• TABLE OF CONTENTS
HIDE
 Cover
 Title Page
 Abstract
 Acknowledgement
 Table of Contents
 List of Tables
 1. Orientation of the study
 2. The evolution of stratifica...
 3. Identification of important...
 4. Incorporating socioeconomic...
 5. Summary and conclusions
 Bibliography
 Appendix. Recommendation domains...














Title: Socioeconomic criteria for defining farmer recommendaton i.e. recommendation domains
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00081541/00001
 Material Information
Title: Socioeconomic criteria for defining farmer recommendaton i.e. recommendation domains implications for farming systems research
Physical Description: viii, 107 leaves : ill. ; 28 cm.
Language: English
Creator: Kelly, Terry C
Publication Date: 1984
 Subjects
Subject: Agricultural systems -- Research   ( lcsh )
Agriculture -- Research   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis (M.S.)--Colorado State University, 1984.
Bibliography: Bibliography: leaves 94-101.
Statement of Responsibility: submitted by Terry C. Kelly.
General Note: Typescript (photocopy)
 Record Information
Bibliographic ID: UF00081541
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 79696380

Table of Contents
    Cover
        Cover
    Title Page
        Page i
        Page ii
    Abstract
        Page iii
    Acknowledgement
        Page iv
    Table of Contents
        Page v
        Page vi
        Page vii
    List of Tables
        Page viii
    1. Orientation of the study
        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
    2. The evolution of stratification
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
    3. Identification of important socioeconomic factors
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
    4. Incorporating socioeconomic factors in farmer stratification
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
    5. Summary and conclusions
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
    Bibliography
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
    Appendix. Recommendation domains for Central Province, Zambia
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
Full Text









THESIS


SOCIOECONOMIC CRITERIA FOR DEFINING FARMER RECOMMENDATORY

DOMAINS: IMPLICATIONS.FOR FARMING SYSTEMS RESEARCH










Submitted by

Terry C. Kelly

Department of Agricultural and Natural Resource Economics











In partial fulfillment of the requirements

for the Degree of Master of Science

Colorado State University

Fort Collins, Colorado

Summer, 1984














THESIS


SOCIOECONOMIC CRITERIA FOR DEFINING FARMER RECOMMENDATION

DOMAINS: IMPLICATIONS FOR FARMING SYSTEMS RESEARCH










Submitted by

Terry C. Kelly
Department of Agricultural and Natural Resource Economics











In partial fulfillment of the requirements

for the Degree of Master of Science

Colorado State University

Fort Collins, Colorado

Summer, 1984











COLORADO STATE UNIVERSITY


May 9, 1984

WE HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER OUR SUPERVISION

BY Terry C. Kelly

ENTITLED Socioeconomic Criteria for Defining Farmer Recommendation

Domains: Implications for Farming Systems Research
BE ACCEPTED AS FULFILLING IN PART REQUIREMENTS FOR THE DEGREE OF

Master of Science


Committee on Graduate Work


Department Chairman













ABSTRACT
SOCIOECONOMIC CRITERIA FOR DEFINING FARMER RECOMMENDATION

DOMAINS: IMPLICATIONS FOR FARMING SYSTEMS RESEARCH

Successful farming systems research and development requires

that relatively homogeneous groups of farmers for whom more or less

the same recommendations can be made be identified. Such groups are

called recommendation domains. This thesis examines the recommendation

domain issue with the intent to suggest means of incorporating appro-

priate socioeconomic variables into the stratification procedure.

The hypothesis is that incorporating socioeconomic and cultural

factors into farmer stratification increases the likelihood that

research recommendations will be successfully adopted by client

farmers.

An assertion is made that those factors which most influence
farmers' decisions regarding adoption of agricultural technologies

are the factors which should be considered when stratifying farmers.

Twelve key socioeconomic factors are identified on this basis, with

others of lesser importance also included in the discussion. These

factors are then incorporated into a suggested procedure for farmer

stratification. A case using actual data from Lesotho illustrates

the viability of this stratification procedure.

Terry C. Kelly
Department of Agricultural and
Natural Resource Economics
Colorado State University
Fort Collins, Colorado 80523
Summer, 1984

iii













ACKNOWLEDGEMENTS
I especially want to thank my wife and colleague, Judith Kidd,

and our young son, Nathan, for being so supportive and understanding,

particularly during those times when I neglected my familial responsi-
bilities. They reminded me that there is life beyond graduate school.

I also with to thank my advisor, Jerry Eckert, and my committee

members, Willis Shaner and Ronald Tinnemneier, for their support, for

their ideas and criticisms, and above all, for the flexibility they

maintained through the ups and downs of this research. A special

thanks goes to Kathy Fiddler who worked beyond the call of duty to

process these manuscripts.

Finally, I'd like to express my appreciation to the United

States Agency for International Development (USAID) and the

International Scholarship Program at Colorado State Unviersity for
providing funding for this research through USAID's Memorandum of

Understanding with CSU.















TABLE OF CONTENTS


1. ORIENTATION OF THE STUDY . . . . . . .

1.1 Introduction . . . . . . . .

1.2 A Farming Systems Research Orientation ..

1.3 Why Define Recommendation Domains? . . .

1.4 Emphasis on the Economic, Social, and Cultural


Page

. . 1

. . 1

. . 2

. . 7


Aspects . . . . . . . . . .

1.5 Proposed Methodology . . . . . . . .

2. THE EVOLUTION OF STRATIFICATION . . . . . .

2.1 Classifying Farming Systems . . . . . . .

2.2 Stratification Emphasizing Technical Criteria . .

2.3 Stratification Emphasizing Human Criteria . . .

2.4 Summary of Stratification Procedures . . . .

2.4.1 Household Resorce Base . . . . . .

2.4.2 Cultural Practices . . . . . . .

2.4.3 Institutional Characteristics . . . .

2.4.4 Household Characteristics . . . .

3. IDENTIFICATION OF IMPORTANT SOCIOECONOMIC FACTORS . .

3.1 Methodological Considerations . . . . . .

3.2 Community Environment . . . . . . . .

3.2.1 Market access . . . . . . . .








TABLE OF CONTENTS

(Continued)

Page


3.2.2 Access to other institutions . . . . 41

3.2.3 Ethnic or class differences . . .. ... 44

3.2.4 Population densities and local employment

characteristics . . . . . . . 45

3.2.5 Group interactions . . . . . . . 46

3.3 Farm-Firm Characteristics . . . . . . . 47

3.3.1 Labor . . . . . . . . . 47

3.3.2 Land . . . . . . . . ... .. 50

3.3.3 Capital equipment . . . . . . . 51

3.3.4 Cash and income . . . . . .... .52

3.3.5 Risk factor . . . . . . . . 53

3.3.6 Interactions within the farming system . . 54

3.4 Household Characteristics . . .. . ... 55

3.4.1 Household composition . . . . . . 56

3.4.2 Household goals . . . . . . . 58

3.4.3 Household orientation . . . . ... 59

3.4.4 Other household characteristics . . . 61

3.5 Checklist of Socioeconomic Factors . . . . 62

4.0 INCORPORATING SOCIOECONOMIC FACTORS IN FARMER

STRATIFICATION . . . . . . . . .. . 67

4.1 A Procedure for Socioeconomic Stratification . .. 67

4.2 The Case of Lesotho's Lowlands . . . . .. 74


. vi








TABLE OF CONTENTS

(Continued)

Page


5. SUMMARY AND CONCLUSIONS . . . . . . . . 87

5.1 Summary .. .. .. .. . .. .. ... 87

5.2 Conclusions . . . . . . . . . 89

5.2.1 Policy Implications . . . . . . 91

5.2.2 Further Research Directions . . . . 92

BIBLIOGRAPHY ......... ....... . . ... 94

APPENDIX. RECOMMENDATION DOMAINS FOR CENTRAL PROVINCE, ZAMBIA 102














LIST OF TABLES


Table


1 Socioeconomic Checklist for Farmer Stratification . . .






LIST OF FIGURES


Figure


1 The hierarchical relationships among regions, farms

and agroecosystems ...................
2 Procedure for Socioeconomic Stratification . . . . .

3 Stratification of rural households in the central

lowlands, Lesotho . . . . . . . . . ..


viii


Page














1. ORIENTATION OF THE STUDY


1.1 Introduction

It is generally agreed that sustained improvement in the pros-

perity of a society requires the active participation of the impov-
erished majority. In Low Income Countries (LICs), at least, the

greatest number of poor are rural and depend primarily upon low

resource, but efficient and complex farming systems to meet their

everyday needs. A necessary condition for amelioration of life in a
society is improvement of these small farm systems. Farming Systems

Research has recently been advanced as an approach toward this end.

Since such research is fairly site-specific, it is necessary to

identify groups of farmers for whom research results generated by FSR

methodologies are applicable. Such groups, or recommendation domains,1

have traditionally been defined on agro-ecological bases with obvious

institutional differences such as farm size or market access only
occasionally considered. More recently, the development community

has recognized the influence of economic and social factors on farmers'

decisions. Consequently, socioeconomic factors have become increasingly

important in all phases of agricultural research, including delineation

of recommendation domains. However, no commonly accepted framework


1Byerlee, Collinson, et al. (1980) define a recommendation domain
as a "group of relatively homogeneous farmers with similar circumstances
for whom we can make more or less the same recommendations."








exists for this stratification process. This research examines the

small-farmer stratification issue with the intent to suggest means of

incorporating appropriate socioeconomic variables.

More specifically, this thesis proposes to 1) review and codify
current approaches to stratification, 2) identify those socioeconomic

factors which are critical to differentiating farming systems under

various circumstances, and 3) develop guidelines for incorporating

these important socioeconomic factors into the basis for stratifying

farmers into homogeneous subgroups.


1.2 A Farming Systems Research Orientation

Recent attempts to address the problems of small farmers in LICs

have resulted in an applied methodology called Farming Systems Research

(FSR); its various aliases include Farming System Approach to Research

(FSAR), Farming Systems Research and Development (FSR&D), and Farming

Systems Research and Extension (FSR/E). Farming Systems Research is
thought of in two modes, "upstream" FSR and "downstream" FSR. Shaner,

et al. (1982, p. 37) characterize "upstream" FSR as being "partly

basic, broadly general, and supportive;" whereas "downstream" FSR is
"site specific, primarily adaptive, and useful without long delay for

target groups of farmers." Target group stratification is employed

primarily in "downstream" Farming Systems Research. For the purposes
of this thesis, then, FSR is defined generally to refer to downstream,

adaptive on-farm research that views the farm household and the

activities managed by it as a whole. FSR begins with farmers' problems

as perceived by farmers. Research is then designed in part by farmers,








managed (again in part) by farmers, and adjusted as farmers' circum-

stances dictate. In other words, client-farmers are involved actively

throughout the entire research process. This study is conducted
within the context of this FSR methodology.

Typically, the clients of FSR are small subsistence farmers. A

small farmer is one whose farm has a very limited resource base,

rather than just being small in land area. "Subsistence" means that

a larger proportion of output from farming and household operations

is retained for the household rather than sold or exchanged in the

market place. Subsistence farms are generally characterized by

1) limited involvement in both the consumption and production sides

of the wider economy, 2) employment of mostly traditional technologies,

and 3) a close relationship between sociocultural considerations and
2
household decision-making (Wharton, 1969)2. These farms are first a
home and, second, a business. In fact, farming may not be the most

important or highest priority activity of the household. Thus,

farming decisions are made for other than just profit maximization
and agronomic reasons, an important point for researchers to remember.

When discussing a farming system, we are referring to a "unique

and reasonably stable arrangement of farming enterprises that the

household manages according to well defined practices in response to

the physical, biological, and socioeconomic environments and in
accordance with the household's goals, preferences, and resources"

(Shaner, et al. 1982, p. 16). It is important to add that "farming


2Wharton develops a more rigorous characterization of subsistence
farming based on various economic, sociocultural, and developmental
criteria.









enterprises" include all household enterprises, some of which are not

specifically farming. Herein lies the strength of FSR: it concen-
trates not only on specific enterprises and activities but also on
the interrelationships among all the farm/household activities.
Hence, FSR takes a holistic view of the farming household.

A farm system is really an arrangement of component subsystems
which function as a unit (Hart, 1980). These subsystems include a
socioeconomic system as well as any number of agroecosystems, each of

which is also an arrangement of subsystems such as soil, crops,

weeds, insects, animals, etc. The farm system is itself a subsystem
of a larger regional system, and so on (Hart, 1980). (see Figure 1)

FSR as a methodology is often characterized by four stages:

1) the diagnostic stage, 2) the design stage, 3) the testing stage,
and 4) the extension stage (Norman, 1983b). Shaner, et al., (1982)

divide the FSR process into five basic activities: 1) target and

research area selection, 2) problem identification and development of
a research base, 3) planning on-farm research, 4) on-farm research

and analysis, and 5) extension of results. Activities 1 and 2 are

roughly equivalent to Norman's diagnostic stage, and this is the

stage of primary concern for this study. A valuable attribute of FSR
is that it is adaptable as it proceeds. While preliminary recommen-
dation domains are delineated early in the diagnostic stage (or in
activity 1 of Shaner's division), as more knowledge of the area and

the farming systems is gathered, the recommendation domains are
further refined.
















































Source: Hart (1980)




Figure 1. The Hierarchical Relationships Among Regions, Farms,
and Agroecosystems.








In the initial phase of FSR, the first step is to gather and
3
collate data on the target area and farmers' circumstances. Byerlee,
et al. (1979) define farmer circumstances as all those factors which

bear on farmers' decisions with respect to technology. They include
natural, technical, economic, and sociocultural factors. An initial

delineation of recommendation domains is done at this time. Analysis

of secondary data is usually followed by an informal or exploratory

survey (sometimes called a sondeo) which may then be followed by a
verification survey. At each step the recommendation domain boundaries

are refined as necessary. The purpose of these steps is to gain as

complete a knowledge of the farmers' circumstances as possible in as

short a time as possible. This information is needed to identify the

appropriate target groups and to design their respective research
activities.

Some final notes on Farming Systems Research need to be made at

this point. Its "bottom up" nature requires that it be collaborative

and multidisciplinary. Economists and other social scientists are as

important to the development of the small farmer as are technical

scientists. Second, because survival is paramount in their decision
making, subsistence farmers have a very low risk threshold and,

hence, will change only in increments and then, only if their survival
is not further threatened. FSR seeks to identify such incremental
improvements to farming systems, instead of jumping to agronomically

and economically optimal solutions or packages. Third, because of


3Target area is a research area usually designated for policy
reasons. This is different than target group or recommendation domain.







this critical situation in which most subsistence farmers find them-

selves, FSR looks for immediate or short term applications to help

alleviate farmers' problems. However, the long term implications of
any recommendation cannot be ignored; the "upstream" mode of FSR is
oriented more to deal with long-term issues. Finally, since FSR, as

used here, concentrates on the short term, many institutional arrange-

ments, social relationships, etc. which are changeable in the long
run by policy makers and/or farmers are treated as fixed parameters.

Only those circumstances which can be changed immediately or in the

near future are considered variables in Farming Systems Research.


