Front Cover
 Title Page
 Table of Contents
 I. Background: Maize in Kenya
 II. Government programs and hybrid...
 III. The pattern of adoption in...
 IV. Analysis of adoption and early...
 V. Conclusions and recommendat...
 Appendix I. Correlation coeffi...
 Appendix II. The 1974 Kenya maize...

Group Title: diffusion of hybrid maize in Western Kenya
Title: The diffusion of hybrid maize in Western Kenya
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00080060/00001
 Material Information
Title: The diffusion of hybrid maize in Western Kenya
Physical Description: vi, 57 p. : ill. ;
Language: English
Creator: Gerhart, John Deuel, 1943-
International Maize and Wheat Improvement Center
Publisher: Centro Internacional de Mejoramiento de Maíz y Trigo
Place of Publication: Mexico City
Publication Date: 1975
Subject: Hybrid corn -- Kenya   ( nal )
Corn -- Kenya   ( nal )
Corn -- Breeding   ( nal )
Genre: international intergovernmental publication   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Kenya
Bibliography: Includes bibliographical references.
Statement of Responsibility: John Gerhart; abridged by CIMMYT.
 Record Information
Bibliographic ID: UF00080060
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 03985408

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Page i
        Page ii
    Table of Contents
        Page iii
        Page iv
        Page v
        Page vi
    I. Background: Maize in Kenya
        Page 1
        Page 2
        Page 3
    II. Government programs and hybrid maize
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
    III. The pattern of adoption in 1973
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
    IV. Analysis of adoption and early adoption
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
    V. Conclusions and recommendations
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
    Appendix I. Correlation coefficients
        Page 52
    Appendix II. The 1974 Kenya maize farmers survey
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
Full Text

g t . ..........

.. .. .... ..
k M 1.
U. oh


.. .. .....

....... .... .. .

.... ... ..... ........

.... ..... W
ip v, 4.n:
.. ............

.. ...... .

22 ,

... ...... ....


........... .. a

... .... .. ........ ...
....... .. .

...... ....



M: 2.S

.. ..... ....

... ......
A, -7

....... ... .... ... .. .


I 9K
Gi, Z 'M
f .. .. ........ .
..... .. ... ..... .. ..


.... .....

..... .. ...

... ... ... .... ..... ...........

gmo w. ,I,.
........ V ... ....
m I.M

K. 4 N.
V =Oo 0 z.
.... .............

Mki. ,-Q

4 .. ......


...... .. ...

J-- N:

Abridged by CIMMYT

John Gerhart

International Maize and Wheat Improvement Center, Apartado Postal 6-641, Mexico 6, D.F. Mexico

Correct citation: Gerhart, John. 1975. The diffusion of
hybrid maize in western Kenya-Abridged by CIMMYT.
Centro Internacional de Mejoramiento de Maiz y Trigo.
Mexico City. vi +57 p.


Foreword v

Adoption of associated maize technology 27

I. Background: Maize in Kenya 1
Agriculture in Kenya's Economy 1
Maize in Kenyan Diets 1

II. Government programs and hybrid maize 4

Maize breeding and agronomic research 4
Extension and farmer training 8

Supply of agricultural inputs 9
Fertilizer distribution 10
Agricultural credit 11

Experimental credit programs 12

Pricing and marketing policies 14

III. The pattern of adoption in 1973 20

Delineating agro-climatic zone 20
Stratification of the sample 21

Sampling procedure 22
The interview 22

Pattern of hybrid seed adoption in 1973 22

Rate of adoption over time 25

IV. Analysis of adoption and early adoption 30
Multivariate model for hybrid adoption 31
Factors affecting the adoption of hybrid maize 32
Adoption and agroclimatic zone 33
Adoption and risk 36

Adoption and farm size 36
Adoption and cash crops 37
Adoption and credit availability 37
Adoption and off-farm work 38
Adoption and information variables 39

Adoption of fertilizers 40

Analysis of early adoption 41

Importance of early adoption 44

V. Conclusions and recommendations 47

Summary of main survey findings 47

Why have hybrids been successful in Kenya? 48

Appendix I. Correlation coefficients 52
Appendix II. The 1974 Kenya maize farmers survey 53


Launching the Studies

The study described in the following chapters is one of
a series aimed at enlarging understanding of the factors
impinging on the adoption of new maize and wheat
technology. Better understanding of the elements shaping
the diffusion of new cereals technology can help govern-
ments and development assistance agencies to increase
farmer income, hence the interest in the topic. Interest
increased as controversy about effects of introducing new
technologies attracted widespread attention to the theme.
CIMMYT, with its mandate defining its role in the
development and diffusion of maize and wheat technology,
quickly assumed a participant's role in the discussions. The
concern and the interest emanating from the critical im-
portance of the theme stimulated CIMMYT to look for a
modus operandi through which patterns of adoption and
the forces shaping those patterns could be identified.
Better understanding of these relationships would influence
CIMMYT efforts to develop new technology, the orientation
of its training program, and the approach taken in counsel-
ing governments about national programs.
In order to better comprehend what influences farmer
response to new technology, CIMMYT set out to facilitate
the research on which this and the other studies of the
series are based. We decided to examine eight cases in
which maize or wheat technology had been introduced to
farmers. In identifying programs for study, we limited
consideration to those in which the technology had been
available to farmers for no less than five years and in
which no less than 100,000 hectares of land might have
been affected. Eight programs were selected for study. For
maize the focus was on Colombia, El Salvador, Kenya west
of Rift Valley, and Mexico's Plan Puebla. For wheat,
programs in India, Iran, Tunisia and Turkey were consider-
ed. CIMMYT's maize and wheat staff participated in the
selection of these programs. With their knowledge of
programs around the world it was possible to choose a
varied set of experiences-e.g. programs with and without
irrigation, with and without effective price guarantees,
with massive extension effort and with virtually none.

To the extent possible, each of the adoption studies was
under the supervision of an indigenous economist. In only
one case was it necessary to turn to an expatriate and
there we had the good fortune to collaborate with a re-
searcher with several years experience in the area. Each
of the collaborators shared CIMMYT's concern for farmer
response to new technology.
Beyond sharing this concern, each collaborator had an
interest in farm level research done in close cooperation
with agricultural scientists. The importance of this interest
emerges from our conviction that agricultural scientists who
are knowledgeable about a particular maize or wheat area
can contribute substantively to research on the cereals
economy of that area. Their special knowledge about the
interaction between plants and their environments is im-
portant in identifying agro-climatic zones, critical periods
for the crop, and activities which are essential to effective
cultivation. Many agricultural scientists played a prominent
role in these studies; each warrants our gratitude for his
As the studies were completed it became apparent that
much could be said for publishing them in a standard
format. With several serving as Ph.D. dissertations and
others as less formal research pieces, a common format
could only be achieved through reworking the original
monographs. In every case but one, then, CIMMYT's
publication is an abridgement of a longer piece. The
Indian study, itself a review of the findings of several other
research efforts, is being published in its entirely with no
effort to recast it in the form of the others.
In making the abridgement we have followed certain
norms. Mathematical proofs have been eliminated, litera-
ture reviews have been included only where they relate to
points which are unique to a given study, and the discussion
of the hypotheses motivating the studies have been dropped.
This last decision arises from recognition of the substantial
commonality of these hypotheses among the studies. This
suggested that, rather than presenting essentially the same
discussion in the text of each abridgement, the hypotheses
could be treated once in an abbreviated form for all studies.
That treatment follows below.

The Hypotheses

While each of the studies examines a somewhat different
set of circumstances all depart from the same general
assumption about farmer behavior. The assumption is
that farmers are income-seeking risk averters who are
sensitive to the nuances of the environment in which they
farm and that they are generally effective in their decision
making. For the six studies based on original survey data
and to a more limited extent for the study of Plan Puebla,
this common point of departure leads to a great deal of
similarity in the motivating hypotheses.
Given a farmer oriented by the assumptions described
above, we might expect to see relationship between the
adoption of elements of the new technology and: 1) char-
acteristics of the farmer-his age, education, family size,
farming experience, off-farm work, percentage of land own-
ed; 2) characteristics of the farm-its agro-climatic region,
competition of industrial crops, relative importance of
cereals, nearness to markets, farm size; 3) characteristics of
government programs-access to credit, access to informa-
tion (through extension agent visits or visits to demonstra-
tion plots).
Some of the relationships between these variables and
the adoption of elements of the new technology are more
arguable, some less. Least arguable are hypotheses relating
adoption to education, farming experiences, percentage of
land owned, more favored climatic regions, relative import-
ance of cereals, nearness to markets, farm size, access to
credit, and access to information. With other things equal
and accepting our assumptions that farmers are income-
seeking, risk-averting, sensitive, and effective maximizers,
virtually no one would argue that any one of these relation-
ships should be negative.
Somewhat more arguable is the relation of age and family
size to adoption. Even here it is likely that only a few
would argue that these relationships might be positive.
Most arguable are the relationships linking adoption to
off-farm work and competition of industrial crops. With
respect to the former, some hold that the relationship is
positive as more off-farm work implies more income, there-
fore a greater capacity to bear risk, hence a greater willing-
ness to adopt new technologies. Others hold the converse,
arguing that more off-farm work implies less interest in the
farm, hence less willingness to put in the time and energy
associated with taking on new technologies. So too for
industrial commodities, where those who see the relation-
ship as positive allude to greater experience with improved
inputs and larger incomes while the contrary view rests on
capital restrictions and the high opportunity cost of labor.

