• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Table of Contents
 Sustainability in agricultural...
 The socioeconomic role of rural...
 Analysis of the competition for...
 Participatory needs assessment:...
 Women farmers' role in managing...
 Farmer-controlled diagnosis and...
 Participation of rural women in...
 Role of farmers in the evaluation...
 Household food security and environmental...






Group Title: Journal for farming systems research-extension.
Title: Journal of farming systems research-extension
ALL VOLUMES CITATION PAGE IMAGE ZOOMABLE PAGE TEXT
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00071921/00009
 Material Information
Title: Journal of farming systems research-extension
Alternate Title: Journal for farming systems research-extension
Abbreviated Title: J. farming syst. res.-ext.
Physical Description: v. : ill. ; 23 cm.
Language: English
Creator: Association of Farming Systems Research-Extension
Publisher: Association of Farming Systems Research-Extension
Place of Publication: Tucson Ariz. USA
Publication Date: 1990-
 Subjects
Subject: Agricultural systems -- Periodicals -- Developing countries   ( lcsh )
Agricultural extension work -- Research -- Periodicals   ( lcsh )
Sustainable agriculture -- Periodicals -- Developing countries   ( lcsh )
Genre: periodical   ( marcgt )
 Notes
Dates or Sequential Designation: Vol. 1, no. 1-
General Note: Title varies slightly.
General Note: Title from cover.
General Note: Latest issue consulted: Vol. 1, no. 2, published in 1990.
Funding: Electronic resources created as part of a prototype UF Institutional Repository and Faculty Papers project by the University of Florida.
 Record Information
Bibliographic ID: UF00071921
Volume ID: VID00009
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 22044949
lccn - sn 90001812
issn - 1051-6786

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Title Page 1
        Title Page 2
    Table of Contents
        Table of Contents
    Sustainability in agricultural development: Trade-offs between productivity, stability, and equitability, by Gordon R. Conway
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
    The socioeconomic role of rural Kom women in the farming system of Cameroon, by Joseph Nkwain Sama
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
    Analysis of the competition for labor by dryland and irrigated crops: The case of rice and millet in Niger, by Ziyou Yu, Robert Deuson, Eric Bomans, and Jess Lowenberg-DeBoer
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
    Participatory needs assessment: A key to FSRE, by A. W. Etling and R. B. Smith
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
    Women farmers' role in managing cassava production in Bandundu, Zaire, by Mbongolo-Ndundu Mputela and Steven E. Kraft
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
    Farmer-controlled diagnosis and experimentation for small rural development organizations, by Bruce Petch and Jane Mt. Pleasant
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
    Participation of rural women in the homestead vegetable farming systems of Bangladesh, by W. A. Shah, Rukshana Yasmin, Rezaul Karim, and M. M. A. Karim
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
    Role of farmers in the evaluation of an improved variety: The case of S35 Sorghum in northern Cameroon, by Mulumba Kamuanga and Martin Fobasso
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
    Household food security and environmental sustainability in farming systems research: Developing sustainable livelihoods, by Michael Drinkwater and Margaret A. McEwan
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
Full Text
Volume 4, Number 2
1994




journal
for Farming Systems
Research- Extension


15 Th oieooi.oeo RrlKi o e ntl
Famn Syte of Caero
Joep 0kanSm
27 Anlyi of til Co ptto oSabrb rln







Journal
for Farming Systems
Research -Extension


Volume 4, Number2, 1994


Published by
the Association for Farming Systems Research-Extension








Journal for Farming Systems Research-Extension


Editor
Timothy R. Frankenberger
Office of Arid Lands Studies
The University of Arizona, Tucson

Associate Editors
Claude Bart, Daniel Goldstein, Jennifer Manthei, and M. Katherine McCaston
Office of Arid Lands Studies
The University of Arizona, Tucson

Production and Layout
Claude Bart, Jennifer Manthei, M. Katherine McCaston,
and Sonia Telesco
Arid Lands Design, Office of Arid Lands Studies
The University of Arizona, Tucson

Sponsors
Ford Foundation
The University of Arizona




The Journal for Farming Systems Research-Extension is published by the Association for
Farming Systems Research-Extension (AFSRE), an international society organized to
promote the development and dissemination of methods and results of participatory on-
farm systems research and extension. The objectives of such research are the development
and adoption through participation by farm household members of improved and
appropriate technologies and management strategies to meet the socioeconomic and
nutritional needs of farm families; to foster the efficient and sustainable use of natural
resources; and to contribute toward meeting global requirements for food, feed, and
fiber.
The purpose of the Journal is to present multidisciplinary reports ofon-farm research-
extension work completed in the field, and discussions on methodology and other issues
of interest to farming systems practitioners, administrators, and trainers. The Journal
serves as a proceedings for the annual international Farming Systems Symposium from
which selected and refereed papers are included. It also welcomes contributed articles
from members of the AFSRE who were unable to attend the symposium. Contributed
articles will be judged by the same review process as invited articles.


ISSN: 1051-6786









Journal for Farming Systems Research-Extension
Volume 4, Number 2, 1994


CONTENTS

1 Sustainability in Agricultural Development: Trade-Offs Between
Productivity, Stability, and Equitability
Gordon R. Conway

15 The Socioeconomic Role of Rural Kom Women in the Farming
System of Cameroon
Joseph Nkwain Sama

27 Analysis of the Competition for Labor by Dryland and Irrigated
Crops: The Case of Rice and Millet in Niger
Ziyou Yu, Robert Deuson, Eric Bomans, and Jess Lowenberg-
DeBoer

45 Participatory Needs Assessment: A Key to FSRE
A. W. Etling and R. B. Smith

57 Women Farmers' Role in Managing Cassava Production in
Bandundu, Zaire
Mbongolo-Ndundu Mputela and Steven E. Kraft

71 Farmer-Controlled Diagnosis and Experimentation for Small Rural
Development Organizations
Bruce Petch and Jane Mt. Pleasant

83 Participation of Rural Women in the Homestead Vegetable
Farming Systems of Bangladesh
W. A. Shah, Rukshana Yasmin, Rezaul Karim, and M. M. A.
Karim

93 Role of Farmers in the Evaluation of an Improved Variety: The
Case of S35 Sorghum in Northern Cameroon
Mulumba Kamuanga and Martin Fobasso

111 Household Food Security and Environmental Sustainability in
Farming Systems Research: Developing Sustainable Livelihoods
Michael Drinkwater and Margaret A. McEwan








Sustainability in Agricultural Development:
Trade-Offs Between Productivity, Stability, and
Equitability

Gordon R. Conway



ABSTRACT
Agroecosystem Analysis (AEA) provides a theoretical and practical
context for useful definitions of productivity, stability, sustainability, and
equitability. Productivity may be defined as any combination ofinput and
output in which the output is a valued agricultural product. Stability
measures the constancy of such productivityin the face ofsmall disturbing
forces in the environment, while sustainability measures the ability of
productivity to withstand major disturbing stress or shock. Sustainability
can be expressed in terms of the inertia, elasticity, amplitude, hysteresis,
or malleability of an agroecosystem. The inevitable trade-offs between
these properties are illustrated quantitatively, using data from British
manorial records ofcereal production, and qualitatively, with a contem-
porary village survey from Ethiopia. The Javanese home garden provides
an example in which the trade-offs are minimized.


INTRODUCTION
The phrase "sustainable agriculture" has acquired diverse meanings. To the
agriculturalist it means maintaining the momentum of the Green Revolution.
To the ecologist it is a way of providing sufficient food without degrading
natural resources. To the economist it represents an efficient long term use of
resources, and to the sociologist and anthropologist it embodies an agricul-
ture that preserves traditional values. Almost anything that is perceived as
"good" from the writer's perspective can fall under the umbrella of sustainable
agriculture-organic farming, the small family farm, indigenous technical
knowledge, biodiversity, integrated pest management, self-sufficiency, recy-
cling, and so on (Conway and Barbier, 1990).
This diversity of interpretation is to be welcomed as part of a process of
gaining consensus for radical change. But it results in concepts and definitions
of little practical value. The often-quoted definition of sustainable develop-

1 Representative, Ford Foundation, New Delhi, and Professor, Centre for Environmental Tech-
nology, Imperial College of Science, Technology, and Medicine, London.






CONWAY


ment proposed by the World Commission on Environment and Develop-
ment-"development that meets the needs of the present without compro-
mising the ability of future generations to meet their own needs" (Brundtland
Report, 1987)-is valuable as a policy statement but is too abstract for
farmers, research scientists, or extension workers trying to design new
agricultural systems and develop new agricultural practices. For them, a
definition is needed that is scientific, open to hypothesis testing and experi-
mentation, and practicable.
In this paper I elaborate on a definition of sustainable agriculture that I and
my colleagues have been using in recent years. The definition arises within a
new paradigm for agricultural development that goes under the name of
Agroecosystem Analysis (AEA). The origins ofAEA lie in efforts to improve
the analysis of natural ecosystems (Walker, et al., 1978), but most of the
concepts and techniques were developed at the University ofChiang Mai in
Thailand beginning in 1978 (Gypmantasiri et al., 1980) and were refined and
elaborated at the University of Khon Kaen, also in Thailand, by a group of
university and government research workers in Indonesia (KEPAS) and by the
Southeast Asia Universities Agroecosystems Network (SUAN). The first
major workshop on sustainable agriculture was held in Indonesia in 1982
(KEPAS, 1984).
Agroecosystem Analysis (AEA) uses the concepts of agroecosystems,
agroecosystem hierarchies, and agroecosystem properties and their trade-offs
to stimulate interdisciplinary analysis. It can be viewed as a form of Farming
Systems Research; however, it has the capacity to extend the analysis, using the
same concepts and techniques, to systems in the agricultural hierarchy above
and below the farm. AEA may involve detailed quantitative analysis based on
experiments or secondary data. But, more commonly, it takes the form of a
rapid field appraisal utilizing a variety of simple descriptive diagrams, prepared
in farms and villages from direct observations and from interviews with rural
people. These diagrams, developed as part ofAEA, have subsequently become
popular techniques of Rapid Rural Appraisal (RRA) (Khon Kaen University,
1987). They have proven to be powerful means of communication, especially
when prepared and analyzed by farmers themselves (Mascarenhas et al.,
1991). Today, AEA lies within a cluster of overlapping approaches that
provides a rich array of concepts and methods for agricultural and rural
analysis.2


2 The literature on AEA concepts includes a number of books and papers [Conway, 1985, 1986,
1987; Conway and Barbier, 1990; Soemarwoto and Conway, 1991; Sajise and Rambo, 1986;
Rerkasem and Rambo, 1988 ]; most of the field-based analyses are described in published reports
[KKU-Ford Cropping Systems Project, 1982a, b; Conway, et al.,1985, 1989; KEPAS, 1985a, b,
1986; Conway and Sajise, 1986; Ethiopian Red Cross, 1988]. Readers are also referred to the
Gatekeeper Series on Sustainable Agriculture and to RRA Notes, both produced by the
International Institute of Environment and Development in London.


Journal for Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 3

CONCEPTS AND DEFINITIONS
Based on the literature cited above, an agroecosystem may be defined as an
ecological and socioeconomic system, comprising domesticated plants and/
or animals and the people who husband them, intended for the purpose of
producing food, fiber, or other agricultural products. Agroecosystems defined
in this way fall into a hierarchy. At the lowest level is the individual plant or
animal, its immediate microenvironment, and the people who tend and
harvest it. The next level is the crop or herd, contained within a field or
paddock, or in a swidden, home garden, or range. These systems, alone or in
various combinations, together with the farm household, comprise a farming
system. The hierarchy continues upwards in a similar fashion, each agroeco-
system forming a component of the agroecosystem at the next level.
Agroecosystems, so defined, are also cybernetic (Wiener, 1948; Ashby,
1956). They have recognizable goals, and strategies to attain them. I suggest
that the primary goal ofan agroecosystem is increased "social value." Broadly,
it is composed of the amounts of goods and services produced by an
agroecosystem, the degree to which they satisfy human needs, and their
allocation among the human population. It also has a time dimension, since
we seek not only increased benefits in the immediate future but also a degree
of security over the longer term.
Thus social value has several measurable components: the present produc-
tion of the agroecosystem, its likely level in the future, and its distribution
among the human population. These are expressed in four agroecosystem
properties: (1) productivity-the output of valued product per unit of
resource input; (2) stability-the constancy of productivity in the face of small
disturbing forces arising from the normal fluctuations and cycles in the
surrounding environment; (3) sustainability-the ability of the agroecosys-
tem to maintain productivity when subject to a major disturbing force; and (4)
equitability-the evenness of distribution of the productivity of the agroeco-
system among the human beneficiaries.


THE MEASUREMENT OF AGROECOSYSTEM PROPERTIES


Productivity
The productivity of plants and animals can be measured as the amount of
new biological material, or biomass, produced per unit of time. But agricul-
ture is only concerned with the portion of the biomass that is useful-the
harvest. Productivity is thus more appropriately measured either as yield or
income, or in terms of the other benefits that derive from the harvest.


Vol. 4, No. 2, 1994






CONWAY


Commonly, yield is measured per hectare or as the total production of
agricultural goods and services per household, region, or nation; a large
number of different measures are possible depending on the nature of the
product and of the resource input being considered. The yield may be
expressed in terms of kilograms of grain, tubers, meat, fish, or any other
consumable or marketable product. Alternatively, it may be expressed in
calories, proteins, or vitamins, or as its monetary value in the market. In the
last case it is measured as income over expenditure (i.e., as profit).
The three basic resource inputs to productivity are land, labor, and capital.
Strictly speaking, energy is subsumed under land (solar energy), labor (human
energy), and capital (fuel energy). Similarly, technological inputs, such as
fertilizers and pesticides, are components of capital, but both energy and
technology can be treated for many purposes as separate inputs. Productivity
can be expressed as tons of grain per hectare, kilograms of nitrogen fertilizer
per hectare, grams of active ingredient of insecticide per hectare, or any other
conceivable combination of input and output.

Stability
Over time, productivity may rise, fall, or remain constant. It will also exhibit
a pattern ofvariability relative to the dominant trend line. The yield of a crop,
for example, is likely to mirror the variability in the climate. Income may also
fluctuate, reflecting not only changes in yield but also variations in the market
price ofinputs, such as labor, fertilizers and pesticides, and of the product. The
latter, in turn, is a function of supply and demand.
Productivity may be defined in any of the ways described above and its
stability measured by the coefficient of variation in productivity.

Sustainability
Stability measures the behavior of an agroecosystem in response to the
normal fluctuations in the surrounding environment. Productivity goes up
and down, but is not seriously threatened. However, agroecosystems are also
subject to major disturbing forces that can cause productivity to fall well below
its previous level. If productivity does fall, it may recover either to its original
level or to a new lower level or, in extreme situations, it may cease altogether.
Sustainability is the ability of an agroecosystem to withstand such disturbing
forces.
The simplest case is an individual crop plant or animal. How does it
withstand forces that threaten its survival? If we consider a plant strictly as a
physical structure, the concepts of mechanics and Newton's laws of motion
apply. A force acting on the plant, such as a low temperature, elicits a physical
or chemical change. The force is termed a stress and the change a strain. If the
change is reversible, the strain is regarded as elastic; if irreversible, the strain
is plastic. When a maize plant is cooled from 30C to 5C it stops growing, but


Journal for Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 5

normal growth resumes when the temperature rises again. If wheat is cooled
to 5C growth continues, although at a slower rate. Both plants have suffered
elastic strain, although the strain is greater in the case of maize. At an even
lower temperature a plant dies; it suffers a plastic strain.
But this physical model is too simple. The responses of biological systems
are more than mechanical. Plants and animals may counter stress through
either evolutionary change or adaptation during the organism's life. A
common example of the latter is when plants of temperate regions harden to
the freezing temperatures of winter through gradual exposure to the increas-
ing cold of the autumn.
More dynamic still are the responses to nonphysical forces, such as attacks
by pests or competition from weeds. Also, the variety of forces and responses
becomes even greater as we move up the agroecosystem hierarchy through the
crop, farm, village, and nation. Productivity may collapse, for example, under
the pressure of economic forces, such as a steep rise in input costs or increased
levels of indebtedness. Social disturbances, such as communal conflict or
political revolution, may also pose a threat.
We can distinguish two kinds of disturbing forces. First, there are relatively
small and predictable forces that act on a regular and sometimes continuous
basis and produce a large cumulative effect. Salinity, toxicity, erosion, pest or
disease attack, indebtedness, and declining market demand are examples.
Such a force constitutes a stress. The other kind of force is very large,
infrequent, and relatively unpredictable, and produces an immediate, large
disturbance or perturbation. This I refer to as a shock. Examples are rare floods
or droughts, an outbreak of a new pest, or the sudden rise in an input price.
In practice, it is usually not difficult to distinguish between a stress and a shock,
providing it is clear which level in the agroecosystem is being considered. A
shock to an individual plant (i.e., being destroyed by a pest) may be only part
of a stress to the whole crop.
There is a continuum between stability and sustainability, but they are
usually distinguishable by qualitatively different patterns of behavior. Stability
refers to the dynamics of an agroecosystem when subject to relatively minor
and commonplace disturbing forces. Typically, the impact of these forces is
small because, out of long association, agroecosystems have developed
adequate defenses. By contrast, sustainability is concerned with forces that are
rarer and less expected, so that agroecosystems are likely to have fewer or less
developed defenses. Often the defense mechanisms are qualitatively different.
An animal experiencing small changes in the outside temperature may respond
with minor changes in circulation, but a sudden shock, such as immersion in
freezing water, may elicit a qualitatively different response, such as the
secretion of adrenaline.
Stability is easily measured from a time series of productivity. However, the
measurement of sustainability is more complex because of the range of forces
and responses that may be encountered. The strength and nature of the shock


Vol. 4, No. 2, 1994






CONWAY


or stress has to be assessed, the pattern of response to disturbance has to be
described, and the degree of resistance or resilience must be quantified. How
far is the productivity depressed, how quickly and to what level does it return,
and in what fashion?
Various responses are possible. First, productivity may remain unaffected
if the agroecosystem adequately resists the disturbing force. One mechanism
of resistance is to escape. For example, the seeds of wild cereals lie dormant in
the ground throughout the Mediterranean summer, thus avoiding the effects
of the extreme heat. Nomadic pastoralists may move their cattle to escape an
impending drought. Alternatively, the agroecosystem may be protected.
Farmers may build a bund to prevent flooding of crop fields. In analogous
fashion, a tariffwall, as has been erected by the European Economic Commu-
nity (EEC), may protect crop production from falling world prices. A measure
of resistance is inertia (i.e., the amount ofstress or shock that can be withstood
without the productivity being affected). For example, where a bund is built
to guard a crop, the inertia can be expressed as the maximum height of flood
against which the bund provides protection.
If resistance is weak, productivity may fall below the usual range of
variation. The question then is: How resilient is the agroecosystem? Will it
return to the previous level or trend of productivity, how quickly, and in what
manner? The return may be smooth, or the disturbance may continue with
violent aftereffects before the previous level or trend of productivity is
regained. Sometimes the productivity does not recover; it may remain at a new
lower level or trend line or, in extreme situations, continue to fall and
disappear altogether.
Various measures of this pattern of resilience are possible (Table 1).
Elasticity, amplitude, and hysteresis measure the speed, likelihood, and
pattern of recovery of productivity following a stress or shock. Malleability
measures the difference between the mean productivity before and after the
disturbance.
Studies of the resilience of ecological systems gained widespread attention
following the work of C.S. Holling (1973, 1985). He stressed the importance
of distinguishing, as I do here, between the behavior of systems under normal
environmental conditions (i.e., their stability), and when subject to a distur-
bance that has the potential to change their state dramatically. He suggested
that ecological and other systems could potentially exist in more than one
steady state and, when subject to disturbance, could "flip" from one state to
another. There is some dispute as to whether multiple steady states exist in
natural ecological systems (Connell and Sousa, 1983), but they commonly
occur in agroecosystems. An example is the collapse ofswidden cultivation
from a regular cropping/fallow cycle to a persistent Imperata grassland cover
following the stress of increased crop harvesting. Here the amplitude of the
sustainability can be measured as the maximum number of crops that can be
taken in the cropping phase before collapse occurs. In an example described


Tournalfor Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 7

Table 1. Measures of agroecosystem sustainability. All measures relate to the sustain-
ability of the productivity trend.

CHARACTERISTIC DEFINITION MEASUREMENT
Inertia resistance of productivity level of stress
trend to change or shock that can be resisted
without affecting trend line
Elasticity rapidity of recovery of time taken for trend
productivity trend line to be recovered
following disturbance
Amplitude zone from which recovery maximum amount of
occurs following disturbance following
stress and shock from
which recovery is possible
Hysteresis degree to which path of difference between disturbance
recovery is exact and recovery paths
reversal of disturbance path
Malleability degree of difference difference between
between system state mean productivities
before and after disturbance
Source: Orians, 1975 and Westman, 1978.

by Trenbath, et al. (1990), when up to eight crops were grown the system
returned to mature forest, but after eight crops it collapsed to permanent
grassland. Eight crops is thus a measure of the amplitude. Another example
is the behavior of grazing systems, investigated by Noy-Meir (1975). Here,
increasing livestock raises productivity but also stresses the vegetation. At a
certain intensity of stress, which is often very close to the maximum livestock
carrying capacity, the vegetation collapses and the grazing system moves to a
new level of productivity, much lower than before. The amplitude can be
measured in terms of the maximum livestock density before collapse.
Finally, the assessment of sustainability requires a measure of the effective-
ness of internal adjustments that agroecosystems make in response to stresses
and shocks. The variety of such adaptations is considerable. Hardening is one
example. Another is the construction of a bund to protect against flooding.
Weeds may be countered by hoeing or by more vigorous crop growth. At the
farm level, the response to growing debt may be to switch to a less risky crop/
livestock combination or to one requiring lower inputs. A village stressed by
the loss of young people emigrating in search of more lucrative work may
respond by adopting more labor-saving techniques of cultivation. Similarly, a
district may counter rising transport costs by a switch to higher value, lower
volume products; a region may respond to a widespread drought by establish-
ing a network of famine relief stores as a protection against future droughts;
and a nation may respond to increasing competition by changing the nature
of its productivity so as to exploit its comparative advantage.


Vol. 4, No. 2, 1994






CONWAY


Equitability
Productivity, stability, and sustainability adequately measure how much an
agroecosystem produces and is likely to produce over time. An African village
agroecosystem that produces a high, stable yield of sorghum using practices
and varieties that are broadly resistant to pests and diseases will have a higher
social value than another village producing lower, less stable or sustainable
yields. However, social value depends not only on the pattern of production
but also on the pattern of consumption. Who benefits from the high, stable,
and sustainable production? How is the harvested sorghum, or the income
from the sorghum, distributed among the people of the village? Is it evenly
shared or do some villagers benefit more than others?
Equitability describes this pattern of distribution of productivity. Again,
productivity may be measured in any of the ways described above. Thus the
product per person-hour of a farm may be shared between the tenant and the
landowner. More commonly, equitability refers to the distribution of the
overall production-that is, the total goods and services produced by an
agroecosystem. In subsistence farming the producers are all consumers, but
the higher the agroecosystem in the hierarchy, and the greater the degree of
commercialization, the more nonproducers benefit.
The most straightforward way of measuring how evenly products or
income are distributed is to construct a histogram. However, this can give a
misleading impression because histograms tend to overemphasize the middle
of the distribution at the expense of the extremes (Atkinson, 1970). At the left,
beyond the origin, are those with negative incomes-bankrupt farmers or
those close to starvation-while to the right the histogram has to be
impossibly extended to incorporate the rich.
There are a number of alternative ways of representing equitability that are
intended to be positive (i.e., objective measures of distribution). The most
commonly used are the Lorenz Curve and the Gini Coefficient (Lorenz,1905;
Gini,1912; Atkinson,1970). However, these measures suffer from several
drawbacks (see the review in Sen, 1973). For example, they tend to give
relatively greater weight to changes in different parts of the range. A more
important criticism is that they refer to distribution of income or product
among people as though everyone were alike. This is clearly untrue; people
differ in their endowments and derive different amounts of individual utility
from their income. A kilogram of rice is valued much more by a person who
is starving than by one who is rich. Furthermore, individuals compare what
they receive with what others receive. Measures of equitability, if they are to
reflect the actual decisions made by social groups (whether they be house-
holds, villages, or nations), must incorporate these notions of social justice.
But in practice this is hard to accomplish because measurement of individuals'
needs and utilities is difficult and laborious. The most satisfactory measure, so
far, is that of Atkinson (1975), who weights the individual incomes before
adding them.


Journal for Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 9

TRADE-OFFS

Each agroecosystem, at each level of the hierarchy, has a social value. Also, one
kind ofagroecosystem may have a greater social value than another and hence
is more sought after. I assume that people seek to maximize social value and,
to this end, will adopt strategies consisting of various combinations of
productivity, stability, sustainability, and equitability. Inevitably, there are
trade-offs between the levels of the properties. For example, a large-scale
irrigation project may achieve greater overall productivity at the expense of
sustainability and equitability. Similarly, too much emphasis on equitability
may inhibit productivity.

Quantitative Analysis
To date, there has been very little quantitative analysis of the trade-offs
between agroecosystem properties. A notable example is the analysis carried
out by Pretty (1990) of the properties of manorial agriculture in England
during the Middle Ages. The study was based on a remarkable set of records
for individual manors extending from 1283 to 1349 A.D. (Titow, 1972).
Three main cereals were grown-oats, wheat, and barley-and their
productivity, stability, and sustainability are summarized in Table 2. Produc-
tivity was measured as the number of seeds harvested per seed sown or as net
yield per hectare (gross yield minus seed retained for the next sowing). On
both measures, wheat and barley outyielded oats but the productivity of oats
was notably more stable. Pretty (1990) has also shown that the price of oats
was more stable than that of wheat and barley.
Sustainability of the cereals was assessed by plotting the pattern of yields
following a significant fall in yield caused, for example, by a very wet year. After
poor harvests, oats recover more rapidly than wheat-they are more elastic
(Table 2). Its greater stability and sustainability meant that the oat crop was
more reliable and hence often provided the mainstay of the peasants' diet,
particularly in the more marginal lands.