1.3 Why Define Recommendation Domains?
Research designed to benefit a particular client group would be

considered successful if results of the research, eg., recommended

technologies, were extensively adopted by the clients. To help

insure successful adoption of technology, recommendations must address

client problems. It is, therefore, necessary to identify client

groups that have the same or similar research needs. Collinson
(1982, p. 8) correctly stated that, "Adaptive on-farm research can
only be done effectively with a particular farm situation, and there-

fore an identified target group of farmers, in mind." Ideally, since
each farm system is unique, research should be performed on an indi-

vidual farm basis. But of course, resources available for research

are not unlimited, making this option unrealistic. On the other

hand, research directed at a large number of diverse farmers would








most likely result in low acceptance rates of the recommended tech-

nologies, again wasting precious research resources. A recommendation

domain approach to research is therefore a compromise. Farmers in a

recommendation domain should have the same researchable problems and

development alternatives, and should react in similar ways to policy

and technological changes (Government of Zambia, 1979; Gilbert, et

al., 1980).

In addition to being necessary for successful on-farm research,

recommendation domains are important for at least two other reasons.

First, correct identification of a domain's circumstances can allow

the research results to be extended to all clients within the domain

as well as extrapolated to farmers with similar circumstances outside

the research area. This could significantly improve the cost effect-

iveness of adaptive on-farm research by expanding the universe for

which the research results are applicable (Harrington, 1980). Benjamin

(1980) argues that it is more cost-effective to identify specific

recommendation domains and extrapolate than to aggregate. Second,

stratification will help to ensure that local circumstances and

interests are considered when planning research and formulating

policy (Government of Zambia, 1979). Policy makers can use this

information to help determine priority groups toward which to focus
their research efforts (Franzel, 1981).

Stratification of farmers in a traget area is essential to

ensure that all farmers in a target area are identified. Adaptive

on-farm research raises the opportunity cost of neglecting farmers

outside the specific target group (Gilbert, et al., 1980). Too often








farmers are omitted from the research because they are not "represen-

tative" of the predominant farming system in the area. Identifying

these farmers as a separate domain ensures that they will at least be

considered when research funds are distributed, not simply neglected

because they have not been identified.

Behnke and Kervin (1983) present a problem with stratification

which researchers must consider. Primarily from their experience in

Botswana, they argue that grouping farmers into homogeneous units may
hide or obscure the economic and social interdependencies of the

community under study. Within a community, target groups will, in

fact, represent economic classes based on differential access to the

means of production. Farmers with quite different but complementary

resource bases often work cooperatively together under a variety of

arrangements so that their resources are used more efficiently.

These multi-household production units cut across target groups.

Behnke and Kervin contend that typification of farmers directs re-

search to formulate recommendations which will inevitably bring about

a realignment of these cooperative arrangements. Benefits to indi-
vidual farmers might be simply a redistribution of productive capacity

within the community with no net increase in prosperity. While this

is a valid criticism, the need to stratify for research purposes is

still important and necessary. The challenge for researchers is to

incorporate these findings into the stratification process. An

attempt to do this is made in this study.

Some are skeptical of the recommendation domain approach from an

applied perspective. Zandstra (from Shaner, 1983, p. 165), for

instance, has difficulty with the concept because the delineation of








a domain depends upon the nature of a future recommendation and this

cannot be fully known until the research has been carried out.
CIMMYT and others, recognizing this point, identify preliminary

recommendation domains and then refine them if necessary as research

progresses (Byerlee, Collinson, et al., 1980). Whether or not re-

searchers actually stratify farmers into homogeneous units, they must
identify a target group for their research. Due to the critical

nature of this identification to the success of research, ability to
accurately identify homogeneous subgroups is crucial. Special method-

ologies, approaches, or guideline criteria need to be developed for

the selection of target groups (TAC Review, 1978). This study is a

first attempt toward this need.


1.4 Emphasis on the Economic, Social, and Cultural Aspects

In the conduct of the research for this paper, two assertions
are made. These assertions cannot conclusively be proved correct or

incorrect, but they make intuitive sense, and logical arguments are

developed and evidence is presented to support them.

One recent way to stratify farmers into target groups is on the
basis of similar farming systems. (More will be presented on this in

Chapter 2.) This would imply that factors which differentiate farming
systems are the appropriate factors to be considered when delineating

recommendation domains. The goal of successful research, though, is
adoption of the recommended technologies by the clients. Factors

which affect farmers' decisions regarding technology are the factors,

then, which differentiate farmers for the purposes of research (Hart,

1983). They may not be the same factors as those which differentiate








farming systems. Gladwin (1976) maintains that only a study of

farmers' decisions on adoption can pinpoint the critical factors or

barriers to adoption. Thus, first assertion presented here is: The

critical factors affecting farmers' decisions concerning adoption of

agricultural technology are the factors which should be considered

when stratifying farmers into recommendation domains.

While it is true that technical elements (environmental and

biological) are important and determine what is potentially feasible

for the farming system, how that system actually evolves is determined

by the human element (Gilbert, et al., 1980). This is the basis for

the second assertion presented here: Socioeconomic and cultural

factors are as important as technical factors, and in some situations,

more important, to farmers' adoption decisions and, therefore, delinea-

tion of recommendation domains.

The following hypothesis is offered, both as an end to which

this study is directed and as the rationale for undertaking this

study in the first place. This hypothesis is: that incorporating

socioeconomic and cultural factors into farmer stratification in
Fanning Systems Research increases the likelihood that research

recommendations will be successfully adopted by client farmers. No

empirical test of this hypothesis is possible here. However, presen-

tation of evidence from others' work and logical arguments should

establish its credibility.

The purpose here is not to diminish the importance of the tech-
nical element in delineating recommendation domains, but rather to

enhance the importance of the human element which has traditionally








been slighted. Little research has been conducted on how these

socioeconomic factors combine with what is ecologically feasible

(Hart, 1983). Only recently have economists and other social scien-

tists been included as full partners on research teams. Thus, the

socioeconomic aspects of identifying recommendation domains are

emphasized in this study because 1) they are extremely important to

the correct identification of target groups and 2) they have until

recently been neglected and research on their importance has generally
been lacking.


1.5 Proposed Methodology

As indicated earlier one of the purposes of this study is to

review and codify the various methods of stratifying farmers into

homogeneous subgroups. To this end Chapter 2 reviews the works of

numerous practitioners of FSR and the procedures employed by a number

of agricultural research centers practicing farming systems research.

Particular attention is given to the extent that socioeconomic criteria

are utilized and to which factors are most commonly considered.

Often an FSR project will not specifically stratify farmers, but

will, by necessity, identify representative farmers or target farmers
to participate in the research. This purposive identification of

farmers is treated in the same manner. Comments and questions are

raised concerning aspects of the different procedures reviewed.

Most of the literature reviewed in Chapter 2 has been gathered

by the various professionals at Colorado State University who are or
have been actively involved with the development of Farming Systems








Research methods. Other works were borrowed from the Farming Systems

Research Library at Kansas State University and from professors and

international students at Colorado State University who specialize in

small farm development.

In order to accomplish the second purpose of this study (to
identify those socioeconomic factors which should be considered when

delineating recommendation domains), and in accordance with my first
assertion, socioeconomic factors are identified which influence

farmers' decisions regarding agricultural technologies. Identifica-

tion of these factors is made in Chapter 3 through a review of selected
literature on economics and technological change and on technological

change in "traditional" societies. Included in this review are

several other reviews of constraints to technology adoption, and

reports on cases of successful and unsuccessful introduction of
agricultural technology. Particular attention is paid to those

factors which appear to explain differences in adoption patterns

among farmers.

A framework for farmer stratification, which is the ultimate
objective of this study, is developed in Chapter 4. The relative

importance accorded to various socioeconomic factors in Chapter 3,

related to methods and reasoning for stratification presented in

Chapter 2, form the basis for this porposed stratification procedure.

In development of this procedure, consideration is given not only to

influence on adoption decisions, but also to practicality and useful-

ness in the field. The viability of the proposed procedure is tested

with a case illustration using actual data from Lesotho.








The multitude of socioeconomic factors which may potentially

affect adoption decisions can result in the inclusion of too many

factors, rendering the resulting stratification framework virtually

useless in practical research situations. And interaction among

various factors can further complicate this framework. The definition

of a recommendation domain and the accompanying research could become

so site specific that it would be impossible to extrapolate the

results to other areas, thereby negating one of the justifications

for identifying recommendation domains in the first place. Navarro

(in Gilbert, et al., 1980) commented that farming systems may be

sufficiently diverse as to "defy meaningful or at least operational

generalization." On the other hand, to omit potentially critical
factors for the sake of simplicity could result in improper identifi-

cation of recommendation domains, something which has happened all
too frequently in the past.

Obviously it is difficult to develop a stratification framework

which is universally applicable given the diverse nature of farming

systems throughout the world. At this time, though, any clarification

of the process of stratifying farmers into homogeneous subgroups will

be an improvement over the general disarray that currently exists.

As Shaner stated in his paper presented at the Second Annual Farming

Systems Symposium at Kansas State University, "alternative approaches

to stratification need to be refined and taught to many so that this

task is not left to a few "renaissance men" who have a gift for

identifying the relevant conditions for effective research" (Shaner,

1983, p. 163).














2. THE EVOLUTION OF STRATIFICATION


2.1 Classifying Farming Systems

Agricultural researchers have realized that in order to initiate

change in the farming sector of LICs they must gain a thorough under-

standing of the complex farming systems which predominate there.

This realization has led to numerous attempts to describe and classify

farming systems in the tropics. As we learned, we began to understand

just how complex and diverse these systems are. We realized that
"peasant" farmers in the non-industrialized world act in an economically

rational manner after all.

T. W. Schultz, in his well-known work, Transforming Traditional
Agriculture (1964), was among the first to point out that small

traditional farmers were efficient in allocating their meager resources.

Development experts began to realize that small farm households make

complex decisions in allocating their various resources to satisfy a

variety of objectives. An understanding of these small farm "subsis-

tence" economies was sought and reported on by Wharton, Mosher,
Ruttan, and others.

One study which describes and classifies tropical farming systems
is Hans Ruthenberg's Farming Systems in the Tropics (1976). According

to natural, economic, and socio-institutional criteria, he defined

three broad types of systems: collecting systems, cultivation systems,








and grazing systems. Ruthenberg excluded collecting systems as being

economically insignificant. He then defined six cultivation systems:

a) shifting cultivation, b) fallow, c) systems with regulated ley4

farming, d) systems with permanent upland cultivation, e) systems

with arable irrigation farming, and f) systems with perennial crops;

and three grazing systems: a) total nomadism, b) semi-nomadism, and

c) ranching. Other types of grazing systems which combine with

cultivation were described as part of the relevant cultivation sys-
tems. Anong the criteria considered by Ruthenburg in this classifi-

cation scheme were type of rotation (the long-term alternation between

various types of land use), intensity of rotation, water supply,

cropping patterns and animal activities, implements used for cultiva-
tion, and degree of commercialization.

Harwood (1979) took a different approach to classifying farming

systems based on their stages of development. On a continuum from

least to most developed he identified hunting and gathering systems,

subsistence-level crop and animal husbandry systems where less than
ten percent of the output is marketed, early consumer systems that

market ten to thirty percent of their output, primary mechanization
systems where mechanical power is generally used only for tillage,

and secondary mechanization systems where mechanical power is used

for virtually all operations.


4
Ley is used wherever several years of arable cropping are fol-
lowed by several years of grass and legumes utilized for livestock
production. Regulated ley systems are often characterized by indivi-
dual grazing, fencing, pasture management, and rotational use of
grassland. Seeding, hay or silage production, and fertilizer applica-
tion indicate intensive types. Unregulated ley systems are more akin
to short-ternn fallow systems. (Rutenberg, 1976)








Neither of these classification schemes is precise enough for

most research purposes. Significant variability may be found within
these various systems resulting from differing technical and human

circumstances. Farming systems which appear to be the same may in

fact be quite different in very important ways. In an area of seem-

ingly similar farming systems, for example, Hildebrand (1979b) found

sufficient variation so that technology generated for one group could

not be transferred to other groups. For effective intervention,

farming systems have to be further defined on the basis of similar

circumstances.

Harrington (1980b) suggests at least two common alternatives for

grouping farming systems into relatively homogeneous units. The

first is on the basis of agro-climatic and socioeconomic factors that

explain important variations in farming systems. The second is on
the basis of the current farming system itself, which is a response

to determining conditions. In either case, the various criteria

employed in stratification are of concern in this chapter.

Farmers may vary geographically and along hierarchical lines.
Geographically, they vary in natural factors such as climate, soils,

topography, etc.; this is the technical element. Hierarchical varia-
tion is from differences in the human element, which can be either

exogeneous (out of the farmer's control) or endogenous (the farmer

has at least partial control) (Norman, 1983b). The human element

includes economic and institutional factors as well as factors which

are social or historical in nature, such as food preferences, customs,

current technology level, and tenure arrangements (Government of








Zambia, 1979). Of course, there is substantial interaction between

technical and human factors.

A wide variety of factors have been used as a basis for strati-

fication at various times. Initially, the technical element was
stressed as critical to grouping, often to the exclusion of the -human

element. More recently, the human element has gained in prominence,
and some suggest that is is more important than the technical element.

This review of stratification will focus on the factors variously

employed by different strategies. Those that emphasize the technical

element are considered first, followed by those focusing more on the

human element.


2.2 Stratification Emphasizing Technical Criteria

At the outset it is important to note that the stratification

strategy employed depends somewhat on the orientation of the project

or organization. While some of the programs reviewed here are not

truly FSR (as defined earlier), they do employ on-farm research

techniques and ultimately consider the impact of recommended techno-

logies on the whole farming system. For this reason they are often

regarded as farming systems research programs.

Some programs have a mandate that covers a geographically and

ecologically diverse area. A few of the International Agricultural
Research Centers (IARCs) have such mandates. The International

Institute for Tropical Agriculture (IITA) in Nigeria, for example, is

charged with developing technologies appropriate for a wide range of

tropical farming systems. IITA researchers develop preliminary

technologies for specific agro-climatic zones which can be adapted by








the fanners to their more specific farming systems. IITA's rationale

for this approach is, given that research resources are inadequate to

cover all systems, the more location specific the research is, the
higher are the opportunity costs of neglecting those systems not
being researched (Menz and Knipscheer, 1980).

Consequently, IITA identifies the main types of farming systems

in its coverage area primarily on the basis of environmental and

biological factors. General socioeconomic criteria of small farm
systems are considered when planning research, but are not used to
differentiate farming systems.

Similar strategies are employed at the International Crops
Research Institute for the Semi-Arid Tropics (ICRISAT) in India and

the International Center for Agricultural Research in the Dry Areas

(ICARDA) in Lebanon. Researchers at ICARDA determined that farming

systems changed across rainfall gradients and they used this approach

to target research on barley production in Syria (Nygaard, 1983).
At ICRISAT, researchers contend that in the semi-arid tropics

water is the most important constraining element. Hence, they use a
watershed-based farming systems approach (Krantz, 1981). Apart from

water, soil order is the major differentiating factor due to the

different hydrological properties of the two prominent soil orders in
the semi-arid tropics (SAT). Farm size and farm resources were also

considered when planning improved soil and water management systems
(Krantz, 1981).