With knowledge of the relationships among these variables,
researchers and policy makers can better develop and
diffuse new technologies. Some of the variables considered,
e.g. age and family size, are beyond the control of these
decision makers. Nonetheless, by incorporating them in the
analysis the effects of variables subject to their control are
more clearly discerned. Knowledge of how these variables,
e.g. agro-climatic zones and extension programs, relate to
adoption can be of critical importance in affecting the
development and diffusion of new technology.
With this rough sketch of the general argument, readers
wanting more detail about the derivation of the hypothe-
sized relationships can turn to the relevant original piece
from which this series of abridgements was drawn. In
all cases the studies feature the effects of agro-climatic
region and farm size on adoption of elements of new
technology. This emphasis is related to the earlier contro-
versy about the effects of new technology where these two
factors played prominent roles.
Before moving into the abridgement, some attention to
the phrase "elements of the new technology" is warranted.
Much has been made of the concept of a package of practi-
ces in the introduction of new technology. We've chosen
to look at this a bit differently, taking the view that the
differences in risk, expected income, and cost of each
element of the technology are large enough to outweigh
the effects of the interaction among these elements. That
is to say, perceptive and prudent decision makers might
well choose to take up only a part of the package rather
than the entire package. For the programs studies, the
two dominant elements in the package are improved seed
and fertilizer. These two were analyzed as dependent var-
iables for each of the studies. Of lesser importance are
such elements as seed treatment, date of planting, method
of planting, use of herbicides, use of pesticides, planting
density, and seed bed preparation. Nevertheless, where
one of these was recommended and where data are adequate,
these are also treated as dependent variables.

What Follows

This report summarizes results of an extensive study of
maize production in Kenya west of the Rift Valley. The
study, which included a 1973 survey of 360 maize farmers,
was a part of the dissertation research of John Gerhart at
Princeton University. Gerhart's dissertation is entitled
"The diffusion of hybrid maize in Western Kenya." His
work was supported by the Ford Foundation.
Appendix II was prepared by Olaf Hesselmark. The
Kenya Maize Board supported Hesselmark's work.

Don Winkelmann
El Batan


In this chapter we look at the importance of agriculture
in the Kenyan economy and the role of maize as the basic
staple in the Kenyan diet. Its central role, especially
in diets, is immediately evident.

Agriculture in Kenya's Economy

Kenya's is basically an agricultural economy. In 1972,
agricultural activities (both subsistence and monerary), ex-
cluding forests and fisheries accounted for some 34.6 per-
cent of Gross Domestic Product, valued at 211.35 million
making it by far the largest sector of the economy (Table
Agriculture accounted for about 60 percent of all exports,
either as raw or processed goods, provided about one-
third of all wage employment, and supported about three-
quarters of the total population.' Moreover, much of
Kenya's growing manufacturing sector is based on the
processing of agricultural commodities, including textiles,
woolens, canned foods, pyrethrum, beef products, sugar
refining, milling, coffee and tea. In addition, the produc-
tion and distribution of agricultural inputs and services
grew from 11 million to 26 million between 1965 and
1973.2 Between 1964 and 1973, agriculture showed an
average real growth rate of 6.0 percent in the monetary
sector and 3.6 percent in the subsistence sector. In 1972
and 1973, however, agriculture grew considerably faster,
with marketed production up 24 percent and 17 percent
respectively. Although the share of agriculture in the total
national product is expected to decline from 35 percent to
29 percent of GDP over the next Plan period, it is certain
to remain the most important sector of the economy for
the foreseeable future.
The continuing importance of the agricultural sector
is reinforced by Kenya's rapid population growth, estimat-
ed at 3.5 percent or higher per year. According to the I LO
employment study of 1972, the number of rural house-
holds in Kenya will increase from 1.7 million to 2.8
million between 1969 and 1985.3 Given continued mi-
gration to the urban areas and increasing demand generat-
ed by rising incomes, food demand will grow by more than

5 percent per year in the rural sector and more than 10
percent per year in the urban sector. Using income elastici-
ties calculated by the Food and Agricultural Organization
of the United Nations and assuming growth rates of 3.5
percent for population and 6 percent for income, the ILO
report projected average annual rates of growth of around
4 percent for maize, millet, sorghum, and other rough
grains; between 4 and 7 percent for fruit; milk, pork, rice,
vegetables, and wheat; and over 7 percent for beef, eggs,
mutton, poultry, sugar, and edible oils. Because of already
existing land shortages in the rainfed areas, these increases
in production will have to come primarily through increased
productivity per unit of land. Only about 7 percent of
Kenya's total land area is classified as good agricultural
land, with adequate, reliable rainfall and good soils and
drainage. Irrigation potential is limited and extremely
expensive, with development costs in the range of 700
to 1000 per hectare, rising to more than 2000 in more
difficult areas.4 Fortunately, however, a considerable
potential for intensification of agriculture exists; in the
case of almost every crop, average yields are less than
half those obtained on the best small farms, husbandry is
poor, planting material can be further improved, and use
of commercial inputs is at extremely low levels.

Maize in Kenyan Diets

Maize is the most important crop in Kenya's agricultural
economy, whether considered in terms of its value, its
role as a basic dietary staple. Table 2 shows the gross
marketed production of Kenya's major agricultural com-
modities from 1967 to 1972. In spite of fluctuations in
prices and weather, substantial increases have been achiev-
ed in the production of temporary crops, most permanent
crops (except sisal and wattle), and livestock products.
The drought in 1970 and 1971 disguised the rapid increase
in maize production deriving from the increase in hybrid
maize usage, the results of which began to be apparent in
1972 and yielded large surpluses in 1973. Also disguised in
these figures is the degree to which different commodities
enter the market economy. It is estimated that only 5 to

Table 1. Kenya's Gross Domestic Product, 1964 and 1972, value and
sectoral shares (Kenya million in constant (1964) prices).

A. Outside Monetary Economy
Total Non-Monetary Segment
B. Monetary Economy
Mining and Quarrying
Manufacturing and Repair
Building and Construction
Electricity and Water
Transport and Comminications
Wholesale and Retail Trade
Banking, Insurance
Ownership of Dwellings
Other services
Domestic services
General Government
Public Administration
Agricultural Services
Other services
Total Monetary Segment
Total Gross Domestic Product

Value Share (%)







Value Share (%)







Source: Economic Survey, 1973, Tables 1.1 and 1.4

15 percent of the maize crop, for example, is marketed
through official channels, depending on yearly surpluses,
leftover stocks on the farm, and local price differentials.
In 1969, for example, if all the maize produced had been
marketed at the then producer price (28 shillings per 200
pound bag), the maize crop would have been worth an
estimated 21 million.5 In 1972 it might have been
worth 50 million. There are well-recognized problems
in estimating the value of subsistence output.6 Nevertheless,
local prices often greatly exceed the official market price.
Whether one uses world prices, f.o.b. Mombasa export
prices, c.i.f. Nairobi import prices, local prices, or the
official price for a given year, maize still ranks as the most
important commodity in value terms.
Perhaps more important, however, is the role of maize
as the basic staple in the Kenyan diet. Table 3 shows the
proportionate areas in the small farm sector planted to
major food crops in 1960 and 1970. In 1970, an estimated
1.1 million hectares were planted to maize, more than
three times the area planted to the next largest food crop
(beans), and about eight times as much as the next most
important grain (sorghum). The dietary importance of
maize is also indicated by what little nutritional data
exist. A 1958 survey of 349 families in Nairobi showed

maize accounting for 80 percent of starchy-staple calories,
the second highest percentage (after Lusaka) of any African
city.7 A WHO-FAO-UNICEF survey in Kenya, although it
was able to sample only seven of its intended 28 locations
in the four years it had available, found maize to be the
major cereal by far in all but one of the seven locations.8
Bruce Jones, analyzing data from the 1969-70 Nairobi Urban
Survey, found that maize was by far the cheapest source of
calories and was second only to cow peas as a source of
inexpensive protein:

Maize flour
Cow peas
Wheat flour


Grams Protein/Shilling

Although the average family spent only about 11 percent of
its food budget on maize (and the lowest income bracket
14 percent), maize accounted for as much as half the
calories consumed.9
The importance of maize established, we turn now to a
discussion of several government programs focused on maize
and on the diffusion of new maize technology.