Table 2. Yields of cereals grown on English manors from 1283-1349.
WHEAT OATS BARLEY
Productivity
Net yield (kg/ha) 385.0 300.0 540.0
Seeds/seed sown 4.0 2.3 3.5
Stability (coefficient of variation)
Net yield 38.8% 31.3% 39.9%
Seeds/seed sown 36.9% 33.6% 37.3%
Sustainability (elasticity)
Time to return to average 4 7+ yrs 1 5 yrs -
Source: Pretty, 1990.


Vol. 4, No. 2, 1994






CONWAY


The manorial agroecosystem of England was extremely long-lived and
Pretty (1990) has concluded that the high degree ofsustainability was marked
by relatively high stability and equitability, but these were obtained at the
expense of productivity. Yields had not significantly increased since Roman
times and remained low for virtually all of the manorial period.

Qualitative Analysis
Most analyses of the trade-offs between agroecosystem properties are
qualitative. They typically consist of rapid assessments of the likely impact of
innovations proposed during AEA exercises. An example is the AEA carried
out in the Wollo province of Ethiopia in 1986 (Ethiopian Red Cross, 1988).
The exercise lasted just under two weeks and was carried out by a multidisci-
plinary team comprising headquarters staff of the Ethiopian Red Cross, the
Ministry of Agriculture, and field staff from the province. Two villages were
analyzed using RRA techniques with an emphasis on semistructured inter-
viewing, analytical games such as preference ranking, and constructing maps,
transects, and seasonal calendars. From these, problems and opportunities
were identified and a shortlist produced of innovations or "best bets" that the
team felt would be appropriate for development in the two villages. These best
bets were then prioritized in terms of their cost, feasibility, and likely impact
on development in terms of the system properties (Table 3).
The best bets were chosen for their potentially high stability and sustain-
ability, but analysis by the team suggested considerable differences in potential
productivity and equitability. Two of the best bets with the highest productive
potential-lowland irrigation and the introduction of rainfed crops-were
judged to have little effect on improving equitability, primarily because they

Table 3. Assessment of best bets for Abicho, Wollo province, Ethiopia.

FEASIBILITY
BET/INNOVATION PRODUCT- STAB- SUSTAIN- EQUIT- COST TECH- SOCIAL
IVITY ILITY ABILITY ABILITY NICAL

1. Lowland irrigation ++ + ++ 0 X XX XXX
2. Gully cropping + + ++ ++ XX XXX XXX
3. New rain-fed crops ++ ++ ++ 0 XX XX XX
4. Upland revegetation + ++ ++ ++ X XXX X
5. New forage crops + ++ ++ + XX XX XX
6. Household water
supply + ++ ++ + XX XX XXX
7. Home garden
development ++ ++ ++ ++ XXX XXX XXX
Key: negative impact, 0 no impact, + positive impact, ++ very positive impact, X high cost or
poor feasibility, XX medium cost or feasibility, XXX low cost or high feasibility.
Source: Ethiopian Red Cross, 1988.


Journal for Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 11


would only benefit certain better-off farmers. Other bets with a high potential
for benefiting poor farmers were considered relatively less productive. The
innovation that scored consistently high across the board was the develop-
ment of home gardens for the new village sites that the government was in the
process of creating. One outcome of the exercise was the establishment, by the
Ethiopian Red Cross in the weeks that followed, of a nursery of home garden
plants.
Because of the short time of the exercise and the pressing need to decide
on appropriate innovations for the impoverished villages, no detailed quanti-
tative analysis was possible. Assessment was based on the collective judgement
of the team members, based on their experience and intuition. Although the
results lacked precision or rigor, they had the advantage of being agreed upon
collectively, which facilitated the subsequent action that was taken.

Minimizing the Trade-offs
The social value of an agroecosystem is the product of the levels of the four
different system properties. However, the product is not a simple arithmetic
addition or multiplication because people in each place and at each period of
time weight the properties differently (Conway, 1985). In the Middle Ages,
high stability and sustainability appear to have been given preference over high
productivity. In the Ethiopian AEA, the team engaged in a fierce dialogue over
the relative weighting between productivity and equitability when deciding
on their final priorities. Nevertheless, in most situations, there is a preference
for agroecosystems in which all properties are high and the trade-offs are
explicitly minimized. In the Ethiopian case, the most promising innovation in
this respect was home-garden development.
Home gardens are one of the oldest forms of farming systems and may have
been the first agricultural system to emerge in hunting and gathering societies.
Today, home or kitchen gardens are particularly well developed in the island
of Java in Indonesia, and these have been explicitly analyzed in terms of
agroecosystem properties by Soemarwoto and Conway (1991).
The immediately notable characteristic of home gardens is their great
diversity relative to their size-in one Javanese home garden, 56 different
species of useful plants were found. Also commonly present is a diversity of
livestock-cattle, goats, chickens, fish in fish ponds, and so on. Closer analysis
shows the high diversity to be matched by high levels of productivity, stability,
sustainability, and equitability (Table 4). Part of the reason for this minimal
trade-off is the inherent diversity, which helps stabilize production, buffers
against stress and shock, and contributes to a more valued level of production.
But equally important is the intimate nature of the home garden. The close
attention that is possible from family labor ensures a high degree of stability
and sustainability and the link between the garden and the traditional culture
leads to an equitable distribution of the diverse products.


Vol. 4, No. 2, 1994






CONWAY


Table 4. System properties of the home garden when compared with a rice field.

HOME GARDEN RICE FIELD

Productivity Higher standing biomass Higher gross income

Stability Higher net income Seasonal production
(lower inputs)

Greater variety of production Vulnerable to climatic and
(food, medicines, fuelwood) disease variation

Year-round production
("living granary")

Higher year-to-year stability

Sustainability Maintenance of social fertility Heavy pest and disease
attack
Protection from soil erosion

Equitability Home gardens in most households Product to land owners

Barter of products
Source: Soermarwoto and Conway, 1991.


CONCLUSION

In many respects, the home garden is a unique agroecosystem. But there are,
undoubtedly, other systems that have similarly high levels of productivity,
stability, sustainability, and equitability. More important, there are technol-
ogies and innovations that can potentially help reduce the trade-offs between
the properties. The methods of analysis described above provide a rigorous yet
straightforward approach to assessing agroecosystem properties and their
trade-offs. Experience in an increasing number of field projects has demon-
strated that this approach is practical and easy to use in the hands of people
with a range of disciplinary skills and backgrounds. As the Ethiopian case study
demonstrates, it can be easily incorporated in day-to-day decision making.
Defining sustainability in terms of preservation or duration, as is commonly
done, has little practical value. Long-term experiments to measure persistence
(i.e., of different cropping systems) are of research interest but take too long
to constitute a practicable analytical method. By contrast, measuring the
ability of an agroecosystem to withstand stress and shock is a subject for
experiments using classical agricultural methods.
The current danger of using sustainability as a loosely defined term to
encompass a wide range of systems and technologies is that benefits may be
obtained at the expense of other, less obvious costs. High sustainability is not


Journal for Farming Systems Research-Extension






SUSTAINABILITY IN AGRICULTURAL DEVELOPMENT 13

the only desirable aspect of agricultural production and in many situations it
may be necessary to trade a degree of sustainability for higher levels of
productivity or equitability. Choices and decisions are necessary and are best
made when the options are clearly apparent.


REFERENCES

Ashby, W.R. 1956. An introduction to cybernetics. London: Chapman and Hall.
Atkinson, A.B. 1970. On the measurement of inequality. Journal of Economic Theory
2:244-263.
Atkinson, A.B. 1975. The economics ofinequality. Oxford: Clarendon Press.
Brundtland Report. 1987. Our common law. World Commission on Environment and
Development. London: Oxford University Press.
Connell, J.H., and W.P. Sousa. 1983. On the evidence needed to judge ecological
stability or persistence. The American Naturalist 121:789-824.
Conway, G.R. 1985. Agroecosystem analysis. Agricultural Administration 20:31-55.
Conway, G.R. 1986. Agroecosystem analysis for research and development. Bangkok:
Winrock International.
Conway, G.R. 1987. The properties of agroecosystems. Agricultural Administration
24:95-117.
Conway, G.R., and E.B. Barbier. 1990. Afterthegreen revolution: Sustainable agriculture
for development. London: Earthscan.
Conway, G.R., Z. Alam, T. Husain, and M.A. Mian. 1985. An agroecosystem analysis for
the northern areas ofPakistan. Aga Khan Rural Support Programme, Gilgit, Pakistan.
Conway, G.R., P.E. Sajise, and W. Knowland. 1989. Lake Buhi: resolving conflicts in a
Philippine development project. Ambio 18:128-35.
Ethiopian Red Cross Society. 1988. Rapid RuralAppraisal: A closer look at rural life in
Wollo. Ethiopian Red Cross Society, Addis Ababa, Ethiopia, and International
Institute for Environment and Development, London, UK.
Gini, C. 1912. Variabilita e Mutabilita. Bologna, Italy.
Gypmantasiri, P., et al. 1980. An inter-disciplinary perspective of cropping systems in the
Chiang Mai Valley: Key questions for research. Faculty of Agriculture, University of
Chiang Mai, Chiang Mai, Thailand.
Holling, C.S. 1973. Resilience and stability of ecological systems. Annual Review of
Ecology and Systematics 4:1-24.
Holling, C.S. 1985. Perceiving and managing the complexity of ecological systems. In
The Science and Praxis of Complexity. Tokyo: United Nations University.
KEPAS. 1984. The sustainability of agricultural intensification in Indonesia: A report of
two workshops of the research group on agroecosystems. Agency for Agricultural
Research and Development, Jakarta, Indonesia.
KEPAS. 198 5a. The critical uplands of Eastern Java: An agroecosystem analysis. Agency
for Agricultural Research and Development, Jakarta, Indonesia.
KEPAS. 1985b. Swampland agroecosystems of Southern Kalimantan. Agency for Agri-
cultural Research and Development, Jakarta, Indonesia.
KEPAS. 1986. Agro-ekosistem daerah kering di Nusa Tenggara Timur. Agency for
Agricultural Research and Development, Jakarta, Indonesia.


Vol. 4, No. 2, 1994






CONWAY


Khon Kaen University. 1987. Proceedings of the international conference on Rapid Rural
Appraisal. Khon Kaen University, Khon Kaen, Thailand.
KKU-Ford Cropping Systems Project. 1982a. An agroecosystem analysis of Northeast
Thailand. Faculty of Agriculture, Khon Kaen University, Khon Kaen, Thailand.
KKU-Ford Cropping Systems Project. 1982b. Tambon and village agricultural systems
in Northeast Thailand. Faculty of Agriculture, Khon Kaen University, Khon Kaen,
Thailand.
Lorenz, M.O. 1905. Methods of measuring the concentration of wealth. Journal of the
American Statistical Association 9.
Mascarhenas, et al. 1991. Participatory Rapid Appraisal. RRA Notes 13. London:
International Institute for Environment and Development.
Noy-Meir, I. 1975. Stability of grazing systems: An application ofpredator-prey graphs.
Journal ofEcology 63:459-481.
Orians, G.H. 1975. Diversity, stability and maturity in natural ecosystems. Pages 64-65
in W.H. van Dobben and R.H. Lowe-McConnel, eds., Unifying concepts in ecology.
The Hague: Junk.
Pretty, J.N. 1990. Sustainable agriculture in the Middle Ages: The English manor. The
Agricultural History Review 38:1-19.
Rerkasem, K, and T. Rambo, eds. 1988. Agroecosystem research for rural development:
Selected papers from the Third SUAN regional research symposium on agroecosystem
research held at Chiangmai, Thailand. University of Chiang Mai, Thailand.
Sajise, P., and T. Rambo, eds. 1986. Agroecosystem research in rural resource manage-
ment and development: Selected papers from the second SUAN regional research
symposium on agroecosystem research held at Baguio City, the Philippines. Univer-
sity of the Philippines, Philippines.
Sen, A. 1973. On economic inequality. New York: W.W. Norton.
Soemarwoto, 0., and G.R. Conway. 1991. The Javanese home garden. Journal for
Farming Systems Research-Extension 2(3):95-118.
Titow, J.Z. 1972. Winchester yields. Cambridge, UK
Trenbath, B.R., G.R. Conway, and I.A. Craig. 1990. Threats to sustainability in
intensified agricultural systems: Analysis and implications for management. Pages
337-365 in S.R. Gleissman, ed., Agroecology: Researching the ecological basis for
sustainable agriculture. New York: Springer-Verlag.
Walker, B.H., G.A. Norton, G.R. Conway, H.N. Comins, and M. Birley. 1978. A
procedure for multidisciplinary ecosystem research: With reference to the South
African Savanna Ecosystem Project. Journal of Applied Ecology 15:481-502.
Westman, W.E. 1978. Measuring the inertia and resilience of ecosystems. BioScience
28:705-710.
Wiener, N. 1948. Cybernetics. Cambridge, Mass.: MIT Press and New York: John Wiley.


Journal for Farming Systems Research-Extension








The Socioeconomic Role of Rural Kom
Women in the Farming System of Cameroon

Joseph Nkwain Sama1



ABSTRACT
Women's access to appropriate resources increases their productivity.
This is true especially in the women-dominated agriculture of Kom
society. Kom, an ethnic group ofCameroon's North West Province, faces
frequent food shortages because its expanding population and increased
cultivation of perennial crops have drastically reduced food-crop land.
Inefficient food usage and corn-based eating habits aggravate the food
deficit. This paper examines women's unrecognized socioeconomic
contribution to the Kom economy and identifies their sociocultural and
physical constraints, and suggests the importance of recognizing the
socioeconomic role ofwomen. It also proposes measures for minimizing
these constraints. Such measures include improved farm-to-market
roads, improved rural schools, intensified rural extension services with
greater focus on the productivity of women, the use of more female
extension agents, and improved inputs and farming systems to increase
income and facilitate food production for Kom farm families.


INTRODUCTION
Women have been observed to do more than half the agricultural work in
female-labor-dominated farming systems (Boserup, 1970). Gladwin and
McMillan (1989) suggest that unless ways are found to help farm women
increase productivity in the short run, a turn around in the socioeconomic
development of developing countries will not be possible. To this end, specific
constraints to women's production and potentials for improvements in
female-labor-dominated farming systems need to be identified. Interventions
to assist them in any given society require prior knowledge of the available
alternatives and of the constraints on such alternatives. Such is the case with
the Kom people, a small, dynamic ethnic group strongly bound by a culture
in which the women's role in agricultural production is crucial.


1 Joseph Nkwain Sama, is an assistant professor in the Department of Agricultural Economics,
Dschang University, Cameroon.






SAMA


Location and Sociocultural Background of Kom
The Kom are located in the Mentchum Division of the North West
Province of Cameroon (Figure 1). Their population numbers over 130,000
and covers an area of less than 100 sq. km of hilly land. The Kom are one of
the four principal ethnic groups of that province. Access to the area is difficult,
but a variety of crops are produced in the valleys, hill slopes, and plains of the
province.
The Kom have a matrilineal inheritance system. Descent is traced through
mothers, but women have no rights to inheritance. Instead of sons inheriting
from fathers, brothers inherit from brothers. If the deceased has no brother,
inheritance passes through his sister to her son. It is thus possible for nephews
to inherit from their uncles.
Each native of Kom can trace his or her heritage to one of three principal
extended families-the Ikui, Itinalah, and Achaff. Marriages can be contracted
within and among these families and each offspring takes his or her lineage
identity from the mother's family. Landed properties in Kom are shared within
the extended families and are jealously preserved and protected from gener-
ation to generation. This complicates the land tenure system (Sama, 1989).





Area: 475,000km2
Population: 9,933,823 (1985)





NIGERIA


ADAMAOUA O
NORD-OUEST
-Kom
CENTRAL
GUEST \ AFRICAN
ouD REPUBLIC


EST


SUD

EQUATORIAL CONGO
GUINEA CONGO
Figure 1. Map of Cameroon showing provinces and the location of Kom
in the North West Province.


Journal for Farming Systems Research-Extension






ROLE OF WOMEN IN KOM


Statement of the Problem and Objectives of Research
Women in general, and African women in particular, play an important role
in food production and in the nutrition and health of the family. Cloud (1986)
pointed out that the access of women to resources such as land, seeds,
fertilizers, technology, credit, and information greatly affects their productiv-
ity. For example, access to appropriate household technology (related to
milling and cooking) can greatly facilitate and improve household efficiency.
Nevertheless, Moock (1986) emphasized that the appropriateness of new
technology depends on who performs what task, controls what resources, and
performs what kinds of family responsibilities. From a recent study in
Cameroon, Koons (1988) concluded that women benefited little and received
less attention than men from the extension staff of the North West Develop-
ment Authority (MIDENO) in the North West Province because of social,
cultural, economic, attitudinal, and circumstantial differences between men
and women. Kom is one of the regions covered by MIDENO.
This region is threatened with acute food shortage resulting from the
reduction of its corn fields by the expanding population and the extension of
perennial crop production (notably coffee and plantains) into the corn fields.
Additionally, food management practices and the corn-based eating habits of
the natives result in frequent and disturbing hungry seasons usually signalled
by the shortage of corn. Potential solutions to this problem lie in the hands
of women because they play a dominant role in household food production.
The objectives of this paper are: (1) to reveal the unrecognized socioeco-
nomic contribution of women in the Kom economy; (2) to identify the
constraints imposed on Kom women by the physical and sociocultural
environment; and (3) to suggest ways to reduce some of these constraints.

Methodology
The Kom are divided into three geographical zones commonly referred to
as the Njinikom, Belo, and Abassakom valleys. Each of these valleys is made
up ofvillages, which in turn are made up of smaller units called quarters. From
a list of the 57 quarters that constitute the Kom ethnic group, 25 were
randomly sampled in order to represent proportionately the three valleys
based on population and number of quarters. Consequently, nine quarters
were sampled from the Belo valley and eight each from the Njinikom and
Abassakom valleys.
From each quarter, a random sample of five households was made and the
head or representative of each household interviewed with the aid of a
structured questionnaire. Two of the five respondents from each quarter were
women. Of the 125 households sampled, 112 representatives were reliably
interviewed-70 men and 42 women. The questionnaires were pretested and
administered by the heads of the agricultural posts in the three valleys after a
short training.


Vol. 4, No. 2, 1994





SAMA


Supplementary general information was obtained by the author from a
panel of 10 volunteer farmers from one of the villages and from the agricultural
delegate (the official representative of the Minister of Agriculture) and other
technical staff of the agricultural delegation. Additional information on the
culture and practices of the area was also gathered from the head of each of the
sampled quarters. The personal experience of the author himself, a Kom native
familiar with its problems and culture, constituted an additional source of
information for the study.
The data were summarized into tables and simple averages and/or frequen-
cies expressed in percentages were used in the analysis.


RESULTS


The Farming System of the Kom People
Land tenure and use. The traditional ruler of Kom (the Fon) is the
custodian of all land, which is shared among the villages and subsequently
among the extended families. Over 90 percent of all individual land acquisi-
tions are through family gifts. The remainder is acquired through borrowing
and gifts from friends, quarter heads, and the Fon. The sale of land is restricted.
Parts of the Kom territory are grazed by cattle owned by Fulani nomads, and
some of the steep and rocky hills are unsuitable for agriculture. The remaining
land is used for agriculture and buildings.
Settlement pattern. Settlement is in homesteads, groupings of which
constitute quarters, which in turn constitute villages. On the average, each
household has a homestead of 1.5 hectares occupied by 15 persons. The
household is the basic production and consumption unit and is mostly
polygamous. The family head has on average two wives and 12 children. Each
household also has on average three food-crop plots with a total area ofabout
1.6 hectares, excluding the homestead. The food-crop plots are dispersed at
distances that average about 4.5 km from the homestead.
Cropping systems. Both crops and livestock are raised in the homesteads,
while staple food crops are grown in the cornfields located away from the
homesteads. The following cropping systems are dominant:
(1) A coffee-based, mixed cropping system in the homesteads, where coffee
(the principal crop), plantains, and a few fruit trees are grown in association.
In the older homesteads, one can also find crops such as maize, coco yams, and
pineapples grown under coffee and other tree crops. In the new homesteads,
other arable crops such as beans, sweet potatoes, cassava, pumpkin, and
assorted vegetables are grown in mixture under the tree crops.
(2) A corn-based, mixed cropping system in the cornfields where corn (the
principal crop) is mix-cropped with beans, groundnut, pumpkin, yams, coco
yams, and vegetables in various combinations and intensities. The cornfields,


Journal for Farming Systems Research-Extension






ROLE OF WOMEN IN KOM


which are located in the valleys, hill-slopes, and plains, are shared by several
households. They are divided into plots of less than halfofa hectare per farmer.
Each cornfield is cultivated at the same time by those who share it, and then
left to fallow for usually two to three years. The length of the fallow period is
rapidly decreasing as the population expands. Two crops of maize-early and
late maize-are grown per year.
Current changes in the settlement pattern are affecting the cropping
systems. Homestead sizes are decreasing as the population increases, and
buildings are increasingly encroaching on the agricultural land within the
homesteads. This decreases the agricultural production potential of the
homesteads. Also, new homesteads encroach on the cornfields as the villages
expand, diminishing the area available for arable crops.
Sources and uses of household revenue. Traditional livestock production
constitutes an important source of household revenue. It involves the raising
of goats, sheep, and poultry within the homestead. Although such animals are
kept mostly for household consumption and for death, birth, and marriage
celebrations, occasionally some are sold to provide needed income. Other
household income is raised from the sale of coffee, plantains, fruits (mangoes,
pears, and kola nuts), and handicrafts, and from trade. Occasionally, staple
food crops are sold when perceived to be in excess of household needs.
Household income is spent on farm inputs, food supplies, health, educa-
tion of children, traditional ceremonies and celebrations, clothing, bride
prices, and other household needs.

Women's Contribution to the Socioeconomic Life of their Households
The gendered division of labor in the household. The household obligations
that the spouses expect from each other are presented in Table 1. It is clear
from the table that both spouses recognize child care, food production, and
cooking as the duties of a wife to her husband and to the household. This was
indicated by more than 50 percent of the respondents of either sex. The other
obligations were recognized by less than 20 percent of the men and 10 percent

Table 1. Wives' obligations to husbands as perceived by respondents.
SUGGESTED OBLIGATIONS PERCENTAGE OF AFFIRMATIVE RESPONSES
MEN WOMEN
Child care 57.5 62.9
Food production 60.0 52.9
Cooking 65.7 59.5
Housekeeping 40.0 19.0
Financial assistance 11.4 44.8
Assistance during sickness 7.1 7.1
Bearing children 15.7 4.8
Source: Survey data.


Vol. 4, No. 2, 1994






SAMA


Table 2. Husbands' obligations to wives as perceived by respondents.
SUGGESTED OBLIGATIONS PERCENTAGE OF AFFIRMATIVE RESPONSES
MEN WOMEN
Education of children 47.1 47.6
General care of household 44.3 40.6
Assistance during sickness 54.3 54.8
Provision of shelter 17.1 27.4
Food production 12.8 16.7
Financial assistance 20.0 27.4
Reproduction 12.8 4.8
Source: Survey data.

of the women. The housekeeping role of women was controversial. Whereas
40 percent of the men expected this service from their wives, only 19 percent
of the women accepted it as a woman's obligation to her husband.
In contrast, Table 2 shows the expected obligations of husbands to their wives.
The table confirms that food production in this community is the affair ofwomen.
Only 13 percent of the men and 17 percent of the women considered men as food
producers. From Tables 1 and 2, it is curious to observe that over 54 percent of
both sexes expected husbands to assist sick wives but only seven percent expected
the same obligation from wives to their husbands. Also, Table 1 shows that 45
percent of women respondents as opposed to 11 percent of the men felt that
women should render financial assistance to their husbands, whereas less than 25
percent of both sexes feel that the husband should render financial assistance to his
wife. This could be explained by the fact that women are not traditionally expected
to own and use money and therefore should not be expected to receive money or
be responsible for any transaction or obligation that involves the use ofmoney. The
Kom traditionally frown on the idea of women earning cash income and/or taking
loans. Over 85 percent of the respondents indicated that Kom people are against
cash crop production or the undertaking of any income-earning activities by
women. Men control the finances of the household and are therefore concerned
mostly with finance-related responsibilities. Food production as the dominant role
of the women is evident from the distribution of farming tasks among the
household members. These are presented in Table 3. The table confirms that men
are mostly active in income-generating activities. They are involved mostly in
coffee, plantains, and livestock production. All other agricultural production
activities of the household are the responsibility of women. The women also
actively assist their husbands in the tedious and labor-intensive aspects of male-
dominated production activities.
Participation of children (boys and girls between 10 and 15 years old) in
household tasks is also gender oriented. Judging from the type of activities they
perform (Table 3), male children tend to assist their fathers whereas female
children assist their mothers. It is also clear from Table 3 that parents rarely engage
their children in activities such as the sale of crops and livestock that involve the
handling of money.


Journal for Farming Systems Research-Extension






ROLE OF WOMEN IN KOM


Table 3. Farming activities and their executors.
ACTIVITIES ExECUTORS AND IDENTIFICATION FREQUENCY (%)
MEN WOMEN BOYs GIRLS
Farm clearing 86.5 5.4 48.2 0.8
Planting coffee 91.0 8.9 25.9 1.7
Planting plantains 86.6 11.5 25.0 3.6
Ridging 8.8 88.5 3.3 35.7
Planting food crops 25.8 85.7 13.3 33.0
Weeding coffee 78.6 87.4 17.8 21.4
Weeding food crops 30.0 58.0 13.3 25.9
Harvesting coffee 80.3 86.6 46.4 43.9
Harvesting plantains 87.2 46.5 30.4 22.3
Harvesting food crops 32.9 78.5 19.6 27.7
Selling cash crops 85.2 11.0 0.8 0.8
Selling food crops 28.5 70.5 4.5 6.3
Selling livestock 88.4 36.7 0.8 0.0
Raising livestock 86.2 40.0 0.8 0.0
Source: Survey data.