Other FSR Programs have a mandate to research one or several

crops. As such, their stratification methods may reflect a bias

toward the determinants which affect particular cropping systems








rather than the whole farm system. Again, these programs normally

consider the effects of recommended changes in the cropping systems

upon the whole farm. In fact, some of the pioneering work on FSR
methodology has come out of such institutions. The International

Rice Research Institute in the Philippines (IRRI) and the International

Maize and Wheat Improvement Center in Mexico (CIMMYT) are two IARCs
which focus on one or two crops and have strong FSR programs.

Classification of rice systems at IRRI focuses primarily on
technical criteria. A region is divided into rainfall zones by the

number of consecutive months of at least 200 mm of rainfall and the

severity of the dry season (Shaner, et al., 1982). Topography, soil

texture, and seasonal temperature fluctuations are considered for

further delineation. Researchers identify an environmental complex,

or a set of sites with similar cropping pattern determinants. They

then define a production complex which is a set of sites with similar

cropping pattern performances. If the environmental and production

complexes differ, then one or more important determinants has been

omitted (Zandstra, Price, et al., 1981). The determinants considered

in addition to the above agro-climatic criteria are duration of

irrigation, remoteness, and length of time the land has been settled.

IRRI's experience has been that Asian rice farmers are generally

willing to participate in researchers' experiments, more so than

farmers in other areas. Consequently, less attention is directed to

socioeconomic factors (Shaner, 1983). While this rationale can be

questioned, IRRI has been successful in transferring rice growing

technologies.









Several national agriculture research programs also have farming
systems research components. At ISRA, Senegal's agricultural research

agency, researchers have divided the country into ecological regions

based on rainfall, irrigation, and soils. These ecological differences
lead to considerable diversity in the farming systems of the country

(ISRA, 1979).

At IDIAP, Panama's agricultural research institute, researchers

in the Caisan Program determined that farmers within the target area
faced similar agroclimatic characteristics, although one locale had

serious market access difficulties (Arauz and Martinez, 1983). This

led to a very different pattern of input use, so that two separate

recommendation domains were identified. Limited resources forced the

research to be concentrated on the more accessible domain. Often

researchers, faced with limited budgets, choose those target farmers

whose circumstances are amenable to directed intervention, resulting

in a more "successful" program. Unfortunately, those excluded from

research by this criterion are invariably the poorest, often female-
headed farm families who might be less likely to adopt the recommen-

dations. Yet, these are the families that desperately need agricultural

assistance.

In the Indonesian Cropping Systems Research Program (CRIA),
target areas are selected for policy reasons, for the potential for

successful research, and for similarity in soils and climates (Effendi

and McIntosh, 1983). Subdistricts are then selected within the

target area primarily according to land use types (rainfed, upland,

perennial crops, etc.). Market accessibility and infrastructure








development considered influential to the cropping pattern are also

taken into account (McIntosh, 1981). Villages identified for research

were selected as being most representative or typical of the subdis-
trict. This was accomplished using a "data matrix" where technical

and economic characteristics form the columns and villages form the

rows. The mean value of each characteristic was then subtracted from

the respective values associated with each village, giving the devia-

tion from the mean for each characteristic for each village. The

villages were ranked for each characteristic, the smallest deviation

assigned "1" and so on. These ranks were then summed and the village

with the smallest rank sum was considered most representative of the

subdistrict (McIntosh, 1981).

A similar method of identifying the most representative village

was used by ICRISAT in their village level studies (Jodha, et al.,

1977). Using the details of forty characteristics, villages were

selected to represent the typical characteristics of the Taluka

(sub-division). It should be noted here that at both ICRISAT and

CRIA, a number of factors, many of them socioeconomic, were used to

identify representative villages. However, stratification into units
of commonality took place when identifying subdistricts, and this was

done using primarily agro-climatic details. Identifying represen-

tative villages simply describes the "most typical" farming system in
the subdistrict.

In the Caqueza Project in Colombia, three zones were delineated

by altitude -- high, middle, and low -- with three distinct farming

systems. Larger farmers were excluded from this delineation since








they were not the focus of the project (Zandstra, et al., 1976). As

with most of the other programs reviewed thus far, a number of other

criteria, mostly socioeconomic, were considered when planning the

research and recommending technologies to be tested. This brings up

a crucial question: If critical socioeconomic factors are going to

be identified, and they must be for successful technology transfer,

why not include them in the delineation of recommendation domains so

that a more precise definition of the client target group can be

obtained? The next section reviews those stratification strategies

which incorporate important socioeconomic and cultural elements.


2.3 Stratification Emphasizing Human Criteria

As more socioeconomic variables are included in the stratifica-

tion equation, the more formalized the procedure becomes. Instead of

being divided into physical areas of commonality, farmers are divided

into groups or strata of commonality. These strata are not likely to

be contiguous geographically. In the Gambian Mixed Farming and

Resource Management Project, ten different strata were identified for

analysis on the basis of the pattern of livesock ownership. Not only

were type and size of herd considered, but how the cattle were herded

was also considered (the Gambia Project, 1983). Obviously, any

number of these strata could be represented in one area.

Many similarities exist among a number of the stratification

strategies reviewed in the next paragraphs. In most, background and

secondary data were reviewed and preliminary recommendation domains

delineated. This was usually followed by an informal or exploratory

survey and a refinement of domain boundaries. Sometimes a formal,








verification survey was administered and recommendation domains were

further refined if necessary. The following strategies considered a

wide variety of factors when grouping, many of which are socioeconomic

and cultural. And most of them incorporated some form of stratifi-

cation into their methodologies.

At the Agricultural Science and Technology Institute (ICTA) in

Guatemala, Peter Hildebrand and his colleagues have developed an
informal survey technique known as Sondeo (from the Spanish verb,

sondear, which means to sound out). The Sondeo, or rapid reconnais-

sance survey, is carried out over a target area by multidisciplinary

teams and is designed to gather as much information on farmers'

circumstances as possible in a very short time (usually one to two

weeks) (Hildebrand, 1979a). This is accomplished through informal

interviews with farmers coupled with regular team meetings to discuss

various findings.

From the information gathered in the Sondeo, the predominant

cropping systems used by potential target farmers were selected for

analysis and later the areas in which these systems are important

were determined. The focus is on the cropping system since it is the

cropping system that ICTA will be modifying with new/improved tech-

nologies. It is assumed that environment and circumstances are

similar if the same crops are grown in approximately the same manner.

The premise is that:

...all farmers who presently use it (a particular cropping
system) have made similar adjustments to a set of restric-
tions which they all face, and since they all made the same
adjustments, they must all be facing the same set of agro-
socioeconomic conditions (Hildebrand, 1979a, p. 3).








A rapid reconnaissance survey is also employed by researchers in
the Cropping Systems Research Project at CATIE (Tropical Agricultural

Research and Training Center) in Costa Rica. Relatively homogeneous

units are defined from information on climate, soil, farm resources,

etc. gathered in the Sondeo (Navarro, 1983).

Researchers at the International Potato Center (CIP) in Peru
first delimit major agro-climatic zones based on all the natural

factors which affect potato production. Different types of potato

farmers are then defined on the basis of the major economic charac-

teristics of their farms (Cortbaoui, 1981; CIP, 1978). When identi-
fying technological alternatives, "the socioeconomic knowledge of the

area will help discard unrealistic non-viable solutions" (Cortbaoui,

1981, p. 4).

As an aid to development planning, Eckert (1982) identified

target groups of farmers in Lesotho according to the resources under
their control. His stratification was based on productive assets,

which include per capital land area, cattle ownership, and capital

equipment, and off-farm incomes. His reasoning was that farmers with
different quantities and types of resources require different develop-

ment directions, and, hence, different research strategies.
The Economics Program at CIMMYT has made considerable progress

in formalizing the "recommendation domain approach" to farming systems

research. The delineation of recommendation domains is an integral

part of CIMMYT's approach. Initial definition of domains is done as

farmers' circumstances are analyzed using secondary data. As the

domains are refined after the exploratory survey, major variations in








farmers' practices are related to the circumstances which are hypoth-

esized to influence a particular practice. After the formal survey,
recommendation domains are further refined by testing to see if the

hypothesized critical circumstance actually does affect the farmer's

management practices. The boundary of a domain should be at the

point where a changing characteristic, (eg. farm size) results in
different management practices (Byerlee, Collinson, et al., 1980).

CIMMYT researchers caution that it is not necessary to delimit
domains too precisely. Recommendations should be general guidelines

to which the farmer can adjust according to his own circumstances.

They do note, however, that more domains will be needed when farmers'

circumstances exhibit greater variation (Byerlee, Collinson, et al.,

1980).

The CIMMYT approach has been used in several projects in Latin
America and Africa. In one study in Ecuador, the farming environment
was characterized in terms of information needed for agricultural

policy. In this case the area in maize, percent of total cropland in

maize, average farm size, and maize yields were considered important
(Winkleman and Moscardi, 1981).

In another Ecuadorean example three sets of farmers were identi-
fied based on natural circumstances: insect patterns and access to

irrigation (insect patterns were related to altitude). In this
study, it was determined that differences in economic circumstances

were slight for virtually all farmers. Hence, no recommendation

domains were defined on the basis of socioeconomic criteria. The

excluded farmers in each set were few in number and small in the








percent of total area given over to maize (Winkleman and Moscardi,

1981).

In a study on maize production in the Peruvian Andes, thirteen

recommendation domains were selected on the basis of altitude, whether

or not irrigation was used, farm size, whether there was significant

marketing of output or not, and in which of three geographic regions
the farmer resided (Benjamin, 1980). Certain farming systems in a

region were omitted from further study because they were insignificant.

In one barley producing area, four recommendation domains were

defined according to three factors: rainfall, farm size, and inter-

cropping with another crop (restricting machinery use). These factors

were determined to cause the major differences in land preparation,

seeding, varieties, disposal, and input use that were found across

the four domains (Byerlee, Collinson, et al., 1980).

In CIMMYT's East Africa Bureau, Michael Collinson and others

have done considerable work with recommendation domains and have made

real efforts to incorporate socioeconomic circumstances into the

stratification process. Included in the exploratory survey were

questions concerning a variety of factors relating to the farming

enterprises, food preferences, output variability and range, main

production methods for each activity, and resource variability
(Collinson, 1982). From this information the following were derived

or identified: 1) labor allocation constraints, 2) resource constraints,

3) farmers' priorities and decision criteria, 4) farmers' management

strategies, 5) potential points of leverage in the farming system,

and 6) approximate levels of return to present cash outlays. Among









other things, this information was used when delimiting recommenda-

tion domains.

As a basis for initial grouping of farmers, Collinson uses

variations in the current farming system. His rationale for this is

twofold: (Government of Zambia, 1979, p. 2)

1. The fanning system is a manifestation of a weighted inter-
action of natural, economic, historical and institutional
factors influencing farmers' decisions. It thus reflects
the balance of those factors important in identifying
homogeneous farmer populations.

2. The existing farming system is the starting point for
development, the base on which productivity improvements
have to grafted.

Using the current farming system also provides a low cost method for

identifying recommendation domains initially; the domains will be

revised as the diagnostic sequence is implemented (Collinson, 1982).

In deriving recommendation domains for Central Province, Zambia,

Collinson made some important observations (Government of Zambia,

1979). The key step is identifying the sources of variation which

are critical in dictating resource allocations in the farming systems.

These key variables vary from area to area. Secondly, it is important

to identify the evolutionary sequence in the development of farming
systems. The three keys to system development in the climatically

homogeneous Central Province have been infrastructure, demand for

maize, and power source. Lastly, Collinson suggests that recommenda-

tion domains be compatible with existing administrative divisions in

order to fit within existing service channels (Government of Zambia,

1979).








Throughout Central Province, Zambia, it was found that farmers

faced relatively homogeneous agro-climatic circumstances. Five
socioeconomic parameters were found to be particularly important in

accounting for variations in the types of farming: 1) power source

(closely correlated to cattle ownership and cultivated area), 2) major
starch staple, 3) main cash source, 4) degree of adoption of two new

cash crops -- cotton and sunflower, and 5) degree of use of purchased
inputs (Government of Zambia, 1979). In the final delineation of

recommendation domains, large-scale commercial farmers (greater than

40 ha) were excluded, emergent farmers (10 to 40 ha) throughout the
province comprised one domain, and traditional farmers (rarely more

than 5 ha) were divided into six domains based on location and differ-

ences in the above five parameters (Government of Zambia, 1979).
(See Appendix.)

While working with CIMMYT East Africa, Franzel (1981) writes
that it is often difficult to decide which differences between farmer

groups are important enough to warrant labelling them as separate

recommendation domains. Certainly, the researchers own judgment
comes into play. He contends, though, that in most circumstances,

"the differences between target groups in an area are readily recog-

nizable" (p. 25).

David Norman and the Kansas State University team in Botswana

essentially followed the CIMMYT strategy for stratification. Initial
typification was based on the existing farming systems (KSU, 1981).

It was determined that all small farmers have one overriding problem,

timeliness of operations, but the solutions will vary according to

the resource base of each farmer (Norman, 1983a). Five particularly









important sources of variation were identified. They were 1) access

to draft power, including ownership and number of cattle, donkeys,

and tractors, and the need to hire and exchange labor, 2) major food

staples, 3) degree of commercialization, 4) size of crop area, and

5) household characteristics, including male- or female-headed, and

condition of equipment and tools (KSU, 1981).

The Kansas State team eventually identified three separate
recommendation domains: 1) wealthy farms with a livestock emphasis,

2) marginal mixed crop and livestock farms, and 3) submarginal mixed

farms with a cropping emphasis (KSU, 1981). Power source and size of
holding were regarded as the most important sources of variation

while the major food staple was found to vary regionally, not across

domains.

Another stratification of rural households in Botswana was done

by J.B. Opschoor (cited in KSU, 1981). He defined three groups
according to the resource base of the farmer and the fanner's ability

to survive in good, normal, and bad years. His purpose was to show

that access to resources and the impact of policy-making affect
farmers' long-term prospects differently.

In a more pedagogical approach, Hart (1980) proposed a method to

initial farming systems characterization based on hierarchical agri-
cultural systems. Like in the Sondeo, multidisciplinary teams were

sent out to first describe and diagram the three sectors of the

regional system -- the primary sector which extracts or produces raw

materials, the secondary sector which processes the output from the

primary sector, and the tertiary sector which involves services

rather than transformation of goods -- and the flows of money,








materials, energy, and information among them. Inconsistencies among

sector diagrams were resolved and a regional diagram was drawn.

Focus of the team then shifted to the primary sector and the

characterization of the farming systems. These were diagrammed, with

special attention given to the agroecosystems found on the farm.

These farm systems studies are essentially descriptions of the differ-

ent combinations of agricultural components listed in the primary

sector study. Emphasis was placed on qualitative description, which

then can be used to identify the phenomena that merit quantitative

studies. As with the Sondeo, this systems approach to characteriza-

tion is designed to be done in the shortest possible time (Hart,

1980).