Table 2. Gross marketed production by commodity, 1968-1973 (Kenya thousands)

1968 1969 1970 1971 1972 1973'
Wheat 6,635 6,583 4,994 5,206 4,160 3,902
Maize 5,405 3,861 2,828 4,276 7,252 8,571
Barley 242 333 392 437 477 975
Rice 429 577 724 725 859 906
Other cereals 218 223 62 102 205 171
Total 12,940 11,577 9,000 10,746 12,953 14,525
Temporary Industrial Crops
Castor and other oil seeds 569 503 556 400 272 227
Pineapples 140 210 242 295 326 412
Pyrethrum 2,622 1,317 1,477 2,523 3,662 3,038
Sugar cane 2,179 2,942 3,509 3,457 3,038 4,453
Tobacco 25 26 35 28 31 30
Cotton 700 834 695 878 980 983
Total 6,305 5,832 6,514 7,481 8,309 9.143
Other Temporary Crops
Pulses 490 428 236 303 753 436
Potatoes 503 698 1,263 1,652 1,723 1,551
Other temporary crops 244 1,657 1,247 1,642 1,773 1,596
Total 1,237 2,783 2,746 3,497 4,249 3,583
Permanent Crops
Coffee 12,266 16,163 21,814 18,922 24,165 32,396
Sisal 2,193 2,250 1,715 1,519 1,862 7,051
Tea 9,335 11,159 13,838 11,803 16,034 16,766
Coconut products 490 484 520 545 572 515
Wattle 433 464 420 423 530 468
Cashew nuts 422 423 1,186 944 638 862
Fruit and other permanent crops 670 702 745 1,025 1,080 972
Total 25,810 31,645 40,238 35,181 44,881 59,030
Total crops 46,291 51,837 58,498 56,905 70,244 86,281
Livestock and Products
Cattle and calves 11,689 12,218 13,324 13,330 16,510 17,132
Sheep, goats and lambs 440 453 475 733 825 825
Pigs 890 614 750 593 631 651
Poultry and Eggs 358 320 998 1,032 1,207 1,207
Wool 533 560 346 220 205 503
Hides and skins 657 675 604 841 1,170 1,205
Dairy products 7,126 6,100 6,806 9,300 10,890 11,319
Total 21,673 20,940 23,303 26,049 31,438 32,842
Unrecorded Marketed Production 3,342 3,425 3,595 3,741 4,100 4,298
GROSS FARM REVENUE 71,306 76,202 85,396 86,695 105,931 123,421

*Provisional. Source: Economic Survey, 1974, p. 54.

Table 3. Percentage of area planted to ten major food
crops grown in the small farm sector, 1960 and 1970.

1960 1970
Crop % of Total % of Total
Maize 44.0 51.4
Pulses 25.7 25.8
Sorghum 7.3 6.8
Millet 5.8 2.9
Cassava 4.4 4.2
Finger millet 4.2 1.8
Bananas 2.7 3.7
English potatoes 2.0 1.5
Sweet potatoes 2.5 1.3
Yams 1.1 0.3

Sources: Kenya Statistical Digest, 1966 (1960 figures); Economic
Survey, 1973 (1970 figures). Note: Due to differences in samp-
ling and classification, small differences (as among potatoes, etc.)
may not be significant.


1. Republic of Kenya, Development Plan, 1970-74, p. 91.

2. Republic of Kenya, Economic Survey, 1974, p. 58.

3. International Labour Office, Employment, In comes, and
Equality (Geneva, 1972), p. 151.

4. International Labour Office, Employment, Incomes and,
Equality (Geneva, 1972), p. 412. No maize is grown under
irrigation in Kenya.

5. Development Plan, p. 237.

6. See, for example, Judith Heyer, "some Problems in the
Valuation of Subsistence Output," Discussion Paper, University
of Nairobi, August, 1965.

7. Marvin P. Miracle, Maize in Tropical Africa (Madison, 1966),
p. 113.

8. M. Bohdal, N.E. Gibbs, W.K. Simmons, Nutrition Survey
and Campaign Against Malnutrition in Kenya, 1964-68, Nairobi
(no date), p. 128.

9. Bruce N. Jones, personal communication, Nairobi, August


In this chapter we consider government programs in research,
extension, farmer training, credit, pricing, marketing, and
input policies in the context of their contribution to the
diffusion of hybrid maize. Along the way, some special
efforts to reach small farmers with new maize technology
are discussed and some conclusions drawn about their
possible impact on the diffusion process.

Maize Breeding and Agronomic Research

Although some maize research trials took place as far back
as the 1920s, these were carried on by isolated research
workers and individual farmers in different locations and
had little, if any, coordinated impact other than as selection
work for yield improvement and disease resistance. "All
a district agricultural officer seeking a sound solution to a
problem had to do," Brown wrote of this period,' "was to
go and look at what some good farmer was doing and there
he had a practical demonstration of what could be done."
Not until 1955, when the first research officer was appoint-
ed to work exclusively on maize at Kitale, did a systematic
program for maize breeding get under way. In 1957 a
program to develop early maturing maize types was started
at Katumani, near Machakos in Eastern Province, and a
program for developing medium maturity varieties was
commenced at Embu in 1965. The details of these pro-
grams have been described by Harrison, Dowker, Eberhart,
and Oagad, and have been briefly summarized by Smith.2
Using conventional breeding methods, inbred lines were
developed from the well-adapted Kenya Flat White com-
plex which had been selected by local farmers over the

proceeding years. The inbred lines were formed into
synthetic type varieties, the first of which, Kitale Synthe-
tic II, was released in 1961. This became very popular and
was grown on about half the large-scale farms in the
Trans-Nzoia area by 1964, when the first conventional
hybrids were commercially released. Because a synthetic
is formed from a large number of inbred lines, it has
considerable genetic variability and the farmer does not
have to buy new seed every year. Hybrids, on the other
hand, are developed from only three or four inbred lines,
and, although they show hybrid vigor in the first genera-
tion cross, yields drop off by perhaps 20 percent in the
second year. Farmers therefore have to buy seed fresh
each season in order to get continuing high yields. For this
reason, the original intention was to develop a joint program
of synthetic and hybrid production in which the synthetics
would be mainly intended for small-scale subsistence farm-
ers and hybrids for large-scale commercial farmers. Accord-
ing to Harrison, "it was felt that small-scale farmers would
not pay the high price needed each year for hybrid seed
and too many would plant second-generation seed.... How-
ever, the impact of the higher-yielding hybrids was so great
(about 30 percent higher than the synthetics) that soon
after their release it became impossible to sell synthetics to
either small-scale or large-scale farmers, and improved syn-
thetics are now used only in [preparing hybrids]"3
The fact that the Kenya Flat White complex, because
of its similarity of origin, had a rather narrow genetic
base presented a complication. The possibility of extract-
ing further inbred lines for continued hybrid production
seemed to be limited. For this reason, a large collection
of genotypes were imported in 1959 from Mexican and