Women's contribution to household income. At first sight, the direct contri-
bution of women to household cash income seems relatively small. Almost all
the household income is farm generated. From Table 4, of the 192,980 francs
average farm income, the contribution from women-controlled crops is about
16 percent. However, this does not mean that the remainder is the contribu-
tion of men. Women also greatly contribute in the production of the male-
controlled farm enterprises. This is true especially for the coffee, plantains, and
livestock enterprises, which provide over 75 percent of the farm income.
The non-cash contribution of women to the total well being of the
household completely overwhelms that of the men, judging from the time
they put into farm work. It is estimated that each woman puts in an average
of 1,643 hours offarm work annually (at the rate of 7.9 hours/day and 4 days/
week), as against the man's 1264 hours (at the rate of 5.4 hours/day and 4.5
days/week). Moreover, the man's work is usually restricted to the supervision
of the other members of the household.
Cash income from food crops (Table 4) represents the value of the surpluses
sold after meeting the household consumption requirements. If the total
production of these crops were valued in monetary terms, the household's
total annual cash income could more than double the present estimates, and
the women's contribution in that case would increase immensely.
Other contributions to household welfare. Although the education of chil-
dren is purported to be the concern of their fathers, this obligation is shown
to be shared by their mothers. The women are responsible for the sociocultural
and moral education of their children. The men do, however, play the major
role in the formal education of children. Even here, however, 52.4 percent of
the women participate in the payment of the children's school fees.


Vol. 4, No. 2, 1994






SAMA


Table 4. Average annual farm income by source.
SOURCE OF INCOME AMOUNT IN FCFA
Coffee 93,100
Plantains 12,050
Pears 8,540
Kola nuts 12,180
Mangoes 5,445
Vegetables* 5,445
Maize* 4,800
Coco yams* 8,525
Beans* 8,245
Irish potatoes* 3,860
Fowl 8,030
Goats/sheep 22,760
Total 192,980
Source: Survey data.
*Food crop.

The women are also very active in every aspect of family life. In fact, virtually
their whole day is occupied and devoted to household activities: cooking and
child care in the morning, trekking long distances to work and/or fetch food
in the food-crop fields, and cooking and childcare in the evenings. This makes
women especially indispensable members of households. Actually, household
functioning depends more on women than on men. This is confirmed by the
fact that over 50 percent of the women expressed confidence in their ability
to manage the household in the absence of their husbands, while only 46
percent of men believed they could do the same in the absence of their wives.

Constraints Faced by Kom Women
In the performance of their numerous, demanding, and all-important
obligations, Kom women are seriously constrained by lack of education and
resources, and by sociocultural and technological factors that significantly
reduce their effectiveness and retard the development of the household,
community, and nation.
Educational constraints. Only 38 percent of the respondents were literate,
and the majority of these respondents were men. In the past, the education of
Kom females was considered uneconomical. At the time of marriage girls fetch
bride prices for their parents and exit the household. Today, some parents still
consider investment in their female children as an unnecessary cost that yields
no future returns.
Although today the majority of the girls attend school, priority is placed on
the education of boys. The education of children renders them employable in
the urban areas and therefore speeds up rural-to-urban migration. As a
consequence, the average age of agricultural producers has been increasing.
This means that the health, nutrition, child care, and food production
responsibilities of the Kom community are being left in the hands of an older

Journal for Farming Systems Research-Extension






ROLE OF WOMEN IN KOM 23
Table 5. Sources of land by sex of recipient.
SOURCE RESPONSE FREQUENCY (%)
MALE RECIPIENTS FEMALE RECIPIENTS
Father 93.8 15.0
Husband 0.0 40.2
Quarter head 29.5 5.4
Village head 4.5 4.5
Fon (chief) 14.3 7.1
Uncle 33.0 14.3
Mother 2.7 33.0
Friend 3.4 5.4
Source: Survey data.

and less-educated fraction of the rural population. Old age and illiteracy are
known to work against efficient use of modern technologies in production,
household nutrition, health, and child care. Among the Kom, therefore, the
results of low investment in female children in the past is affecting agricultural
production today. At the same time, increased access to education tends to
encourage the migration of youths to urban centers. It could be concluded
that failure to invest in female education in the past seriously constrains the
productivity of Kom women today, and that increased investment in female
education today has similar effects on productivity today and in the future
because of the youth migration incentive.
Resource constraints. Land, labor, and capital, the traditional production
resources, are either in short supply or available under very difficult conditions
to Kom women. Food-crop land is scarce, fragmented, and dispersed, and its
possession and use are uncertain. As indicated earlier, each household has an
average of 1.6 hectares of food-crop land divided into several tiny plots located
at average distances of 4.5 km from the homesteads. The hilly topography
renders daily trekking difficult and slow, and further reduces the time and
energy available to women for farm work.
The farm land available to the individual or household is fragmented and
dispersed because the individual acquires it from several persons whose lands
are also geographically dispersed (Table 5). Because most land is family
property, the individual cannot undertake any meaningful or permanent
development on it. This is a drawback to modern agricultural development in
the area. Table 5 also brings out the current land acquisition trends operating
in Kom. Fathers generally provide land for their sons, who alternatively may
obtain family land from their uncles. Husbands usually acquire and make
available farm land to their wives. Village or quarter heads may provide both
building and farm land to those male children of the village or quarter whose
parents do not have enough land, and female children may acquire some of
their farm land by sharing land with their mothers.


Vol. 4, No. 2, 1994






SAMA


Out of the 15 members ofan average household, eight constitute the active
labor force and include the husband, two wives, and five children. The man's
role in food production is mostly limited to farm clearing (the initial stage of
land preparation for planting), and he is aided in this task by the male children.
The rest of the food production labor is supplied by the women. Farm labor
is further reduced when the female children attend school and also by the
women's other household tasks.
Additionally, as discussed above, women have no easy access to the money
necessary to pay for labor and other production inputs. Women must
therefore depend on their own labor and the use of traditional production
methods that require little capital. As a consequence, efficient tools and
modern inputs are not available to support improvement in productivity.
Technological constraints. The author's personal knowledge and under-
standing of the situation in Kom is consistent with Koons' (1988) conclusion
that sociocultural, economic, and attitudinal conditions work against wom-
en's ability to benefit equitably from MIDENO's extension staff in the North
West Province of Cameroon. However, this study did not attempt to verify
whether women equitably receive such benefits. The facts are that women are
less literate than men and that the extension agents carrying the extension
packages are mostly men. Also, these agents are sometimes handicapped in
their ability to communicate by language barrier. Most of the women
understand only the Kom language, whereas the non-Kom extension agents
use either English or pidgin English, understood by most of the men.
Furthermore, most of the modern technology to be delivered involves a cost
that the women, given their disadvantaged position in the household, cannot
directly bear. Consequently, they have little access to improved farming
methods.
The absence of home economics extension agents and of modern kitchen
equipment deprives the women of the technology they so desperately need for
child care, family health, and household nutrition. The absence of modern
technology in the household means that life within the household is poor and
tedious. Also, the absence of good farm-to-market roads and oftransportation
to facilitate the daily treks to the farms and the markets further robs the
women of much needed time and energy for farming.


CONCLUSION AND RECOMMENDATIONS
It is essential to recognize the economic importance of women in Kom society
and to take steps to enhance their productivity in agriculture and other
household activities. The men and women of the Kom community, as well as
policy makers at all levels, need to focus attention on the total activities of the
farm households and to discuss needed steps publicly. A fuller appreciation of
the constraints women face in performing their responsibilities will help


Journal for Farming Systems Research-Extension






ROLE OF WOMEN IN KOM


participants focus on what is needed to begin relaxing these constraints in
order to make women even more effective in their many important roles. The
increased efficiency of women will not only increase the productivity of the
economy but will also enrich the material well-being of families in the rural
sector.
Several items need explicit attention to improve the effectiveness and
quality of life in rural households. These include: (1) greater access to electrical
power and clean water, (2) improved farm-to-market roads, (3) increased and
improved rural schools with smaller class sizes, (4) intensified rural extension
services employing more women extension agents with greater focus on the
productivity of women in their household and farm production activities, and
(5) improved inputs (including tools and equipment) and farming systems to
facilitate both food and income production for farm families.
An improved quality of life in the rural sector will also provide greater
encouragement for rural youth to remain in rural areas and engage in
agriculture and other production activities. Such encouragement is very
important for the long-term economic well-being of the rural sector in
Cameroon.


ACKNOWLEDGMENTS

The author sincerely thanks Dr. Max Langham of the Food and Resources
Department of the University of Florida for editing a draft of this paper and
for the useful comments and suggestions he made to improve its quality.


REFERENCES
Gladwin, C.H., and D. McMillan. 1989. Economic Development and Cultural Change
37(2):346-369.
Boserup, E. 1970. Women's role in economic development. New York: St. Martin's Press.
Cloud, K. 1986. Women's productivity in agricultural households: How can we think
about it? Pages 11-18 in Women as food producers in developing countries. Africa
Studies Center, University of California at Los Angeles, OEF International, Los
Angeles.
Koons, A.S. 1988. Being rural women in the North West Province: A presentation of
more ways in which women are to men. Paper presented at the Conference on
Development in Cameroon: The role of food and agriculture. University of Florida.
Gainesville, April 7-9.
Moock, J.L. 1986. Understanding Africa's rural households and farming systems. Boul-
der: Westview Press.
Sama, J.N. 1989. A socio-economic survey of agricultural related problems in the North
West Province of Cameroon. Department of Agricultural Economics, Dschang
University Center. Mimeo.


Vol. 4, No. 2, 1994








Analysis of the Competition for Labor by
Dryland and Irrigated Crops:
The Case of Rice and Millet in Niger

Ziyou Tu, Robert Deuson, Eric Bomans, and Jess Lowenberg-DeBoer



ABSTRACT
The efforts to develop rice production in the irrigated perimeters along
the Niger River have not lived up to expectations. A key problem is that
farmers appear to neglect rice production in favor of allocating labor to
dryland crops, especially millet, particularly when millet weeding and rice
transplanting occur simultaneously. A hypothesis that considers the
marginal value product of labor as the main determinant of labor
allocation is tested. The production functions of rice and millet are
estimated individually with ordinary least squares from the farm survey
data collected in the rainy season of 1985 from five villages. The results
indicate that the marginal product of weeding millet is not higher than
that of transplanting rice. Four examples are used to illustrate the
structure ofthe agronomic and economic tradeoffs in the labor allocation
decision. These examples indicate thatin order to achieve high total grain
yields and revenues, rice must be transplanted as early as possible at the
beginning of the rainy period. Because of the high demand for labor in
a short period, it would be difficult to achieve the highest rice yields with
only family labor. The data support the hypothesis that Nigerien farmers
allocate their labor to maximize millet yields, rather than to maximize
revenue. The loss of revenue linked to this focus on millet production
appears to be small.


INTRODUCTION
Rice production in the irrigated perimeters along the Niger River is an
important part of the food self-sufficiency and agricultural development plan
of the government of Niger (Rep. du Niger, 1987). But in spite of large
investments in infrastructure, research, and extension, the production on
these perimeters has not lived up to expectations. One of the key problems has
been the competition for labor between rainfed and irrigated crops (Anders,
1 Ziyou Yu is a graduate student, and Robert Deuson and Jess Lowenberg-DeBoer are assistant
professors in the Department of Agricultural Economics, Purdue University. Eric Bomans is
agronomist and team leader of the SASEAC Project in Burundi.






YU, ET AL.


et al.,1984; Maikor6ma,1986; Rassas and Loute,1989; Rep. du Niger,1989a).
During the rainy season, farmers are observed to neglect rice production in
favor of allocating labor to dryland crops, primarily millet. The hypothesis that
is tested in this analysis is that millet production is favored because the
marginal value product of labor is higher for millet than it is for rice in the
critical periods. The analysis uses production functions estimated with ordi-
nary least squares (OLS) from farm survey data. The results are useful to policy
makers, researchers, and extension workers concerned with the Niger River
region and similar areas in West Africa.
The Republic of Niger is a landlocked country in the Sahelian region of
West Africa. It shares borders with Nigeria and Benin on the south, Chad on
the east, Burkina Faso and Mali on the west, and Algeria and Libya on the
north. More than six percent of Niger is in the Sahara Desert zone with less
than 200 mm of rain per year. Most of the country's agricultural land is in a
narrow band within 150 km of the Nigerian border. The average annual
rainfall in this band is in the 300-800 mm range. Millet is planted on over seven
percent of the dryland crop area, often intercropped with cowpeas and
sorghum. The average millet yield in the 1980s was 409 kg/ha (Rep. du
Niger, 1989b).
Currently, less than one percent of the crop area of Niger is irrigated. The
predominant irrigated crop is rice, with smaller areas devoted to cotton,
sorghum, onions, and other crops. Two rice crops are produced per year with
the crop calendar determined by climate and river flow. A rainy season crop
is transplanted in July or August and harvested in November or December. A
dry season crop is transplanted in December or January and harvested in April
or May. In 1988 irrigated rice occupied about 5000 ha in each season (Rep.
du Niger, 1989a), but this area is expected to rise rapidly with the construction
of new perimeters. Potentially 140,000 ha could be irrigated in the rice
growing area of the Niger River valley (Rep. du Niger, 1989a). Average yields
for irrigated rice have been about 4000 kg/ha in the 1980s (Rassas and
Loutte, 1989), substantially less than the average of 6000 kg/ha commonly
observed in Asia.
Farmers in Niger also grow about 12,000 ha of traditional rice varieties in
naturally occurring swamps and depressions without the benefit of controlled
irrigation. Traditional rice can be very profitable, but the naturally suitable
areas are limited and, because water is not controlled, crop failure is common.
Because of the small cultivated area and simple labor methods, there is
relatively little competition for labor between traditional and irrigated rice
(Bomans, 1986). This article does not concern itself with traditional rice-
growing practices.
Currently, rice makes up about four percent of total cereals production of
Niger, but it is more important in the national strategy than the level of
production would indicate. Because rice yields can be good even in drought
years when the millet crop fails, irrigated rice has the potential of stabilizing


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


the national grain supply. In addition, domestically produced rice can help
satisfy the growing demand for rice in urban areas, and at the same time
provide a source of cash income for farmers. Domestically produced rice now
satisfies only about 50 percent of the demand in Niger (Rep. du Niger,
1989a).
The competition for labor affects rice yields in several ways. In the rainy
season, late planting of nurseries and transplanting of seedlings can push the
rice maturity period into the cool season when night temperatures can drop
as low as five degrees centigrade. The low temperatures cause rice grains to
abort. In some cases, rice seedlings are left in the nursery too long before
transplanting, which reduces tillering and hence yield. Weeding may be done
late and incompletely, resulting in heavy weed pressure. An indirect effect of
labor competition is that farmers choose lower yielding but flexible rice
varieties, instead of the higher yielding varieties that are more sensitive to
transplanting date, nursery time, and weed competition.
Alternative hypotheses about why farmers allocate labor to millet instead
of rice focus on food preferences, relative profitability, and risk (Anders, et al.,
1984; Maikor6ma, 1986; Rassas and Loutte, 1989). Relatively little of the
irrigated rice is consumed by farm families. Millet is the staple food of their
diet. Rice is consumed mainly on holidays and ceremonial occasions, such as
weddings. For these events, the traditional rice varieties are preferred.
Irrigated rice is often treated as a cash crop produced for urban consumers. In
this situation, if the farmer's primary objective is to secure the family millet
supply, with cash income as a secondary goal, labor would be allocated first to
millet and only to irrigated rice when subsistence needs for millet are satisfied.
Still, several budgeting studies have shown rice to be moderately more
profitable than millet (Anders, et al., 1984; Bomans, 1986; Maikor6ma, 1986;
Rassas and Loutte, 1989), but it is not clear that the additional profits are
sufficient to compensate for the extra financial and marketing risk. With
irrigated rice, yield risk is relatively low, but financial risk is high because of the
need for purchased inputs. Traditionally, millet production in Niger uses few
purchased inputs, so the financial risk is negligible. Rice is purchased by the
parastatal company Riz du Niger (RINI) at official prices that are often higher
than world market prices, but RINI does not buy all of the production. About
one-third of the irrigated rice production must be sold through private
marketing channels that are not well developed for this product. In contrast,
marketing channels for millet and the traditional rice varieties are well
developed. Thus, we argue that farmers allocate labor to millet production
because rice is not sufficiently more profitable than dryland crops to justify risk
and marketing problems.


Vol. 4, No. 2, 1994






YU, ET AL.


DATA AND METHODS

The data for this study were collected during the 1985 rainy season in five
villages on the east bank of the Niger River, about 50 km north of the capital
city of Niamey. They are: Koutoukal6 Kado, Koutoukal6 Kourthey, Kout-
oukal6 Tegui, Koutoukale Zeno, and Zamakwara Zeno. The Koutoukal6
perimeter has been in operation since 1966. These villages were chosen on the
basis of the following criteria: (1) isolation from main urban centers, (2)
experience with irrigated agriculture to avoid the "learning curve" bias
common with introduction of a new technique, and (3) relative ease of access,
to permit regular and frequent supervisory visits.

Table 1. Land, labor, crop mix, dates of field operations, yields, costs and returns,
Koutoukalk perimeter, Niger, rainy season, 1985.

ITEM AND UNITS MEAN1 MINIMUM MAXIMUM


FAMILY SIZE:
Men 1.90
Women 2.10
Boys 2.30
Girls 2.10
Elderly 0.30
LAND PER FARM:
Monocrop millet, ha 2.20
Millet intercrop, ha 1.12
Other dryland crops, ha 0.20
Irrigated rice, ha 0.50
YIELDS:
Monocrop millet, kg/ha 367.80
Irrigated rice, kg/ha 3,789.90
CASH COSTS FOR IRRIGATED RICE:
Water use, FCFA/ha 58,171.00
Fertilizer, FCFA/ha 15,059.80
Hired labor, FCFA/ha 7,618.30
TOTAL LABOR USE:
Monocrop millet, PDE/ha 63.10
Irrigated rice, PDE/ha 364.60
GROSS RETURNS PER HECTARE:
Millet, FCFA/ha 29,560.50
Rice, FCFA/ha 373,858.00
GROSS RETURNS PER DAY:
Millet, FCFA/ha 782.50
Rice, FCFA/ha 2,452.10


3.00
3.20
3.10
3.00
1.00


0.60
0.30
1.00
1.00
0.00

0.10
0.09
0.02
0.03


104.40
2,343.70

58,171.00
0.00
1,250.00

8.20
206.90


990.10
5,207.30

58,171.00
26,556.30
39,750.00

296.40
812.80


2,077.70 171,053.00
108,355.10 272,727.20


49.80
706.60


2,795.50
8,635.50


Journal for Farming Systems Research-Extension


1 The arithmetic means reported here may differ from the geometric means reported in Table 4.
2 Gross returns were calculated by Bomans, 1986. Costs include hired labor, fertilizer, and
cooperative charges, which include water use and rice seed. The opportunity cost of family labor
is not subtracted. The money unit is measured in Francs CFA (FCFA). The value of the FCFA
at the time of this analysis was about 300 FCFA per US$1.






RICE AND MILLET IN NIGER


The sample size was 50 farmers, with a total of 1890 hectares in 272 fields.
In each village, 20 volunteers were identified and a sample of 10 farmers per
village was chosen at random from among the volunteers. A completely
random sample was not used because experience in the study area indicated
that farmers forced to participate in such a study provide data of dubious
quality. Nevertheless, this approach yielded a very complete data set. Out of
the 516 farmer-months of data, only 10 farmer-months of data were lost, and
that was due to enumerator error. One field enumerator was assigned to each
village.
All data collected were identified by field and farm. The data include both
physical inputs and output, and some accounting data ofexpenses. Labor time
records include working hours, type of worker, operation, and salaries for
hired labor. Worker types include family labor, hired labor, and community
labor. Labor is divided into three groups: adult male, adult female, and child.
The worker type "communal labor" is used for unpaid labor provided by
friends and neighbors.
The descriptive statistics (Table 1) show a farming system in which the
majority of the crop area is devoted to millet and other dryland crops, but the
majority of the labor use, grain production, and income comes from the
irrigated area. Only 12 percent of the farm area is in irrigated rice, but it uses
65 percent of the labor, yields 62 percent ofthe grain, and produces 69 percent
of the total value of crops.
Most of the work on both dryland crops and rice is done by family labor.
Overall, about 16 percent of all labor is hired. About 62 percent of all hired
labor is used in rice production, with about 72 percent of all hired labor for
rice production being used in harvest and postharvest operations. Most of the
hired labor for dryland crops is used in millet weeding. Except for millet
planting and rice threshing, women seldom do field work at Koutoukal6 .
The planting, transplanting, and weeding date statistics (Table 2) show that
the periods for these operations overlap. The labor bottleneck that causes the
most problems is in late July and early August, when both millet weeding and
rice transplanting should be done (Bomans, 1986; Maikor6ma, 1986). The
date of millet planting is determined primarily by the onset of rains. Though
millet planting and replanting extends over a long period, most of the millet
area is planted for the first time within a few days of the onset of the rains, and
thus the conflict with rice operations is relatively small. Rice weeding is done
mainly during a period before millet harvest, but after millet weeding.
It can be assumed that all farmers in the sample use the same technologies,
though with varying input levels. They all used the traditional millet produc-
tion system. All large perimeters in Niger are managed by ONAHA (Office
National des Am6nagements Hydro-Agricoles), which prescribes recom-
mended production practices. Though these recommendations are not fol-
lowed to the letter, they create a certain consistency in rice production
technology among farmers, for instance, in variety choice.


Vol. 4, No. 2, 1994





YU, ET AL.


Table 2. Critical periods in rainy season rice and millet production, Koutoukal, Ni-
ger, 1985.
OPERATION JUNE JULY AUG. SEPT. OCT.

Millet planting and replanting ... June 18-Aug. 9...
mean = June 29
Millet first weeding ...July 8-Aug. 13...
mean = July 28
Rice transplanting ...July 24-Sept. 20...
mean = Aug. 22
Rice weeding ...Aug. ?7-Oct. 24...
mean = Sept. 24
1 The mean reported here is the weighted average in which the weights are the proportions of the
total labor allocated to an operation expended on a given day.

The Model
A production function approach was chosen as the simplest means to test
the hypothesis posed. The popular dual approach was not used because the
hypothesis concerns the magnitude of the marginal value of the physical
production response to labor allocation and because the available data were
primarily in the form of physical inputs and outputs. The dual approach would
estimate the production function characteristics implied by the structure of
monetary costs and returns found in accounting data. In general, relatively
good data on physical inputs and output can be collected in Niger. Accounting
data on expenses and returns are usually less reliable than physical input and
output data. The primary input, family and community labor, is hard to value.
A large part of the production is consumed at home. Farmers are reluctant to
discuss their cash transactions. Also, the dual approach often entails multi-
equation estimation.
A programing model analysis would be a useful approach, but it would not
be the simplest method to test the hypothesis. Development of a representa-
tive farm programing model would have required substantially more time,
data, and expertise than were available. A complete analysis of the manage-
ment effects of the production response to labor timing would require a
representative farm programing model. Estimation of the yield responses to
various inputs, as demonstrated in this study, is a step toward a more complete
analysis.
The decision framework for this analysis assumes the maximization of
expected results in terms of food value or monetary value. It is commonly
observed that farmers in Niger want both food self-sufficiency and cash
income; thus, both aspects are considered. In mathematical terms, the
problems are identical. In one case, the inputs and outputs are measured in
terms of food value (kilograms of grain, calories, etc.), and in the other case
they are measured in terms of money.


Journalfor Farming Systems Research-Extension






RICE AND MILLET IN NIGER


Zellner, et al. (1966) show that simple, single-equation production func-
tion estimates with survey data can be consistent and unbiased if it can be
assumed that the production process is risky and that this fact is recognized in
the choice of input levels-assumptions that seem to hold for Niger.
Two equations are used to describe rice and millet production separately.
The assumption is that the production of rice and millet are statistically
independent while economically related. Statistical independence of the two
crops is plausible because they are grown in substantially different conditions.
Rice is irrigated and millet is rainfed. Rice is grown on clay soil near the river
and millet on sandy uplands, even though the same insects and diseases do not
affect both crops. Labor allocation also affects both production systems in the
busy season. Particularly, in July/August, rice transplanting and millet
weeding are crossed. Allocating more labor to rice transplanting must reduce
the labor used in millet weeding. However, once the labor is allocated in the
rice production, statistically, it will not have an effect in millet production.

Functional Form
A Cobb-Douglas production function is used for simplicity and to reduce
the number of parameters to be estimated. This functional form is used as a
convenient summary representation of the production process. Real produc-
tion systems are, of course, much more complicated. The form is assumed to
be adequate for the range of data observed, but not to represent the entire
possible range of input levels. The properties of this functional form are well
known and have been discussed in detail by Heady and Dillon (1961), and by
Beattie and Taylor (1985). All data are transformed to natural logarithms;
thus linear estimation using OLS is possible. The estimated equations were:
YRi = alphao + alpha,*RHAi + alpha2*RLABi + alpha3*FERi +
alpha,*RPDATEi + alpha,*WDATE, + epsilon r,
YMj = betao + beta,*MHA + beta2*MLAB + beta3*MPDATE. +
beta4*WDATEJ + epsilon m,
where:
YRi is the rice yield (kg/ha) from field i,
YM. is the millet yield (kg/ha) from field j,
RHAi is the area (ha) of rice field i,
MHA is the area (ha) of millet field j,
RLABi is the total labor (PDE) used in rice field i,
MLABj is the total labor (PDE) used in rice field j,
FERi is the total fertilizer expense for rice field i,
RPDATEi is the rice transplanting date for field i,
MPDATE. is the millet planting date for field j,
RWDATEi is the date of first weeding for rice field i, and
MWDATE, is the date of first weeding for millet field j,
epsilon r, and epsilon m, are error terms for the rice and millet functions,
respectively.


Vol. 4, No. 2, 1994






YU, ET AL.