Employing Hart's approach in La Esperanza, Honduras, a study

group of graduate students identified five farm system types:

1) subsistence farms with less than 0.5 ha of potatoes, 2) small

commercial farms with 0.5 to 4.0 ha of potatoes, 3) large commercial

farms with more than 4.0 ha of potatoes, 4) cattle farms, and 5) fruit

farms (Hart, 1980).
Hart contends that classification of farm systems should be

according to those factors which affect farmers' decisions with

respect to technology (Hart, 1983). He identified six determinants

of farmers' design and control decisions. They were 1) the ecological
environment, 2) the socioeconomic environment, 3) agricultural re-

sources, 4) household-socioeconomic objectives, 5) flows of material

and energy among agroecosystems, and 6) the particular agroecosystems

of concern. Using these determinants as criteria, the limits or
boundaries of each type of farm system were defined based on the








points where the level of a determinant triggers a decision (Fart,

1983).

A final approach to stratification is presented, one with
definite emphasis on socioeconomic criteria. Carol Kervin in 1979

typified female-headed rural households in Botswana based on five

variables (cited in KSU, 1981). They were 1) cattle ownership,

2) employment status, 3) marital status, 4) availability of household
or extended family labor, and 5) remittances sent into the household
from outside the community. Using these criteria, Kervin identified

six different strata according to the household's dependence on

agriculture, ranging from the high dependency of owners of large

cattle herds and arable land to the independency of the wage-employed
and households with high outside remittances.


2.4 Summary of Stratification Procedures

It is tempting to evaluate various projects and programs on
their stratification methods. Far too little information has been

collected, however, to make a fair evaluation. And evaluation is not

the intent of this study; rather the purpose is to identify those
factors crucial to farmer stratification and clarify procedures for

inclusion of these factors in the stratification process. Hence, the

various procedures reviewed above are not compared.

This summary, then, is of the factors, and methods, identified
and the frequency they have been employed. Physical and climatic

factors are not included. Socioeconomic factors thus far identified

are divided into four groups: 1) household resource base, 2) cultural









practices, 3) institutional characteristics, and 4) household charac-

teristics. Each is summarized in turn.

2.4.1 Household Resource Base. Socioeconomic factors most

frequently used as criteria in farmer stratification are the re-

sources of the household. These include physical farm resources,

labor, cash sources, and resource flows. Of the physical resources,

farm size was most frequently considered, though related variables

such as cultivated areas of particular crops and percent of the total

area cultivated in a certain crop were considered explicitly only by

CIMMYT. This no doubt reflects CIMMYT's two crop emphasis. Another

physical resource commonly considered is power source and the closely

related variable of livestock ownership. This is testimony to the

importance of timeliness in farming operations. Several methods

considered the availability of farm implements and their condition.

Surprisingly, only two procedures reviewed considered labor

availability, though I would suspect from its importance that others

consider it also -- it just didn't show up in the literature reviewed.
Source of cash and off-farm income were considered important to

stratification by three of the procedures. Only Hart's systems

analysis approach explicitly considered resource flows within the

farming system.

2.4.2 Cultural Practices. The most frequent criterion used to

stratify farmers is the particular cropping system employed by the

farmer. This is said to be a reflection of the combined agro-

socioeconomic factors. While not explicitly detailed, one must

assume that cropping system includes such practices as intercropping,

relay cropping, rotation and fallow systems, major crops cultivated,








and irrigation practices. In Africa, the major food staple produced

is used as a differentiating factor. Emphasis on whether the farming
system is primarily crop- or livestock-based is used in several of

the procedures, and the cropping emphasis of many of the projects

implies consideration of this criterion. Method of cattle herding

was considered significant in the Gambia.

2.4.3 Institutional Characteristics. These include access to

factor and produce markets, credit, transportation and delivery

systems, communication channels, extension services, and farmer

organizations. While most strategies consider the institutional

climate generally, only market access was explicitly used as a

stratification criterion. Market orientation as a criterion employed

by CIMMYT and Kansas State University in Africa, however, implies

that some of the other factors affecting market access (transportation,

extension, etc.) were considered as well. Market orientation includes

marketed output, use of purchased inputs, and adoption of cash crops.

2.4.4 Household Characteristics. Household characteristics, as

a group, were the most seldom considered of all the factors. Objec-

tives and priorities of the household are conspicuously absent in

stratification procedures -- only Hart considered them. Gender,

employment status, and presence or absence of the household head (or
the chief decision maker regarding the farming system) were con-

sidered by the Kansas State team and Kervin in Botswana. Only CIMMYT

considered the household's food preferences and only Kervin looked

specifically at the degree of household's dependency on agriculture.

Two further comments are necessary here. Although evidence on
the use of socioeconomic criteria for farmer stratification in some









programs was not found, nowhere was it stated that they did not use
such criteria, and, if questioned they would probably indicate that

they do. However, such stratification was not formalized in their

research procedures. Second, those agencies which stratify according

to current farming or cropping systems give implicit acknowledgement

to the multitude of determinants which affect these systems. But

implicit consideration of factors still leaves the stratification

process very subjective and open to the intuition of the researcher.
It is argued here that in order to make this process more objective,

certain critical factors must be explicitly considered in the strati-

fication procedure.

One can conclude from this review that the state of the art of

stratification is in a bit of disarray. General agreement does not
appear, although the approach employed by CIMMYT seems to be the one

with the largest following. This stands to reason since CIMMYT has

developed the most definitive approach to famner stratification for

research purposes. Lack of agreement may also indicate that, like

farming systems research, stratification is too site specific to be

conceptualized generally. The following sections of this thesis
attempt to dispel this notion, at least partially, by identifying

important factors and developing guidelines for stratification.













3. IDENTIFICATION OF IMPORTANT SOCIOECONOMIC FACTORS


3.1 Methodological Considerations

In keeping with the assertions presented earlier, this chapter
identifies those socioeconomic factors which are believed to be most

influential to farmers' decisions regarding adoption of agricultural
technologies and, thus, which should be considered for farmer strati-

fication. A suggested procedure for incorporating these factors into

the stratification process is presented in Chapter 4.

Numerous factors have been hypothesized as influential to farmers'

adoption decisions. For purposes of stratification, only those

factors which may potentially result in differential adoption among
farmers need to be considered. Decisions of the field team as to

what constitutes a difference or how much variation is significant

are necessarily subjective and depend on the particular situation.
Therefore, the factors included here are those which should be con-

sidered when delineating recommendation domains; however, they will

not all be important in every situation. The discussion also includes

reasons why certain other factors are not as important for stratifi-
cation. The included factors are summarized in a checklist at the

end of this chapter.

I have used influence on adoption decisions as the basis for

inclusion of a factor here. However, assuming that the ultimate goal

of small farmer development is improved welfare of the farm family








and rural society, technological solutions are not necessarily the

answer. Inequities among different rural classes may actually be

exacerbated by certain technologies (Behnke and Kervin, 1983; Ryan

and Binswanger, 1979). The plight of certain household members

(i.e., women and children) may also be worsened with the introduction

of certain types of technology. These issues must be kept in mind

when stratifying farm families. Their importance is reflected, I

think, in the emphasis on socio-cultural issues in this discussion.

Adoption of a recommended technology requires that the recommen-

dation be technically correct, economically justified, and socially

and culturally acceptable. The technical requirement, though not

included in this study must not be construed as unimportant. Perrin

and Winklemann (1976) found agroclimatic and topographic factors to

be most important in explaining differential adoption rates among

farmers. Low (1982) in Swaziland found that adoption of hydrid maize

was positively related to suitability of an area for maize production.

Others similarly agree on the importance of the technical element.
I
Widespread agreement also exists on the importance of economics

to adoption; indeed, many consider it to be of primary importance

(Foster, 1973; Wilson, 1977; Gladwin, 1976; Beal and Sibley, 1967;

and others). Gladwin found non-profitability to be the major reason

why Mexican farmers decided not to fertilize twice as was recommended.

Beal and Sibley cited research supporting the importance of economic

motivation to adoption of agricultural technology. Interestingly,

though, in their empirical study of adoption by Guatemalan Indians,

the data did not support a positive relationship between adoption and

economic motivation. They attribute these findings in part to the









alien nature of economic gain and accumulation to the Indian culture,

supporting the importance of the socio-cultural requirement. They

also suggest that perhaps the questions were not properly presented,

and they cite the need for further study before the theory supporting
the importance of economic motivation is rejected.

One possible conclusion from this is that economic motivation is

not as important in very traditional societies. However, the concept
of economic gain is widespread enough that its importance cannot be

disregarded. Thus, the general conclusion here is that economics is

significantly influential to farmers' adoption decisions; this is

supported overwhelmingly in the literature.

A certain amount of caution should be exercised when interpreting

evidence on the relationship between various factors and technology

adoption. Few studies provide information on the effectiveness of
on-farm use of improved technology (Schutjer and Van Der Veen, 1976).

When a technology is reported as being adopted, the degree of adoption

and the extent to which farmers apply the new practice is usually
unknown. Yet, such information would seem important in determining

"successful" adoption and adoption differences.

Remember that the rationale for farmer stratification is to
identify groups of farm families for whom we can make more or less

the same recommendations. Although each farming system is unique,

research cannot be farm specific. Thus, research recommendations

need to be sufficiently general so as to be applicable to a group of

similar famners. Farmers can then adapt the recommendation to their

particular circumstances. It would not be cost effective to delineate
recommendation domains too precisely. Decisions must be made on how








much variation warrants a separate domain. But, this determination

cannot be completely scientific: certain factors may be more important
than others in a particular instance; some systems are more flexible

than others; or a certain parameter may be very rigid for one family,

yet easily changed for another. Thus, much of stratification remains

an art which a set of criteria cannot replace.

Finally, to add some order to the discussion, I have chosen to
categorize the various socioeconomic factors according to the way

they might be identified by field personnel. First I consider the

characteristics of the community in which the farmer operates. Next,
I look at the factors of the farm as a firm, and finally, I consider

the characteristics of the household not specific to the farming
operation, although the close relationship between these last two

make this distinction somewhat arbitrary. In fact, the interrelated-

ness of the farm to the household and both these systems to the

larger community system make any categorization scheme difficult.


3.2 Community Environment

The fanner must operate within a larger community environment.
This includes not only the agro-ecological environment, but institu-

tional, demographic, and sociocultural environments as well.

It is incorrect to assume that the institutional environment
poses an equal constraint to all farmers in an area and therefore is

unimportant to stratification. Access to institutions may vary

within an area for physical reasons (e.g., distance from road),

because of resources differences (e.g., ownership of an ox-cart or

vehicle), and due to farmers' perceptions.








Institutions are significant, man-made elements (such as practices,

regulations, and organizations) in a culture which guide behavior and
which center on fundamental needs, activities, or values. As used

here, institution refers primarily to those elements bearing directly

on agricultural activity, such as markets, transportation, credit,

and communication systems. Land tenure is considered with land
resources in section 3.3.

3.2.1 Market access. Access to product, input, labor, and food
markets depends on physical factors and psychological factors. Of

concern to stratification is whether different farmers have differential

access to these markets. Although it is not possible from the literature

to document the extent or impact of differential market access on

rates of adoption, general agreement exists that such differential

access is a major barrier to adoption by many farmers (Schutjer and

Van Der Veen, 1976). Therefore, significant differences in market

access may require different recommendation domains.

Work patterns, landholding and cropping patterns, choice of

crop, use of purchased inputs, and cultivation methods are all in-

fluenced by market access (Norman, et al., 1982). Prices paid and

received by fanners may vary due to differences in market access.

Local factor markets make it possible for farmers to better allocate

their resources, particularly with regard to labor and power source,

and they serve to mitigate unequal factor endowments (Ryan and Binswanger,

1979). Markets are needed to supply food and consumer goods. And,

of course, adequate market access is required before farmers will

produce beyond their own needs.








Physical access to markets depends on existence of markets,
infrastructure development in the area, and farmers' own resources.

Assuming markets do exist, an obvious way to ascertain how physically

accessible they are to the farmer is to look at local roads and
transportation, their condition particularly during the wet months,

and the distance from the farm to the road and to market centers.

Also important is whether the farmer owns a vehicle, ox-cart, or

other means of transporting supplies and produce, or whether such
transport is locally available.

If the farmer has seemingly good access to markets yet has low
market activity, then maybe his "psychological" access is bad. He

may perceive that markets are inaccessible to him, or he may believe

that he is treated unfairly (Beal and Sibley, 1967). This could be

due to a number of factors. Different farmers may face different

costs and remunerations depending on who they deal with, whether they

belong to a cooperative or other farmer organization, existence of

quantity discounts and bonuses, and their ethnic or class origin.
Different farmers may face different market transactions costs, or

they may have different perceptions of their market accessibility.
Finally, farmers' differing goals and motiviations may affect whether

or not they wish to produce for market sale.

3.2.2 Access to other institutions. In addition to market
access, access to extension and credit are frequently considered

essential to adoption of recommended technologies. With both, however,

evidence supporting this relationship is at best mixed.








Access to market information is essential to farmer participa-
tion in the market. Access to other information is necessary as
well, particularly information on new technologies and methods. For

instance, Gladwin (1976) found that lack of knowledge on the recom-

mendation was most important in explaining why Mexican farmers decided

not to increase plant populations as was recommended. Ability to

adopt new practices and participate effectively in the market place

depends on timely information and knowledge. Differences in access

to such information can result in differential adoption of technology.

If these differences are not easily corrected, then separate recommen-
dation domains may be required.

Information to farmers can come via different sources. As a

source of information on new technologies, extension is believed to

be important. No evidence was found, however, that indicates that
extension is positively related to adoption rates. In an AID study

of 51 extension agencies in 1,560 farming communities in Latin America,

it was concluded that "the assistance of the extension services was
neither sufficient or necessary for modernization," (cited in Hatch,

1976, p. 2). In the Philippines, Barlow, et al. (1983) found that

once information on new varieties was known, the presence of extension
or research personnel was not required for further adoption. Their

presence may, however, accelerate adoption.

Past use of recommended practices is one key to the adequacy of

information sources. Depending on the channels of communication in

the area, certain factors, such as ownership of a radio and membership
in effective farmer organizations, may clue researchers to the degree

of access to information. Also, a farmer's negative attitude toward








government or unscientific behavior may indicate inadequate access to

information.

Evidence on the importance of credit to technology adoption is

also inconclusive. In Plan Puebla in Mexico, Gladwin (1976) found

lack of credit to be an important factor in farmers' decisions on

fertilizer use. Valdez and Franklin (1979) noted the importance of a

market for renewal of long-term credit, and Norman, et al. (1982)

found that savings and credit can help to overcome cash flow problems

created by the introduction of new technology. On the other hand,

Scobie and Franklin (1977) found that membership in a supervised
credit program, which entailed restrictions on input use, made no

significant difference to adoption of technology in Guatemala.

Perrin and Winklemann (1976) concluded that, while use of credit was

found to be significantly related to adoption of new varieties,

credit programs are not necessarily critical to farmer adoption

except in the case where new technology is marginally profitable.

