHERLANDS 6 RueGrents
APO boxtOO Gandhi Nagar Medical
Books
Europe.
BV(MBE) Cas pO 381
G. .O Sx OM nabrelo%WO .Noorderwal
33. CH 1211CGare 1l
BElGIUM Apeelay Chamber, P.O.B0x736 7241 BLLoche. TANZANU
BELcaI d Wallace Street NIGERIUA OxodUriversity
Pubb dRom d02 tions
Pra
BUf
Bombav4O1 Lruvernmy PrfssLimited P 0. Box5299
.4v.du Ro 202 cut DD6FveMilbS
1060OnLqels 8/1-91Chowymghw Law Dar es Ssi_
[OWBrussels 8/I-B.~Caottaghee Thfee Crowns BukidigJericho D
La6 Pnat Bwag
5m THAILAN
BLAZIL Cdan CentrlDepaiWsent Store
Publwoe Tecn Internudoamis71%.ICA).Swarup Nagar NORWAY
Llda. 306SibmRoad
Kaimpu 200 TanunKa Johan. A.S. Bangkok
RuaPe aut Gomude, 29 SivaganRoad P 0. Box1177 TRM AD TOBAGO
Nungambakkam Sentrutn S teasStudies Unit
CANADA Madras 60004 Oslo1 ;5 Easternm.UinRoad
le Duseur 9 t INDONESIA PAJaSTAN Curepe
BC.ePlle 0Aere 1U35E6 Pt Indira
Lauted .MrraBok Agena Ttiidid, West Indies
Boudwmk,
Queec 14B5E6 P SamRAftulanp 37 65,Shabralh-eQuaid-e-AzamTUIA
CHILE JakartPusat P.O.BoxNo.729 Societe Tunimenne deDiusion
EditonlRenaaneto POBoxIS Lahore 3 3 Avenue deCarthae
.Uinflores354 IRELAND PANAMA Tunis
Santiago TDCPublihsen Edi f LibnuaCutural RTliKEY
COSTA RICA 12NorthFredekSreet Panamena. S.A. Haset Kitpen A.S.
Liberem Tnlos DubiinI Av 7. Espana 16
CalIe11-13 469. IstiklWl
Caddesi
Av.Femandez Gue1
l Pm oi
IrALY P;uuma Zone I bevoglu-lstanbul
sA lose LicosaConmmmnaa Sawmo SPA EiRU DUGANDA
ViaLalmrrnorl45 Elhtorl DesuiooSA Uganda Bookshop
CYPRUS 50121 Florence Aprtdo 3524 Artn:Mr.Marin-Luther
MEMRB Gaiwango
P0 Box2098 LiBOx7143 P
JAPANP .Bx74
Nicosa EasternBook SevcelESS) IUPP KESampla
37-3.Hongo 3-Chome,Bunko-ku .NationalBook Store LNffD ARABEMIRATES
DENMAIK 113 701RizalAvenue MEM
SamludsLtterstur Tokyo M.etr.Mani P 0 Box6097
RoDenoemC Atke11 JORDAN PORTUGAL Sharah
DK-1970Copenhagen V UIEMRS Livua Portuga UNITED KINGDOM AND
EGYW, ARAEREULICOf P 0 Sox 3143 RueDoCanno 70-74 NORTHERN
A: Aham MIELAND
(abai 1200 LuibO Uicroutno Ltd.
Gal Street Amman SAUDI ARABIA P 0. Box3
CALMo KENYA Iaw Book Store Alton.Hampshire GU34ZPG
FINLND ArncA Book Servie(E.A.)Ltd. P.0. Box3196 United Kingdom
Akaxemnen Kaiakauppa P 0 Box45245 Riyadh11471 VEtEZUELA
P.OBox128 Nairobt SINGAPORE, TAIWAN, BURMA Librma
SF-091I defEste
HeSF4010lo KOREA. REPLBUC OF Inkrnutaon PubltionwPnvate. Ltd. Aptdo. 60.337
PanKorea BookCorporation 0246IstFl.,P*e.Fu Industit Bldg
.. Caaas1060I A
FRANCE P 0 Box101.Kwangwhamun 24NewIndustuRald
WotidBinkPublicato WESlERN SAMOA
Seoul Sinpoe We6yBooDbhop
66Avenue d'lena SPAIN Apix
Pans
i1 Mwudi.Prensa Libmi,S.A.
Casrtelo37
2SOOI Madnd
WORLD BANK TECHNICAL PAPERS (continued)

NcI. 25. Industrialization in Sub-Saharan Africa: Strategies and Performance

No. 26. Small Enterprise Development: Economic Issues from African Experience

Nc. 27. Farming Systems in Africa: The Great Lakes Highlands of Zaire, Rwanda,
and Burundi (also available in French as 27F)

No. 28. Technical Assistance and Aid Agency Staff: Alternative Techniques
for Greater Effectiveness

No. 29. Handpumps Testing and Development: Progress Report on Field and Laboratory Testing

NO. 30. Recycling from Municipal Refuse: A State-of-the-Art Review and Annotated Bibliography

No. 31. Remanufacturing: The Experience of the United States and Implications
for Developing Countries

No. 32. World Refinery Industry: Need for Restructuring

No. 33. Guidelines for Calculating Financial and Economic Rates of Return for DFC Projects

No. 34. Energy Efficiency in the Pulp and Paper Industry with Emphasis on Developing Countries

No. 35. Potential for Energy Efficiency in the Fertilizer Industry

No. 36. Aguaculture: A Component of Low Cost Sanitation Technology

No. 37. Municipal Waste Processing in Europe:

and Energy Recovery Projects

No. 38. Bulk Shipping and Terminal Logistics

No. 39. Cocoa Production: Present Constraii

No. 40. Irrigation Design and Management: I

No. 41. Fuel Peat in Developing Countries

No. 42. Administrative and Operational Proc(

and Area Upgrading
The World Bank

Headquarters EuropeanOffice Tokyo Office

1818 H Street, N.W. 66, avenue d'lena KokusaiBuilding
Washington,D.C.20433, U.S.A. 75116 Paris,France 1-1 Marunouchi3-chome
Telephone:(202) 477-1234 Telephone:(1) 723-54.21 Chiyoda-ku,Tokyo 100. Japan
Telex: WUI64145 WORLDBANK Telex:842-620628 Telephone:(03) 214-5001
RCA 248423 WORLDBK Telex: 781-26838
CableAddress: INTBAFRAD
WASHINGTONDC

ISSN 0253-7494
Cover design by Bill Fraser ISBN 0-8213-0566-2