The total labor per field is the sum of family, hired, and communal labors
in units of PDE (person-day equivalent). The PDE is defined as an adult male
working eight hours per day. An adult female working eight hours is counted
as 0.85 PDE and a child as 0.65 PDE.
The weighted average date (WAD) of a field operation is used as a single
valued index of the timeliness of that operation. The point of reference for
calculating the WAD is the earliest date for a given operation in the data set.
The earliest date is counted as day 1 of the period suitable for that operation.
The weights used in calculating the WAD are the proportions of total labor
for that operation used on a given day. For example, if the earliest rice
transplanting date in the data set is July 24 and the farmer started transplanting
the rice on July 31, the first day of work is counted as day 8. If the field was
finished on August 3, that could be counted as day 11. If on July 31 and
August 2 and 3, one man and nine boys worked (6.85 PDE/day), the total
labor is 28.8 PDE. The weighted average date of transplanting is:
WAD = 8*(7.55/28.8) + (*(7.55/28.8) + 10*(6.85/28.8) + 11*(6.85/
28.8) = 9.45
In general, the calculation formula of weighted average date is:
WAD = Dx*L/TL + D2*L,/TL + .... + D *Ln/TL = E Dk*L/TL
TL=L1+L2 L =~ Lk
where:
Dk = the Julian date of kth operation day, k=l, 2, ..., n;
Lk = the labor used on kth day for a given operation;
TL = the total labor used for that operation.
The labor variables were chosen to embody the maximum information in
a parsimonious form. Based on the results reported by Maikor6ma (1986) and
Bomans (1986), completion of millet planting, rice transplanting, and weed-
ing of both crops were identified as the most important factors in timing of
field operations. Measuring the planting, transplanting, and weeding dates
from the earliest planting, transplanting, and weeding dates in the sample
controls for environmental conditions, such as the onset of rains, that
determine the cropping calendar in any given year. It is expected that the
estimated coefficient for total labor input (RLAB, MLAB) will be positive. It
is unlikely that farmers would use labor beyond the point of positive marginal
product, even if that labor is low cost. The planting, transplanting, and
weeding variables are expected to have negative signs. Agronomic reasoning
suggests that delays in these operations in the range of dates observed in the
sample will have negative effects on yield. Very early planting dates, which
could have negative effects because of high temperatures, are not observed in
the sample.
Other variables were chosen so as to control for as much of the remaining
variation as possible, given the data collected. The field size variable is
intended to capture management problems associated with labor utilization
and water distribution. Field observations suggest that both labor and water


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


management become more difficult with larger field sizes. Thus, the field size
coefficient is expected to have a negative sign for both millet and rice. The
fertilizer coefficient is expected to be positive. The fertilizer coefficient was
estimated only for rice because no fertilizer use was observed for millet.
Only the millet monocrop data were used in the estimation of the millet
production function. Monocrop millet occupied 65 percent of the total millet
area in the sample. Intercropping with sorghum, cowpeas, and sesame
introduces complicating factors and, unfortunately, the small number of
observations on each crop association (millet-cowpeas, millet-sorghum, mil-
let-sorghum-cowpeas, etc.) does not permit reliable estimates for these
combinations.

Interpretation
Analysis focuses on the millet weeding and rice transplanting coefficients
with respect to both date and labor usage. Other estimates will be considered
to the extent that they complete the labor allocation picture. The marginal
physical products (MPP) and marginal value products (MVP) are calculated.
The MPP can be interpreted as the yield or food value impact of a one-unit
change in a factor of production. Mathematically, the MPP is the first
derivative of the production function with respect to the factor. The MVP can
be interpreted as the monetary value of the yield change due to a one-unit
change in a factor. The MVP is equal to the MPP multiplied by the output
price. Consistent with the logic of the logarithmic transformation of the data,
the central tendency of the marginal products is calculated at the geometric
means instead of the arithmetic means (Heady and Dillon, 1961).
Accordingly, the MPPs and MVPs of millet weeding date and rice trans-
planting date can be directly observed from the first derivative of the
production function. They should be negative because in the MPP and MVP
calculation the denominator (expected yield or monetary value of the expect-
ed yield) will decrease as time passes and the numerator (the weighted average
date) increases. However, comparison of the MPPs and MVPs calculated on
a calendar date basis is of limited relevance because the amount of labor
represented by a one calendar day change in date differs between the
operations. Therefore, the MPPs and MVPs of labor used in millet weeding
and rice transplanting are calculated by dividing the MPPs and MVPs of millet
weeding and rice transplanting date by their respective labor requirements per
hectare of crop. In this study, these requirement are 12 PDE/ha for the first
weeding of millet and 59 PDE/ha for rice transplanting.
The MPP and MVP of millet weeding date and rice transplanting date are
used to measure the yield/ha or food value/ha change due to the change of
operation date such as delay (or advance) of one calendar day. The implication
of MPPs and MVPs of labor used in millet weeding and rice transplanting are
the crop yield or food value change due to the change of one unit labor (one
PDE) used in weeding or transplanting.


Vol. 4, No. 2, 1994






YU, ET AL.


In the typical one-crop profit maximization problem, the optimality
conditions require that the MVP of an input be equal to the cost of that input.
In the two-crop case, the additional requirement is added that the MVPs of
the two factors be equal. These typical conditions do not hold true with
respect to the millet weeding and rice transplanting date variables, because
they are expected to have a negative impact on yield; a larger date value results
in a lower yield. Mathematically, this means that the function is convex in the
transplanting and weeding dates, not concave as is required for maximization.
Hence, in this case, the equality of the MVP is a condition for a minimum.
The maximum for this type of problem often takes an all-or-nothing form
(corner solution). This type of solution is found interactively by checking
candidate solutions and choosing the maximum. For example, if during the
overlap period the marginal value product of the millet weeding date is
consistently higher than that of rice transplanting, a candidate for the
maximizing solution may be to allocate all labor to millet weeding until it is
finished and then start on rice transplanting. If the opposite situation occurs
and the rice transplanting marginal value product is higher, then a candidate
solution would be to allocate all labor to rice early in the overlap period and
weed the millet only after the rice is in the paddy.
These maximization results can be used in interpreting the magnitudes of
the estimated coefficients. For the variables with positive coefficients, such as
fertilizer and labor, average marginal products that are approximately equal to
the opportunity cost of the input indicate that input levels are such that they
maximize expected results. Given the tendency of the Nigerien farmer to
allocate labor to millet weeding instead of to rice transplanting, one would
expect to find that the marginal products of millet weeding are larger than
those office transplanting, ifmaximization of food or monetary value is indeed
the objective. If the magnitudes of the marginal products do not correspond
to the hypothesis, it may be that they reflect an optimization, but not of the
total food or monetary value. The alternative hypotheses suggest that the
objective may be either a maximization of millet production or a risk-adjusted
optimization.

RESULTS AND DISCUSSION
The estimation results are reported in Table 3. The F statistics are
significant at the five percent level for both crops. All of the estimated
coefficients have the expected signs, except for the coefficient ofthe total labor
used in rice production, and that coefficient is not significantly different from
zero, even at the 10 percent level. For rice, the coefficient estimates for field
size, fertilizer, and transplanting day are statistically significant at the five
percent level. For millet, field size and total labor use coefficients are
significant at the five percent level. The planting day and weeding day are
significant at the 10 percent level for millet. Because of the logarithmic


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


transformation, the estimated coefficients can be directly interpreted as
elasticities, that is, as the percentage change in yield for a one percent change
in the input level. For example, a one percent change in either the millet
weeding date or the rice transplanting date results in about a 0.14 percent
change in yield. Because of the higher rice yields, the per calendar day response
of rice to transplanting date appears more dramatic than that of the response
of millet to weeding date (Figure 1), but the percentage response is almost the
same for the two.
The average MPP and MVP of rice transplanting date are greater than the
average MPP and MVP of millet weeding date (Table 4). Each day translates
into a 17.8 kg/ha yield loss for rice, but only a 3.6 kg/ha loss for millet. In
monetary terms, this is about 1387 FCFA per calendar day for rice transplant-
ing and about 307 FCFA per day for millet weeding. These impacts reflect
only the change in yield and revenue due to labor allocation.
For both crops, labor appears to be near the point of zero marginal return.
In the case of rice, the coefficient is negative and not significantly different
from zero. For millet, the marginal physical product of about 1.5 kg per
hectare per day approaches the amount of grain needed to feed a laborer for
one day. The marginal value product of labor in millet is only 128 FCFA/day,
substantially less than the average wage rate for adult males in the area in 1985
of about 400 FCFA/day (Bomans, 1986).

Table 3. Elasticity estimates for rice and millet production functions in terms of yield
per hectare, Niger, 1985.

INDEPENDENT VARIABLE DEPENDENT VARIABLE
RICE MILLET
(T-STATISTIC) (T-STATISTIC)
Intercept 9.2099 5.6683
(16.373)** (11.333)**
Field size -0.2564 -0.1537
(-2.184)** (-2.078)**
Labor -0.1491 0.2511
(-1.448) (2.584)**
Fertilizer 0.0270
(2.352)**
Planting or transplanting date -0.1396 -0.2513
(-2.322)** (-1.722)*
Weeding date -0.277 -0.1389
(-0.499) (-1.612)*
F 2.80 6.822
MSE 0.038 0.2747
N 54 113
*Significance at the 0.10 level.
** Significance at the 0.05 level.


Vol. 4, No. 2, 1994






YU, ET AL.


It should be remembered, however, that for some of the family labor used
in millet production the opportunity cost may be lower than the adult male
wage. Women and children in the study area seldom work for wages, but some
evidence indicates that earnings from traditional activities, such as basket
making, may be in the range of 100 to 200 FCFA per day. The near zero
marginal return to labor at the mean input levels does not necessarily indicate
inefficient use of labor, but may simply be a reflection of the low marginal cost
of family labor.
The MPPs and MVPs of labor (per PDE) used in rice transplanting and
millet weeding were calculated by dividing the MPPs and MVPs by their labor
coefficients (59 PDE/ha for rice transplanting and 12 PDE/ha for millet
weeding). At the geometric means, the MMPs and the MVPs of millet
weeding are about equal (Table 5). If feasible dates are defined as those that
allow enough time to complete both tasks given the labor supply, for many
feasible combinations of weighted average millet weeding and rice transplant-
ing dates the absolute marginal value product of millet weeding is smaller than
that of rice transplanting (Figure 2). In Figure 2, the MVPs are increasing
(decreasing absolute value) and approaching zero as the operation date
becomes later. This is because of the relatively short optimal period for these
field operations. Biologically, after a certain date, the damage to the crop's
yield due to the delay of operation is already made. For example, the yield of
millet can be increased substantially by weeding on July 8 rather than 9, but
there is only a small difference between weeding on July 28 or July 29.



6

4


8-Jul 18-Jul 28-Jul 7-Aig 17-Aug 27-Aug 6-Sep 16-Sep
] rice O millet
Figure 1. Rice response to transplanting date and millet response to weeding date for
Koutoukal6, Niger, 1985, estimated with a Cobb-Douglas production function.


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


Feasible date combinations depend on the amount of labor available, the
work rate, and the area. In creating a mental image of the distribution of
weeding or transplanting dates, it is useful to assume that the labor per day in
a given activity is about equal; thus about half of the labor needed to
accomplish a task would have to occur before the weighted average date. For
example, ifan average farmer has 2.16 hectares ofmonocrop millet and it takes
12 PDE to weed one hectare, the weeding will require almost 26 days. If the
farmer has one PDE per day of labor available, millet weeding would have to
start about 13 days before the weighted average date. It should be noted that
the PDE concept assumes an eight-hour workday, but in fact the farmers at
Koutoukal often work four to five hours in the fields and spend the rest of
the day in household chores, taking care of livestock, and other activities. The
relatively short field work day can also be linked to poor health and nutrition.
Thus, the one PDE labor used in the example may represent the work of two
individuals.
To illustrate the structure of the agronomic and economic tradeoffs in the
labor allocation decision, the MPPs, MVPs, yields, and returns for several
feasible combinations of weeding and transplant dates are given in Table 6.
The calculations assume 0.54 ha of irrigated rice and 2.16 ha of monocrop
millet. It is further assumed that one PDE of labor is available each calendar
day for either millet weeding or rice transplanting. All inputs other than the
millet weeding or rice transplanting day are held at their geometric mean
levels. The work schedule for Example 1 has all labor allocated to millet until
millet weeding is finished, and then it allocates labor to rice transplanting.


8-Jul 18-Jul 28-Jul 7-Aug 17-Aug 27-Aug 6-Sep 16-Sep
[] rice 0 millet
Figure 2. Marginal value product of the rice transplanting date and millet weeding
date on a per-person day basis for Koutoukal6, Niger, 1985, estimated with a
Cobb-Douglas production function.


Vol. 4, No. 2, 1994






YU, ET AL.


Table 4. Input means, estimated marginal physical products, and estimated marginal
value products for rice and millet, Koutoukal6, Niger, 1985.

MEANS AND MARGINAL PRODUCTS BY FACTOR UNIT RICE MILLET

GEOMETRIC MEAN INPUT LEVELS:
Field size ha 0.46 0.81
Labor PDE/ha 346.45 52.87
Fertilizer FCFA/ha 8,059.71 0
Planting or transplanting date day 29.12 10.95
Weeding date day 28.69 12.19
MARGINAL PHYSICAL PRODUCTS:
Field size kg/ha -2,088.81 -59.43
Labor kg/ha -1.60 1.49
Fertilizer kg/ha 0.01
Planting or transplanting date kg/ha -17.79 -7.20
Weeding date kg/ha -358 -3.57
MARGINAL VALUE PRODUCTS:
Field size FCFA/ha -162,927.00 -5,110.99
Labor FCFA/ha -124.60 128.09
Fertilizer FCFA/ha 0.97 -
Planting or transplanting date FCFA/ha -1,387.62 -619.03
Weeding date FCFA/ha -278.99 -307.15
The marginal physical products are calculated as dY/dX=e*MY/MX, where Y is the yield, X is
the input, e is the estimated coefficient, MY is the mean yield in the sample, and MX is the mean
input for the sample.
The marginal value products are the marginal physical products multiplied by the average local
market price in 1985: paddy rice, 78 FCFA/kg; millet, 86 FCFA/kg. The markets used were
at Karma and Namaro.

Table 5. Marginal physical product and marginal value product of labor (per PDE)
used in each operation.

OPERATION MPP (KG/PDE) MVP (FCFA/PDE)
Millet weeding -0.30 -26
Rice transplanting -0.31 -24

Example 1 has weighted average millet weeding and transplanting days
close to those found in the sample. Example 2 is like Example 1 except that
the beginning of rice transplanting is moved up one day from July 24 to July
23, and July 24 is allocated to millet weeding. Example 3 is like Example 2
except that instead of moving up rice transplanting one day, it is moved up 10
days. Example 4 shows the effect of giving priority to rice so that rice is
transplanted at the earliest transplant day and millet operations are done only
in between times.
The results listed in Table 6 indicate that as labor is allocated to rice
transplanting earlier in the season, total grain production and net revenue rise
modestly, but expected millet yields fall. The marginal products of rice


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


Table 6. Expected marginal products, yields, and returns from millet and rice produc-
tion under alternative weeding and transplanting schedules.'

SAMPLE EXAMPLE EXAMPLE EXAMPLE EXAMPLE
ITEM AVERAGES 1 2 3 4
DATES:
Rice transplanting Aug. 22 Aug. 19 Aug. 18 Aug. 15 Aug. 9
Millet weeding July 20 July 21 July 21 July 24 Aug. 2
YIELDS:
Millet, kg/ha 314 311 311 299 284
Rice, kg/ha 3,711 3,770 3,791 3,859 4,035
MPPs:
Rice, kg/PDE -0.31 -0.35 -0.36 -0.42 -0.60
Millet, kg/PDE -0.30 -0.28 -0.28 -0.22 -0.13
MVPs:
Rice, FCFA/PDE -24 -27 -28 -33 -47
Millet, FCFA/PDE -26 -24 -24 -19 -11

TOTAL GRAIN, KG 2,682 2,708 2,719 2,736 2,792
NET REVENUE,2 FCFA 178,871 180,799 181,684 182,876 186,945

1 Calculated with the estimated rice and millet production functions, assuming 0.54 ha of irri-
gated rice, 2.16 ha of millet monocrop, and labor available is 1 PDE per calendar day. The
transplanting and weeding dates used to calculate the weighted average dates are:
Example 1: millet weeding July 8-Aug. 2, rice transplanting Aug. 3-Sept. 3
Example 2: millet weeding July 8-Aug. 1 and on Aug. 3, rice transplanting on Aug. 2 and
from Aug.4-Scpt. 3
Example 3: millet weeding July 8-23 and Aug. 3-12, rice transplanting July 24-Aug. 2 and
Aug. 13-Sept. 3
Example 4: millet weeding July 8-23 and Aug. 25-Sept. 3, rice transplanting July 24-Aug.
24.
2 Net revenue is value of grain produced minus the cash costs. Subtracted costs are fertilizer for
rice (8,060 FCFA/ha) and the cooperative charges for water use and other costs (58,171
FCFA/ha). It is assumed that all labor is family or community labor. Millet seed is produced
by the farmer; rice seed is included in the cooperative charge. In addition, because there is no
fertilizer used in millet production, cash costs for millet were considered as zero.

transplanting are above those of millet weeding for all the examples. Because
of the negative sign of the millet weeding and rice transplanting response, the
equality of the marginal products of millet weeding and rice transplanting
suggests that the observed labor allocation (the sample averages that are the
results from Table 5) may be minimizing total grain production and net
revenue while maximizing millet production. Although the examples support
the idea that the sample labor allocation is at a minimum production and net
revenue, they also indicate that the cost of choosing the maximum millet
production is relatively small under the assumed labor constraints. The
difference in total grain production between Example 1 and Example 4 is only
110 kg and in the net revenue is 8074 FCFA. But the proportion of millet in
the total grain supply between Example 1 and Example 4 is reduced from 25


Vol. 4, No. 2, 1994






YU, ET AL.


percent to 22 percent while the rice supply is increased from 75 percent to 78
percent. The strategy observed in farmers fields appears to maximize the
proportion of millet in the total grain supply.
An alternative labor allocation only shows a substantial advantage if rice can
be transplanted the first week of the production period. For example, if28 days
of labor are hired so that rice can be transplanted in the period from July 25
to 28, then the estimated total grain production rises by 736 kg to 3418 kg
and the expected net revenue is up 46,064 FCFA, after subtracting the labor
cost. As in all the examples, this estimate assumes that total labor and other
inputs are held constant.

Other Variables
Because the fertilizer variable is measured in terms of money spent on this
factor, the marginal value product is a cost-benefit ratio. It estimates the return
for each franc spent on nitrogen fertilizer. The estimate indicates that at the
margin, farmers receive about 0.97 franc for each franc invested in fertilizer.
Thus, the MVP is approximately equal to the input cost, and returns are
maximized.
The field size variable marginal products are included in Table 3 for
completeness, but because it is, in effect, a proxy for management factors, that
coefficient must be interpreted with more than the usual caution. In partic-
ular, the marginal cost of the field size variable is not clear. It is not the cost
of land or the land rental rate, because the area assumed by the production
function is fixed at one hectare. In any case, it may be said that these
management effects appear to be important for both crops, but more
important for rice than for millet. This may be due to the role of field leveling
in irrigation water management and to the fact that it is hard to level a larger
field.


CONCLUSIONS AND IMPLICATIONS
The hypothesis that farmers in Niger allocate labor to millet production
instead of rice production because the marginal product of labor is higher in
millet is not supported by this research. At the average input levels, the
absolute marginal products of labor for millet weeding and rice transplanting
are about equal. It is difficult to find a feasible combination of millet weeding
and rice transplanting dates for which the MPP and the MVP of millet weeding
is substantially above those of rice transplanting.
The alternative hypothesis that Nigerien farmers allocate labor to maximize
millet production is supported by the results. Examples using the estimated
production functions suggest that it is easy to find alternative labor allocation
patterns that modestly increase total grain production and net revenue, but
most involve lower expected millet yields. The examples also suggest that yield


Journal for Farming Systems Research-Extension






RICE AND MILLET IN NIGER


increases due to alternative labor allocation would be modest unless labor
hiring could permit rice to be transplanted very early in the transplanting
period. Without hired labor, it would be difficult for most families to
transplant rice early enough in the period to achieve the highest yields. The
examples indicate that with family labor alone, even if the farmer puts an
absolute priority on rice transplanting, the yield gains over the current system
are modest. To achieve the high yields, the estimates suggest that the rice must
be transplanted quickly at the beginning of the period.
Analysis of the structure of the labor allocation problem during the millet
weeding and rice transplanting period at Koutoukal6 suggests that food
preferences for millet may play a modest role in reducing rice yields and
returns, but the heart of the problem is that for a short period there is a high
demand for labor used in transplanting irrigated rice. Even if there were no
competition from millet production, this demand could create labor prob-
lems.
Two suggestions have been made to reduce this problem: (1) enlarge the
supply of labor or increase the effectiveness of the labor supply, or (2) change
the cropping system to reduce labor needs. It would not be easy to greatly
enlarge or improve the efficiency of the overall labor supply in rural Niger.
Only the second suggestion has been seriously considered. Potential changes
in the cropping system range from direct seeding of rice to switching crops.
Sorghum has been mentioned as an alternative crop in the Niger River
perimeters (Bomans, 1986). Because it has labor requirements similar to
millet, sorghum production would substantially reduce the overall labor
requirement in the critical millet weeding period. Rotation of rice with
sorghum could break pest cycles. Potentially, improved sorghum could
exceed current rice yields, but it is not clear that the market could absorb a
large increase in production. Sorghum is part of the traditional Nigerien diet.
Increased sorghum production would help meet the goal of food self-
sufficiency, but the growing urban market does not demand sorghum as it
does rice.

ACKNOWLEDGMENTS

This research was supported by the European Development Fund (FED);
Commission for the European Economic Community (EEC), project No.
5605.33.40.007; and by the U.S. Agency for International Development
(USAID), Science and Technology Bureau, Technology of Soil and Moisture
Management Project, under USDA PASA No. BST-4021-P-AG-108D-00.
The data were collected with the help of the National Office of Irrigated
Perimeters (ONAHA), Ministry of Agriculture, Republic of Niger. The views
expressed by the authors in this document do not necessarily represent those
of the organizations cited.


Vol. 4, No. 2, 1994






YU, ET AL.


The authors are grateful to Drs. J. Binkley and D. Brown for their helpful
comments, to Dr. Patrick Jomini for his help in using the SAS procedures, and
to Mr. Abdoulaye Bonkoula for sharing his insights on the agronomic
mechanisms for the yield effects of delayed planting and weeding.


REFERENCES

Anders, G., W. Firestone, M. Gould, E. Malek, E. Simmons, M. Versel, T. Ware, and T.
Zalla. 1984. Niger irrigation subsector assessment, volumes 1 and 2. USAID, Niamey,
Niger.
Beattie, B., and C.R. Taylor. 1985. The economics ofproduction. New York: John Wiley
and Sons.
Bomans, E. 1986. Temps de travaux et revenues des exploitations agricole de la Vall6 e du
Fleuve Niger, tome II, Koutoukal6, Hivernage, 1985. Cellule Suivi-Evaluation,
Division de Mise en Valeur d'Office National des Amenagements Hydro-Agricole,
Ministere du D6veloppement Rural, Republique du Niger.
Heady, E.O., and J. Dillon. 1961. Agriculturalproductionfunctions. Ames, Iowa: Iowa
State University Press.
Maikor6ma, Z.B. 1986. Gestion de deux syst6mes de culture dans les exploitations
agricoles du fleuve Niger. M6moire de fin d'etudes, Ecole Nationale Superieure
d'Agronomie de Rennes, Chaire d'Economie Rural.
Rassas, B., and T. Loutte. 1989. Niger: Rice and cotton policy. Technical Report No. 106
(Draft). USAID, Niamey, Niger.
Republique du Niger. 1987. Plan de d6veloppement economique et social du Niger.
Ministere du Plan.
Republique du Niger. 1989a. Seminaire national sur le d6veloppement de l'irrigation au
Niger, Rapport de Synth6se. Birni N'Konni, Niger.
Republique du Niger. 1989b. Rapport annuel des statistiques de l'agriculture et de
l'environnement. Minist6re d'Agriculture et de l'Environnement, Direction de la
Statistique de l'Agriculture et de l'Environnement.
Zellner, A., J. Kmenta, and J. Dreze. 1966. Specification and estimation of Cobb-
Douglas production function models. Econometrica 34(4):784-795.


Journal for Farming Systems Research-Extension








Participatory Needs Assessment:
A Key to FSRE

A.W. Etling and R.B. Smith



ABSTRACT
Needs assessment is a critical factor in Farming Systems Research-
Extension (FSRE) projects. Participation of local farmers and villagers is
essential to identifying their needs accurately, planning, and securing the
commitment necessary for successful implementation. However, needs
assessments often neglected in rural development work. When included,
the needs assessment is usually done superficially by local field staffer by
outside "experts" using research methodology.
This article reviews two needs assessment methodologies, "rapid rural
appraisal" and "participatory rural assessment," which balance the field
perspective with academic rigor. These methodologies have been used in
FSRE to help communities develop natural resource management plans.
Advantages and disadvantages of the methodologies are discussed, and
specific steps in conducting a participatory rural assessment are described.
The example used is an adaptation of participatory rural assessment that
was piloted in Costa Rica to identify the training needs of rural develop-
ment workers. Various data-collection techniques-previously used to
assess natural resource (technical) needs-were used to assess training
(human) needs. This new needs assessment methodology is easy to
implement and has the potential to strengthen training programs for
development workers so that they can do a better job of guiding FSRE
projects.

INTRODUCTION
Needs assessment is on the verge of establishing itself as a promising
international development tool. Achievement of this status depends, to a great
extent, on the abilities of academics and practitioners to understand and
interpret the opportunities and realities faced by developing nations and
donors. In our development work experience, needs assessment is rarely given
the recognition it deserves, and its potential contributions are seldom consid-
ered early or frequently enough (Butler and Butler, 1987).

1Assistant Professor and graduate student, respectively, Department of Agricultural and Exten-
sion Education, The Pennsylvania State University at University Park.