Tinnermeier (1983) concludes that while some form of access to

capital is important, reliability of credit is more important than

cost (interest rates). Informal credit markets are often more reliable

than are formal markets. Even access to informal credit may not be a

requirement for adoption. Farmers will often find a way to adopt a

technology which is profitable. In the Philippines, introduction of

new profitable varieties resulted in the creation of an informal

credit market (Barlow, et al., 1983). Schutjer and Van Der Veen

(1976) concluded that access to institutional credit does not seem to

be a prerequisite for technology adoption except in the case of lumpy








investments such as tractors and tubewells. They also suggested,

however, that more evidence on this relationship is needed to make a

definite conclusion.

3.2.3 Ethnic or class differences. Certain social characteris-
tics in an area bear upon farmers' decisions on technology. Foster

(1973) identifies several important social barriers to change. These

include locus of authority -- roles of village and religious leaders

and public acceptance of each leader's authority; group organization --

formal and informal farmer groups, group rivalries, acceptance of
innovators, and peer pressure; and social structure -- caste and

class barriers and societal configuration (automomy, hierarchical,

central control, etc.).

Different ethnic groups generally have significantly different

cultural and social structures resulting in differing attitudes

toward change. Their goals and obligations may differ and they may

have different perceptions of the institutions and resources available

to them. Different classes or castes often have significantly different
resource endowments as well as different perceptions. All lead to

potentially different adoption behavior. For stratification purposes

the criteria are straightforward: different ethnic or class groups
should comprise different recommendations; that is unless the field
team determines rather conclusively that groups of differing ethnic

background are still very similar in their adoption behavior. Strati-

fying by ethnic group will account for many of the sociocultural

differences which are potentially significant.








One must also consider the historical development of a society,
particularly in the past several decades, and the resultant breakdown

of social institutions due to technological change. Expansion,

splitting, or combining of ethnic groups influence technological

development and resource allocation (Collinson, 1972). Particularly

for marginal producers, it is important to know how agricultural

production has changed over time and how intense or rapid has been

environmental degradation (Garrett, 1983).

Census and population data are the easiest sources from which to

determine if there are different ethnic groups in the area. Local

leaders, officials, and merchants can also identify different ethnic

and class groups. Obvious variations in housing and compound structure

and differences in local lines of authority may also indicate ethnic

or class differences. Usually, this information is not difficult to

obtain.

3.2.4 Population densities and local employment characteristics.

Increased population densities result in land becoming constraining

relative to labor, which necessitates different technological recommen-

dations (Norman, et al., 1982; Ryan and Binswanger, 1979; Spencer and

Byerlee, 1976; and Cleave, 1976). It also results in less fallow

land, decreasing soil fertility, and more fragmented farms (Norman,

et al., 1982; Collinson, 1972).

Population densities also affect off-farm employment opportuni-

ties and supply of local labor, both of which affect local wage

levels. All three combine to influence farmers' management decisions

(Norman, et al., 1982). Farmers' decisions to produce a surplus are








affected by the wage they could earn off-farm relative to returns on

marketable produce (Low, 1982). Local wage levels also affect methods

of planting, cultivating, harvesting, etc. (Barlow, et al., 1983).

While population densities and local employment characteristics
are the same for a group of farmers in a given locale, they critically

affect farmers' adoption. They should be incorporated in the defini-

tion of recommendation domains since some domains may include farmers

from different areas, and since research results are likely to be
extrapolated to other areas. This information is normally derived

from analysis of secondary data and included in the preliminary

definition of recommendation domains.

3.2.5 Group interactions. As Behnke and Kervin (1983) pointed
out, analyzing homogeneous groups in isolation often masks important

interactions among different economic classes. Reciprocal arrange-

ments for labor, tractors, and draft animals are commonplace among

farmers in low income countries. Many farmers depend on these

arrangements for timely completion of agricultural and non-agricultural

activities. Not only do these arrangements affect their participants'
decisions, technological solutions may significantly alter such

arrangements to the detriment of one or more parties. Different

farmer groups may also interact in the control and use of communal

resources. This is particularly common in much of Africa where

traditional tenure arrangements still prevail.

While the benefits of inter-group arrangements probably appear

in other factors used in stratification, the ensuing obligations may

not. There, the existence and nature of cooperative arrangements

should be considered when stratifying. Behnke and Kervin suggest









using supra-households as the basis for research when such arrange-

ments exist. Unfortunately, this would probably complicate an already

difficult research task. The researcher will have to use her5 training

and knowledge of the situation to decide how to stratify farmers who

are partners in inter-group arrangements.


3.3 Farm-Firm Characteristics

There is wide agreement that farm resources are very important in
explaining differential adoption rates among farmers. Beal and

Sibley (1967) found that farm firm variables explained 25% of the

variance in adoption among the group of Guatemalan Indians they

studied. Garrett (1983), Eckert (1982), and others have concluded

that different research strategies are required for farmers with

different resource levels. The farm resources discussed here are

labor, land, capital equipment, and cash. The risk factor associated

with a fanner's resources interacting with the institutional and

ecological environments and interactions within the farming system
are also discussed in this section.

3.3.1 Labor. The supply of labor, both from within the house-

hold and from outside, relative to the scale of operation is a most
crucial determinant to farmers' decisions (Norman, et al., 1982).

The farmer's own labor to land ratio determines the need for outside

labor and these combine to constrain the level of year-round agricul-

tural activity to the amount of land which can be worked during


5The feminine program is used occasionally to reflect and
emphasize the fact that women play a significant role in agriculture
in most low income countries.









bottleneck periods (Norman, et al., 1982; Zulberti, et al., 1979).

Availability of labor determines whether land-augmenting or labor-

augmenting technologies are acceptable, and appropriate solutions

depend on correct identification of labor bottlenecks (Barlow, et

al., 1983; Spencer and Byerlee, 1976; Morss, et al., 1976). Labor

availability is clearly a function of household size, which also

determines household food requirements. This influences the agricul-

tural activity of the household. Norman, et al. (1982) found a

positive relation between household size and cultivated area and

yields.

Type of labor available relative to off-farm employment oppor-

tunities is also important in labor-use decisions (Low, 1982; Zulberti,

et al., 1979). In Swaziland, where off-farm employment opportunities

do exist, Low found that production of subsistence goods was done

first by those members with the lowest opportunity costs. In fact,

he found that households adopted high yielding varieties of maize in

amounts just sufficient for household needs but did not produce for
market because labor was better used in wage employment. The returns

to wage labor were higher than returns to labor used in commercial
production of these new varieties, though the cost of buying maize

for own consumption was high enough that labor was used in subsistence
production. Consequently, introduction of high yielding varieties of

maize did not result in an increase in surplus production as was

hoped.

Cleave (1976) calls labor the most complex of all factors be-

cause it depends on biological factors (age, sex), technological

factors (training) and economic factors (opportunity costs). He also








considers it to be of overwhelming importance to farmers' decisions.

The literature supports the conclusion that quantity and type of

labor available and its relationship to the labor market are important

in explaining differential adoption, though empirical data confirming

this relationship are limited.

As a stratification criteria, labor is rather complex. Labor-

surplus households and labor-deficit households might warrant separate
recommendation domains. Farmers with the same labor bottlenecks can
be included in one domain, barring other significant differences.

And areas with different labor market features (supply and demand)

and wage rates may require separate domains. When considering the

type of labor available to the farmer, stratification becomes more

difficult; specific circumstances will dictate domain boundaries.

Quite possibly, farmers with a supply of skilled labor will react

differently to a set of recommendations than will a farmer having

primarily unskilled labor. And a household with a large number of

children would be different from one composed largely of adults. The
number of female workers relative to male workers might be a differen-

tiating factor as well.

A number of factors influence labor availability. In the community,
population densities, seasonal migration patterns, employment oppor-

tunities, level of unemployment, and prevailing wage rates all affect

the labor market. On the farm, household size and composition, role

distinctions within the household, farm size, returns from surplus

production relative to returns from off-farm employment, ability to

hire labor, and additional food requirements of hired labor all

influence the farmer's labor decisions.








3.3.2 Land. Given the frequency that farm size is used as a

differentiating factor, it is surprising to find the lack of agreement
on its relationship to technology adoption. While some studies have

found size or area cultivated positively related to adoption (Beal

and Sibley, 1967; Arauz, and Martinez, 1983), others have found no

significant relationship (Barlow, et al., 1983; Ryan and Binswanger,
1979). Schutjer and Van Der Veen (1976) concluded that size of land

holding is more closely correlated with lumpy, indivisible technologies.
However, the introduction of custom work and a wider variety of

tractor sizes, and the improvement of markets and other institutional

mechanisms have partially overcome this indivisibility problem. Size

of holding is often closely correlated with ownership of other re-

sources as well, complicating the relationship between farm size and

technology adoption. Thus, the influence of farm size on adoption is

primarily through intervening variables such as institutional access,
wealth, power, and social status; size per se has little correlation

to adoption (Schutjer and Van Der Veen, 1976; Perrin and Windkemann,

1976). However, economies of size in transactions costs and the

ability to obtain quantity discounts can result in differential

adoption rates of even completely divisible technologies, leading

small famners to sometimes lag in adoption (Perrin and Winklemann,

1976).

Evidence of the relationship of tenancy to adoption is also

mixed. Schutjer and Van Der Veen (1976) conclude that what influence

there is, is indirect through variation in access to credit, markets,

and technical information. Morss, et al., (1976) found that social

control of land use and competing uses for fallow land constrain








adoption, and Scandizzo (1979) noted that landlords and tenants

favored different technologies.

Fragmentation of holdings becomes more important in irrigated

systems (Ryan and Binswanger, 1979), and with the introduction and

use of farm machinery. Group action required to effectively manage

an irrigation system is more difficult when there are many small

fields. Delivery systems and tractorization programs are less effec-

tive when fields are highly fragmented. And the distance that farmers
must travel between fields influences their management of those

fields.

While its effects on farmers' decisions are primarily through

intervening variables, farm size may reflect one or more other factors
and is easily understood as a basis for stratification. However,

farm size alone is usually not sufficient to explain differences

among farmers unless the variation is wide; rather farm size relative

to other resources, particularly labor, is the more crucial element.

Differing land/labor ratios might be an appropriate criterion for

stratification. Differing tenancy arrangements would be a basis for

stratifying farmers if the differences were substantial enough to

result in significantly differing adoption decisions. Fragmentation

may be significant in explaining differences among fanners, parti-

cularly in irrigated areas and in areas with mechanization programs.

3.3.3 Capital equipment. Timing of farming operations is
crucial to successful cultivation, particularly in rain-fed agricul-

ture. Besides labor, the critical variable affecting timing is

power source. Adequate power is also necessary to increase the

extent or intensity of farming operations. Ownership of or access to








tractors or draft animals, and their condition and health, are important

differentiating factors among farming systems (KSU, 1981; Barlow, et

al., 1983; Byerlee, et al., 1980; Eckert, 1982; Banta, 1980).

Sometimes variation in power source among farmers is obvious,
other times it is not. Farmers have often been stratified according

to the number of draft animals owned. The difficulty is in deciding

at what number to separate into another domain: e.g., are five head

of cattle significantly different from twenty head as far as power is
concerned? Obviously, a farmer with access to a tractor will have

different technological capabilities than a farmer with a span of

oxen. The difference is not so obvious when comparing a two-wheeled

hand tractor to draft animals, or when comparing oxen to horses. A

farmer with access to good dry season forage may have a decided

advantage over one without such access.

When using power source as a basis for stratification, the

crucial question is whether power is available when needed and in the

quantity needed. Significant variation in source of power among

farmers may result from ownership of a tractor or draft animals,

ownership of necessary implements, availability of hired traction and

the farmer's ability to hire it, and condition of the tractor and

implements and health of draft animals at the time they are needed.

Shared arrangements for draft power and access to communal or govern-

ment tractor programs are also important considerations.

3.3.4 Cash and income. One would expect that availability of

cash, from whatever source, is necessary for technology adoption.

Although Schutjer and Van Der Veen (1976) found there was no clear

relationship of cash to adoption, several more recent studies have








concluded that cash availability and farmers' willingness to expend

cash were clearly constraining to adoption of even profitable tech-

nologies (Barlow, et al., 1983; Byerlee, et al., 1980; Zulberti, et

al., 1979). Farm income and per capital income were also found to be

positively related to adoption (Morss, et al., 1976; Beal and Sibley,

1967). This supports the importance given to off-farm income sources

by several of the stratification stragegies reviewed in Chapter 2.

Important for stratification are the farmer's ability to purchase

or hire necessary inputs, his ability to meet financial obligations

and make additional investments, and his seasonal cash flow. Factors

which determine whether cash is constraining are the farm family's

income, both from farming operations and from off-farm employment,

and the family's cash requirement. Cash is required for farming

needs and households needs such as additional food, clothing, school

fees, and consumer goods. Household needs are dependent upon house-

hold size and stage of development as well as kinship and social
obligations. Access to borrowing, either formal or informal, is more

important for families with very limited resources who often find

credit access most difficult. Seasonal cash constraints may be

indicated by sale of crops and livestock at harvest and buying food

later at higher prices, working off-farm at a time of labor shortage

on the farm, and borrowing short-term at unfavorable rates.

3.3.5 Risk factor. The level and quality of farm resources

interacting with the ecological and institutional environment determine
the riskiness of the situation in which the farmer operates. Degree

of variability in factors beyond the farmer's control is reflected in









the risk factor, composed of yield risk due to ecological variability

and price risk due to market and price level variability (Byerlee,

Collinson, et al., 1980; Ryan and Binswanger, 1979; Cleave, 1976).

The farmer's resource base determines her ability to absorb risk. It

is generally agreed that the risk factor associated with a farmer's

environment impinges on adoption of agricultural technology: techno-

logies which increase risk are less likely to be adopted than technolo-

gies which decrease risk (Zulberti, et al., 1979; Valdez and Franklin,

1979). However, the risk factor should be reflected in the analysis

of the farmer's technical and institutional environments and resource

base. Therefore, it should not be necessary to include it as a

factor, for stratification, although the general riskiness of a

farmer's environment should be kept in mind. The risk orientation of

the farmer is looked at in conjunction with household orientation in

section 3.4.3.

3.3.6 Interactions within the farming system. By looking

strictly at components of a farming system, important interactions
can easily be overlooked. FSR takes a holistic view of the farming

system which requires an understanding of the dynamic interactions

and relationships within the famning system which bear upon farmers'

decisions. These interactions include competition for and allocation

of scarce resources (input-input relationships), complementarity

among enterprises where the output of one is the input for another

(input-output relationships), and the relationships among different

enterprises to satisfy the various needs and objectives of the house-

hold (output-output relationships) (Byerlee, Collinson, et al., 1980;







Moreno and Saunders, 1978). Banta (1980) suggests that an understanding

of input-input relationships (resource ratios) is particularly important.

And it has become widely accepted that relationships between enterprises,

especially between crops and livestock, are important to farmers'
adoption decisions. There have been many cases where a new crop

variety was not acceptable because the residue was not suitable for

fodder.