Colombian germ-plasm banks to widen the genetic base of
the Kenyan breeding program. Though nothing significant
had come from earlier importations from temperate coun-
tries (U.S., Europe, South Africa, Australia),-the Latin
American material came from areas with similar ecological
conditions to Kenya, close to the equator and with a wide
range of altitudes. Moreover, material from the "original"
or center-of-origin stock tends to have greater genetic
diversity than varieties previously removed and adapted
elsewhere. After careful pre-selection for late-maturing,
high altitude, disease resistant types, 124 test crosses were
grown in 1961 in a yield trial. Outstanding among the
crosses was one between Kitale Synthetic II and Ecuador
573, a long-eared, high-altitude variety with tall stalks,
and late maturity. This varietal hybrid was released as
Hybrid 611 in 1964, the same year as the conventional
hybrids, since no inbreeding had been required. Although
the Ecuador 573 itself had yielded only 75 percent of the
Kitale Synthetic II, the two varieties were sufficiently
diverse to give excellent heterosis, and the first cross yield-
ed 40 percent more than the Kitale Synthetic II. By
improving both sides of the parent populations through
recurrent selection while keeping the genetic variability
within each, continuing improvements in yield have been
made.4 The breakthrough in yields, however, had already
been made by 1964, at least for the high altitude areas, and
within only a few years the reputation of "hybrid" was
firmly established.
How did the new technology compare with the old
technology on farmer's fields? Unfortunately, very few
reliable data exist on actual on-farm yields in different
localities, since measurement is costly and difficult, farm-
ers generally don't know how large their fields are, some
maize is picked green before harvest time, and husbandry
practices differ considerably. Simply asking the farmer
will give poor (generally high) results. Using research
station results is often misleading since conditions are
controlled. Even non-treated "control" experiments often
given erroneous results due to residues of fertilizer left
in the soil from previous years, insect control on neighbor-
ing plots, or merely from the advantageous location of
most research stations. As we shall see below, even a
deliberate attempt to approximate "poor" husbandry in
the district husbandry trials yielded almost 50 percent more
than the estimated average on-farm yield.
In the district variety trials, the effect of hybrid seed
(assuming good husbandry in all cases) was found to be
an increase of 30 to 80 percent depending mainly on altitLu
de. (Chapter III, Figure 6) The effect of using the total
package of recommended practices as compared with none
of the recommended practices was an increase in yield of
307 percent and an estimated increase in profitability of
532/- shillings per acre at a price of 25/- per 200 pounds
bag. Most farmers, of course, would fall somewhere be-
tween these estimates, using some of the recommended
practices and not others. A 1971 study of 153 farms in a
high potential area, for example, found an average yield

of 18.3 bags for 137 hybrid users and 12.6 bags for 16
non-hybrid users, although many of the hybrid users were
specially selected participants in a government program
and probably had higher levels of husbandry and input
usage than the average farmer.5 Nevertheless, a 50 per-
cent increase is probably not an unreasonable average in
that area. A follow-up on 36 farmers from the present
study had too few non-hybrid users to give a compara-
tive result, but found larger differences in yields due to
different husbandry practices.6 A comparison of hybrid
yields on demonstration plots situated in farmer's fields
(see Chapter III) showed substantial differences between
hybrid yields in different agroclimatic zones, but again had
no information on non-hybrid use. We can conclude that,
although it is the combined package of practices that pro-
duces the most dramatic results, the use of hybrid seed
alone will raise yields substantially, probably by as much
as 50 percent under good conditions.
Plant breeding, of course, is only part of the research
story. An active program of agronomic research has also
been essential to determine proper fertilizer application,
spacing, time of planting, and other recommendations.
Although agronomic experiments go back as far as 1910,
it is only since 1963 that A.Y. Allan, supported by funds
from the Rockefeller Foundation and the British govern-
ment, has developed a systematic program of agronomic
research, closely tied to the breeding activities at Kitale.
Prior to that time, the effects of husbandry, fertilizer
application, and plant population had been examined singly
or at best in pairs. Allan set up a series of district-maize-
variety and district-cultural trials using a 33 factorial design
with different levels of nitrogen, phosphate and plant
population from which annual recommendations for farm-
ers were prepared.7 At that time, the average yield of
maize in Kenya was estimated to be about five to six
bags (1000 to 1200 pounds) per acre. In the district
variety trials, however, the unimproved local maize used
in the control plots (with carefully supervised husbandry)
yielded an average of 4000 pounds (20 bags) over the
1964-66 period. Clearly hybrids alone were not required
to increase yields.
To test the effects of poor husbandry, Allan established
a series of 26 factorial trials in which each of the six
factors (time of planting plant population, type of seed,
standard of weeding, and use of nitrogen and phosphate)
were deliberately set at a "high" and a "low" level. The
results of these experiments are shown in Table 4. Time
of planting and type of seed were the most important
factors in determining yield, followed by standard of
weeding and plant population. Fertilizer application was
not important, and, in the absence of good husbandry
practices, was actually unprofitable. It is interesting that
even the "low" level of the practices applied on the re-
search plots yielded 1760 pounds per acre, considerably
above the estimated national average of 1200 pounds. The
results were illustrated clearly in a "maize diamond" (Fi-
gure 1) in which physical inputs (seed, nitrogen, phosphate)

Table 4. Effects of husbandry and input use on maize yields.

Yields Added Return Added Cost
Factor Treatment Ibs/Acre Shillings/Acre Shillings/Acre2
Time of planting Start of rains 5200 270 Very little
4 weeks later 3040
Plants per acre 16,000 4580 115 8
8,000 3770
Type of seed Hybrid 4860 175 12
Local 3380
Amount of weeding Three times, early 4640 130 20
Once, late 3600
Phosphate per acre 50 Ib. 4160 10 32
None 4080
Nitrogen per acre 70 Ib. 4380 65 72
None 3860

1/ At 1966 price of 25/- per 200 pound bag. 2/ Based on costs of inputs required and
estimated labor costs. Source: A.Y. Allan, "District Husbandry Trials in Western
Kenya, 1966 and 1967." Quoted in M.N. Harrison, "Maize Improvement in East
Africa" in C.L.A. Leakey, Crop Improvement in East Africa, 1970. p. 45.

and poor husbandry are compared with good husbandry
(early planting, weeding, proper spacing). The physical
inputs alone produced a 66 percent increase over the
original average of all practices taken at a low level, while
the good husbandry practices produced a 148 percent
increase. All six practices taken at a high level produced a
307 percent increase. The implication is clear that it is
dangerous to recommend the use of expensive fertilizer
in the absence of high levels of husbandry. A "second-best"
formula is implied, in which improved seed genotype is
combined with improved husbandry practices for a low-
cost, high-return solution. This is what most farmers in
western Kenya have, in fact, discovered for themselves
(see Chapter III) although the extension service has con-
tinued to recommend the full package of practices.
Probably the most important of the agronomic find-
ings was the importance of early planting. Several explana-
tions of this have been advanced, including the onset of
fungal leaf infections, soil nitrogen fluctuations, and the
moisture requirements of maize. Allan concluded that the
effect of late planting was due primarily to poor root
aeration in the early stages of growth. But simply, maize
needs a small amount of moisture in its early stages of
growth, when the growing point is still below the surface
of the ground, and then considerable moisture in the later
stages when it is tasselling and filling out the cob. Plant-
ing after the rains have started means that the plant starts
off in very wet, cold, poorly aerated soil, and then may
suffer later from water shortages when the rains are tail-
ing off. The problem with this fact is that many, particular-
ly small-scale, farmers have difficulty in planting early due
to the hardness of the soil, delays in getting seed and
fertilizer, and a fear that the rains will be late and the dry-
planted seed will be wasted. As we see in Chapter III,
planting before the rains had the lowest adoption level of
any cultural practice in all three agroclimatic zones of the

study. The costs of late planting, however, are enormous.
In nine different experiments between 1966 and 1968,
Allan found that each day's delay in planting after the
rains started resulted in a loss in yield of an average of
75.9 pounds per acre per day (or 7.59 bags for a 20 day
delay after the start of the rains). E.R. Watts found in
Embu District of Eastern Province that the loss was as
great as 6.2 bags per acre per week, yet only 58 percent of
"progressive" farmers and 26 percent of a control group
had planted before the rains due to the hardness of the
soil and a lack of plows.8
The possibilities for extended agronomic and breeding
research are, of course, practically limitless. DeWilde,
writing in 1966, complained that, although suitable recom-
mendations existed for the highland mixed-farming areas,
additional field traisl under "widely varying ecological
conditions" were needed to determine a package of im-
provements for less favored areas.9 Recognizing this need,
the Kenya Maize Research Section recently began, with
British assistance, an expanded agronomic research program
in key maize growing districts. Belshaw and Hall, in an
excellent discussion of research problems in Africa, recom-
mended that more emphasis be placed on the economics of
small-farm operations, including particularly analysis of labor
constraints and of the reasons why small farmers persist in
such practices as inter-cropping (planting mixed stands).' 0
Without implying that the initial success in breeding
high yielding hybrids has solved all the research problems
for hybrid maize, it is nevertheless useful to ask what in
the Kenyan situation produced a breakthrough at all. What,
if anything, can be generalized from the Kitale research
experience? Several things are apparent. First, demand
for the research product was immediately present. Kitale
is centered in the middle of the biggest maize growing
area in Africa and the maize industry pervades the town
and the research station. This is in direct contrast to