ETLING & SMITH


According to Butler and Butler (1987:262), Farming Systems Research-
Extension (FSRE) focuses on the farm family's goals and resources, and the
interrelationships among these goals and resources. The primary aim of FSRE
is to increase farm productivity in ways that are useful and acceptable to the
farm family. This aim can only be realized if the farmer, along with other
members of the local community, participates in problem identification
through a practical yet reliable needs assessment.
Needs assessment, the "systematic process whereby relevant needs are
documented" (Fear et al., 1978), has often been conducted in one of two
"extreme" approaches, according to Chambers (1981). The brief site visit by
an outside expert is the first approach. At its worst, this approach is unsystem-
atic, incomplete, and more reflective of the expert's biases and experience than
of local needs. The second approach is the academic or research study. Use of
appropriate research methods to ensure validity and reliability makes this
approach expensive in terms of time and other resources. The data often do
not get summarized or read by the decision-makers. Also, farmers' needs have
often changed by the time the research is completed.
Both of these extreme approaches to needs assessment tend to neglect the
participation of client groups that is desirable in FSRE. Neither approach
depends on the local community, which can be important in providing
personal support to farmers as they interact with outside experts. Further-
more, local community support is often critical to implementing outsiders'
recommendations.
A middle approach to needs assessment was needed to complement the
holistic rural development emphasis embodied in FSRE. This middle ap-
proach evolved through "rapid rural appraisal" (Smith, 1991).

RAPID RURAL APPRAISAL
Also known as "rapid reconnaissance," "exploratory survey," and "sondeo,"
rapid rural appraisal (RRA) uses a wide range of needs assessment techniques.
Mainly, it emphasizes careful observation coupled with semistructured inter-
views of farmers, local leaders, and officials during one or more brief visits.
RRA has been used for agricultural marketing appraisal (Holtzman, 1986),
natural resource appraisal (Stocking and Abel, 1981), participatory research
for small farmers (Swift, 1981), as well as FSRE (Bartlett and Ikeorgu, 1981;
Collinson, 1981; Hildebrand, 1981; Conway, 1986; Abalu, et al., 1987;
McCracken, 1988).
McCracken (1988:164) lists five characteristics that qualify a needs assess-
ment as RRA: (1) quick-will be completed within a few weeks; (2) team
effort-two or more researchers are involved; (3) multidisciplinary-team
members come from different disciplines; (4) interactive-team members
share their different disciplinary perspectives during the appraisal; and (5)
repetitive-techniques are repeated when doubts or inconsistencies arise.


Journal for Farming Systems Research-Extension






PARTICIPATORY NEEDS ASSESSMENT


According to Butler and Butler (1987:264), weaknesses may arise if team
members fail to appreciate each others' disciplines, if team members insist on
rigid control of variables, or if the process is allowed to become too open-
ended.

PARTICIPATORY RURAL ASSESSMENT:
NEEDS ASSESSMENT BY THOSE IN NEED
In response to these and other criticisms of RRA, modifications were
introduced and a new needs assessment methodology-participatory rural
assessment (PRA)-was developed. PRA is RRA with full participation of the
community.
According to Ford (1989), PRA is designed to focus on rural communities,
systematize rural participation, and help communities establish resource
management plans. PRA is useful in remote rural communities that are often
ignored by macrodevelopment strategies of national planning offices. Focus-
ing on natural resource management, PRA involves specialists from various
disciplines and representatives of different organizations who may not other-
wise come together for a needs assessment focused on a particular community.

Theoretical Steps of PRA
Local villagers cooperate actively in each ofthe steps ofPRA, which include:
(1) site selection, (2) preliminary visits by the PRA team, (3) data collection,
(4) data synthesis and analysis, (5) ranking problems, (6) ranking opportuni-
ties, (7) adopting a village resource management plan, and (8) implementa-
tion of the plan. The PRA team is usually composed of four to six specialists.
At least half are technical officers assigned to the community or area to be
studied. The specializations may include plant science, animal science, com-
munity development, forestry, health, etc., based on the characteristics of the
local area. In Kenya, where PRA has worked effectively, the needs assessment
focuses on a particular village.
Data collection emphasizes spatial, temporal, social, and technical data.
Spatial data come from a village sketch map compiled in cooperation with
village leaders, a village transect (depicting land uses), and simple farm
sketches. Six to eight farms are identified and sketches are prepared by team
members and household heads to show distances, land use on typical farms,
ecological variety, income variation, and ethnic distribution around the
village. Temporal data include a time line of events important to local
residents, trend lines ofa 40-year pattern ofchanges in resources (rainfall, crop
production, soil loss, deforestation, health, population), and a seasonal
calendar (land use, food surplus, food shortages, disease, cash availability).
Social data are derived from farm interviews and discussions of village
institutions. The interviews are carried out at those households where the farm
sketches are compiled. Village institutions are described in diagrams that


Vol. 4, No. 2, 1994






ETLING & SMITH


express their relative importance and relationships to each other. These
diagrams result from discussions among groups of residents. Technical data
show the economic and technical potential of resources (soil, water, etc.)
needed for agriculture.
After the data are collected, the PRA team works with community
representatives to organize the data and compile lists of problems and
opportunities for possible action. Then villagers are assembled to discuss and
rank the listed problems and create a priority list. Next, village groups rank the
opportunities that seem to address the most severe problems. In ranking the
opportunities, villagers are encouraged to consider the feasibility of imple-
menting the opportunities and their likelihood of contributing to stability,
equity, productivity, and sustainability.
The highest ranking opportunities are written into a plan that describes
each action to be taken, the committee or individual responsible, resources
needed, and a deadline for completion. This plan becomes the basic work plan
for all elements of the community. It can also take the form of a contract
among village groups, technical officers, and external groups such as donors
or international agencies. Implementation of the plan is usually guided by a
village leader. The actual work is performed by the community's self-help
groups.

Advantages and Disadvantages of PRA
Advantages of PRA include: (1) use of visual materials that are easy for
villagers to understand; (2) promotion of systematic participation of villagers,
village groups, and interested agencies; (3) provision for interactive problem
analysis and interdisciplinary problem solving; (4) identification of village-
based priorities; (5) application in the field quickly and inexpensively; (6)
strengthening of rural institutions; (7) helping communities prepare orga-
nized proposals for external support; and (8) motivation of participants to
action.
One disadvantage of PRA is that it ends with implementation and omits
evaluation, which could easily be added as the last step. A village meeting
could be called annually to discuss progress in implementing the village
resource management plan and update priorities for the coming year. Another
disadvantage of PRA, until recently, is that it has focused primarily on
technical and resource needs. The lack of leadership and program manage-
ment skills of those implementing FSRE programs are often greater problems
than the lack of resources or technical expertise. For this reason, a recent
adaptation of PRA to assess the training needs of extension workers in Costa
Rica may have great importance for future applications of FSRE.


Journal for Farming Systems Research-Extension






PARTICIPATORY NEEDS ASSESSMENT


USING PRA TO ASSESS TRAINING NEEDS: AN EXAMPLE
FROM COSTA RICA
When the training staff of the United States Agency for International
Development (USAID) in Costa Rica decided to organize a training program
for youth development professionals, they recognized the importance of
needs assessment. For several years, USAID had been providing nine-month
scholarships for rural Costa Rican high school students that were members of
4-S (similar to 4-H) to study in the US. These students, called Central
American Peace Scholars (CAPS), lived with 4-H host families in several states
in the US. They returned to Costa Rica with many ideas for improving their
communities through the 4-S organization. Such innovation was encouraged
by the CAPS program. The 4-S professionals, however, seemed unprepared
to encourage and support the returning high school students' new ideas. In
some cases, the 4-S professionals perceived the new ideas as threatening; in
others, these ideas were not priorities. Sometimes the professionals simply
lacked the knowledge, skills, and experience needed to help the students
implement new ideas for youth development.
Noting the frustrations that resulted, USAID decided to provide a two-
month professional development workshop in the US for 20 4-S agents.
Through competitive proposals, the National 4-H Council was chosen to (1)
conduct a needs assessment of the 4-S agents to determine the content of the
training, and (2) design, implement, and evaluate the training program.
In Costa Rica, the needs assessment process adapted from PRA included
seven of the eight theoretical steps (step 2 was not possible in this case).

Step 1
In Step 1, the site was determined by USAID to be the entire country.
Twenty participants were chosen to represent the geographic and social
diversity of Costa Rica. In the US, a needs assessment team was selected
according to USAID guidelines; it was interdisciplinary and featured diverse
yet complementary experience. Of the five individuals comprising the team,
four had work experience in cooperative extension including 4-H, community
development, family living, and agriculture. One member had not worked for
extension before. Team members had 4-H experience at the national, state,
county, and local levels. One was a 4-H parent and another a former 4-H
member. Functional expertise of team members included administration, staff
development, coordination of international exchange programs, experience
in extension methodology, and work with extension in Central America.
Three of the five team members were fluent in Spanish.

Step 2
Due to the nature of this needs assessment, a preliminary team visit was not
possible. This PRA step was skipped without any apparent negative results.


Vol. 4, No. 2, 1994









Figure 1. Logic model of the training program.


MAG] MEP

4-S Professionals

SAIDD


Rough Draft of
Training Itinerary
in US


PLANNING
1. Needs
2. Priorities
3. Objectives
4. Resources -
Inventory


PROGRAM
5. Plan


- -


* Assess
Needs
* Determine
Objectives
* Conduct
Orientation
for 4-H


TRAINING
6. Program in
US


7. Evaluate


INCORPORATE
CAPS
Ideas into 4-S
Clubs



Begin some
Community
Clubs



Train Peers


DEVELOP
Youth Life
Skills


IMPROVE
Costa Rican
Society


DEVELOP
stronger 4-S







SAbility to
Generate Funds


* Support MEP
& MAG


Ir 1






PARTICIPATORY NEEDS ASSESSMENT


Step 3
In Step 3, data collection was adapted to determine the training needs of
the 4-S professionals. Data were collected in six stages. Stage I was the
construction of a "logic model" (Figure 1) to show the flow of events that
brought about the need for the training, the training plan, and projected
effects of the training. This logic model, used in "evaluability assessment"
(Smith, 1989), another needs assessment methodology, is useful in under-
standing the chain of causes and effects in complex organizations.
The logic model begins with the 4-S professionals who are hired by the
Ministry of Agriculture (MAG) and the Ministry of Public Education (MEP).
They, along with USAID and the National 4-H Council needs assessment
team, engaged collaboratively in planning (including the needs assessment)
for the training program. The numbered items in the planning box indicate
the first four phases of a standard program planning process. As a result of the


NICARAGUA


E] E


COSTA RICA


Figure 2. Costa Rica: Work sites of
the 4-S professionals who participated in the training.


Vol. 4, No. 2, 1994






ETLING & SMITH


planning process, an outline for the training program was developed. Imple-
mentation of the training program in the US would lead to increased
competence and ability, on the part of the 4-S professionals, to incorporate
new ideas from returning CAPS youth into the 4-S clubs located where the 4-
S agents worked (Figure 2). In turn, these ideas would develop a stronger 4-
S organization with greater capacity to develop life skills in Costa Rican youth.
In Stage II of data collection, characteristics of the participants were
described through bio-data (application) forms and individual interviews with
each participant. They were asked about their 4-S responsibilities, communi-
ties, future plans for 4-S, and conditions that inhibited their work. The clear
impression that emerged from this stage of the data collection was that the
group was very heterogeneous. Those employed by MEP worked in schools
and had programs that were quite different from the MAG employees, who
worked out of county or area offices and tended to focus on community
groups outside the schools. Furthermore, each participant had a unique mix
of projects and activities underway. These facts indicated that individualized
training would be necessary to accommodate individual differences. The
training would also need to allow for considerable interaction so that
participants could share ideas.
Stage III of the data collection involved a systematic review of documents
pertinent to the organizational structure in which the participants worked.
Three documents were especially helpful at this stage: (1) a manual for
organizing 4-S Clubs, (2) a 4-S information brochure, and (3) the 4-S
newsletter, "Intercambio," which is sent to each 4-S agent (Fundaci6n
Nacional de Clubes 4-S, n.d., 1989, 1990). These documents gave the needs
assessment team an idea of the resources and relationships pertinent to the
participants. From these sources emerged an understanding of the institution-
al linkages that were important to the success of the individual participants in
promoting 4-S clubs.
Stage IVofdata collection was a nominal group process conducted with the
participants to identify their perceptions of their own training needs. The
question posed was: What do you hope to gain from this training program
(including skills, knowledge, and attitudes)? Each participant was given a 4"
x 6" index card to record answers to the question silently. A Spanish-speaking
member of the needs assessment team led participants through the nominal
group process while a Costa Rican recorded the answers on newsprint. After
each participant's answers were recorded (round-robin style), the answers
were discussed and debated. Then each participant was given another index
card and instructed to vote for the top three priorities of all the answers on the
newsprint. The secret ballots were tallied and a priority list of training needs
was identified (see Results below). The priority list was presented to the
participants and their comments were solicited.


Journal for Farming Systems Research-Extension






PARTICIPATORY NEEDS ASSESSMENT


Only the participants were allowed to discuss training needs. Aside from the
needs assessment team member who conducted the session and the 4-S
administrator who recorded answers, only three observers were allowed in the
room: two representatives of USAID and a 4-S administrator who wanted to
learn how to conduct a nominal group process. Discussion was lively and
spontaneous and participants seemed happy with the training priorities
identified. USAID and 4-S officials were also satisfied because the priorities
complemented (yet did not exactly duplicate) USAID and 4-S expectations
for the training program. This was the single most critical stage of the needs
assessment. However, the overall strategy of complementary needs assessment
techniques used in the other five data collection stages strengthened the
nominal group process.
Stage V of data collection consisted of structured discussions among
representatives of the organizations that were stakeholders in the participant
training. Top administrators of4-S, USAID in Costa Rica, and the Ministries
of Agriculture and Public Education were all interviewed in groups. In each,
the primary question was: What is your (organization's) view of the training
needs of 4-S agents?
Stage VI of data collection was observation of the home and work
environments of the participants. Visits were made to the schools where two
participants worked, and the PRA team met families in their homes.

Steps 4-8
In Step 4, data were analyzed and synthesized by team members into a list
of training needs (Step 5). This list was discussed, refined, and prioritized
(Step 6) in a group discussion with the 20 participants and USAID officials
responsible for funding the training.
A training plan was then developed (Step 7) that addressed the needs
identified. The training consisted ofa two-month program (Step 8) that took
the participants from Washington, DC to North Carolina A&T State Univer-
sity, Clemson, University of Tennessee, University of Georgia, the annual
conference of the National Association of Extension 4-H Agents in West
Virginia, Ohio State University, and Penn State University.

RESULTS
The needs identified for the professional development training of 4-S profes-
sionals were: (1) leadership skills (communication, motivation, program
planning, curriculum development, and teaching skills); (2) competence in
conducting leadership training for new 4-S professionals (including recruit-
ment and management of volunteers); (3) community development skills; (4)
subject matter in agriculture and family living; (5) skills in promoting
entrepreneurship; (6) ideas to promote the personal development of 4-S
members; (7) skills in promoting interorganizational cooperation at the


Vol. 4, No. 2, 1994





ETLING & SMITH


community level; (8) organizational development skills to expand the impact of
4-S; (9) networking skills for sharing ideas among other youth development
professionals; apd (10) skills in integrating 4-S members' individual projects with
the dub projects.
These clearly identified needs were consistent with the expectations of the
cooperating organizations. As a result of this participatory needs assessment
process, an efficient and cost-effective training program was designed that satisfied
both the participants and the donor (USAID). The needs identified provided the
basis for evaluating the training program at its conclusion. Commitment to the
training was high not only among the participants, but also at the National 4-H
Council, which implemented the training, and at USAID, which funded it.
Expectations among participants, trainers, and the donor were always clear and
consistent.
At the end of the training, evaluation was conducted using a written question-
naire and group interview. Ayearlater, the questionnaire and group interview were
repeated in Costa Rica, and a nominal group process was conducted to determine
future training needs. This evaluation documented that the participants had
improved markedly the skills, knowledge, and attitudes necessary to help returning
CAPS youth implement community development projects. A year after the
training, most of the participants had implemented new community projects of
their own. Perhaps the most significant result of the training was that the
participants formed their own "professional association of4-S agents" to help each
other in their professional improvement.
Nine months after the final evaluation, however, the professional association
was inactive due to lack of support from the participants' supervisors, who
cancelled a national meeting of the association. Although this fact may have many
causes, perhaps the agents' training and the association that resulted were
considered a threat to supervisors who did not receive the training.

RECOMMENDATIONS
The needs assessment team's experience in Costa Rica indicates the importance of
flexibility in carrying out a needs assessment. The stages of data collection are not
necessarily sequential; team members were involved in several stages of data
collection at the same time. Multiple techniques and sources for data collection
mean that inconsistencies in the data can be expected and can be the focus of
follow-up or subsequent stages in the data collection.
Time is needed for the needs assessment team to analyze data each day and to
discuss the progress of the needs assessment and the activities for the coming day.
Time is also needed to respond to unanticipated requests that may be irrelevant to
the needs assessment process but important to the donor or participants. In Costa
Rica, the needs assessment team was asked to provide participants with orientation
to the training sites in the US.


Journal for Farming Systems Research-Extension






PARTICIPATORY NEEDS ASSESSMENT


Team members who had never worked on formal needs assessment were able
to understand the process and perform each of the steps without special training.
Skills needed to assess the training needs of local 4-S agents were: (1) ability to
conduct interviews, (2) ability to conduct group discussions with diverse individ-
uals, and (3) experience in group dynamics. In this instance, fluency in Spanish and
experience in conducting the nominal group process were also important.

EDUCATIONAL IMPORTANCE
Potentially the greatest benefit of this needs assessment is the new methodology
developed in response to the request from USAID Costa Rica. PRA has been used
successfully in several developing countries, involving collaboration with local
villagers in their communities in order to identify problems and opportunities for
FSRE projects (Ford, 1989). Now a participatory needs assessment has been
adapted from PRA to identify the training needs of professional educators in
developing countries. This new adaptation should further strengthen FSRE
projects that often depend on local professionals to help the villages to carry out
their plans. The participatory needs assessment process is easy to understand and
implement. It can result in plans that all stakeholders help to develop and that they
therefore have a greater commitment to implement.
The particular data collection techniques used in Costa Rica included tech-
niques tested over decades (group discussion, individual interviews, nominal
group process, and observation), as well as a new technique-construction of a
logic model that clarifies complex relationships among individuals and organiza-
tions in local communities. These techniques were combined in a methodology
that emphasized the collaboration of the local beneficiaries in their communities.
The advantages of the participatory needs assessment in Costa Rica confirmed
and expanded the list of advantages of PRA. Active participation by those whose
needs were being assessed was achieved. The donor and sponsoring organizations
were sensitized to the particular needs of the individuals most involved in the
program. The needs assessmentprovided for interactive problem solving. All ofthe
stakeholders participated in defining problems, analyzing alternatives, and prior-
itizing solutions. The assessment was inexpensive and completed quickly.
The total time on-site in Costa Rica was one week. An objective research effort,
needed to produce the same results, would have taken much longer and cost much
more. Some of the steps (site selection, participant selection, and review of
documents) were started prior to the site visit. Finally, this participatory needs
assessment led to action. The two-month training program was designed,
collaboratively, around the needs identified, and completed within three months
of identifying the needs.
In addition to these advantages, a selection process was used that ensured a
diverse, interdisciplinary assessment team, an important factor in conducting this
collaborative needs assessment. The model was easily understood by all team
members, easy to implement, and flexible. It has great promise for use by


Vol. 4, No. 2, 1994






ETLING & SMITH


institutions planning development work in other countries where collaboration is
desirable. It is particularly appropriate for FSRE projects that depend on the
participation of all stakeholders (professional educators as well as villagers) in
identifying pertinent needs in order to increase farm productivity and strengthen
local communities.


REFERENCES

Abalu, G.O.I., N.M. Fisher, and Y. Abdullahi. 1987. Rapid rural appraisal for generating
appropriate technologies for peasant farmers: Some experiences from Northern Nigeria.
Agricultural Systems 25(4):311-324.
Bartlett, C.D.S., and J.E. Ikeorgu. 1981. A project to identify suitable innovations for small
farmers in Nigeria. Agricultural Administration 8(6):451-462.
Butler, L.M., and R.O. Butler. 1987. Needs assessment in international development. In
Johnson, D.E., et al., (eds.), Needs assessment. Theoryand practice. Ames, Iowa: Iowa State
University Press.
Chambers, R. 1981. Rapid rural appraisal: Rationale and repertoire. Public Administration
and Development (1):95-106.
Collinson, M. 1981. A low cost approach to understanding small farmers. Agricultural
Administration 8(6):433-50.
Conway, G.R. 1986. Agroecosystem analysis for research and development. Winrock Interna-
tional, Institute for Agricultural Development, Bangkok.
Fear, F.A., et. al. 1978. Needs assessment in community development: A resource book.
Department of Sociology and Anthropology, Iowa State University, Ames, Iowa.
Ford, R 1989. An introduction to participatory rural appraisal for rural resource management.
Program for International Development, Clark University, Worcester, Massachusetts.
Fundaci6n Nacional de Clubes 4-S. (n.d). Manual para establecer clubes 4-S. San Jos6, Costa
Rica.
Fundaci6n Nacional de Clubes 4-S. 1989. Informaci6n bbsica sobre 4-S. San Jos6, Costa Rica.
Fundaci6n Nacional de Clubes 4-S. 1990. Intercambio 1. San Jos6, Costa Rica.
Hildebrand, P. 1981. Combining disciplines in rapid appraisal: The sondeo approach.
Agricultural Administration 8(6):423-32.
Holtzman, J.S. 1986. Rapid reconnaissance guidelines for agricultural marketing and food
system research in developing countries. Working Paper 30. Michigan State University, East
Lansing, Michigan.
McCracken, J. 1988. A working framework for rapid rural appraisal: Lessons from a Fiji
experience. Agricultural Administration and Extension 29(3):163-184.
Smith, R. 1991. Rapid rural appraisal: A promising needs assessment paradigm for grassroots
development. In Proceedings of the 7th Annual Conference of the Association for Interna-
tionalAgricultural and Extension Education, St. Louis, Missouri, March 28-30.
Smith, M.F. 1989. Evaluability assessment: A practical approach. Boston, Massachusetts:
Kluwer Academic Publishers.
Stocking, M., and N. Abel. 1981. Ecological and environmental indicators for the rapid
appraisal of natural resources. Agricultural Administration 8(6):473-484.
Swift, J. 1981. Rapid appraisal and cost effective participatory research in dry pastoral areas of
West Africa. Agricultural Administration 8(6):485-492.


Journal for Farming Systems Research-Extension









Women Farmers' Role in Managing Cassava

Production in Bandundu, Zaire

Mbongolo-Ndundu Mputela and Steven E. Kraft1


ABSTRACT

Women farmers in many developing countries contribute more than 60
percentofthe effort involved in production of food crops. Consequently,
in countries such as Zaire, it is important to analyze the contribution of
women to food production and to relate it to national policies of food
security.
To enhance the productive capability ofwomen farmers, it is necessary to
provide them with information that they can use in making better
allocative and technical decisions. These decisions relate to the use ofland,
labor, and capital. In addition, there are decisions related to credit, market
access, and the use of "improved inputs" such as fertilizer and seed. Data
derived from a farm-level survey of women farmers in Bandundu are
useful in assessing what women farmers are doing now and for making
recommendations for the future. In this study, a Cobb-Douglas produc-
tion function is estimated, based on cross-sectional data collected from
women farmers in three subregions of Bandundu during the summer of
1990. The production ofcassava is analyzed using data on eight variable
inputs alongwith information ofthe women's use ofcredit and extension
training.
Female labor, male labor, seed, tools, market access, credit, and training
were found to have a positive impact on cassava production in Bandundu.
Results are interpreted in terms offarm-level managerial decisionmaking
and macropolicy.


INTRODUCTION

Zaire is a vast country with social, cultural, and economic diversities. Because
of the vastness of the territory and the diversity of resources, Zaire has a
complex food problem. The annual growth rate of food production (2
percent) is lower than the annual growth rate of the population (3.3 percent).

1 Mputela Mbongolo-Ndundu has a MS in agribusiness economics from Southern Illinois
University. She has worked on promoting women's economic issues in Zaire. She has also worked
on the marketing of agricultural products for the Government of Zaire. Steven Kraft has a Ph.D.
in agricultural economics from Cornell University with an emphasis in land-resource economics.
He currently works in the areas of soil and water conservation policy and implementation.





MPUTELA & KRAFT


In addition, although wages and salaries remain very low, there is a continuous
increase in the price of food, especially in urban areas. The result is a spread
of malnutrition. Furthermore, Zaire's capacity to import food is hampered by
even lower prices for exported raw materials, and a heavy debt burden. Finally,
the share of foreign exchange earned by the agricultural sector is declining.
One of the main constraints on food production is the low productivity of
traditional farming. The labor division between sexes in Zaire indicates that
women farmers play an important role in traditional farming. Hence any effort
to improve food production needs to concentrate on women farmers and one
of their main crops, cassava (Fresco, 1982; Tshibaka, 1990).
Women farmers in Zaire have little control over the economic and social
environment in which they work. Nonetheless, most women farmers decide
to produce cassava, and they decide how to allocate their limited resources in
their farm operations to achieve their goals. How they allocate these resources
is frequently determined by whether they have access to information about
alternative production processes and the socioeconomic environment in
which they operate.
A major step toward the resolution of the current food crisis in Zaire is
enhanced availability of accurate information about the relationship between
farm inputs such as land, labor, seed, and fertilizer, as well as from outputs.
Information from the analysis of production functions allows women farmers
to make choices regarding the best use of their limited resources. This paper
gives the result of the estimation of a Cobb-Douglas production function
using data from our survey of 360 women farmers in Bandundu, Zaire, in
1990. Based on the results, we offer suggestions to improve women farmers'
management of cassava production.