For stratification purposes, significant differences in system
relationships which potentially influence adoption decisions should

be identified. In more traditional systems, these relationships are

more complex and their identification is more important. Differences

may be identified by flow-charting the farming system (Hart, 1980,

1983) and by establishing the disposition of output from each enterprise.

Some differences may be accounted for when distinguishing ethnic
groups.


3.4 Household Characteristics

The household6 characteristics discussed in this section, while

closely related to farm-firm characteristics, are those which, to be

identified, require examination of the household as well as the
farming system. They frequently explain why farmers, operating under

seemingly similar circumstances, react differently to recommended

changes. Yet these household characteristics are often most elusive
to the research team and, hence, are seldom considered not only when

stratifying farmers but when planning and designing research as well.


6Household, as used throughout this thesis, refers to a group of
people, not necessarily related, who produce and consume as a unit.








3.4.1 Household composition. Compositon of a household includes
its size, make-up, and organization. Differences in composition can

account for differences in labor availability, in household require-

ments, and in goals and motivations, all of which influence farmers'
decisions.

Composition and organization of the household are important to
farmers' decisions through delineation of control and responsibility
within the household. Roles are clearly defined in subsistence

agriculture, usually on the basis of sex and age (Cleave, 1976).

Women's contribution to both agricultural production and off-farm

income are often considerable. Control of certain resources and the
returns generated therefrom are often divided by sex and are usually

independent of each other (Garrett, 1983; Norman, et al., 1982).

Clearly, role distinction within the household has a major influence

on decisions concerning technologies as well as on effects of tech-

nological change (Spencer and Byerlee, 1976). This latter point is

particularly important since the effects of change on women have so

often been overlooked; and these effects are often detrimental.

Norman, et al. (1982) noted that whether the household was
complex or nuclear affected decisions on technology adoption. A

change from complex to nuclear units resulted in less communal and

more individual control of fields. The resulting decentralized
decision-making created problems in introducing technologies when

fields were not controled by the family head. Fresco (1979), Kervin

(in KSU, 1981) and others have argued that households headed by women

should be a separate domain since they are usually resource poor and

they differ from male-headed households in decisions regarding

technology.








Stage of development of the household, though hard to ascertain,
influences its composition and, thus, its resources, needs, and

goals. Fortes (cited in Low, 1982, p. 24) characterizes the domestic

development cycle into five stages: establishment, expansion, con-

solidation, fission, and decline. Households at different stages in

their development cycle may be expected to exhibit different produc-

tion characteristics. Those early in the cycle normally have strong

desires to accumulate, emphasizing surplus production, and have high

consumer/worker ratios resulting in high demands on workers and
greater intensity of labor input per hectare (Low, 1983). Households

in the middle of their cycle are larger and have a definite resource

advantage, while households late in the cycle have less labor avail-

able, are often female-headed, and have little desire to accumulate
for the future (Low, 1983). The stage of development, then, can

account for differences in household's goals, resources, and employ-
ment prospects.

Composition of the household may provide a basis for stratifica-
tion along several lines. In many societies, particularly in Africa,

a number of households are headed by women, at least in a de facto
sense. In these cases, gender would be a basis for stratification.

Complex households might be dfferentiated from nuclear households.

Households in the early stages of development, those in the middle or

consolidation stage, and those in later stages may warrant three

different domains. Role distinctions within households, if signi-

ficantly different, might also form a basis for stratification.

Identification of differences in household composition will come

out of farmer interviews. Particular attention should be paid to








work patterns within the household for both agricultural and non-

agricultural tasks. Roles of various household members in the different

activities should be identified. An accurate assessment of this

implies that members other than the household head need to be inter-

viewed as well.

3.4.2 Household goals. Decisions made by a farmer relate

household goals and objectives to the specific situation in which the

household operates. While compatibility with farmers' circumstances

is necessary for adoption, compatibility with farmers' goals provides

the sufficient condition for adoption (Norman, et al., 1982). Gilbert,

et al. (1980) refer to goals and motivations as "the motor that
drives the entire system" (p. 9).

One view is that the desire for economic gain ultimately outweighs
other goals; that food preferences, risk aversion, etc., simply

modify the economic goal (Byerlee, et al., 1980; Foster, 1973).

Others contend that non-economic motivations are paramount. Collinson

(1982), for one, gives the priorities of farmers as 1) social and

cultural obligations, 2) reliable supply of preferred foods, 3) cash

for additional basic needs, and 4) extra cash. He suggests that this

order of priorities is fairly firm, though the weight given to each

will vary depending on how close the household is to subsistence.

Others argue that providing for subsistence -- satisfaction of

the household's basic food requirement -- is probably the most

important goal of small farm families (Norman, 1983; Norman, et al.,

1982; Cleave, 1976). This subsistence need may even lead marginal

farmers to value their wage employment more highly than their farming

operation, resulting in resources being directed off-farm (Behnke and








Kervin, 1983). Other goals which are influential to farmers' deci-

sions are food preferences, kinship and social obligations, and

desire for social status, power, and prestige (Norman, et al., 1982;

Anthonio, 1977; Cleave, 1976).

The important point is not whether economic or non-economic

goals are primary -- they both are important. Rather, the point is

that differing goals result in different decisions. When stratifying

farmers, identifying differences in the relative weights given to
various goals is the key. Of particular concern is the importance

attached to non-economic goals. Subsistence needs, security, and

food preferences dictate cropping patterns, choice of variety, plant-

ing dates, and storage and marketing strategies. The strength of

these goals may vary according to the family's level of subsistence,

its stage of development, and its involvement in the market. A high

emphasis placed on food security may be identified by diverse cropping

patterns and use of traditional varieties and technologies, particu-

larly on the main staple crop. Substantial household expenditures

for school expenses, health care, etc., or a household in its early

stages of development probably indicate a greater emphasis on economic

gain, particularly if household food requirements have already been

met.

3.4.3 Household orientation. The orientation of the household

toward market production, toward government, and toward control over

nature (scientific orientation) may help explain low market activity

and low adoption rates. Orientation reflects and influences household

goals and motivations. Importance of societal norms diminishes as









farmers acquire a more positive market and scientific orientation

(Norman, et al., 1982; Cleave, 1976).

Among the most difficult factors to identify are farmers'

attitudes and perceptions, yet they can dramatically influence adop-
tion decisions.. As perceived by an outside observer, resources might

be present at optimum levels; yet farmers may perceive these resources

as inadequate or even non-existent. Beal and Sibley (1967) cited
studies which indicate that adoption is related to farmers perceptions

and the results of their own study in Guatemala generally supported

this. Foster (1973) argued that differential cross-cultural percep-

tions are important barriers to change. Farmers' attitudes towards

government and institutions, and their attitude toward control over

nature, or scientific orientation, all affect adoption decisions
(Foster, 1973; Beal and Sibley, 1967).

The problem is that farmers' attitudes and perceptions are

difficult to ascertain. Market orientation, which is more easily

determined, can serve as a proxy. Market orientation has been shown

to be positively correlated to adoption rates (Norman, et al., 1982;

Wilson, 1977; Cleave, 1976; Beal and Sibley, 1967). Thus, the degree

of market orientation should be a basis for stratifying farmers.

Several factors can be used to identify a household's market
orientation. Historical use of recommended technologies and past

participation in development schemes have been found to explain

variation in adoption (Barlow, et al., 1983; Beal and Sibley, 1967)

and it gives an indication of the degree of market orientation.
Extent of cash cropping, use of purchased inputs, number of consumer









goods purchased, food purchases, income, and participation in local

organizations are also good indications.

A determinant of market orientation is the household's degree of

averseness to risk, or its risk orientation. Risk avoidance is

generally concluded to be a constraining factor in adoption of agri-

cultural technologies by small farmers (Norman, et al., 1982; Zulberti,

et al., 1979). The closer a household is to a minimum societal

standard for survival, the more strongly risk avoidance affects their

behavior (Wharton, 1968). The question is whether there is sufficient

variation in the risk orientation of small farmers to include it as a

criterion for stratification.

While some studies have reported the degree of risk aversion to

be negatively related to technology adoption (Moscardi, 1979 and Beal
and Sibley, 1967, for example), a number of others have concluded

that the degree of risk aversion is not responsible for differential

rates of technology adoption (Norman, et al., 1982; Sanders and Dias

de Hollanda, 1979; Roumasset, 1979; Schutjer and Van Der Veen, 1976).

The argument is that small farmers are at least moderately risk

averse, but that variation in their risk orientation is minimal.

Where such variation exists, it does not appear to contribute to

adoption differences. And, of course, some differences in risk
orientation are accounted for in differences in market orientation.

3.4.4 Other household characteristics. Certain cultural traits

of the household influence adoption of agricultural technology. To

be accepted, technology must be compatible with norms of modesty,

work values, customary motor patterns and body positions, and supersti-

tions and taboos (Foster, 1973). While these may be the cause of








sane variation in adoption decisions, they generally do not vary

within ethnic groups and it would probably not be cost effective to

ascertain such information for stratification.


3.5 Checklist of Socioeconomic Factors

Theoretical evidence and arguments suggest the existence of

certain relationships between different factors and adoption of

agricultural technology. These relationships have been reviewed

above. However, empirical data supporting the theories are scarce.

Drawing on what evidence there is and on the conclusions of others,

this chapter has identified a number of socioeconomic factors which

appear to explain differences in small farmers' adoption behavior.

The most important of these factors or characteristics are

summarized in a checklist (Table 1) at the end of this chapter. For

each factor, possible criteria for stratification are given and key
contributing factors which may explain or help to identify variation

in the factor are presented. Inclusion of a factor in this checklist

was dependent primarily on its apparent importance in explaining
differential adoption among small farmers. Other considerations for

inclusion were the extent of influence of a factor on other factors

(e.g., household goals), the frequency or likelihood that a factor or

characteristic may be overlooked by the field team (e.g., interactions),
and the relative ease with which variation in a factor might be

identified.

This checklist is intended to be just that, a checklist. It is

not meant to be a blueprint for stratification, nor is a blueprint
possible given the diverse nature of small farming systems around the





63

world and the site specificity of farming systems research. Rather,
the intention of this discussion is to assist researchers in identi-

fying sources of variation among fanners which are significant enough

to warrant separate recommendation domains. A possible procedure for

incorporating these factors into the stratification process is

presented in the next chapter. The procedure is then illustrated

with the case of Lesotho's lowlands.











Table 1. Socioeconomic Checklist for Farmer Stratification


Community Environment:
MARKET ACCESS









INFORMATION SOURCE


Factor or Characteristic Criteria for Stratification Key Contributing Factors


ETHNIC/CLASS DIFFERENCES

INTERACTION AMONG GROUPS


Differential access physically and
psychologically to product, factor,
and food markets







Differential access to market
information and information on
new technologies, provided that
inadequate access is not easily
corrected

Different ethnic groups and socio-
economic classes
Existence and nature of reciprocal,
cooperative, and competive arrange-
ments among different farmer groups


Distance from roads and markets
Infrastructure development, including transportation
Market information
Farmer's cash and transportation resources
Farm size & ethnic differences resulting in different
prices and transactions costs
Farmer's past A current market activity: extent of cash
cropping, use of purchases inputs, food purchases, and
consumer goods
Membership in local organizations
Farmer's perceptions of market access
Past use of recommended technologies
Extension
Membership in local organizations
Radio ownership
Farmer's information source behavior
Farmer's attitude toward government


Arrangements for sharing and exchanging resources
Communal work arrangements
Control and allocation of common resources









Table 1. Socioeconomic Checklist for Farmer Stratification (cont'd)


Factor or Characteristic Criteria for Stratification Key Contributing Factors


Farm Resources:
LABOR







POWER SOURCE


Population densities and unemployment rates
Employment opportunities and wage rates
Seasonal migration patterns
Household size, particularly relative to farm size
Household composition
Cash resources to hire labor
Returns from surplus production relative to returns
from off-farm employment
Fanner's perceptions of the local labor market
Livestock or tractor ownership
Availability of hired tractor
Cash resources to hire traction
Shared arrangements for traction power
Condition of machinery and implements
Health of animals at the times when they are needed
Income, Including off-farm income
Household cash requirements
Seasonal receipts and payments
Early sale of harvest, off-farm work, and short-term
borrowing may Indicate seasonal cash constraints
Access to capital (credit), particularly for resource-poor
farmers
Cropping patterns, intensity, and rotation may Indicate
land constraint and degree of fragmentation
Farm area, area under cultivation, area in a particular
crop
Household size and food requirements
Tenancy agreements


Labor-surplus or labor-deficit
households; different labor bottle-
neck periods; significantly different
types of labor available; and areas
with different labor market
characteristics



Differences in timely availability
of adequate power; significant
differences in the type, quantity,
and condition of power source


Differences in farmers' abilities to
purchase Inputs and meet financial
obligations;, significantly different
cash flow problems; and significant
differences in off-farm remittances


Hide differences in farm size,
particularly relative to labor and
other resources; significantly
different tenancy arrangements; and
significant differences in degree
of fragmentation, especially in
irrigated systems


LIQUIDITY


LAND











Table 1. Socioeconomic Checklist for Farmer Stratification (cont'd)


Factor or Characteristic Criteria for Stratification Key Contributing Factors


FARMING SYSTEM INTERACTIONS






Household Characteristics:
HOUSEHOLD'S GOALS







HOUSEHOLD C(OPOSITION




HOUSEHOLD ORIENTATION


Significantly different system
interactions which could potentially
affect adoption decisions





Differences in the relative weights
given to various goals, particularly
between economic and non-economic
goals





Gender of household head; complex
or nuclear in organization; and
significant difference in the roles
of household members

Low or high market orientation of the
household


Crop-livestock relationships
Competition for resources among activities; input-input
Complementarity among activities; input-output
Relationships among activities to provide for the various
household goals, eg., cash-cropping or food-cropping;
output-output
Disposition of output from various activities

Level of subsistence
Household food requirements
Stage of household development
Household's market orientation
Household's cash requirements
Kinship, religious, and social obligations
Security emphasis may be identifed by cropping diversity
and use of traditional varieties and methods, particularly
on food crops
Stage of household development
Roles of men, women, and children in the household
and in the production of agricultural output
Who controls what resources
Means of household decision-making

Present and past use of recommended technologies
Participation in development programs
Current and past market activity: extent of cash cropping,
use of purchased inputs, food purchases, and acquisition
of consumer goods
Income
Scientific orientation (control over nature)
Attitude toward government














4. INCORPORATING SOCIOECONOMIC FACTORS IN FARMER STRATIFICATION


The preceding discussion illustrates the numerous factors which
are potential sources of variation in adoption of technology among

farmers. However, not all of these factors are significant in any
one situation.. A stratification process needs to be devised through

which field teams can identify those factors which are significant in

their particular situation. Such a process must be simple enough to

be applicable in the field, yet thorough enough to ensure that critical

elements are not overlooked.

An examination of Table 1 reveals the interrelatedness among

factors. It also reveals that certain factors, such as market orienta-

tion, reflect a number of other circumstances, many of which are

difficult to identify in the field. These revelations are useful in

developing a simplified procedure for delineating fanner target
groups. The purpose of this chapter, then, is to disentangle and

prioritize the myriad factors that influence farmers' adoption deci-

sions into a workable framework for farmer stratification. This

framework is illustrated in Section 4.2 for the case of Lesotho's
lowlands.