many places in Africa where research stations are placed in
cool hilltop conditions removed from the growers who
are expected to benefit from the results. People in Kitale
literally eat, drink, and talk maize. Secondly, the Kenyan
case was patently not an attempt to transplant technology
unalloyed from a developed, tropical environment. Fifty
years of local adaptation and selection had taken place to
produce Kenya Flat White, an appropriate parent stock.
Thirdly, the importation of additional genetic material,
when it came, was done systematically and carefully, taking
advantage of a world-wide network of scientists who were
able to preselect for Kenyan conditions from an analogous,
but diverse, Latin American population and to use that
imported material in a way that would produce results in
the quickest and most productive way. In other words, a
thorough knowledge of local conditions and extensive local
research and breeding was a prerequisite to effective utili-
zation of international research work. Fourth, the quality

and continuity of the Kenyan maize program staff is probab-
ly unmatched in any national research program in Africa.
From its inception in 1955 until 1973, the maize breeding
program had only two directors, Michael Harrison and
Festus Ogada, who themselves had three or four years of
overlapping experience at Kitale. Their chief agronomist,
A.Y. Allan, had more than a dozen years experience in
Kenya. The prograqp was supported by three American
scientists coming out of the same (Iowa State) program
over a period of nine years, each familiar with the others'
methods and experience. In a governmental bureaucracy
where two years is considered a long posting, this continui-
ty in leadership (which has also largely been matched in
supporting staff) is most remarkable. Finally, and perhaps
most importantly, the Kenya Maize Research Program has
never been isolated from the other components of a na-
tional maize program. It has worked closely with seed
production staff, government extension personnel, and

Fig. 1. "Maize Diamonds" in Kenya: Husbandry, physical inputs, and yield. Source:
1966 District Husbandry Trials, A.Y. Allan, "Maize Diamonds", Kenya Farmer, January,

A. Bad husbandry, local maize seed, no fertilizer

8.8 bags/acre

B. Bad husbandry,
hybrid seed,

14.6 bags acre
(66% increase)

C. Good husbandry,
local seed,
no fertilizer.

21.8 bags/acre
(148% increase)

35.8 bags/acre
(307% increase)

D. Good husbandry, hybrid seed, fertilizer.

farmers themselves. As we look further at the other com-
ponents of the system, we can see how important this
constant communication and feedback have become.

Extension and Farmer Training

The literature and practice of agricultural extension efforts
is, of course, enormous.'1 In Kenya, government interest
in bringing new information to African farmers goes back
at least to 1911, when, according to Fearn, "instruction
was given in the use of the wheel-plough to sons of chiefs
and others at Kibos, some four or five miles from Kisumu."'2
Unfortunately, the chiefly bias reflected in that early ex-
tension effort has persisted to the present day; extension
contacts and farmer training are disproportionately dis-
tributed toward the larger and wealthier farmers. A study
of farmers in one part of Central Province found that all
of the most "progressive" 25 percent of the farmers had
been visited by an extension agent during the previous
year as opposed to 41 percent of "the least progressive"
18 percent of the farmers.13 Leonard, in a detailed study
of extension agents and clients in Western Province (part
of the area covered in the present study), found that 57
percent of all extension visits were paid to "progressive"
farmers (who constituted10 percent of all farmers) and
only 6 percent of visits were paid to "non-innovative"
farmers (who constituted 47 percent of all farmers).14
The present study found that 56.5 percent of the farmers
over 50 acres in size had received extension visits in the
past year, as opposed to 35 percent of all farms under 50
acres and only 27 percent of farms in the lakeside districts.
Attendance at Farmers Training Center courses was even
more highly skewed, with 48.6 percent of the largest farm
owners having attended and only 8 percent of the farmers
in the lakeside area. Although these percentages may be
quite high by African standards, and one might even
argue from the congruence of visits and progressiveness
that the extension service is demonstrably effective, never-
theless the fact remains that the extension services favor
the more progressive, and hence, in most cases, the weal-
thier farmers. This obviously has important implications
for the pattern of diffusion of new technology.
Concentrating extension efforts on wealthier, more pro-
gressive farmers has been justified on the grounds that,
since it is impossible to reach the majority of farmers with
a limited extension network, a larger proportion of total
farm area can be influenced by concentrating on larger
farmers, who would also serve as models for the rest of
the farm community to emulate. The argument has been
supported to some extent by the attention paid by acade-
mic investigators to "innovators" or "progressive farmers"
who tend, for a variety of reasons, to be better educated
and to have larger farms.'5 Recently, however, a healthy
corrective view has emerged, in Kenya at least, which puts
emphasis on reaching those farmers who have been left
behind by changes in agricultural development. In this

view, attention should be devoted to those farmers who
have not adopted new farm practices and more emphasis
should be placed on group and mass media extension
techniques.' 6
The case of hybrid maize provides an excellent test
of the extension service at its best since, unlike many
cases where the extension service has little or nothing to
offer the farmers, in this case the service had a proven
and profitable package of recommendations. As of 1972,
the agricultural field services had some 2600 agricultural
and animal health assistants (AAs and AHAs) with certifica-
te level (post-secondary) training and some 5500 junior
agricultural and animal health assistants (JAAs and JAHAs)
with only on-the-job training of a limited nature. This
represented about one trained extension agent per 700
farms (one to 138 on settlement schemes), although far
from all the trained staff were available for actual extension
work, many being occupied with administrative and statis-
tical duties. In addition, there were a total of 32 district
farmer training centers, although, according to the ILO
report, only 27 of these were operating even at modest
levels and these suffered from "chaotic" financing, inex-
perienced staff and frequent changes in leadership (12 of 27
principals changing, for example, in 1971).' 7 The quality
of the field staff performance is also questionable. Leonard,
for example, found that out of a possible high of 19
points on an index of knowledge of hybrid maize recom-
mendations, the average score of 269 extension agents
was only 13.7. Whereas 91 percent of the agents knew
what fertilizers to recommend, only 59 percent knew how
much to apply. Moreover, agents visited an average of
only 7.2 farmers a week, which accounted for 35 percent
of their work time, 13 percent devoted to demonstrations
and 29 percent to administrative and statistical chores.'
Although these figures are from only one in-depth study,
they correspond rather closely to a Tanzanian study that
found only 28 percent of extension agents' time was spent
visiting farmers and 14 percent in demonstrations and
group activities.19 Although the staffing levels, by African
standards, are not bad, the frequency, distribution, and
content of extension contact with farmers leave much to
be desired. Incidentally, there is in Kenya an Agricultural
Information Centre, started in 1966, which produces post-
ers, handouts, films, and radio programs for mass dissemina-
tion. At the time of a 1971 visits, however, the centre
lacked adequate staff or funding to do its work well. By
far the most useful of all mass media methods are the
printed leaflets, in English and Swahili, enclosed in every
package of seed sold by the Kenya Seed Company.
In spite of these deficiencies, however, one must con-
clude that the extension service did play an important part
in putting across the basic message about hybrid maize.
In 1965 over 5000 demonstration plots were organized in
the country, mostly in the western portion. In recent
years with FAO support, the extension service has conduct-
ed hundreds of fertilizer demonstrations each year in all
the maize growing districts of western Kenya. In the present

survey, 25 percent of farmers claimed to have attended one
personally, and such attendance was positively related to
adoption and early adoption (especially in the high density
districts where most of the demonstrations were held).
Moreover, some 35.4 percent of 11 farmers said that they
first heard of hybrid from extension agents (Table 5) and
64 percent said that they would go to an extension agent
for advice if they had a question about their maize. (In
two earlier surveys, 42 percent of farmers in Vihiga Divi-
sion of Kakamega District reported an extension visit;
37 percent, demonstration attendance; and 7 percent,
attendance at an FTC course. In Kisii, 89 percent reported
an extension visit and 40 percent had seen demonstration
plots.20) These findings indicate that, whatever its short-
comings, the extension service has played an important
part in the diffusion of hybrid maize, with the result that
by the time of the present survey, only 9 of 360 farmers
(2.5 percent) could actually say they had never heard of
hybrid maize.