BACKGROUND
The Zairian agricultural sector accounts for a large percentage of that nation's
gross domestic product (29.5 percent), compared to 3 percent of GDP in the
developed countries. But food production in Zaire has been increasing at an
estimated rate of 2 percent per year. This increase has not kept up with the
population growth rate of 3.3 percent per year (Mputu and Elengesa, 1987).
The index of per capital food production (86) is far below the index of the
developed countries such as the US (111), France (118), and even Niger (92),
an important producer of cassava (FAO, 1984:51).
Food production in Zaire continues to utilize traditional agricultural
practices, which means that labor is a main determinant of production. The
percentage of the economically active population engaged in agriculture still
remains very high (74 percent) compared to the active population engaged in
agriculture in developed countries such as the US, where it represents only 2.1
percent (FAO, 1984:8). Labor is rarely enhanced by use of enough traditional


Journal for Farming Systems Research-Extension






WOMEN FARMERS IN ZAIRE


and modern inputs such as improved seed, improved tools, and chemical
fertilizer to increase productivity. For example, Zaire has an abundance of
available land, but land available for permanent crop production per farm
operator is less than 1 hectare; the average size in the US is 85.52 hectares per
farmer (FAO, 1984). Furthermore, Zairian fertilizer consumption per hectare
of available land is only 1.3 kg of plant nutrient, whereas in a developed
country such as Belgium fertilizer consumption has reached a level of499 kgs.
Increasing agricultural exports is one of the few means available to Zaire for
earning essential foreign exchange to acquire the technology necessary for the
modernization of its agricultural sector. However, the share of foreign
exchange earnings contributed by farming has declined from 38.9 percent in
1959 to 16 percent in 1987 (Tshibaka, 1986). In 1980, Zaire's main export
food crops were coffee and palm oil. The world prices for these exports have
been very low, limiting the earning capacity of the country. In addition, tariffs
and other barriers put up by the developed countries hamper the efforts of
small traders to sell Zairian food crops such as cassava.
Increased food needs and poor performance in the domestic food produc-
tion sector have led the Zairian government to increase food imports. But the
ratio of the value of exports to the value of imports in Zaire's food trade is very
low (14.4 percent) compared to the percentage in developed countries such
as the US (343 percent). In addition, if food imports are sold in domestic
markets at prices below domestic foods, this direct competition hampers the
ability of farmers to sell their products. As a result, Zaire is far from reaching
its objective of food self-sufficiency.
Women play an instrumental role in the Zairian agricultural sector as
significant actors in traditional agricultural production. This study is limited
to the role ofwomen in the production of the main food crop in Zaire, cassava.
The producer role of women is very pertinent for cassava, one of the most
widely consumed crops in Zaire (Tshibaka and Lumpunga, 1983). However,
a woman's role as farm manager is invisible because of many constraints. First,
the concept of "family farm," with its implicit assumption that men are the
heads of farm production, and married women are considered "helping
workers" that are equivalent to "unpaid workers," reduces the impact of
women in the food sector because married women farmers do not have
independent access to the farm resources. The second constraint is the concept
of "cash crop" versus "food crop," with the assumption that women farmers
produce only food crops. As a result, the impact of development in terms of
the expansion in trade, technological progress, and returns to labor do not
benefit women, and reduce their economic role (Norris, 1990).
When women are indeed considered farm managers, they have been found
to be more technically efficient than men in food production (Mosch, 1976).
The analysis of production function seems a good approach to helping African
women farmers with their decisions for allocating their available resources.
Previous analyses of production functions focused on the labor-productivity


Vol. 4, No. 2, 1994






MPUTELA & KRAFT


in Africa's agricultural sectors. One study found that there is a possibility of
increasing agriculture output per worker through adequate investment in
education, research, and the supply of modern technical inputs (Kawagoe, et
al., 1985). Another study found that it is possible to raise yields by increasing
labor inputs per hectare. The study recommended a land reform giving the
rural poor access to more land. Land redistribution would, if thoroughly
implemented, have immediate beneficial effects in terms of enhanced output
(Giovani, 1984). The result of a Cobb-Douglas production function analysis
in the Zaire region, using cross-sectional data, demonstrated that easy access
to land, along with education, research, and extension services such as weather
forecasting, improved the productivity of labor (Tshibaka, 1990). Another
application of the Cobb-Douglas production function showed a positive and
significant relationship between yield per hectare and credit in kind (Kante,
1989).


METHODOLOGIES
Bandundu, the study area selected, is one of the largest agricultural regions
near Zaire's capital city, Kinshasa, and is a region with social, cultural, and
economic diversity. Bandundu has three subregions, two located in a savanna
area and the third in a forest area. The three cities were chosen for the survey
based on agricultural development programs for the Bandundu region.
The sample size for this study consisted of a total of 360 women farmers
randomly selected. The sampling unit of the study is the adult woman farmer.
We did not use the family farm because of the assumption that the husband
is the head of the family farm. The 120 women farmers were selected in each
location from the list of family farms compiled by the local agriculture service.

The systematic random sampling procedure involved selecting every

Ith = Total family farms
120

For example, if there were 840 farms, every seventh family farm on the list
would be drawn. However, only the woman farmer was designated to be
interviewed.
The instrument used for data collection was a questionnaire specifically
designed to collect quantitative data about women farmers in Bandundu. The
questionnaire had three main parts. The first involved the identification of
female farmers according to social and economic characteristics such as age,
marital status, spouse information, education level, and services received from
agricultural projects in the area. The second part involved the production
decisions made by women farmers, especially regarding which food crops to


Journal for Farming Systems Research-Extension





WOMEN FARMERS IN ZAIRE


produce and the use of inputs such as labor, land, seed, fertilizer, tools and
equipment, training, credit, and market access. The last part dealt with the
quantity and value of food crops produced and marketed. The basic question-
naire in English was translated into three languages-French, Lingala, and
Kikongo-to assure that everyone interviewed was asked the same questions
and to avoid misinterpretations by the interviewers.
The interviewers were recruited from the local women of the survey area.
The criteria used to select the interviewers were: (1) sex: only females were
selected as interviewers in order to solicit freer responses from women farmers;
(2) locality: four interviewers were selected in each location to gain the
confidence of women farmers and to explain the questionnaire in the local
language, which is sometimes different from the official language of the
location; (3) level of education: only interviewers with 12 years of education
were selected to be sure that they understood the ideas underlying the study;
and (4) experience: experience in school teaching or agricultural extension
services was required. The investigator used three days to train the interview-
ers and one day to pretest the questionnaire. Each interviewer arranged the
interview schedules with the women farmers, and ten days were used for the
administration of the questionnaire in each location. The interviewers were
motivated by a ten-day salary that was high compared to their monthly salary.
Every day, when they brought back the questionnaire of the day, the
investigator immediately checked for completeness before giving them the
questionnaires for the next day.
Two methods of analysis were used in this study. The first was the
descriptive statistics that classify and summarize the data of women farmers in
Bandundu. The second method used was multiple regression to estimate the
Cobb-Douglas production function. Dillon and Handaker (1980:106) sug-
gest that Cobb-Douglas production function is appropriate for small farms,
for the case of more than one variable input, and is also supported by the
previous applications of the Cobb-Douglas production function in Africa,
specifically in Zaire (Upton, 1987, Kante, 1989, Tshibaka, 1990,).
The equation for the Cobb-Douglas production function is:
Y= aXI blX2 b2x3 b3 -X bn
where the a, bl, b2, b3, -bn are the coefficients to be estimated, and the Xs are
a vector of independent variables. The Cobb-Douglas production function is
a multiplicative relationship between output and the various inputs. It
becomes a linear function in its logarithmic form, and can be easily analyzed
by using multiple regression analysis.
Relationships in the model to verify in this study were whether the
productivity of land (yield of cassava or value of cassava per hectare) was
positively or negatively related to women's labor per hectare, husband's labor
per hectare, seed per hectare, cost of tools per hectare, natural fertilizer,
market access, training, association of credit, and the location of women's
farms. In this study, we were interested in the productivity per unit of land


Vol. 4, No. 2, 1994





MPUTELA & KRAFT


because land is considered the most limiting factor of production for women
farmers and because this form has the advantage of reducing the problems of
multicollinearity among the independent variable (Intriligator, 1978:269).

Social and Economic Characteristics of Women Farmers in Bandundu
The descriptive statistics indicate that the average age of women farmers in
Bandundu is 44 years, and 90 percent are married. The female farmer has, on
average, a low level of education (i.e., three years of primary school).
The results of the study show that 53 percent of harvested cassava is
consumed and 47 percent is sold. In short, cassava is not only a subsistence
crop, but also a cash crop.
The allocation of time among farming operations indicates that transport-
ing (57 days/year), and processing (30 days/year) are the most time-
consuming operations. The other operations require 17 to 26 days/year,
except the tree felling operation, on which women farmers spend less time (6
days/year).
There is great variability in the husbands' time allocation among the three
study sites. For example, in Feshi, women farmers work alone in the various
farming operations, while in Inongo, located in the forest zone, husbands
work with their wives in each operation, and especially in land clearing and tree
felling.
The average size of a cassava field for a woman farmer is less than one
hectare, and 82 percent of women farmers have one hectare or less of land for
cultivation. Most women farmers are not the owners of the lands they
cultivate. However, they report that it is possible to buy land. Unfortunately,
they don't have enough money for that. Most women farmers in Bandundu
use cuttings from the mature plants of their fields for the next planting. New
varieties of cutting, such as F100 produced by the national cassava program,
are used only in the demonstration fields of the extension services.
Most women farmers in Bandundu use only natural fertilization from the
fallow period. Fifty-nine percent of women farmers leave land idle for less than
six years, because with the fallow system a woman farmer needs to change the
location of her field and this implies a negotiation with the traditional
landowners and the potential for an increased distance between her fields and
the market. A woman farmer in Bandundu needs to walk an average of nine
km to sell her cassava. Many women farmers say that they are not able to
replace their old manual tools after many years of utilization because they are
not available in the rural area, and the cost of tools is high relative to their cash
income. Unfortunately, women farmers in Bandundu have no access to
commercial banks, and use credit associations to finance their farming activity.
Better trained female farmers may increase their ability and willingness to
allocate resources efficiently, but there was a variation among the study sites.
In two locations, only 30 percent of women farmers have access to the advice
of extension workers.


Journal for Farming Systems Research-Extension






WOMEN FARMERS IN ZAIRE 63

Table 1. Descriptive statistics: Distribution of output and inputs per hectare in Band-
undu in 1990.


VARIABLE


1. Kg HAPA (in kgs)






2. TOTPA (in Z)







3. LABWPA (days)





4. LABHPA (days)



5. Colla H
(LABHPA)
(LABWPA)




6. BOTTPA (cutting)








7. A Cost PA (z)


1 -1000
1,001 2,000
2,001 3,000
3,001 4,000
4,001 5,000
5,001 6,000
More than 6,000
1 50,000
50,001- 100,000
100,001 150,000
150,001 200,000
200,001 250,000
250,001 300,000
300,001 350,000
more than 350,000
0-200
201 400
401 600
601 800
801 1,000
more than 1,000
0- 100
101 200
201 300
more than 300
0 -0.3
0.31 0.6
0.61 0.9
0.91- 1.2
1.21 1.5
1.51 1.8
more than 1.8
1- 10
11 -20
21 30
31 40
41 50
51 -60
61 70
71 80
more than 80
1 3,000
3,001 6,000
6,001 9,000
9,001 12,000
12,001 15,000
15,001 18,000
more than 18,000


PERCENT
52.0
21.0
10.5
4.0
3.5
4.0
5.0
51.0
17.0
9.0
5.0
4.0
3.0
2.5
8.5
34.5
32.0
15.0
9.0
6.0
3.5
77.0
9.0
6.0
8.0
74.0
12.0
5.0
4.0
3.0
0.5
1.5
31.0
39.0
10.0
6.0
4.0
3.0
2.0
2.0
3.0
58.5
20.0
12.0
3.5
3.0
1.0
2.0


Source: Survey data


Vol. 4, No. 2, 1994


CUMULATIVE PERCENT
52.0
73.0
83.5
87.5
91.0
95.0
100.0
51.0
68.0
77.0
82.0
86.0
89.0
91.5
100.0
34.5
66.5
81.5
90.5
96.5
100
77.0
86.0
92.0
100.0
74.0
86.0
91.0
95.0
98.0
98.5
100.0
31.0
70.0
80.0
86.0
90.0
93.0
95.0
97.0
100.0
58.5
78.5
90.5
94.0
97.0
98.0
o00.0





MPUTELA & KRAFT


Production Inputs and Outputs
The average yield per hectare of cassava for women farmers in Bandundu
is estimated at 1,243 kgs/ha. There is a variation among the three locations.
The productivity of cassava in Bandundu is very low compared to the 4t/ha
found in other cassava producing countries (Cock, 1985). Women farmers in
the sample averaged 362 woman farmer days per hectare of cassava. This
average is slightly greater than the 200 man days per hectare in Nigeria and
300 man days per hectare in Indonesia (Cock, 1985). Total husband labor is
103 days per hectare, which is less than the 200 man days per hectare in
Nigeria.
Table 1 summarizes the distribution of output and inputs per hectare in
Bandundu in 1990. Pearson correlation coefficients indicate a positive corre-
lation between cassava harvested per hectare and inputs such as women labor
per hectare, husband labor per hectare, number of packages of cuttings per
hectare, cost of tools per hectare, and market access.

Production Function
The estimated model of the relationship between the quantity of cassava
harvested per hectare and the set of inputs is:
In Kghapa = 1.71+.17 In LABWPA + .09 Ln LABHPA + .34 Ln BOTTPA
- .18 Ln Fall + .29 Ln ACost +.21 Ln Dist + .56 Ln Train + .30 Ln Likelem
+.13 Ln Loc.
The dependent variable Kghapa is the quantity of cassava harvested per
hectare, and the independent variables are LABWPA = labor per hectare,
HABHPA = husband labor per hectare, BOTTPA = number of packages of
cutting per hectare, Fall = fallow period, Acost = cost of tools per hectare, Dist
= access to market, train = training, LIKELEM = association of credit, and Loc
= location.
The model to estimate the relationship between the value of cassava
harvested per hectare and the same set of inputs is:
Ln tot PA = 4.59 + 0 14 Ln LABWPA + .13 Ln LABHPA +.21 Ln Bottpa
.29 Ln Fall + .56 Ln Costpa +.08 Ln Dist + .42 Ln Train + .41 Ln Likelem
+.06 Ln LOC.
Table 2 shows the regression solution of the models.
Each of the coefficients in the model is positive except the fallow period
with value and quantity of cassava per hectare. The coefficient of determina-
tion for the overall model with the value of cassava per hectare is R2 = 0.58
percent and the adjusted coefficient of determination R2 = 0.57 percent may
be considered as a good fit for a cross sectional study (Johnson, 1987).
However, the model with quantity of cassava harvested per hectare as the
dependent variable has R2 = 0.52 percent and R2 = 0.51 percent implying that
the result must be interpreted with caution (Upton, 1987).


Journal for Farming Systems Research-Extension






WOMEN FARMERS IN ZAIRE


Table 2. Cobb-Douglas production function of women farmers in Bandundu in 1990.

VARIABLES OUTPUT: LN TOTPA
PARAMETER STANDARD ERROR T FOR HO PROB
ESTIMATE PARAMETER = 0
Intercep 4.593 0.700 6.559 0.00
LN LABWPA 0.141 0.086 1.638 0.10
LN LABHPA 0.137 0.034 4.023 0.00
LN BOTTPA 0.214 0.104 2.047 0.04
LN FALL -0.293 0.196 -2.146 0.03
LN A Cost 0.567 0.102 5.544 0.00
LN Dist 0.088 0.092 0.951 0.34
TRAIN 0.423 0.143 2.942 0.00
LIKELEM 0.419 0.137 3.054 0.00
LOC 0.062 0105 0.592 0.55
R = 0.58; R =0.57; N = 356

VARIABLES OUTPUT: LN KGHAPA
PARAMETER STANDARD ERROR T FOR Ho PROB
ESTIMATE PARAMETER = 0
Intercep 1.711 0.633 2.701 0.00
LN LABWPA 0.171 0.077 2.201 0.00
LN LABHPA 0.098 0.039 3.179 0.00
LN BOTTPA 0.347 0.094 3.662 0.00
LN FALL -0.185 0.123 -1.502 0.19
LN A Cost 0.293 0.092 3.173 0.00
LN Dist 0.212 0.083 2.548 0.01
TRAIN 0.562 0.130 4.322 0.00
LIKELEM 0.305 0.124 2.463 0.03
LOC 0.137 0.095 1.449 0.14
R2 = .52; R = .51; N = 355, LABWPA= women labor/ha, LABHPA=husband labor/ha, BOT-
TPA= cutting/ha, FALL- fallow period, A Cost = annual cost of tools/ha, Dist = market ac-
cess, TRAIN= training, LIKELEM = association of credit, LOC = location, TOTPA =
value/ha, KgHaPA = quantity/ha.
Source: Survey data.

T tests were performed to assess whether the individual regression coeffi-
cients were significant. The coefficients for women labor per hectare, husband
labor per hectare, number of cutting per hectare, cost of tools per hectare,
training, and credit association are all statistically significant in the two models
at 10 percent level.
The coefficient fallow period is significant in the model with the value of
cassava harvested per hectare, while the coefficient for access to market is
significant in the model with the quantity of cassava harvested per hectare.
However, the coefficients for location are not significant in the two models.
This result suggests that despite the ecological, social, economic, and cultural
diversity ofBandundu, the similarities of the factors affecting the productivity
of land for women farmers are more striking than the differences. The results
of the study indicate a diminishing marginal productivity of factors, which is
relevant with the principle of economic efficiency (Dillon and Handaker,
Vol. 4, No. 2, 1994






MPUTELA & KRAFT


Table 3. Cobb-Douglas type of production function of women farmers in Bandundu
in 1990.

VARIABLES OUTPUT: LN TOTPA
PARAMETER STANDARD ERROR T FOR Ho PROB
ESTIMATE PARAMETER = 0
Intercep 4.572 0.708 6.410 0.00
LN Colla H 0.100 0.032 3.607 0.00
LN BOTTPA 0.306 0.101 3.021 0.00
LN FALL -0.262 0.138 -1.898 0.05
LN A Cost 0.706 0.092 7.654 0.00
LN Dist 0.063 0.093 0.687 0.49
TRAIN 0.463 0.144 3.197 0.00
LIKELEM 0.487 0.137 3.556 0.00
LOC 0.15 0.102 1.474 0.14
R =.57; R2 =.56; N = 356

VARIABLES OUTPUT: LN KGHAPA
PARAMETER STANDARD ERROR T FOR Ho PROB
ESTIMATE PARAMETER = 0
Intercep 1.558 0.641 2.585 0.01
LN Colla H 0.062 0.029 2.122 0.03
LN BOTTPA 0.436 0.091 4.744 0.00
LN FALL -0.155 0.125 -1.241 0.21
LN A Cost 0.428 0.083 5.123 0.00
LN Dist 0.189 0.084 2.250 0.02
TRAIN 0.601 0.131 4.577 0.00
LIKELEM 0.371 0.124 2.993 0.00
LOC 0.223 0.092 2.412 0.00
R2 = .50, R = .49, N = 356, Colla H = husband labor/ha, BOTTPA = cutting/ha, FALL=
fallow period, A Cost = annual cost of tools/ha, Dist = market access, TRAIN = training,
LIKELEM = association of credit, LOC = location, TOTPA = value/ha, KgHaPA = quanti-
ty/ha.
Source: Survey data.

1980). The sum of the coefficients for woman labor/ha, husband labor/ha
seed/ha, and cost of tools/ha is .06 (value/ha) .91 (quantity/ha), suggesting
constant return to scale. This means that, for one hectare of land, increasing
these inputs by the fixed proportion will increase cassava harvested by the same
fixed proportion.
The adjusted coefficient for the categorical variable training indicates that
increasing the participation of women farmers irb training will increase the
productivity of cassava by 53 percent to 75 percent, all other inputs held
constant. The specific result of both models is thatwomen farmers' association
of credit is positively and significantly related to the productivity of cassava. An
increase in the participation of women farmers in credit association will
increase the quantity or the value of cassava harvested per hectare by 35
percent to 53 percent, all other inputs held constant.


Journal for Farming Systems Research-Extension






WOMEN FARMERS IN ZAIRE


To correct for some degree of multicollinearity between female labor and
male labor, and to test for the importance of collaboration of male labor in
cassava production in Bandundu, the ratio of husband labor per hectare and
woman labor per hectare (Colla H) was used in the model. The result suggests
that increasing the collaboration of male labor may increase the productivity
of cassava in Bandundu in terms of quantity or value of cassava harvested per
hectare. Table 3 shows the regression solution.


CONCLUSIONS
Despite its enormous resources such as land, labor, raw materials, and
hydroelectric energy, Zaire is still far from reaching its objective of self-
sufficiency in food. The ratio of food self-sufficiency defined as number of
calories domestically produced as a percentage of total calories supplied is
estimated at 96 percent, which is very low compared to the ratio in developed
countries such as the US with 131 percent. Food self-sufficiency entails food
security, defined as the ability to produce an adequate quantity of food
throughout the year, and remains a challenge for Zaire. The key to designing
an effective food security program is to have an empirical understanding of the
impact of various factors on the food production sector. Production functions
analysis provides a valuable tool for making suggestions and recommenda-
tions to farmers and government concerning food production decisions.
The hypothesis of this study is that women farmers in Bandundu can be
more productive if they have access to agricultural inputs. The problem was
to determine what inputs are more effective in increasing the production of
cassava, which is the basic food crop in Zaire, providing 75 percent of daily
calorie intake.
The first step for this study was to define a framework in which to collect
data about women farmers. A questionnaire was designed to collect the
information for the estimation of a production function. The questionnaire
was administered in three locations of Bandundu (Zaire) in a limited time
period. The results of the descriptive statistics indicate that women farmers in
Bandundu have a low productivity defined as the quantity or the value of
cassava harvested per hectare. For example, 73 percent of women farmers have
a yield of less than 2t/ha of cassava, which is very low compared to the world
average of about 9t/ha (Cock 1985). This low productivity can be explained
by the lack of control over land, the limited land cultivated by female farmers
in Bandundu, and the limited use of agricultural inputs.
Land, the basic factor of production in agriculture, is the most limited
resource for women farmers who have no right to the lands they cultivate; 82
percent of women farmers use less than one hectare, which is very small
compared to 20 hectares in other cassava growing countries (Cock 1985).
Women farmers, though not landowners, are still able to sell 47 percent of


Vol. 4, No. 2, 1994






MPUTELA & KRAFT


cassava harvested. The results indicate that women farmers in Bandundu use
more time in transporting and processing cassava than in the productive
operations such as preparing and hoeing, planting seeds, and weeding. They
have limited access to male labor. Seventy-seven percent of husband farmers
contribute less than 100 days/ha. Women farmers do not use improved
varieties and the number of cuttings per hectare used are less than in the other
countries that grow cassava (Cock 1985). The lack of availability of new tools
and cash credit to purchase inputs also explain this low productivity.
The result of analysis of variance shows that the means of cassava harvested
(kg/ha) in different locations were significantly different from each other at
10 percent level.
A Cobb-Douglas production function was used in this study of 360 women
farmers because it is the best form for small farm operations, convenient for
more than three independent variables, and allows the marginal productivity
of a given input to depend on the levels of all inputs employed.
Regression results demonstrate that the coefficients for woman labor,
husband labor, number of cuttings, cost of tools, training, and credit are all
significant and positively related to the value and the quantity of cassava
harvested. This means that women farmers in Bandundu not only need to have
the right to the land they cultivate, but they also need husband labor,
improved seed, fixed capital, access to extension services and access to credit
to make the land produce. The significant negative coefficients of fallow
period with yield and income suggest that women farmers also need chemical
fertilizer to use with natural fertilizer because increasing the number of years
of fallow after seven years might not improve the productivity of cassava.
Access to market is significant only for quantity of cassava.
This case of cassava production analysis confirms the conclusion of the
study ofGittinger et al. (1990) that increasing women's economic opportu-
nities is not a "zero sum game," because not only women farmers and their
husbands gain, but these gains also contribute to national food security.
The differences in the ecological, social, cultural, and economic diversities
of Bandundu are among the factors affecting women farmers' contribution to
cassava production. Direct access to traditional as well as modern factors of
production may improve the economic role of women farmers in Bandundu.
This research has to be extended to other regions in order to draw any general
conclusion for the country. Nevertheless, the results of the study suggest the
following recommendations:
* more attention has to be paid to women farmers in agricultural programs
of food self-sufficiency;
* each woman farmer needs to be secured with more land for food
production;
* further research is needed to reduce the time women use in cassava
processing;


Journal for Farming Systems Research-Extension






WOMEN FARMERS IN ZAIRE


* more women farmers have to be trained, and one of the solutions is to
train and motivate women extension agents to advise women farmers;
* women farmers' credit associations should be used as channels to finance
women's activities, and thus to facilitate their access to inputs; and
* means of transportation have to be available for women farmers and
extension services should be used to reduce time women farmers use to
transport their cassava from their fields to the local market or the local
storage facilities.


REFERENCES

Cock, J.H. 1985. Cassava: New potential for a neglected crop. Boulder, Colorado:
Westview Press.
Dillon, J.L., and J.B. Handaker. 1980. Production function and analysis. Farm Manage-
ment Research 13:103-118.
Food and Agriculture Organization (FAO). 1984. Socio-economic indicators relating to
the agricultural sector and rural development. Economic and social development
paper. Rome: FAO.
Fresco, L. 1982. Women and cassava production: An approach to improving agricultural
productivity in rural Zaire. USAID/Kinshasa.
Giovani, A.C. 1984. Farm size, land yields and the agricultural production function: An
analysis for fifteen developing countries. World Development 13(4):513-534.
Gittinger, J.P., S. Chernick, N.R. Horensters, and K. Salts. 1990. Household food
security and the role of women. World Bank Discussion Paper 96:1-37.
Intriligator, M.D. 1978. Economic models, techniques, application. Englewood Cliffs,
New Jersey: Prentice Hall.
Johnson, A., C. Martin, J.B. Johnson, and R.C. Buse. 1987. Econometrics: Basic and
applied. Mecheillon, Inc.
Kante, B. 1989. Production and estimation of production functions. In Effectiveness of
agricultural credit in Mali, Case study: The office du Niger. The Department of
Agribusiness, Economics SIU-C.
Kawagoe, I., Y. Hayani, and V.W. Rulton. 1985. The intercountry agricultural produc-
tion function and productivity differences among countries. Journal of Development
Economics 19:113-131.
Mosch, P.R. 1976. The efficiency ofwomen as farm managers: Kenya. American Journal
of Agricultural Economics 58:832-835.
Mputu, D., and M. Elengesa. 1987. La femme Zairaise et 1'autosuffisance alimentaire.
Pp. 1-7.
Norris, M.E. 1990. The impact of development on women: A specific factors analysis.
Research paper. Pp. 1-27.
Tshibaka, B.T. 1986. The effects of trade and exchange rate policies on agriculture in
Zaire. Research Report 56:11-54. Washington, DC.
Tshibaka, B.T. 1990. Prospects for increasing agricultural labor productivity under the
current resource base and technology. In Division and allocation of labor in a rural
household: Economy and the implications for the productivity of agricultural labor.
Washington, DC: IFPRI.