4.1 A Procedure for Socioeconomic Stratification

FSR practitioners generally follow a three-tiered procedure for

identifying farmer circumstances and researchable problems. Delineating







recommendation domains is part of all three steps, but the third step

,is primarily for verification. The first step is to study and analyze

secondary data in order to gain a good understanding of the general

topographic, climatic, demographic, and institutional environments of

the target area. Preliminary target groups are identified from

obvious variations in these data.

The exploratory, or informal survey is the second step. Primarily

qualitative information is gathered by quick inspection of the research
area and through informal interviews with farmers, merchants, agri-

cultural officers, and village leaders. Most of the information used

to delimit recommendation domains is obtained here. Step three, the

verification or formal survey, is primarily to gather more quantita-

tive information to verify problems and variations hypothesized from

the exploratory survey. It also provides researchers with a more

solid base for developing technologies to be tested on farmers'
fields.

Stratification starts with the analysis of secondary data.

Farms can be distinguished according to obvious variations in agro-

ecologic characteristics, access to markets and other institutions,

and farm size. Local population densities and labor market charac-

teristics should be identified for purposes of comparison to farmers'
individual characteristics later on. Significant variation in popu-

lation densities and labor market characteristics across the target

area should be noted as potential stratification criteria. Different

land tenure arrangements in the target area should be identified as

to their potential for differentiating farmers. Different ethnic
groups may also be identified here. Information gathered from









secondary data sources help guide researchers in the design and

implementation of the exploratory survey. At this point strati-

fication is very tentative; farms as physical units and farming areas

may be stratified, farm households, though, are not.

Stratification of farm households really begins with information

gathered in the exploratory survey, though information from secondary

data sources is still very useful. It is easiest if farmers are

first grouped according to parameters which have the fewest alterna-

tives and are easiest to identify. Also, other factors may only be

important for certain groups, such as small farmers with low market
orientation. With this in mind, the following stratification pro-

cedure is proposed. (See Figure 2)

Farm size is a good stratification criterion with which to

begin. It is relatively easy to identify and is easily understood by

research and extension personnel. Differing farm sizes reflect

different resource endowments, different goals and motivations, and

perhaps different risk orientation. At this point, the distinction

should only be made between large, commercially oriented farmers and

medium and small farmers. Large farmers generally have high market

orientation, indicating good institutional access, and likely have
had past experience with introduced technologies, suggesting a willing-

ness to experiment and change. Separating out "large" farmers at

this point is an attempt to capture all of these differences which

usually occur together. Differences within this large farmer group,

resulting in different technological requirements, are primarily

agro-economic rather than socio-cultural. The actual distinction

between medium and small farmers and large farmers is relative and











Topography and climate
Population densities
Institutional environment Preliminary Recommendation Domains
Land tenure Derived from Secondary Data Sources
Local labor markets
Seasonal migration patterns



Total farm area __ -Farm Size Large with high
Cultivated area ---- market activity

medium
& small

Market access
Market activity
Past technological -------- Market Orientation g
experience
Income Low
Membership in farmer
organization

Ethnic Group


Female- or Male-Headed

Ownership of tractor
or draft animals /


Hired traction
Condition of machinery
Health of animals


Cropping patterns
and intensity


SPower Source Differences


___ . . . . |----


System of rotation Current Farming System Differences
Fragmentation and Differences in Potential for Change
Animal husbandry
System interactions
Changes over time


Possible Sources of
Variation

Landholding Labor Constraint Different
Off-farm income Land Constraint Technological
Household size Cash Constraint Requireents
Household composition Verify Food Needs &
State of development --Sources of--- Preferences
Role distinctions Variation Cash Requireients
Differential control Security
of resources Social Obligations
Decision-making Tenure Arrangements
Perceptions
Attitudes


Interactions among groups - - ----- -- RECO1ENDATION DAINS


Verified by formal survey
and refined if necessary


Note: Box indicates where a stratification decision is made.

Figure 2. Procedure for Socioeconomic Stratification.


.








must be made in light of the particular situation. Further division

of small- and medium-sized farmers on the basis of size is not

necessary at this point because absolute farm size is not as impor-

tant as farm size related to other farm resources. Further division

along other lines is probably more useful.

Farmers with small- and medium-sized operations can then be
stratified according to their degree of market orientation. Market

orientation reflects two important factors, market access and resource

endowments, as well as a farmer's level of subsistence, motivations,

and risk orientation. A high degree of market orientation indicates

good institutional access, at least adequate resources to overcome

major constraints, high economic motivation vis-a-vis subsistence or

security goals, and limited aversion to risk. Low market orientation

probably indicates the opposite is true in at least one of these

areas.

Farmers with high market orientation are apparently not as

influenced by social and cultural norms as are more subsistence

oriented farmers (Norman, et al., 1982; and others). Their circum-

stances are probably more akin to those of larger farmers, and they

may be grouped with them at this point. If research is to be targeted

at this group, further distinction can be made based on agro-economic

variations which indicate different technological requirements.

When deciding whether a farmer has a high or low degree of

market orientation, researchers should be conservative and, when

there is doubt, include the farmer in the low group. An error in
this decision would be less costly to the fanrer if he is wrongly









included in the "low" group rather than if he is wrongly included in

the "high" group.

Small and medium-sized farmers with low market orientation are

usually the focus of development research in low income countries.
Many socioeconomic and cultural factors influence adoption decisions

in this group, so stratification becomes more important and more

difficult. Thus, an early distinction should be made according to

ethnic group. Many important socio-cultural differences such as food
preferences and social obligations may be explained by ethnic origin.

Female- and male-headed households should also be separated into

different recommendation domains. Many contemporary Women in Develop-

ment experts contend that female-headed households react differently
than do those headed by males (Kervin, in KSU, 1981; Fresco, 1979).

For one, women-headed households generally are more resource-poor.
However, even given similar resource endowments, it is argued that

women tend to react differently than men because of their socialization.

For example, they probably act more cautiously than do men and they
generally place a greater emphasis on household needs and are less

likely to make investments in the farm. Further, in most cultures,

women probably have lesser institutional access and may not have the
respect accorded to men.

Now is the time to begin careful examination of the different
farming systems. A first source of variation to be looked for is

difference in power source. Power is critical to most farming opera-

tions, it is commonly used in current stratification strategies, it
is relatively easy to identify, and it has a limited number of alter-

natives. Source of power available to the farmer is the major concern









here, although a famner's choice of power reflects other constraints

in the cropping system and household as well.

Other differences in farmers' current famning systems need to be

identified. These include differences in enterprises, differences in

management practices such as cropping patterns, intensity, rotation,

fragmentation of fields, and animal husbandry, and differences in

interactions within the system. Hypotheses on the sources of these

variations are developed. Variation may be due to constraints on

labor, land, or cash; they may be due to differing food needs and

preferences, cash requirements, security needs, or social obliga-

tions; or they may be the result of different tenure arrangements.

These hypotheses then need to be verified by examining household

characteristics, including household size and composition, stage of

household development, role distinctions and differential control of

resources, differences in decision-making, and famner's perceptions
and attitudes.

A final consideration is interaction between the different
groups of farmers identified by the field team. Farmers involved in

reciprocal and cooperative arrangements may need to be grouped

differently, depending on the nature of the agreements and their

influence on the farmer's decisions.
At this point the field team should be fairly confident of their

delineation of recommendation domains and the research staff can

begin to formulate recommendations for each domain. Necessarily,

decisions must be made on whether variation is significant or not.

This is where the "artistic" skill of the researcher is important.
Following the formal survey, recommendation domain boundaries should








be refined if necessary. However, with a good exploratory survey,

little adjustment is usually needed (Collinson, 1982).

This procedure for stratifying farmers into recommendation

domains is summarized in Figure 2. The order of the decision criteria
may vary in some situations to reflect differences in the relative

importance of certain factors. It should be obvious to the reader

that, should farmers be stratified according to each of the potential

differentiating factors, an unworkably large number of domains will
result. This should not happen for three reasons. First, signifi-

cant sources of variation will probably be limited to just a few
factors in any given situation. Second, policy decisions regarding

groups to be targeted and groups which are too small to be signifi-

cant will limit the number of recommendation domains which are

ultimately considered. And third, the type and scope of potential

recommendations may allow two or more recommendation domains to be

combined into one. The following case should illustrate these points.


4.2 The Case of Lesotho's Lowlands

Lesotho's lowlands are generally north-south in orientation and
are low only relative to the mountains. They lie 5,000 to 6,000 feet

in altitude and are characterized by deep gullies (dongas), severely

eroded hillsides, sandstone cliffs, little flat land, and, thus, a

very limited arable area. The climate has four seasons with frost

and occasional light snow in the winter and irregular rainfall in the

summer, coming mostly in violent thunderstorms with frequent hail.








Most of the soils are marginally suited for cultivation and the

grasslands are severely overgrazed.

Agriculture is primarily small, subsistence farming. The main

staple is maize with sorghum also produced for domestic consumption.

Until recently, traditional land tenure prevailed where farmers

maintained only usufruct rights to the land and allocation of these

rights was through local chiefs. Although a new Land Act was recently

adopted which provides for long-term leaseholds that allow farmers to

borrow against their holdings, farming still reflects the traditional

tenure system. Most households farm two or three scattered fields of
varying quality. As a result of traditional land allocation there are

very few "large" holdings. In fact, due to repeated subdivision of

holdings, most rural Basotho cultivate quantities of land inadequate

to meet subsistence needs.

Insufficient returns from farming are supplemented by remittances

from migrant labor in the Republic of South Africa. Half the male

work force and about 20% of the female work force are absent from
Lesotho at any one time. A full 60% of the households receive income

from external labor migration and another 10% receive income from
wage employment in Lesotho (van de Wiel, 1977). While the future

prospects for migrant labor are uncertain as South Africa undergoes

internal changes, off-farm income will continue to be significant for

rural Basotho households.

The following example of a stratification of farmers in lowland

Lesotho is presented for illustration only. While an accurate repre-

sentation of farming is this area is attempted, certain assumptions








are made in the absence of data that a research team would normally

acquire. These assumptions are made based in part on my own experience

in Lesotho. The data used in this case are from Toward the Year 2000:

Strategies for Lesotho's Agriculture by Eckert (1982) unless otherwise

specified. These data were compiled from a survey of the Thaba Bosiu

Project area but Eckert concluded that they were fairly representative

of Lesotho's lowlands.

Analysis of available secondary data is the first step in deter-

mining recommendation domains. For the lowlands of Lesotho, this

analysis yields the following information. During the summer growing

season, the northern part receives significantly more rainfall than

the central and southern areas. In the south, rainfall is more

variable and humidity is lower. The length of the growing season

(defined by the average number of frost-free days) declines signifi-

cantly from north to south. In the north, there is a predominance of

richer, deeper soils with higher moisture holding capacities. In

general, then, growing conditions become poorer with movement from

north to south. Yield data confirm that there is a comparative

advantage for maize in the north and for sorghum in the south. These

several differences acting together probably indicate that the low-

lands of Lesotho should be preliminarily divided into north, central,

and south regions. Their boundaries should follow current district
boundaries for easier administration.

Certain other potential sources of variation are also revealed

from the analysis of secondary data. This information helps to

direct the research team in its search for significant variation

among farmers. The high incidence of migrant labor means that








off-farm income is probably significant for stratification. Van der

Wiel (1977) estimated that 34% of the households in Lesotho are

headed by women and that 68% are managed by women in'their husband's

absence. Gender of the household head, then, is likely significant

as well. Since Lesotho is basically populated by one ethnic group,

this distinction is not necessary; and since there are virtually no

large farms, farm size is not important either. (This would not be

the case where accumulation of landholdings is possible.)

The importance of cattle to farming in Lesotho also becomes

apparent from secondary analysis. Ownership of cattle allow a house-
hold to be self-sufficient in draft power and have control over the

timing of plowing and planting. This is particularly important in
light of Lesotho's single growing season and uncertainty of the first

spring rains. Finally, it should be noted that 16% of rural lowland

households do not have access to land, and that a significant number

are not accessible by road, limiting potential market activity.

The extent that potential variation is significant, and the
number of households in different possible groups can be ascertained

from data gathered in informal surveys. (However, in this case, data

were gathered in formal surveys.) This information is used to define

possible recommendation domains while policy will dictate the ultimate

delineation. Although this illustration is for the central lowlands,

similar procedures could be employed in the northern and southern

regions. The decision tree in Figure 3 portrays this procedure for

stratification of farmers in the central lowlands of Lesotho.

Although most agricultural production in Lesotho is for
subsistence, there are a number of farm households which are active









1.1 Landless 16%ab
(0%)


1.2 High market
orientation


2.1 labor surplus 5%
(<0.4 ha/cap.) (6%)
18X
18% 3.1
(21%) 2.2 land surplus 13% <
(>0.4 ha/cap.) (151) 3.2


cattle only 4%
(5X)
cattle and tractor 9%
(10%)


3.3

2.3 Hale-headed 41%
(48%)

3.4
1.3 Low market 66
orientation (79%)
3.5


2.4 Female-headed 26%
(31%)
3.6


No cattle 27%
(32%)


cattle 14% ---
(17%)

cattle 7%
(8%)



No cattle 19%
(23%)\


4.1 No off-fam 6%
income (9%)
5.1

4.2 off-farm 21%c 5
income (25%) 5
5.3
4.3 off-farm 14%c 5.3
income (17%) 5.4



4.4 off-farm 13% ----- 5.5
/ Incoe (16%) .
5.6


4.5


labor-surplus 12% (14%)

land-surplus 9% (11%)
labor-surplus 10% (12%)
land-surplus 4% (4%)



labor-surplus 9% (10%)
land-surplus 5% (5%)


no off-farm 13%
income (15%)


a
bPercentages are of all rural households.
cPercentages in parentheses are of landed rural households.
An estimated half of 4:2 and 4.3 are female managed, or a total of 18% of all rural households.
Note: Percentages may not add up due to rounding.

Figure 3. Stratification of Rural Households in the Central Lowlands, Lesotho.








in agricultural markets. Ideally, these households would be identi-

fied by primary data on the extent of their market involvement. In

the absence of such information, farm resources are used to identify

those households which are likely to be highly market oriented.

Households which have off-farm income and own cattle with yoke and

plow, cultivator, harrow, and planter, and households with off-farm

income and which have cattle with yoke and plow and adequate land

(greater than 0.4 ha. per household member) to produce a surplus are

considered to have high market orientation. Some of these households

have access to tractors as well.

The highly market-oriented farm households comprise 21% of rural

landed households or 18% (21% of 84%) of all rural households.

(Henceforth, percentages given are of all rural households.) They

are primarily headed by males so gender is not considered for this

group.

Eckert (1982) considered that the amount of land per household

member was significant to farming in Lesotho. He estimated that at

least 0.4 ha. per capital was necessary to be self-sufficient in

cereals and have the potential to produce a surplus. This figure

could be lowered slightly if the farmer had other resources adequate

to farm intensively. Households with less than 0.4 ha. per capital

are described as labor-surplus while those with greater than 0.4 ha.
per capital are land-surplus.