Supply of Agricultural Inputs

Hybrid seed production in Kenya is carried out by the
Kenya Seed Company, a locally-owned commercial enter-
prise in which the government, through the Agricultural
Development Corporation, is a major share-holder. The
commercial aggressiveness and efficiency of this enter-
prise is one of the major reasons for the Kenyan hybrid
maize success story. The KSC was started in 1956 by
a group of European farmers in the Kitale area to produce
improved varieties of grass seed which had been selected
from indigenous Kenyan grasses at the National Agricultural
Research Station at Kitale. The founder of the company,
Mr. W, Heilbuth, is still chairman of the company's board
of directors. In the early 1960s, with Jomo Kenyatta's
release from detention and the prospect of Independence
on the horizon, European farmers stopped investing in
improved pastures, and the company began to incur losses.
Then, in 1962, a Dutchman, W.H. Verburght, was re-
cruited as manager just at the time that the chief maize

breeder at Kitale, Michael Harrison, was looking for some-
one to produce hybrid seed. Up to that time it was wide-
ly thought in Kenya that small farmers would not buy
hybrid seed every year. Verburght, however, saw that with
a declining number of European farmers, selling seed to
small farmers was the company's best hope for success,
and with Harrison's prodding and E.J.R. Hazelden, an
aggressive former research officer, as sales manager, the
KSC began pushing seed to anyone who would buy it.
Since the Kenya Farmers Association at that time was
oriented almost exclusively toward European farmers, the
KSC got the West Kenya Marketing Board and Dalgety, a
trading company engaged in buying hides and skins in
African areas, to handle the seed, as well as the Bungoma
Farmer's Cooperative Society. The KSC then began ap-
pointing stockists, small-scale African storekeepers select-
ed for their location, reputation, and interest, who alone
would be allowed to buy seed at wholesale prices from
the KFA or Dalgety's outlets. This network spread until,
by 1972, the KSC had over a thousand registered stockists.
Their distribution is shown in Map 1 in relation to popula-
tion. The commercial maize-growing districts (Trans Nzoia,
Bungoma, and Uasin Gishu) have less than 3000 persons
per stockist, whereas the densely populated high potential
districts (Kakamega, Kericho, and Kisii) together with
Nandi, have between 3000 and 5000, and the lakeside
districts (Busia, Siaya, Kisumu, and South Nyanza) have
10,000 or more persons per outlet. As the KFA and Dalge-
ty gradually took over this promotion work on their own,
the KSC reduced its staff to the present three salesmen
(and the KFA increased to sixteen).
Seed is sold nationwide at a fixed price of 20 shillings
for a 20 pound bag, enough to plant one acre. The whole-
sale price is 17.75 shillings and the total markup is divided
so that each stage in the distribution chain, from factory to
wholesaler to stockist, receives a larger share, and con-
sequently, an incentive to push the product. This approach,
modelled, according to Verburght, on the marketing of
Wilkinson razor blades, has made "every stockist an ex-
tension agent." After an initial bad experience with credit,
thw KFA now sells mainly for cash, although some stockists

Table 5. "From whom did you first hear about hybrid
maize? "(All answers in percentages)

Source Zone 1 Zone 2 Zone 3 Zone 4 Average
Extension agent 30.5 33.3 43.2 40.2 35.4
Dealer/stockist 15.8 6.7 4.5 4.1 9.9
Friend/neighbor 42.1 50.0 44.3 18.0 44.7
Employer 3.2 0 1.1 2.7 1.7
show/field day 0 0 1.1 4.1 0.4
Newspaper 2.1 1.1 0 5.5 1.2
Can't recall, other 6.3 8.9 5.7 11.1 6.7
100.0 100.0 100.0 100.0 100.0

pass on credit at their own responsibility to local custom-
A recent parliamentary report, complained that stockists
often ran shott of inputs because they could not afford
to carry inventories on a cash basis and recommended that
a credit scheme for stockists and a subsidy incentive be
introduced to encourage stockists to have inputs on hand
in good time for planting.22 The committee also recom-
mended subsidizing the price of seed and fertilizers to
lower their price to the farmer. Seed company officials
generally are opposed to a subsidy on the grounds that
farmers will use greater care in planting seed they have
paid for and seed is a small cost relative to the value of
output per acre. What seems more important, however,
is guaranteeing that stockists and wholesalers have ade-
quate profit incentives to insure that stocks are available in
time for planting in the rural areas. For this reason,
reducing the price of seed should be viewed with some
Hybrid seed itself is produced by some 80 to 100
registered farmers in the Kitale area under close super-
vision (including weekly field visits to inspect weeding.
detasseling, etc.) by the Kenya Seed Company and a go-
vernment seed inspection unit. Farmers in 1973 re-
ceived a price of 100/- for a 100 kg. bag of clean seed
maize, which -at an average yield of 18 bags per acre-
gives an adjusted gross margin of 320/- per acre for seed
maize which compares with 213/- per acre for commercial
maize at an equivalent yield and perhaps 190/- per acre for
wheat, a competitive crop in some portions of the maize
growing area.23 Although seed growers complained that
this was a lower profit margin than they had had in the
past, given the trouble of growing seed maize,24 a more
serious long-term threat to seed maize growing is the con-
tinued break-up of large farms by land"companies" for
squatter settlement. To ensure the isolation from other
maize needed to grow seed correctly, a larger share of
seed growing may have to be taken over by state farms in
the area.
After harvesting, the seed is inspected, fumigated, and
cleaned, graded and treated with a fungicide/insecticide

Table 6. Imports of manufactured fertilizer into Kenya.

Year (Metric Tons) Value 000
1963 38,621 n.a.
1964 55,364 n.a.
1965 70,498 n.a.
1966 71,279 n.a.
1967 81,476 1601
1968 84,701 1875
1969 105,413 2272
1970 141,215 3041

Source: Annual Trade Report, E.A. Community, Economic
Survey, 1973.

dressing before being packed. The dressing includes a
bright blue-green coloring which only the company is all-
owed to import, to prevent counterfeiting by stockists.
Although the KSC has in some years run short of seed due
to the rapid uptake of hybrids by small farmers, it now
tries to carry over from one year to the next 20-50 percent
of yearly sales to safeguard against sharply increased de-
mand or a poor seed production season. In spite of these
difficulties, however, the widespread sale of high quality
seed at 20/- per bag is one of the strongest links in the
hybrid maize chain in Kenya. The situation compares
very favorably with most Latin American and Asian coun-
tries, where government-run seed agencies both grow and
inspect their own seed.25 Although the Kenyan govern-
ment has a share in the ownership of the Kenya Seed
Company, it has been wise enough to leave the produc-
tion and distribution of inputs to commercial operations,
while itself conducting a separate seed inspection unit.
This production, distribution and inspection system may
be the major reason why Kenya, almost alone among the
poor countries of the world, has been successful with
hybrids as opposed to synthetic maize varieties.

Fertilizer Distribution

No basic fertilizers are manufactured in Kenya, although
several companies maintain storage, blending, and bagging
factories designed to package a wide range of fertilizer
mixes with a comparatively small tonnage of fertilizers
each.26 Some phosphates are imported from Uganda but
most fertilizers are imported directly from overseas. Ferti-
lizer imports for 1963-70 are shown in Table 6.
Fertilizers face the same distribution problems as hybrid
maize seed, only more so. They 'o- more expensive, bulk-
iers, more specialized in use, anc are used more exclusively
by the large farm sector. Demand for the products is limit-
ed to a relatively short period during the year, just before
planting, and yet timely availability is all-important to
sales. Stockists, however, have neither the capital, the
storage space, or the transport to keep and move large
quantities easily. The result of these problems is that
fertilizer distribution has developed more slowly and less
broadly than seed distribution, although essentially using
the same channels, namely, wholesale firms such as the
Kenya Farmers Association and Dalgety, cooperative socie-
ties in those districts (mostly in Central Kenya) where they
are strong, and a broad system of selected and registered
stockists.27 In spite of the generally low level of fertilizer
use in some areas (only 4 percent of farmers, for example,
in Zone 3 of the survey area), overall usage is increasing
rapidly as shown in Table 6. This increase has been helped
by the FAO-supported fertilizer demonstrations which show
average benefit/cost ratios for both phosphates and phos-
phates plus nitrogen of 2.4, 3.2, and 4.0 for 1969, 1970,
and 1971 respectively as maize prices were increased from
25/- to 30/- to 35/- per bag while fertilizer costs held