Vol. 4, No. 2, 1994






MPUTELA & KRAFT


Tshibaka, B.T., and K Lumpungu. 1983. Trends and prospects for cassava in Zaire.
Working Paper on Cassava 4:1-45. Washington, DC: IFPRI.
Upton, M. 1987. Estimation of production function. In African farm management.
Cambridge: Cambridge University Press. Pp. 137-173.


Journal for Farming Systems Research-Extension








Farmer-Controlled Diagnosis and
Experimentation for Small Rural
Development Organizations

Bruce Petch and Jane Mt. Pleasant



ABSTRACT

Small rural development organizations working in highly variable agro-
ecosystems must develop multiple technologies appropriate to the com-
plex conditions of the different settings in which they work. With little or
no access to institutional facilities or expertise, and few resources for
conducting conventional agronomic research, new approaches are re-
quired. This paper describes efforts to meet this challenge. New methods
to facilitate farmer participation in the generation and transfer of agricul-
tural knowledge were developed and tested in coordination with five rural
development organizations in seven remote upland areas in Indonesia
and the Philippines. Three topics are discussed: (1) farmer diagnosis of
crop production constraints; (2) farmer-controlled experimentation; and
(3) redefining roles of agriculturalists.
The diagnosis of crop production constraints focused on helping farmers
analyze their own situations rather than bringing in a multidisciplinary
team ofexperts. Not all ofthe techniques were successful,and someraised
additional problems. Involving farmers in agricultural experimentation is
extolled in the farming systems literature, but implementation can be
extraordinarily difficult. When farmers were involved in the planning and
design of experiments, the resulting heterogeneous set of treatments
greatly complicated efforts to synthesize information and draw conclu-
sions. Additionally, development organizations had to choose between
introducing new technologies and encouraging the development of local
methods. The complexity and implications of both of these issues are
illustrated with examples from farmer-controlled experiments in several
villages in one province. Agriculturalists working in areas where farmers,
rather than experts, have primary responsibility for technology develop-
ment are forced to redefine their roles. Several ways in which agricultur-
alists can contribute effectively in these situations are suggested.

1 Former graduate student, International Agriculture Program, and Assistant Professor, Depart-
ment of Soil, Crop, and Atmospheric Sciences, respectively, Cornell University, Ithaca, New
York, USA.






PETCH & MT. PLEASANT


INTRODUCTION
Critics of conventional research-extension approaches have called for "rever-
sals" in relationships between scientists and resource-poor farmers (Chambers
and Jiggins, 1987). Expanding the participation of these farmers in the
generation and transfer of agricultural knowledge requires fundamental
changes in the fabric of the agricultural research-extension system (Farrington
and Martin, 1988). The implications of increased farmer participation are not
trivial and successful adoption of new models of researcher-farmer interaction
is likely to be some time in coming. Even if such reversals are possible on a large
scale, the ratio of "converted" scientists to resource-poor farmers would still
be very small. This is an especially relevant concern in hilly and mountainous
areas where the diversity of agroecosystems complicates extrapolation of
experiences gained in one or two locations.
In Southeast Asia, small rural development organizations (mostly nongov-
ernmental) are at the forefront of efforts to develop improved farming systems
for upland areas. These organizations in remote upland areas of Indonesia and
the Philippines cannot effect fundamental changes in the research-extension
system, nor can they wait for others to implement them. Without regular
access to scientists or research facilities, they must diagnose agricultural
production problems in a range of agroecosystems and assist farmers in
generating locally appropriate technologies.
This paper describes some of the issues faced by small rural development
organizations in upland areas of Indonesia and the Philippines in developing
techniques that can be used by farmers and field workers to address system-
atically the constraints they face, with minimal input from outsiders. The
experiences and approaches described herein reflect the situation faced by
small NGOs in remote areas generally far from (or lacking vehicle access to)
research institutions, universities, and other institutional sources of expertise.
In other locations and in different types of organizations in both countries,
access to such facilities is much better and therefore the options for facilitating
technology development are greatly expanded.

Responding to the Variable Impact of Contour Hedgerows
Until recently, many organizations in the region built their programs
around contour planting of widely spaced, dense rows of leguminous trees
(also known as hedgerow intercropping, alley cropping, or sloping agricultur-
al land technology [SALT]). The appropriateness of this technology has been
proven in many areas, and thousands of upland farmers in the region have
adopted it. However, in some locations adoption has been very slow. Some
adopters complain that the effect on crop production has been negligible or
even negative. The reasons for such variability have not been assessed
systematically in the region. A review of research conducted in various
locations in the tropics indicates that competition between crops and hedg-


Journal for Farming Systems Research-Extension






FARMER-CONTROLLED EXPERIMENTATION


erows for nutrients and moisture is often a problem. Furthermore, although
the hedgerows contribute ample quantities of nitrogen to the cropping
system, deficiencies in other nutrients (notably phosphorus) are not necessar-
ily overcome (Petch and Mt. Pleasant, 1991).
The realization that the effectiveness of contour hedgerows is highly
variable has led to attempts to develop complementary or alternative crop
production technologies. As has already been learned, the heterogeneous
nature of the agroecosystems of the region works against the "discovery" of
any single technology or combination of technologies that would be appro-
priate and effective throughout an organization's area of service. Yet these
small organizations lack the capability to carry out conventional agronomic
research in a range of locations. This fact, combined with the desire to enhance
the self-reliance of local communities, necessitates that farmers be largely
responsible for technology development. The challenge addressed by the
work described herein is to develop procedures that small rural development
organizations can use to facilitate the generation of effective crop production
technologies by farmers.


STUDY AREA AND METHODS
This paper is based on work done in the provinces of Nusa Tenggara Timur
(Indonesia) and Cebu and Cavite (Philippines) from June to December 1990.
The locations of these provinces are illustrated in Figure 1. The climate in the
areas studied is generally subhumid. Topography is hilly to mountainous;
calcareous soils predominate. Major food crops are maize, upland rice, and
cassava. Seasonal food shortages are common in many locations, and famines
occur occasionally.
Five rural development organizations participated in the work described
herein. Fieldwork was done in seven villages served by the organizations.
Emphasis was on developing new approaches and procedures, rather than on
formal research. Discussions were held with farmers, fieldworkers and pro-
gram leaders regarding existing procedures for problem analysis and technol-
ogy development, focusing on food-crop production. Existing
technology-testing programs were modified, and new ones were established.
Some new procedures were designed and, in a few cases, tested. A detailed
report (Petch, 1991) was compiled as part of the first author's graduate
studies.

Participatory Diagnosis of Crop Production Constraints
Typically, diagnostic methods are applied to enhance researchers' under-
standing of local farming systems. As Gubbels (1988) points out, farmers may
already have the capability to analyze their own agricultural problems. He
suggests that the focus of "diagnosis" of farming systems by outsiders should


Vol. 4, No. 2, 1994





FETCH & MT. PLEASANT


SLowland Rice Areas

] Uplands


INDONESIA




NUSA TENGGARA TIMUR c

Figure 1. Location of study sites in the Philippines and Indonesia.
Source: Garrity and Sajise, 1990.


Journal for Farming Systems Research-Extension






FARMER-CONTROLLED EXPERIMENTATION


be to "learn only what is essential to know in order to effectively guide peasant
farmers to undertake their own analysis" (1988:11). We sought to develop
methods in which researchers facilitate farmers' analysis of their own situation
and problems. The primary goal was to help farmers overcome the constraints
they face; enhancing outsiders' understanding was secondary. In the event, we
were often perceived to be gathering information for our own purposes and
we found it difficult to avoid playing the role expected. Furthermore,
especially when outsiders are new to an area, a minimum understanding of
local farming systems is necessary before meaningful dialogue can take place.
Nevertheless, there were times when we were truly facilitators helping farmers
to analyze the problems they face.
Many of the techniques used were modifications of methods described in
recent farming systems research literature. One important difference from
most approaches was that the activities were carried out not by a multidisci-
plinary team of specialists, but by a pair offield workers and a group offarmers
in each location. The methods used included: (1) discussion of changes over
time in farming systems with older farmers; (2) tour of local farms to stimulate
analysis of issues; (3) discussion of constraints to crop production while
working with farmer groups; and (4) community meeting to consider jointly
possible solutions to the problems raised. Some progress was made in helping
farmers assess crop production constraints systematically. Table 1 provides an
example of the constraints identified in one location.

Difficulties Encountered
Many problems were encountered in applying these participatory tech-
niques. Two notable difficulties were: (1) influential individuals tended to
dominate discussions, despite efforts made by facilitators to constrain them;
and (2) in one instance, raising the issue of unclear land tenure led to a
proposal by landowners to adopt arrangements that would have been clearer
but less fair. Because the landowners are also the village leaders, the proposal
was difficult to reject and created a new problem that program staff had to
address.
The capacity of the community and the rural development organization to
address the problems raised is also a matter of concern. Establishing priorities
is essential (Lightfoot, et al., 1987). Even so, there is some risk of engendering
disappointment and cynicism when a wide range of problems are put forward
but follow-up action is limited to only one or two issues.

Table 1. Crop production constraints in an Indonesian village.
PROBLEM SUGGESTED SOLUTIONS
infertile soil intensified use of Sesbaniagrandiflora; alley cropping
livestock as pests fence pastures; keep animals in pens
tenure not clear fixed percentage of crop for landowners
untimely weeding establish a fixed weekly schedule for group work


Vol. 4, No. 2, 1994






PETCH & MT. PLEASANT


In addition, the value ofspending time on diagnostic activities was questioned
by program staff in some locations. They felt that local farmers and field staff
(who are often local farmers themselves, or at least village based) already know
what the problems are. This perspective results in a tendency, rightly or
wrongly, to attempt to solve apparent problems directly, rather than to
diagnose or analyze them. In a sense, the assumption is made that farmers in
an area have diagnosed their problems through years of experience. To
suggest that outsiders can better appraise the situation in a few days smells of
"expert arrogance."
In a different institutional environment (national agricultural research
services in Botswana and Cameroon), Baker states that "large investments in
formal diagnostic surveys or participatory activities such as group treks, farmer
groups, or village meetings" are not necessarily a prerequisite to "farmer-
based experimentation" (1991:144). Nevertheless, farmer-participatory ap-
proaches have focused more on diagnosis than on technology development.
However, there are at least two factors that could affect the relevance of
nonsystematic diagnosis made by outsiders or by a small number of local field
workers and/or farmers: (1) The diagnosis may reflect the situation and
perceptions of only one segment of the community, for example only relatively
wealthy or productive farmers, or only male farmers; (2) Especially in isolated
locations, farmers are often unaware of the potential for improvement of a
particular parameter and thus may not perceive a problem to exist. For
example, a tenant farmer who pays 60 percent of his harvest to the landowner
may not see it as a problem if all other tenants in the area do the same.
In both cases, the perspective ofan outsider can contribute to ensuring that
the problems being addressed are indeed relevant to a large portion of the
farmers in an area. If there is widespread agreement on the nature and priority
of problems, diagnostic activities may be superfluous.


FARMER-CONTROLLED EXPERIMENTATION
The abundant rhetoric about farmer participation in research found in recent
farming systems literature is not matched by serious consideration of the
logistical issues involved. In the case of small rural development organizations
working in heterogeneous environments, the relative participation of re-
searchers and farmers in experimentation is determined more by lack of
research staff than by ideological considerations. The main challenge con-
fronting these organizations is not to get more farmer participation but rather
to expedite the generation and dissemination of knowledge and technologies
with a very small number of technical and professional staff.
Heinrich and Masikara (1991), based on several years of experience in
Botswana, provide some useful guidelines for organizing farmer-managed
trials when the researcher-to-farmer ratio is very low: (1) work with groups of


Journal for Farming Systems Research-Extension






FARMER-CONTROLLED EXPERIMENTATION


farmers rather than individuals; (2) do as much preparatory work as possible
before the cropping season; (3) use trial designs that can be implemented by
farmers and readily assessed by both farmers and researchers; (4) minimize the
quantitative data collected and find efficient ways of collecting relevant
qualitative data; and (5) make the data analysis simple enough to be handled
by available staff and equipment. They also caution that such a participatory
approach to experimentation "is not well-suited to collecting large amounts
of empirical, technical data on specific technologies" (1991:8). The primary
objective ofHeinrich and Masikara was to include the perspectives of farmers
in developing new technologies.
Gubbels (1988) reports a similar approach to farmer experimentation that
World Neighbors has used in Mali. In the case described, each village selected
"pilot farmers" to test a new variety of seed on their own land. Results were
analyzed in a meeting of representatives from all the villages involved. The
recommendations for extension that were compiled at the meeting were not
based on "scientifically rigorous experimental data," but rather on the
observations of participating farmers and their knowledge of local conditions.
In the programs described in this paper, two key issues complicate efforts
by program staff to facilitate the process of technology development by
farmers and to synthesize information gained for the benefit of farmers in
other areas: (1) heterogeneity of experimental design and treatments due to
farmer control of the research process; and (2) the "dilemma" between
introducing new technologies or facilitating indigenous technology develop-
ment. These issues are illustrated by experiences in the province of Cebu.

Effect of Farmer Control on On-farm Trials in Cebu
In Cebu, in-row tillage (cultivation of 50 cm planting strips, leaving
interrow space uncultivated) and herbaceous green manure crops were
selected for testing by a local rural development organization, based on
experience with the technologies in Central America. Initially, a conventional
agronomic approach was planned in which identical trials would be imple-
mented in the three districts in which the organization worked. But when
farmer-cooperators in each district were involved in planning and design,
there was considerable variation in treatments and experimental designs
between districts, and to a lesser degree between cooperators in each district.
The variation reflected both the agroecological variability of the area and the
heterogeneity of farmer preferences and interests. The result was a rather ad
hoc experimental program (summarized in Table 2) that did not adhere to the
norms of agronomic research (a statistician's nightmare!).
In the program area of Guba, where soils are mostly stone free, friable, and
slightly acidic to neutral, herbaceous green manures were enthusiastically
included in the experimental program. The green manures were perceived by
farmers as a means of providing nitrogen and organic matter as a supplement


Vol. 4, No. 2, 1994






PETCH & MT. PLEASANT


Table 2. Adaptation of experimental program by farmers in three locations in Cebu,
the Philippines (simplified for purposes of illustration).

LOCATION TECHNOLOGIES INTRODUCED TECHNOLOGY COMBINATIONS
FOR TESTING SELECTED BY FARMERS
FOR TESTING
Guba green manures, green manures +
in-row tillage alley cropping + fertilizer
Argao green manures, green manures +
in-row tillage animal manure + fertilizer
Pinamungajan green manures, in-row tillage + goat manure,
in-row tillage in-row tillage + green manures

to alley cropping, which had already been widely adopted in the area. They
hoped that the green manures would reduce the amount of chemical fertilizer
they needed to get acceptable maize yields. There was little interest in in-row
tillage among participating farmers.
In the program area ofArgao, soils are shallow over limestone, and alkaline.
Alley cropping has never been as effective or as widely adopted as in Guba.
Farmers in Argao had been experimenting with different combinations of goat
manure, chicken manure, and various fertilizer formulations. They were
interested in trying green manures as another element in their mix of manures
and fertilizers. As in Guba, in-row tillage generated little interest.
In the third program area, Pinamungajan, the dominant soil type is
comprised of more than 50 percent stones. Maize yields are generally far lower
than in the other two sites, and the potential for improving production seemed
limited given the soil condition. Attempts at alley cropping had produced
small, spindly hedgerows that contributed nothing to the soil in the alleys. But
when stones were removed from and animal manure added to narrow strips
in an in-row tillage system, maize yields were spectacular (compared to past
experience). The experimental plan thus became focused on various modifi-
cations to in-row tillage.

Introduce New Technologies or Facilitate Indigenous Technology
Development?
Variability in the modification of introduced technologies is only one
dimension of the complexity of farmer experimentation. An additional
dimension is the experiments that farmers do on their own initiative. Although
the current fashion is to reject the technology transfer approach and instead
facilitate development of indigenous technologies, provision of new ideas and
improved access to resources remains important to farmers. As Baker states,
"what most farmers want and need is new technological options, not catalysts
to their own innovation processes" (1991: 128). Nevertheless, the folly of
ignoring farmers' knowledge and experiments is well documented (e.g.,


Journal for Farming Systems Research-Extension






FARMER-CONTROLLED EXPERIMENTATION


Richards, 1985). Furthermore, the objectives of rural development organiza-
tions are not only technological. They also aim to enhance the self-reliance and
confidence of farmers who, in many locations, have long been led to believe
that they are backward, lazy, or both. Provision of new ideas should at least
be accompanied by understanding of, and respect for, existing farming
methods and farmer innovation. Ideally, an approach combining technology
transfer with support for local knowledge seems desirable.
In practice, the holistic alternative of incorporating introduced technolo-
gies with indigenous technologies and farmer experimentation is difficult to
monitor. For example, while visiting the introduced green manure trials of one
farmer-cooperator in Cebu, we passed by the following experiments (among
others): fertilizer + prunings of Gliricidia sepium + Cassia siamea versus
fertilizer alone; prunings versus fertilizer; and goat manure + fertilizer versus
fertilizer alone.
For staff already overburdened and confused by the range of modifications
to the technologies they introduced, the addition of a whole set of farmer-
specific experiments is difficult to take into account. In a different institutional
environment, the "FSR team" might be able to put together a coherent
picture of on-farm technology development processes. However, for one or
two practitioners working with tens or even hundreds of farmers, a different
approach is necessary.

REDEFINING ROLES
The agriculturists hired to "do research" in such situations are confused. None
of the experimental approaches they learned at university or research stations
seems to fit the rather ad hoc and site-specific trials that result from involving
many farmers in different locations in the planning process. Furthermore,
because farmers are supposed to be taking primary responsibility for technol-
ogy development on their own farms, the role of and need for researchers from
outside is unclear. The conventional role of protesting technologies that are
then disseminated among farmers is inappropriate. This lack of clear direction
and purpose has affected the motivation of the agriculturists. In some cases
they have slipped into routine program administration tasks; in others they
have set up neat randomized block variety trials in their backyardss," and
elsewhere they have undertaken surveys peripheral to program direction.
There is, however, a gradual discernment of how professional agriculturists
can most effectively contribute to a technology-development program in
which farmers have primary responsibility. The following roles seem to be
appropriate:
1. Assisting farmers and village-based program staff in analyzing the
problems theyface and assessingpossible solutions. Given the doubts raised above
regarding diagnostic activities, this role may not always be considered useful.


Vol. 4, No. 2, 1994






PETCH & MT. PLEASANT


2. Introducing new technologies toprogramstaffandfarmers. This role is not
greatly different from the conventional role of extension agents, except that
the researcher has neither the time, the facilities, nor the mandate to pretest
new technologies. Therefore, the research must provide farmers and field staff
with balanced information on the likely advantages and disadvantages of a
particular technology and avoid the promotional approach that has prevailed
in the past. The agriculturists should also share insights resulting from their
broader exposure to the world outside of the village, for example by providing
information on the likely trend in fertilizer prices.
3. Enhancing local capacity to test and develop technologies. If agriculturists
can see through the plethora of experimental designs that they have been told
constitute "research," they can provide useful advice on technology develop-
ment and basic principles of experimentation. A simple example was observed
on a steeply sloping field. The farmer planned to have treatments running
parallel to the contour. The agriculturist recommended that the treatments
run perpendicular to the contour, given that the soil on the upper slope was
much thinner than on the lower slope.
4. Providing farmers and program leaders with insights on variation in the
performance of different technology combinations between locations (Kirkby,
1981). Such insights are essential to avoid "reinventing the wheel" in each
new program area. To gain these insights, the research advisors have no
alternative but to synthesize the incomplete and largely qualitative data that
are being generated. As described above, such data will come both from
testing of introduced technologies and from farmer-initiated experiments, as
well as from observations of "normal practice." Initially, efforts will have to
focus on procedures for collecting and recording such information in a
systematic manner. The increasing use of laptop computers by small rural
development organizations in Southeast Asia should facilitate this process. So
far, the computers are used almost exclusively for word processing. Setting up
a simple database to keep track of the diverse information available may be a
useful first step. The use of nonparametric statistics and simple techniques
such as modified stability analysis (Hildebrand, 1990) might be considered at
a later stage.


CONCLUSION
The experiences described above represent a very preliminary attempt to
develop technology development procedures that do not rely on "experts."
Although the geographic scope was limited, the general situation is far from
unique. In the tropics, many farmers are seldom if ever reached by national
agricultural research services, international centers, or bilateral aid projects.


Journal for Farming Systems Research-Extension






FARMER-CONTROLLED EXPERIMENTATION


While waiting for "reversals" of scientists' attitudes to occur, perhaps more
farming systems research practitioners could focus their attention on develop-
ing approaches that can be implemented where there are no "experts."


ACKNOWLEDGMENTS

The Canadian International Development Agency provided financial support
for the research upon which this paper is based. Cuso (a private Canadian
development organization) provided logistical and financial support in South-
east Asia. The examples described from Indonesia and the Philippines are
based on the work of the Geo Meno Foundation and the Mag-Uugmad
Foundation, respectively. Discussions with Larry Fisher and John Jackson of
World Neighbors Southeast Asia were of particular importance to the evolu-
tion of the ideas presented. The assistance provided by the above organiza-
tions and individuals is gratefully acknowledged.


REFERENCES

Baker, D. 1991. Reorientation, not reversal: African farmer-based experimentation.
Journal of Farming Systems Research-Extension 2:125-147.
Chambers, R., and J. Jiggins. 1987. Agricultural research for resource-poor farmers, Part
I: Transfer of technology and farming systems research. Agricultural Administration
and Extension 27:35-52.
Farrington, J., and A. Martin. 1988. Farmer participation in agricultural research: A
review of concepts and practices. London: Overseas Development Institute. 79 pp.
Garrity, D.P., and P.E. Sajise. 1990. Sustainable land use systems research in Southeast
Asia: A regional assessment. Paper presented at the workshop on sustainable land use
systems research and development, New Delhi, India.
Gubbels, P. 1988. Peasant farmer agricultural self-development. ILEIA Newsletter
4(3):11-14.
Heinrich, G.M., and S. Masikara. 1991. Trial designs and logistics for farmer-implement-
ed technology assessments with large numbers of farmers: Some approaches used in
Botswana. Paper presented at the 11th Annual Symposium of the Association for
Farming Systems Research-Extension, Michigan State University, East Lansing,
Michigan, October 5-10.
Hildebrand, P.E. 1990. Modified stability analysis and on-farm research to breed specific
adaptability for ecological diversity. Pages 169-180. In M.S. Kang, (ed.) Genotype-by-
environmentinteractionandplant breeding. Baton Rouge: Louisiana State University
Press.
Kirkby, RA. 1981. The study of agronomic practices and maize varieties appropriate to
the circumstances of small farmers in highland Ecuador. Unpublished Ph.D. disser-
tation, Comell University, Ithaca, NY.


Vol. 4, No. 2, 1994






PETCH & MT. PLEASANT


Lightfoot, C., O. de Guia, Jr., A. Aliman, and F. Ocado. 1987. Participatory methods for
identifying, analyzing, and solving systems problems. Paper presented at the 7th
Annual Farming Systems Research-Extension Symposium, University of Arkansas,
Fayetteville.
Petch, R-B. 1991. Working with upland farmers to grow more food: Strategies from
Indonesia and the Philippines. Unpublished MPS project report, Cornell University,
Ithaca, NY.
Petch, R.B., and J. Mt. Pleasant. 1991. Agronomic limitations ofalley cropping:A review.
Department of Soil, Crop and Atmospheric Sciences. Research Series R91-8, Cornell
University, Ithaca, NY.
Richards, P. 1985. Indigenous agricultural revolution: Ecology and food production in
West Africa. London: Hutchinson. 192 pp.


Journal for Farming Systems Research-Extension








Participation of Rural Women
in the Homestead Vegetable Farming
Systems of Bangladesh'

W.A. Shah, Rukshana Tasmin, Rezaul Karim, and M.M.A. Karim2



ABSTRACT
Due to increasingly limited access to land and highly inadequate nutrition
among rural women in Bangladesh, the analysis of and strategies to
improve women's participation in vegetable production activities on
farmer homesteads are urgently needed. This report is based on research
carried out by the On-Farm Research Division of the Bangladesh
Agricultural Research Institute from 1987 to 1989. The nature and
extent ofwomen's participation in vegetable farming systems is a function
of various social and socioeconomic factors, which suggest a number of
policy implications.


INTRODUCTION
The word "homestead" brings to mind two different images: urban home-
steads (homes and gardens in city settings) and rural homesteads (with homes,
sheds for cattle and poultry, and fruit and vegetable gardens). Rural home-
steads are common in many developing countries. In Bangladesh, homesteads
occupy about five percent of the total cultivable land (Taherunnesa, 1986).
Homestead areas vary with farm size and have a highly skewed distribution
(Hussain, et al., 1988). Of the 12 million rural homesteads in Bangladesh, 56
percent have only 0.004 to 0.04 ha, approximately 17 percent have 0.45 to
0.81 ha, and less than 10 percent own more than 0.81 ha (Islam and Rahman,
1989). The remaining 17 percent own no land. Thus two million people do
not own their own homesteads; however, they usually reside on someone
else's land as sharecroppers or wage laborers.
During the past few decades, the thrust of agricultural research and
development has been on field crops, and the importance ofsmallholdings was
mostly ignored. Only a few contemporary studies (Hussain, 1980; Gill and
Sultana, 1982; Islam and Ahmed, 1986-87) highlight the importance of the
1 Paper presented at the Tenth Annual Association for Farming Systems Research-Extension
Symposium, Michigan State University, East Lansing, MI, October 14-19, 1990.
Farming Systems Research team members, On-Farm Division, Bangladesh Agricultural Research
Institute, Ishurdi, Bangladesh.