Less than one-third of the households with high market orienta-

tion were labor-surplus. This group was 5% of all rural households.

Land-surplus, highly market-oriented households were 13% of all rural

households.









Further distinction of market-oriented farmers might be made
according to whether or not they have access to tractors. From a

baseline study, tractors were found to have been used on 15% of the

fields in one project area (van der Wiel, 1977). Assuming (1) this
is representative of the lowlands in general, (2) that only a small

portion of these fields would not be included in my grouping of
market-oriented farmers (to include them would raise the percent of

landed households with high market orientation to, say, 25%), and (3)
that the skewness of land distribution means that 15% of the fields

represents less than 15% of the households, I estimate that about

half of the market-oriented farmers have access to tractor power.

In the case of labor-surplus households, tractor power is probably
not significant since their fanning is labor intensive. The propor-

tion of households using tractors in labor intensive operations is

likely to be small. Therefore, maybe two-thirds of the land-surplus,

market-oriented households used tractors, or about 9% of all rural

households.
At this point, the stratification process has yielded three

recommendation domains for market-oriented households: (See Figure

3.)

1. labor-surplus, market-oriented households: 5% (2.17 in

Figure 3)
2. land-surplus, market-oriented households with no tractor

access: 4% (3.1 in Figure 2)

3. land-surplus, market-oriented households with tractor

access: 9% (3.2 in Figure 3).


7Refers to the cell number in Figure 3.








Franzel (1981) stated that a group of fanners comprising at

least 10% of the farmers in an area is significant enough to be a

separate recommendation domain. This seems to be rather arbitrary,

but research budgets certainly limit the size and number of domains
to be targeted. In this case, policy considerations may dictate that

the first two are too small to be considered. In the case of market-
oriented farmers, per capital landholding is more important in determining

cash cropping potential than is type of traction, particularly since

all these farmers do have animal power. In Lesotho, farmers produce

cereals (mainly wheat), legumes, and, in some cases, vegetables for

market. Commercial production of cereals requires adequate land

resources; one would expect labor-surplus households to have a compar-

ative advantage in commercial production of legumes and vegetables

which require a significantly larger labor input per hectare.

For this reason, it would be appropriate to combine the two

land-surplus domains, leaving two domains of market-oriented farmers:

1. labor-surplus, market-oriented farmers: 5% (2.1 in Figure 3)

2. land-surplus, market-oriented farmers: 13% (2.2 in Figure 3).

Farmers in each of these domains have different cash cropping potential.

Given the non-land resources of farmers in the first domain, they

likely farm intensively and may cultivate other fields through a
sharecropping agreement. Farmers in the second domain are more

likely to farm extensively. For this group, the fact that a signi-

ficant proportion do not have access to a tractor should be remembered

since this affects their ability to plant their crop on time.

It should be noted that the number of market-oriented households
might actually be greater than indicated here. Substantial off-farm








earnings, sharecropping, and cattle exchange make it possible for

some households, who otherwise couldn't, to purchase inputs and food

and sell produce.

For households with low market orientation (66% of all rural

households, or 79% of landed households), more factors normally need
to be considered for stratification. Lesotho is no exception.

Gender of the head-of-household is an important distinguishing feature.

(See Figure 3.) I estimate from Eckert's (1982) data that male-

headed households with low market orientation comprise 41% of all

rural households; female-headed of the same group make up 26%.

Ownership of cattle is important for a household's control over

land preparation and timely planting. Male-headed households should

be divided into those with cattle (implying self-sufficiency in draft

power) and those without. The number of female-headed households
with cattle is small (7% of all rural households). And since women

do not normally handle livestock, which means that plowing must still

be contracted, differentiation based on cattle ownership is not

significant for households headed by women. (This distinction is

still shown in Figure 3 in order to present more information.)

Since many non-market-oriented farmers in Lesotho can be con-

sidered as part-time farmers, off-farm income is particularly

important. The data indicate that there are very few (if any) house-

holds in Lesotho that have cattle and no off-farm income. This is
reasonable since the main reason Basotho migrate to work off the farm

is to accumulate assets for the future, and the first assets they tend

to acquire with their earnings are cattle. Therefore, only male-

headed households with no cattle and female-headed households are









divided according to whether they have a source of off-farm income.

(See Figure 3.)

Since non-market-oriented farmers generally do not farm inten-

sively, relative per capital land areas are not as important for this

group. However, this information may become important as technolo-

gical improvements are adopted and some farmers become more active in

the market. For this reason, those households which can potentially
become market oriented (those with cattle and off-farm income, 4.3

and part of 4.4 in Figure 3) are further divided into labor-surplus

and land-surplus households. Even households without cattle may have

surplus land which can potentially produce food for Lesotho's food-
deficit economy. These households should be identified to aid develop-

ment planning (4.1, 4.2, 4.5, and part of 4.4 in Figure 3).

Also important for this group of potential market actors is

whether the farm is accessible by road. Road access is necessary for

the farmer to market significant surplus production. Much of Lesotho's

lowlands are not accessible by road but the number of households
farming in these areas is unknown. Hence, road access is not used in

this stratification. However, its importance must be considered when

planning research for this group of farmers.

Although per capital landholding is not used to define recommen-

dation domains initially, the proportions which are labor surplus and

land surplus are noted for relevant domains in the event that further

delineation along these lines is found to be desirable. The stratifi-
cation of households with low market orientation yields the following

recommendation domains: (See Figure 3.)









1. male-headed households with no cattle and no off-farm

income: 6% (4.1 in Figure 3.1; primarily land-surplus)

2. male-headed households with no cattle but with off-farm

income: 21% (4.2 in Figure 3; labor-surplus, 12%; land-
surplus, 9%)

3. male-headed households with cattle and off-farm incomes:

14% (4.3 in Figure 3; labor-surplus, 10%; land-surplus, 4%)
4. female-headed households with off-farm incomes: 13% (4.4

in Figure 3; about half have cattle; labor-surplus, 9%;
land-surplus, 5%)

5. female-headed households with no cattle and no off-farm

incomes: 13% (4.5 in Figure 3; primarily land-surplus).

According to van der Wiel (1977), a large number of households

headed by absentee males are actually managed by women. Thus, it is

estimated that half of the male-headed households with low market

orientation and with off-farm incomes (4.2 and 4.3 in Figure 3) are

actually managed by women. (See page 76.) In Lesotho, this generally

means that women make the day-to-day decisions but not the larger,

one-time investment decisions. This group of male-headed, female-

managed households comprise about 18% of all rural households. This

group is significant since decisions to adopt recommended techno-
logies are most likely made by a household member who is frequently

absent and who is not actively involved in the daily operation of the

farm.

Two of the recommendation domains of non-market-oriented farmers

are probably not in a position to benefit directly from agricultural
research. These are the male-headed and female-headed groups with no








cattle or off-farm incomes (4.1 and 4.5 in Figure 3). They are

probably welfare households and can be combined into one domain.

Potential programs for this group might involve increased off-farm

labor opportunities. Under the new land act, they may also be able

to lease their underutilized landholdings to those farmers with

superior non-land resources.

From this stratification of rural households in the central
lowlands of Lesotho, eight recommendations domains have been identi-

fied. Two of the groups should be of vital concern to development

planners but would not be targeted for agricultural development:

1. Landless households: 16% (1.1 in Figure 3)

2. Households with no cattle and no off-farm income that are

primarily labor deficit: 19% (4.1 and 4.5 in Figure 3).
The significance of these two groups is that 35% of all rural house-

holds are not likely to benefit from efforts to increase agricultural

production.

Two other recommendation domains have been identified as market
oriented:

3. Labor-surplus, primarily male-headed households: 5% (2.1
in Figure 3)

4. Land-surplus, primarily male-headed households: 13% (2.2
in Figure 3; households with access to tractor power are

9%, 3.2 in Figure 3).
These households all have cattle and off-farm incomes and they have

sufficient amounts of other resources to likely be active in the
marketplace. These are generally the more well-off households in








Lesotho. Efforts to increase aggregate food production will probably

be targeted at these groups.

Households with low market orientation are divided into four
recommendation domains:

5. Male-headed and managed households with no cattle but with

off-farm incomes: 10% (half of 4.2 in Figure 3; over half

of these have surplus labor)

6. Male-headed and managed households with cattle and off-farm

incomes: 7% (half of 4.3 in Figure 3; over 70% of these
have surplus labor)

7. Male-headed but female-managed households with off-farm

incomes: 18% (5.1 in Figure 3)
8. Female-headed households with off-farm incomes: 13% (4.4

in Figure 3; almost two-thirds of these have surplus labor).

Together, these groups make up 47% of all rural households. They are

primarily part-time farmers, relying on supplemental off-farm remit-

tances. Efforts to improve the welfare of rural households through

agricultural development should be directed at these groups. Those
with cattle and landholdings surplus to subsistence needs might be

expected to produce a surplus with improved cultural practices.

It must be stressed that stratification of farmers into recommen-

dation domains is an iterative process. The decision tree in Figure 3
provides a substantial amount of information which allows domain

boundaries to be adjusted as new information is obtained, as produc-

tion constraints are identified, and as development goals and direc-

tions are refined.














5. SUMMARY AND CONCLUSIONS


5.1 Summary

This thesis sought to: 1) review and codify current FSR strati-

fication approaches, 2) identify socioeconomic factors which are

important when stratifying small farmers, and 3) develop guidelines

for systematically including socioeconomic factors in small farmer

stratification.

Rationale for stratifying farmers is threefold. One, effective

research can only be conducted for a similar group of farmers. Two,

extrapolating research results to other similar farmers improves the

cost-effectiveness of farming systems research. And, three, recom-

mendation domains can aid in formulating policy and development

alternatives. A recommendation domain is a relatively homogeneous

group of farmers for whom more or less the same recommendations can

be made. Farmers in the same domain should have the same researchable

problems and development alternatives, and should react in the same

way to policy and technological changes.

Two assertions are made regarding the identification of factors

to be considered in farmer stratification: one is that critical

factors influencing farmers' decision concerning adoption of agri-

cultural technology are the factors which should be considered for
stratification. The other is that socioeconomic and cultural factors








are as important, and in some cases more important than technical

factors in explaining differences in farmers' adoption decisions.

Both assertions are intuitively acceptable and evidence presented in

this study supports them.

In light of these assertions, the following hypothesis was

offered: that incorporating socioeconomic and cultural factors into

the fanner stratification process in Farming Systems Research in-

creases the likelihood that research recommendations will be success-

fully adopted by client farmers. Evidence and theory presented in

this paper support this hypothesis. An empirical illustration using

Lesotho data demonstrated the feasibility of this precept.

An extensive sample of FSR literature was reviewed to determine

which stratification strategies were commonly employed. Two observa-

tions emerged. First, while socioeconomic factors were often con-

sidered when identifying representative farmers and planning research,

stratification was frequently done along agro-climatic lines. Second,

as stratification becomes more of an integral component in the research
process, socioeconomic factors have been more frequently considered.

Socioeconomic factors appearing most frequently in the literature

were identified. Variations in farmers' current cropping systems

were frequently employed as stratification criteria. These variations

were primarily differences in farmers' management practices and

production. The fanner's resource base was frequently used, with
farm size and power source most often mentioned. The institutional

climate, markets in particular, was considered in most strategies.

Household characteristics, including farmers' goals, were rarely
considered.









Various socioeconomic factors were analyzed which, hypothetically
influence farmers' decisions concerning agricultural technology.

Twelve factors which appear to be most influential for farmer strati-

fication were isolated. These were 1) market access, 2) information
source, 3) ethnic or class differences, 4) interactions among farmer

groups, 5) labor, 6) power source, 7) liquidity, 8) land, 9) farming

system interactions, 10) household goals, 11) household composition,
and 12) household orientation. In addition to influence on adoption

behavior, extent of influence on other factors, the likelihood that

the factor might be overlooked, and ease in identifying variation
were considered as well when selecting these twelve factors. Other

factors (such as access to institutional credit and extension) were

not included because the evidence on their relationship to farmers'

decisions was mixed with general conclusions being that they did not

account for significant variation in farmers' practices. Or they

were not included because they are implicitly accounted for in one or

more of the included factors. Risk factor was not included for this
reason.

Finally, a simplified procedure for socioeconomic stratification

of farmers was developed on the basis of conclusions derived from the

analysis of these twelve socioeconomic factors and from the review of

current stratification strategies. This procedure was shown to be
feasible by illustrating the stratification of lowland farmers in

Lesotho.


5.2 Conclusions

Several conclusions can be drawn from this study. First and

foremost is that socioeconomic and cultural considerations are very









important to farmers' technological decisions and must be considered

in any farmer stratification scheme. Currently, they are not suffic-

iently considered. The approaches that are the best in this regard

are those patterned after the CIMMYT strategy, particularly those

employed in Eastern Africa. It seems apparent that as stratification

increases in importance and becomes more formalized in approach, the

importance of the socioeconomic element will also increase as a basis

for stratification.

The relative importance of certain household characteristics has

been demonstrated here. They need not necessarily be considered as

stratification criteria, but they must at least be examined as possible

explanations of farming system variations. Variation itself is not

what is important; rather, the causes of that variation are what

affect farmers' reactions to potential changes. If, for example,

labor is constraining and a new method is proposed to alleviate that

constraint, the reasons for that labor constraint determine whether

or not the new method is acceptable. While all possible sources of
variation cannot be included in the stratification criteria, the

research team, together with local farmers and agricultural officers,

must decide which sources are important in any given situation. In

other words, many factors must be examined before the few, most
important ones are isolated. The complexity of the stratification

issue must be understood before a simpler procedure can be success-

fully employed.

Another conclusion drawn here, and in other recent farming

systems work, is the importance of system interactions. These include
interactions within a single farming system as well as interactions








between different farming systems and between a farming system and

the larger community and regional systems. The nature of such inter-

actions can explain why some farmers adopt a recommendation while

others do not. The interactions are more complex in traditional

farming systems, thus, an understanding of them is more important.

5.2.1 Policy Implications. Recommendation domains are discussed
in this thesis primarily as a tool for research. It is argued that

correctly identifying homogeneous farmer groups improves research
results (adoption of recommendations) and makes research more cost-

effective because of improved adoption rates and increased ability to
extrapolate research results. A recommendation domain approach may

have significant policy implications as well. Entire countries or
regions could ultimately be stratified into recommendation domains.

Scarce resources could be allocated according to stated policy goals

to reach either the poorest people, or the most people, or to achieve

self-sufficiency in a certain crop or increase the production of an

export commodity. Clear identification of the different groups of

producers make design of research strategies and implementation of

policy decisions easier.

In the same vein, clear identification of target groups, and
inclusion of all rural households into one or another groups, can

help ensure that no household is overlooked. This may force finan-

cially strapped governments to make some hard choices, but at least

no families are left out by default. Even if a distinct group com-

prises less than 10% of the households in an area and does not there-

fore qualify as its own domain, as Franzel (1981) suggests, surely

there are other similar households in other areas. Together they may




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