constant. The combined effect of the demonstrations
and the increasing profitability of fertilizers obviously ac-
counts for much of the increase in small farm use. (About
40 percent of total fertilizer imports are used on maize).
Whether recent increases in fertilizer use will continue in
the face of substantial increases in fertilizer costs remains
to be seen. It should be remembered that these benefit/
cost ratios were obtained on demonstration plots with
relatively high levels of husbandry.
One reason why the fertilizer companies have not been
as successful as the Kenya Seed Company in promoting
their products in the small farm sector is that the bulki-
ness of fertilizer packages (sold in 50 kg. and 100 kg. units)
makes them difficult to transport in areas where roads
are poor and distances to a stockist are far. Some stock-
ists have taken to breaking up packages and selling by the
kilo, but this has aroused fears among some farmers that
the fertilizer has been adulterated. Several papers and
reports, therefore, have called for packaging in smaller
and more manageable units.28 Recent price increases in
fertilizers have also coincided with a reduction in the
government subsidy for fertilizers, making continued adop-
tion by new users even more difficult. It is doubtful, any-
way, that the subsidy was responsible for much of the
increase in demand since 80 percent of the fertilizer im-
ported is used on large-scale farms and estates, where use
has been stagnant in recent years, whereas the increase in
use is primarily on small-scale farms. The ILO Report
recommended discontinuation of the subsidy on equity
grounds, but a government working party and the parliament-
ary committee recommended it be continued and increased.
The parliamentary committee also charged that "the ferti-
lizer distributors in Kenya operate a cartel with strong
links with parent companies in Europe" which prevented
them from buying from cheaper Persian Gulf sources,
and asked that this situation be investigated.29 It also
expressed convern that protection being granted to East
African fertilizer producers "should not have the effect
of increasing the cost of fertilizers to maize farmers,"
given the importance of increasing fertilizer use in the
region. Similar problems also affect the distribution of
insecticides. One representative pointed out that, although
60 percent of his total sales force was employed in servicing
the smallholder areas, this effort accounted for only 20
percent of total turnover.30 Most of the proposals to solve
these problems have come in the form of stockist credit
schemes, which are discussed in the following section.

Agricultural Credit

Seasonal credit for maize was initiated by KFA in the
early 1930's as declining world prices and locust plagues
created adverse circumstances for farmers. With pressure
to increase production during World War II seasonal credit
was expanded. Under the Guaranteed Minimum Return
(GMR) system, a farmer applies to the Agricultural Finance

Corporation (AFC) for a loan for a specific crop (either
wheat or maize) and a specific acreage. After a -'umber-
some system of approvals, the AFC authorizes the Kenya
Farmers Association (KFA) to make available inputs suf-
ficient to cover that acreage. The farmer sells his crop
through the KFA (acting as agent for thw Wheat or Maize
and Produce Boards) and the outstanding loan is deducted
from the payment. In case of a crop failure, the farmer is
protected against loss, subject to an inspection of his fields
by an agent of the AFC (hence, the title GMR). Unfortuna-
tely, like many other services, this system was designed to
serve the interests of large-scale European farmers and has
proved ill-adapted to current Kenyan farming conditions.
As an excellent recent review of farm-credit in Kenya
points out,31 there are some 1.2 million smallholdings
and only 3500 large farms and ranches in Kenya. Although
ownership of these large holdings has been shifted fairly
successfully in recent years from European to African
bands, some of the institutions designed to serve the large-
farm sector have not as yet made the transition towards

Map 1. Western Kenya, distribution of registed stock-
ists, 1972. Sources: Kenya Seed Company and Kenya
Population Census, 1969.

serving the African farm sector as a whole. Most of the
current credit institutions, including some experimental
programs discussed below, are in the painful process of
trying to make such a transition.
The first government loans to smallholders were made
in 1948 by the Africa Land Utilization and Settlement
Board, later the African Land Development Board
(ALDEV).32 In 1963, the lending activities of the two
Boards of Agriculture (for African and European areas)
were combined to form the Agricultural Finance Corpora-
tion, referred to above. Under the AFC, three basic types
of loans are made:
-Long-term loans for land purchase, of dairy cattle
and permanent improvements. These loans are made to
farms with a minimum of ten acres and require land titles
as security. These are estimated to account for 43 percent
of all farm credit.
-Medium-term loans for farm development, machinery,
and crops, available to all farmers but generally requiring
title deeds for security. These are estimated to account
for 31 percent of farm credit.
-Short-term loans, such as the GMR for planting maize
and wheat, and other loans for the planting of pyrethrum
and tea. These are estimated to account for about 26
percent of farm credit, although most of this comes from
commercial banks and merchant suppliers, with the GMR
itself accounting for only about 3 percent of total farm
Exbluding funds for land purchase provided under the
Agricultural Settlement Fund, Donaldson and von Pischke's
estimates show that about 23 percent of all farm credit
from both public and private lending institutions went to
some 200,000 small farmers whereas 77 percent went to
about 3500 large farm borrowers.33 GMR loans were
divided about equally between large and small farmers,
although the definition of small in this case is not clear,
since GMR has a minimum acreage requirement of 15
acres. (Prior to 1964, in fact, the minimum acreage was
100 acres). In 1968, a survey in four districts of western
Kenya showed that, in 1966, and 1967, the total of such
loans was only 150 in Nandi District, 151 in Busia, 55 in
South Nyanza, and 18 in Kisii.34 Even at best, therefore,
only 1.5 percent of official credit went for maize grown
by small farmers as of 1972. From these data it is clear
that the diffusion of hybrid maize among smallholders
in Kenya could not have been dependent on credit, at
least from official sources.
Credit from non-official sources is also extremely limit-
ed in Kenya particularly since maize trading is prohibited
except through official marketing channels. It is probably
safe to say that official credit in Kenya is largely restrict-
ed to wealthy farmers who are in a position to obtain
credit from commercial sources. For this reason, numerous
reports have recommended that commercial banks be en-
couraged (or required) to lend more to the agricultural
sector and that the government concentrate its activities
on aiding small farmers. A number of schemes to do this

are being tried in Kenya, which offer some preliminary
lessons about the problems involved in promoting small-
holder development of hybrid maize.

Experimental Credit Programs

Three types of approaches have been tried in recent years to
improve the provision of seasonal credit to small-scale
farmers: cooperative credit, supervised individual credit,
and stockist credit. All three are intended to address
themselves to the administrative problems which make
smallholder lending more difficult than lending to large,
established enterprises. These problems may be grouped
simply under the headings security, supervision and recove-
ry. Small-scale farmers often lack title deeds to their land
as security for a loan. A better security for the lender,
in the view of many experts, is an assurance that the funds
borrowed will be spent on a profitable activity but this
requires farmer training and/or extension supervision to
see that proper husbandry practices are used. Finally, the
recovery of loans of small amounts from large numbers of
poor farmers presents difficult and expensive administrative
problems, especially where the crop for which the loan has
been made is one which can be consumed on the farm or
sold locally, as well as through official channels. The
only substantial, successful smallholder lending in the past
has been by the Kenya Tea Development Authority, which
has a monopoly on purchasing the farmer's output and can
easily effect recovery. Another problem is that using
agricultural agents as loan collectors undercuts their sub-
sequent effectiveness as advisors and motivators of the
farmers they serve. Each of the methods recently tried
in Kenya deals with some but not all of these problems.
Cooperative credit. The Cooperative Production Credit
Scheme (CPCS) of the ministry of Cooperatives and Social
Services was initiated in 1970 following a major report on
cooperative credit by S. Lindquist in 1967 and the establish-
ment of a Cooperative Bank in 1968. In 1969 some nine
cooperative societies were participating in a pilot project
which lent 25,000 to approximately 2000 members, and
by March 1971, 52 societies had joined the project and
had been allocated some 600,000 for lending to approxi-
mately 96,000 members.3 5 This compares with a total of
about 4500 smallholders receiving GMR seasonal credit
through the Agricultural Finance Corporation. Most of
loans, however, were used for coffee picking and husbandry
improvements, which reflects the concentration of viable
cooperatives in the coffee growing areas. Under the
CPCS, loans are made largely in kind from the cooperative
society's own store or from another supplier who bills
the cooperative. Loans are secured by the crop proceeds
of the farmer and two guarantors and are granted at an
interest rate of from 8 to 12 percent. Loan money is
obtained from the Cooperative Bank (at 8 percent) or
from the societies' own funds (at a lower rate), allowing a
blended cost of funds of about 6 percent to the society.36