SHAH, ET AL.


use of homestead land for agricultural production. In Bangladesh, population
pressure has caused the percentage of landless families to increase, posing a
serious threat to rural development. The technologies generated for field
crops are of little use to large groups of landless families. Therefore, these farm
households need appropriate strategies for income generation and/or expen-
diture savings.
The nutritional status of rural people is of great concern to the government
of Bangladesh. The present average food intake of rural women is deficient by
almost 40 percent (INFS, 1977) and 85 percent of rural women suffer from
iron and protein deficiencies (Khan, 1988). Ensuring a sustainable and
balanced diet has become one of the primary concerns of policy makers.
However, there are several controversial issues relating to nutrition and
income. Income elasticity and demand for nutritionally rich food by poor
families are extremely low (Lipton, 1988). Bouis and Haddad (1988) ob-
served that a rise in household income of 20 percent for poor farm families led
to a rise in nutritional food intake of only one percent. These propositions
were not tested in this study. However, we assumed that increased ingestion
of vegetables increases nutritional status and health, and thus labor supply
(Lipton, 1988).
Researchers at the Bangladesh Agricultural Research Institute (BARI) On-
Farm Research Division (OFRD) have performed a series of vegetable-
production research activities on farmer homesteads. The results demonstrate
that it is possible to increase the cultivation of different types of vegetables and
that this can generate a considerable amount of income (Shah, et al., 1990).
Women in rural areas of Bangladesh perform a broad spectrum of agricul-
tural activities, including fruit, vegetable, poultry, livestock, and fisheries
production. Therefore, there is a need to develop technologies that benefit
rural women. Although OFRD has studied homestead vegetable production
systems, the role of women has not been assessed. The purpose of our study
has been to analyze the nature and extent of women's participation in
homestead production systems and includes the following objectives:
1. To describe the socioeconomic characteristics of the study's respon-
dents;
2. To determine the relationship between the nature and extent ofwomen's
participation in homestead farming systems; and
3. To derive a development policy for homestead farming systems.


RESEARCH METHODS
From 1987 to 1989, OFRD initiated a program to improve homestead areas
by introducing and testing different vegetable and fruit species. The farm sizes
of cooperating families were initially categorized as landless (less than 0.2 ha),
marginal (0.21-0.50 ha), small (0.51-1.0 ha), medium (1.01-1.99 ha), and


Journal for Farming Systems Research-Extension






BANGLADESHI WOMEN & VEGETABLE FARMING


large (above 2.0 ha). Each family grew vegetables in five I x 6m beds. The
homestead vegetable farming systems (HVFS), locally known as the "Ka-
likapur Model," comprised five vegetable cropping patterns:
1. Red amaranthus (Amaranthus gangeticus)-Red amaranthus-Indian
spinach (Basella alba and B. rubra)-Radish (Raphanus sativus)-Tomato
(Lycopenican esculentum Mill);
2. Okra (Abelmoschus esculentus)-Brinjal (Solanum melongena)-Red ama-
ranthus-Red amaranthus;
3. Red amaranthus-Radish and Indian Spinach-Garlic (Allium sativum);
4. Kang kong (Ipomoea reptans)-Red amaranthus-Batisak (Brassica chien-
ensis)-Orion-bitter gourd; and
5. Red amaranthus-Brinjal-Red amaranthus-Cabbage (Brassica oleracea
var. capitata).
In a 1989 questionnaire survey, all 50 female farmers from the cooperating
households were interviewed by a female enumerator to determine their
perceptions of the usefulness of the model and the extent of their participation
in the different activities of vegetable production.

RESULTS AND DISCUSSION

Socioeconomic Distributions
The majority of cooperating households in the program fall into the
landless (30 percent) and marginal (24 percent) categories. Women farmers
make up 18.9 percent of household members in both the small- and medium-
size farm cooperators. Furthermore, if the farm categories are regrouped into
small (less than 1.0 ha), medium (1.0-2.0 ha), and large (more than 2.0 ha),
73 percent of the women are in the small-farm category. We conclude that
technology-transfer programs in Bangladesh should focus on such small farms
in order to reach the greatest number of women (Table 1).
For the survey sample, family size averages 6.24, which is close to the
national average. The majority of women respondents (70 percent) have
between five and nine people in their families (Table 2). The mean number of
years ofschooling for the women is about one and a half. Only 16 percent have
between one and five years of schooling, and 73 percent of the women have

Table 1. Distribution of farm categories among the respondents.
FARM CATEGORY FREQUENCY % OF TOTAL
Landless (less than 0.20 ha) 11 29.8
Marginal (0.21-0.50 ha) 9 24.3
Small (0.51-1.00 ha) 7 18.9
Medium (1.01-1.99 ha) 7 18.9
Large (above 2.00 ha) 3 8.1
TOTAL 37 100.0


Vol. 4, No. 2, 1994






SHAH, ET AL.


Table 2. Distribution of family size among the respondents.
FAMILY SIZE FREQUENCY % OF TOTAL
Less than 5 members 7 19
5-9 members 26 70
Above 9 members 4 11

Table 3. Distribution of years of schooling among the respondents.
YEARS OF SCHOOLING FREQUENCY % OF TOTAL
No schooling 27 73
1-5 years 6 16
6-10 years 4 11
Above 10 years 0 0
TOTAL 37 100
MEAN 1.51 -

no schooling (Table 3). In 1974, about 89 percent of women had no formal
schooling (Khan, 1988); thus our study indicates a possible rise in women's
literacy rates in the rural areas of Bangladesh.

Social Obligations
Certain social customs discourage active participation of women outside of
the home and potentially limit participation of women in HVFS activities.
Purdah, the seclusion of women from public observation, is practiced widely
by women from Moslem families, which make up about 80 percent of the
families in Bangladesh. In the purdah system, women are not allowed to work
outside the home. Nevertheless, 62 percent of the women interviewed in this
study felt no social obligation restricting their participation in HVFS activities.
A total of 29.7 percent reported that some social obstacles existed, but that
they participated because of financial need. About eight percent did not
participate due to social dictates (Table 4). This suggests that the traditional
social barriers have changed substantially and more women have access to
work outside of their homes.

Husband's Attitude
The government's second five-year plan (1980-85) for development
emphasized the importance of congenial socioeconomic conditions to ensure
greater participation of women in economic activities. This called for a change
in traditional social attitudes; more specifically, a change in men's attitudes
toward women.
The results of our study support the generalization that, in rural families,
the husband dictates the activities of his wife. The majority of husbands (78.4
percent) allowed their wives to participate in HVFS. However, about one-


Journal for Farming Systems Research-Extension






BANGLADESHI WOMEN & VEGETABLE FARMING


Table 4. Distribution of social obligation of women respondents.

SOCIAL OBLIGATION FREQUENCY % OF TOTAL
No obligation 23 62.16
Obligation exists, but is
not followed due to necessity 11 29.73
Obligation exists and is always observed 3 8.11

TOTAL 37 100.00

Table 5. Husband's attitude toward wife's participation.

ATrITUDE FREQUENCY % OF TOTAL
Does not allow to participate 6 16.22
Sometimes allows to participate 2 5.41
Always allows to participate 29 78.37

TOTAL 37 100.00

Table 6. Women's level of awareness about homestead vegetable production systems.

LEVEL OF AWARENESS FREQUENCY % OF TOTAL
Not aware 7 19
Moderately aware 11 29
Aware 19 52

TOTAL 37 100.00

sixth (16 percent) of the women were not allowed by their husbands to
participate (Table 5) because of religious restrictions such as purdah.

Awareness
The study assessed the degree of awareness of women about newly
introduced homestead technologies. The series of on-farm activities promot-
ing different vegetable production possibilities directly affected the awareness
and motivation ofwomen with regard to HVFS. The results indicate that, after
the program, more than 50 percent of the women were aware of the
importance of homestead vegetable production (Table 6).

Type of Participation
Three categories characterize the participation of women in homesteading
activities: (1) no participation, (2) participation as initiator or motivator (no
physical participation in any of the activities), and (3) physical participation in
activities. The majority of women respondents participated physically in
harvesting, irrigating gardens, and storing seeds (Table 7). The women were


Vol. 4, No. 2, 1994






SHAH, ET AL.


Table 7. Nature of women's participation in homestead vegetable production.

ACTIVITIES No PARTICIPATION As INITIATORS PARTICIPATED
/MOTIVATORS PHYSICALLY
Land preparation 54 5 41
Weed management 46 0 54
Fertilization 65 3 32
Irrigation 16 11 73
Harvesting 11 8 81
Storing seeds 16 0 84
Others (e.g., like fencing) 57 0 43


Table 8. Nature of participation in relationship to selected variables.

SELECTED VARIABLES CORRELATION COEFFICIENTS
Age -0.2925
Family size -0.2106
Farm size -0.4345
Years of schooling -0.1177
Family income -0.2300
Social obligation -0.0203
Husband's attitude 0.4883
Awareness 0.84762
'Significant at 5% level.
2Significant at 1% level.

not as involved in fertilization and fence-making as they were in other
activities. The results indicate that women participated very little as motiva-
tors.
Eight variables were correlated using Pearson's Product Moment with the
level of women's participation in HVFS (Table 8). The results reveal that farm
size, husband's attitude, and awareness about vegetable production technol-
ogies were significantly related to women's participation.
Among socioeconomic factors (including age, family size, farm size, and
years of schooling), farm size was the only variable inversely related to
women's participation (Table 8). A similar negative relationship was also
observed in rice-farming areas in the Philippines (Shah, 1989) and India
(Agarwal, 1985). This association may be related to family affluence: in
wealthy families (with larger farms), family labor tends to decrease. This is
particularly true for household women, who are replaced by hired women.
Two sociocultural factors (social obligation and husband's attitude) were
included in the correlation. Social obligations and cultural values in the
traditional Bangladesh society obstruct women's participation in farming
activities. The correlation test suggests that no significant association exists
between the level of women's participation and social obligations, whereas
husbands' attitudes are significantly associated.


Journal for Farming Systems Research-Extension






BANGLADESHI WOMEN & VEGETABLE FARMING


The positive association between husband's attitude and the type of
participation by his wife suggests that as the husband becomes aware of the
need and usefulness of her participation, the wife's involvement in homestead
farming systems tends to increase. Because the majority of rural households
fall into the landless, marginal, and small-farm categories, their subsistence
needs override cultural values and attitudes. Masood (1988) notes that
perceived economic benefit is an important determinant in women's partic-
ipation in rural development programs.
Furthermore, the level of awareness is significantly related to the level of
women's participation in homestead farming. This positive association sug-
gests that as the level of awareness about newly introduced HVFS increases,
the participation of women will also increase. Awareness is endogenous in
nature, reflecting individual behavior through psychological processes, and is
influenced by cognition of what an individual thinks, feels, believes, and
anticipates (Shah, 1989). Therefore, level of awareness can also be increased
through personal contacts, radio programs, and other popular means of mass
communication, and, to a lesser degree, through leaflets and booklets.

Extent of Participation
Participation was measured at three levels: (1) never, (2) sometimes, and
(3) always. More than 92 percent of the women respondents participated in
some kind of activity in the HVFS (Table 9). This suggests that the employ-
ment of rural women in income-generating and/or expenditure-saving
activities is possible in rural development programs. However, it may be
necessary to estimate the willingness and ability of women to supply labor for
homestead farming in order to determine the availability of women for labor,
and also to determine the social, nutritional, and economic advantages of
activities related to newly developed HVFS.
Statistical analysis (Pearson's Product Moment) indicates that age, hus-
band's attitude, and awareness are significantly related to the extent of
women's participation in homesteading activities (Table 10). The age of
women was negatively related, suggesting that as the ages of rural women
increase, the extent of their participation decreases. Younger women (25-34
years old) participated more than older women. Husband's attitude and the
level of awareness correlated positively with the extent of women's participa-
tion.

Table 9. Extent of women's participation in homestead vegetable production.
EXTENT OF PARTICIPATION FREQUENCY % OF TOTAL
Never participated 3 8.11
Sometimes participated 14 37.84
Always participated 20 54.05
TOTAL 37 100.00


Vol. 4, No. 2, 1994






SHAH, ET AL.


Table 10. Relationship between the extent of women's participation and selected vari
ables.

VARIABLES CORRELATION COEFFICIENTS
Age1 -0.4259
Family size 0.0565
Farm category -0.2171
Years of schooling -0.0551
Family income -0.0893
Social obligation -0.1192
Husband's attitude 0.3208
Awareness 0.6814
ISignificant at 5% level of probability.


CONCLUSIONS AND POLICY IMPLICATIONS

The importance of sound plans and policies as a means of maximizing the use
of scarce resources in socioeconomic development processes is obvious.
However, rigorous analysis of data may not protect planners adequately
against the pursuit of flawed policies or blindness to better alternatives. The
difficulty in collecting reliable data and of monitoring programs and policies
in the field often hinders the application of more appropriate alternatives.
Results generated in this study, which was conducted in a controlled
setting, may not provide sufficient evidence for wider implications regarding
the integration of women into homestead vegetable farming. Rather, the
experience of the researchers involved highlights the possibilities for further
research on such issues. However, based on our findings, the following
recommendations are made:
1. In a broad sense, homestead vegetable farming not only maximizes the
use of scarce resources, but also leads to increased income generation,
improvement of nutritional status, and employment of rural women.
2. The study revealed that women are interested in adopting such technol-
ogy. However, transfer of an innovative technology from one location to
another is complicated. At this stage, similar tests should be carried out in
representative farms households in other settings in order to determine how
social and cultural constraints effect homestead vegetable gardening in other
areas. Simultaneously, plans should be undertaken for technology dissemina-
tion.
3. Both husband's attitude toward his wife's participation and women's
awareness of homestead vegetable farming technology were related signifi-
cantly to the nature and magnitude of participation. Information dissemina-
tion regarding the importance of women's participation in homesteading
activities may be accomplished through the mass media and existing extension
services. Training programs for field-level extension workers may also increase
the awareness of such programs.


Journal for Farming Systems Research-Extension






BANGLADESHI WOMEN & VEGETABLE FARMING


4. The HVFS may be extended to similar agroecological zones, which will
require the involvement of existing extension services.
5. The HVFS technology needs further refinement in diverse agroecolog-
ical conditions. Furthermore, the effect of supply of and demand for female
labor should be studied carefully.
6. Marketing facilities for HVFS products, women's access to credit
markets, and technology adoption need to be studied.


ACKNOWLEDGMENTS

The authors thank Mr. Kenneth D. Swann, Communication Specialist,
PacMar, Inc.; Dr. M. Zainul Abedin, Chief Scientific Officer and Head, BARI
OFRD; and Drs. Kamal Uddin Ahmad and R.N. Mallick, FSR Technology
Specialists, Checchi and Company Consulting, Inc. for their suggestions,
comments, and editing of this paper. We also thank Dr. M.H. Mondal,
Director General, BARI, and Dr. Hamizuddin Ahmed, Director, Training
and Communication, BARI, for allowing us to present this paper at the Tenth
Annual AFSRE Symposium. Finally, we are grateful to the Symposium
Committee for providing travel support.


REFERENCES

Agarwal, B. 1985. Rural women and high yielding rice technology in India. Pages 164-
171. Women in Rice Farming Systems, International Rice Research Institute, Los
Bafios, Philippines.
Bouis, H., and L. Haddad. 1988. Comparing caloric-income elasticities. IFPRI, Wash-
ington, D.C. Mimeograph.
Gill, G.J., and W. Sultana. 1982. Women's role in small farm production and resource
management in Bangladesh. The ADC Inc., Dhaka. Pages 15-21.
Hussain, A. 1980. Vegetable gardening in the homestead area of Bangladesh. Fifth
Workshop for the District Representatives of UNICEF, Dhaka..
Hussain, M.S., M.Z. Abedin, M.A. Quddus, S.M. Hossain, Banu, and Ahmed. 1988.
Women's contribution to homestead agricultural production systems in Bangladesh.
Bangladesh Academy for Rural Development, Comilla.
Institute of Nutrition and Food Science (INFS). 1977. Nutrition survey of rural
Bangladesh. University of Dhaka, Bangladesh.
Islam, M., and D. Ahmed. 1986-87. Analysis of homestead production and utilization
systems. BARI, OFRD Farming Systems Research Report. Mimeograph. 28 pp.
Islam, M., and Rahman. 1989. Homestead farming in rural Bangladesh: A case study.
Economic Affairs 34(2):89.
Khan, S. 1988. The 50percent: Women in development and policy in Bangladesh. Dhaka:
The University Press Limited.
Lipton, M. 1988. Attacking under-nutrition and poverty: Some issues of adoption and
sustainability. International Food Policy Research Institute, Washington, D.C.


Vol. 4, No. 2, 1994






SHAH, ET AL.


Masood, F. 1988. Women in traditional irrigated farming systems. Women in Rice
Farming Systems, International Rice Research Institute, Los Bahos, Philippines.
Shah, W.A. 1989. Women's participation in rice farming systems of Nueva Ecija,
Philippines. Unpublished Ph.D. dissertation, Central Luzon State University, Mu-
noz, fhilippines
Shah, W.A., Rezaul Karim, and M.A. Karim. 1990. Economics of homestead vegetable
farming systems: Kalikapur model. BARI, OFRD Farming Systems Research Report.
Mimeograph. 5 pp.
Shah, W.A., Salima Jahan Nury, and M.A. Karim. 1990. Women's participation in
vegetable seed storage methods and practices in farm households. BARI, OFRD
Farming Systems Research Report. Mimeograph. 31 pages.
Taherunnesa, A.A. 1986. Home-based agricultural production in rural Bangladesh.
Association of Development Agencies in Bangladesh (ADAB) News 13(5):2-9.


Journal for Farming Systems Research-Extension








Role of Farmers in the Evaluation of an
Improved Variety:
The Case ofS35 Sorghum in Northern
Cameroon 1

Mulumba Kamuanga and Martin Fobasso2


ABSTRACT
Efforts to develop high-yielding, stable, early-maturing sorghum variet-
ies in northern Cameroon produced an apparently resounding success
with on-farm tests of the variety S35 from 1983-1985. This caused great
excitement and led to the recommendation of S35 to the regional
extension agency in 1986. In subsequent years, under more favorable
rainfall conditions, farmers reported serious agronomic problems with
the variety. However, informal surveys revealed some farmer-initiated
strategies for incorporating S35 into their traditional cropping systems.
In 1990 an adoption survey to measure research impact revealed that 13
percent of farmers had adopted the new variety. Results of regression
models suggestthatfarmers tend to adopt S35 more on its own merit than
as part ofa package ofrecommendations. Adoption rate is higher among
farmers who planted improved sorghum three to five years ago, and
location in drought-prone zone is determinant. Reasons for nonadop-
tion, together with these results, now form the basis for revising breeding
objectives in order to respond better to farmers' needs.

INTRODUCTION
On-farm testing of improved varieties and agronomic practices developed by
the Institute of Agronomic Research (IRA) at Maroua, northern Cameroon,
has been conducted by two USAID-funded projects since 1979. The Semi-
Arid Food Grain and Development Project (SAFGRAD) conducted on-farm
trials on food crops until 1986. As SAFGRAD phased out, the National
Cereals Research and Extension Project (NCRE), with technical assistance

1 Paper presented at the 11 thAnnual Symposium of the Association for Farming Systems Research-
Extension, Michigan State University, East Lansing, MI, Oct. 5-10, 1991.
Agricultural Economist and Agronomist, Testing and Liaison Unit, National Cereals Research
and Extension Project (NCRE/IITA/USAID), Institute of Agronomic Research, B.P. 33
Maroua, Cameroon. The authors are grateful to Doyle Baker and two anonymous reviewers for
comments received on an early version of this paper. Standard disclaimers apply.






KAMUANGA & FOBASSO


from the International Institute of Tropical Agriculture (IITA), took respon-
sibility for on-farm testing in northern Cameroon in an effort to promote
farming systems research and extension (FSRE) through its Testing and
Liaison Unit (TLU). Maroua TLU is now responsible for farming systems
diagnosis, on-farm research, and research-extension linkages.
The target zone of Maroua TLU is the region above the 10th parallel,
which includes much of the cotton-growing portion of Far North Province
and the Mayo Louti Division of North Province (Figure 1), an area frequently
referred to as the Center North Zone. This region has about 210,000 farm
families that grow cotton for cash and sorghum as the staple food.
Agronomic themes tested by Maroua TLU are jointly decided by IRA
researchers and SODECOTON, a government parastatal in charge of cotton
development and food crops extension in the North and Far North provinces.
Between 1984 and 1988, nearly 840 on-farm tests were conducted success-


CHAD


NIGERIA


*" 0
0 0


Maroua
o 0


0 0


CHAD


Garoua


Figure 1. Northern Cameroon. Location of TLU Test Sites.


Journal for Farming Systems Research-Extension






S35 SORGHUM IN CAMEROON


fully, 56 percent of which were concerned with sorghum. Emphasis was on
variety tests in response to constraints on increased sorghum production faced
by farmers in the region (e.g., poor and erratic rainfall, weed [Striga
hermonthica] infestation, low soil fertility, and labor scarcity), which could be
globally addressed by developing short-cycle (85-95 day), drought-tolerant
varieties.
The sorghum breeding program at IRA Maroua concentrated on the
selection and multilocational trials ofshort-cycle, open-pollinated varieties, all
medium in height (2.5m) with white grain, and resistant to diseases, insects,
and Striga (Dangi, et al., 1989). Collaboration between TLU and the IRA/
NCRE breeding program resulted in some of these varieties being tested in
farmers' fields over the 1984-1989 period. Ofparticular interest is S35, a new,
high-yielding, early-maturing sorghum variety released to extension in 1986.
Although S35 adoption has remained controversial, it has provided evidence
of successful research efforts in developing and extending a new cultivar that
is playing a particular role in a region characterized by a wide range of well-
adapted local materials. Measuring the impact of such efforts is an objective
well in line with TLU orientation.
The purpose of this paper is: (1) to discuss the role collaborating farmers
played in the evaluation and diffusion of S35, and (2) to examine the extent
and patterns of its adoption in the region. We then propose revisions in
sorghum breeding objectives at IRA in light of the feedback from farmers.
Suggestions for extension, to further adoption of improved sorghum varieties
in northern Cameroon, are also discussed.

BACKGROUND
The sorghum variety S35 originated in India. It was selected by the IRA/
NCRE sorghum breeding program from hundreds of lines sent to Cameroon
in 1982 by Dr. N.G.P. Rao, then working in Zaria, Nigeria. These lines had
general characteristics believed to be important for sorghum improvement in
semiarid zones of Africa-earliness, high potential yield, medium height (2-
2.5m), cream-colored grain, and resistance to diseases (Johnson, 1988;
Dangi, et al., 1989). The variety S35 is nonphotosensitive, with a cycle of 90
days in northern Cameroon.

On-Farm Tests
On-farm testing ofS35 began in 1983 with 11 researcher-managed trials
showing no significant difference in yield from the local varieties. From 1984-
1987, the new variety was tested in 240 farmers' fields, in addition to
multilocational trials in 17 locations. Results summarized in Table 1 indicate
that S35 outyielded local varieties by a surprising 85 percent in 1984, a year
of extremely low rainfall in northern Cameroon. During years of "normal


Vol. 4, No. 2, 1994






KAMUANGA & FOBASSO


Table 1. Grain yield (kg/ha) of introduced and local sorghum varieties in on-farm
tests across sites in Northern Cameroon (1984-1987).

YEAR 1984 1985 1986 1987 ALL TESTS 1988
Rainfall1 (mm) 528.7 729.0 772.6 614.0 830.5
Yields of S35 1,333.0 1,689.0 1,866.0 1,888.0 1,694.0 -
Yields of locals 719.0 1,539.0 1,721.0 1,825.0 1,451.0 1,4662
Yields of other
introduced
varieties 784.0 1,202.0 2,185.0 1,974.0 1,536.0 1,119
Number of sites 88 79 38 35 240 60
Yield difference:
S35 over locals
(%) 85.4 9.7 8.4 3.5 26.7 -
1 Mean total across selected sites.
2 Mean yield of the best local variety, Gueling. Mean yield of local Djigari varieties in 1988 was
1,336 kg/ha.
3 Mean yields of E35-1 and 38-3 in 1984; S34, S36, S20, and 82-S-50 in 1985; CS54 and CS61
in 1986 and 1987.
4 Mean yield of CS54.
Source: NCRE/TLU, IRA/SAFGRAD Annual Reports.

rainfall," as in 1986, S35 maintained its high yield but was not significantly
different in yield from the local varieties.
Surveys conducted at the time of harvest in 1986 and 1988 revealed that
53 percent of collaborating farmers preferred S35 over other introduced
cultivars. The majority (67 percent) cited the white-colored flour and taste of
S35 as reasons for their preference (TLU, 1986). Despite mixed results in on-
farm testing and farmers' assessment, by early 1987 researchers and extension-
ists alike believed that S35 was on its way to adoption. In 1986, for example,
the Government's Seed Multiplication Project at Garoua produced more than
20 tons of S35 seed in response to increased demand. The same year,
SODECOTON extended the variety on more than 600 hectares.

Farmer Feedback
Feedback from farmers who planted S35 over the 1983-1990 period as
TLU adopters or collaborators in on-farm tests has been collected in various
ways, including questionnaire surveys at time of harvest and informal or
prearranged interviews with farmers. The following summarizes the most
frequent observations (TLU, 1986; Johnson, 1988; Kamuanga et al., 1991).
Recommended planting dates (June 15-July 10) are considered late by
farmers because the average date of planting traditional varieties is May 25 and
much earlier in years when usable rains fall in May.
The short cycle of S35 is an obvious advantage, contributing to drought
avoidance; the same characteristic, however, leads to increased susceptibility
to grain mold when planting is too early in the season.


Journal for Farming Systems Research-Extension




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

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