The loans are then deducted from the payout of the
cooperative to the farmer at the end of the crop season.
The CPCS has several distinct advantages over other
methods. First, the security of a commercial cash crop
(such as coffee or pyrethrum) can be used to cover food
crops (such as maize) on which collection would be dif-
ficult. Second, the existing staff and accounts machinery
of the cooperative can be used to handle the credit ad-
ministration without involving extension staff in debt
collection. Third, an excellent cooperative education pro-
gram exists in Kenya to train society officials and employe-
es in management of loan funds. Fourth, very large numb-
ers of farmers can be reached through the cooperative system.
The system, however, also has several drawbacks. First, it
operates effectively only where cooperatives are strong,
namely in the coffee and pyrethrum areas which are
comparatively the wealthiest in the rural sector. Farmers
in marginal areas or in districts where cooperatives are
not functioning, who are most in need of seasonal credit
to finance their input purchases, are generally excluded.
Second, the program does not include close farmer super-
vision although some liaison with agricultural extension
staff of course takes place. Third, only societies who have
substantially cleared themselves of past debts are allowed
to participate in the scheme. Since many societies have
long-term unsecured loans outstanding (usually to their
own officials), this has unavoidably restricted participa-
tion in the system. Nevertheless, for the areas where
cooperative societies are strong due to the presence of
high-value cash crops, this is the most promising credit
system to date.
Supervised Credit. Two experiments in providing super-
vised credit to individual farmers have been tried under
the Kenya Special Rural Development Program (SRDP)
in Tetu Division of Nyeri District (Central Province) and
Vihiga Division of Kakamega District (Western Province).3
Both of these have been well-documented by academic
researchers and project personnel, and there are some
interesting differences between them. Both projects are
in high-potential areas of extremely high population density
where the intensification of farming through the adoption
of hybrid maize would both improve family incomes and
release land for growing of cash crops.
The Vihiga maize credit program began in 1971 when
63 selected farmers were given AFC credit vouchers rede-
emable for inputs through stockists in the area. Farmers
were selected on the basis of "credit worthiness" from
a random sample of some 600 farmers in the division, and
hence, represented wealthier and better-than-average farm-
ers, though of course on a lesser scale than the AFC's normal
activities. Credit was given for two to four acres' worth
of inputs. Farmers received intensive extension advice
and obtained excellent yields (18-20 bags per acre). An
88 percent repayment rate was achieved, but only after
considerable pressure from the chiefs, the AFC, and in
some cases local extension staff. In 1972, 323 farmers
received loans, but only a 58 percent repayment rate was

achieved. In 1973, 920 loans at an average of 271/- each
were made. The AFC estimated servicing costs (exclud-
ing repayment losses) at 2.2 percent. An analysis of loan
repayments in 1972 showed that profitability of the in-
puts was similar between those who repaid and those who
did not, but that repayers had more outside sources of
income (50 percent vs. 26 percent for non-repayers).38
Although the AFC has been able to streamline its loan-
making and recovery operations considerably on the basis
of the project's experience, it nevertheless seems clear
that the administrative costs of supervising and recovering
individual loans of such small magnitude makes the ex-
pansion of supervised credit programs extremely difficult.
The Tetu project differed from the Vihiga project not
in the nature of the loan and the inputs used, but in the
selection and training of farmers. After a careful baseline
study, less progressive farmers were selected for participa-
tion in the program and were taken to the local Farmers
Training Center for a three-day course in hybrid maize
growing together with the extension staff from their im-
mediate area. Participants were offered vouchers for inputs
suitable for one acre of hybrid maize. After harvest, the
AFC with the help of the government administration col-
lected loan repayments, and farmers who repaid were
eligible to receive another loan. Seventy-eight percent
of farmers in the first FTC couse repaid the loan, although
again only after intensive effort by the AFC and the local
chiefs. The attractive part of this program is that poorer
farmers are selected for participation and use is made of
group extension methods, thereby reducing the costs of
supervision. The costs of surveying and selecting the
farmers were considerable, however, and the costs of col-
lection raise the same problems as the Vihiga program. The
widespread adoption of hybrid maize by other farmers on
a purely cash basis in both the Vihiga and Tetu project
areas raises questions as to how much of a constraint
credit really is and indicates that the greatest benefits of
supervised credit schemes to date may have been in the
demonstration effect they had on surrounding farmers.
Stockist Credit. Faced with the limited (though subs-
tantial) regional coverage of a cooperative-based credit
program on the one hand, and the problems of debt col-
lection and loan security for giving supervised credit to
individual farmers on the other, government and various
aid agencies have looked to stockists credit as another
means of improving the utilization of commercial inputs
in maize farming. The most promising program along this
line has arisen out of the FAO fertilizer demonstration
plots and a pilot credit scheme in the Migori Special Rural
Development Program in South Nyanza District of Nyanza
Province.39 In this experiment, credit was provided by
the Kenya Commercial Bank to eight selected stockists
chosen from a list of 15 prepared by the agricultural field
staff. (One of the stockist was in fact a cooperative
union). The loans varied in size from 3000/- to 5000/- at 9
percent interest and were paid by the bank to the KFA
depot in the province from which inputs were delivered

to the stockists. The loans were secured by mortgages on
the land and inventories of the various stockists. The
stockists were therefore able to purchase adequate inputs
and have them available in good time for planting and
were encouraged to pass on part of the credit to their
customers on their own. Because of adequate security,
careful selection of stockists, and the small number of
units involved, administrative costs were low and the bank
experienced no bad debts. On the basis of this experiment,
a proposal to extend the scheme to 400 stockist in 20
districts was accepted by the Danish aid agency to provide
credit for seasonal purchases of inputs and construction of
improved storage facilities totalling 500,000. This program
should go a long way to improving the availability of inputs
in these rural areas in adequate quantities, but does not
provide supervisory services for the use of the inputs nor
does it reach the farmers to whom credit is most likely to
be a real constraint. Careful evaluation will be needed to
see if greater availability of credit to stockists does result
in improved private lending to their customers. In this
connection, it should be remembered that the cost of seed
(at 20 shillings per acre) is not a prohibitive investment
for very many farmers and that fertilizer (at 100 shillings
or more per acre) is not considered profitable unless used
with good husbandry practices. It is likely, therefore,
that improved credit systems are not a necessity for increas-
ed adoption of hybrid maize at present although they may
become more important at a later stage when achieving
still higher yields is required.

Pricing and Marketing Policies

Of all the institutions and policies affecting the Kenyan
maize industry, none arouse as much interest, outrage, or
concern as pricing and marketing policies. Since the
first official maize report in 1922, periodic maize crises
have provoked at least eleven commissions of inquiry,
working parties or select committees to investigate and
make recommendations on the pricing and marketing of
maize.40 Most of these investigations, moreover, have been
competent undertakings which have conscientiously taken
evidence, conducted expert analysis, and made responsible
recommendations, only to see maize policy continue to be
made on an ad hoc, emergency basis, with the authorities
acting too late to meet the immediate crisis and then taking
little remedial action until the next crisis occurs. The.
latest case is an excellent example. Following a maize short-
age in 1971, a select parliamentary committee was ap-
pointed in July 1972, which prepared an excellent report
that was unanimously adopted by Parliament in December
1973. Hardly a month later, the government rejected a
relaxation of maize controls, the most important recom-
mendation in the report, and proceeded instead to increase
controls over the sale and movement of maize. As in
previous crises, most of the blame was placed on traders
who were "sabotaging the distribution system."41 Road

blocks were instituted by the police to "stamp out" move-
ment of maize, rice, and beans, and new permits were
introduced to control the movement of crops from farm to
market and from millers to retail stores.42
The factors which cause this kind of periodic chaos
are actually quite straightforward, but this unfortunately
does not make thier solution any easier. The basic pro-
blem can be expressed in a very few statements:43
-Maize is the basic staple of about 90 percent of the
population and the demand for it is highly price inelastic.
It would be hard to find a more politically sensitive crop.
-There are regular but unpredictable changes in the
weather in East Africa which affect the size of the maize
crop, as well as the availability of other foods.
-Only 5 to 15 percent of total maize production is
normally marketed through official marketing channels,
depending on the size of the crop. In a good year, when
maize is in surplus on farms and back market prices are low,
the official marketing body will receive two or three times
as much maize as in a bad year, when black market prices
are high. Moreover, storage costs are reasonably high
(about 5.60 shillings per bag per year in 1966) so it is
costly to hold stocks from year to year.
-Because maize is bulky relative to its value, transport
costs are high. There is a big gap between the price which
Kenya receives for exports and the price it pays for im-
ports and, until the recent increase in the world price of
maize, Kenya could neither import maize more cheaply
than it could produce it nor export at a profit.
In the face of these conditions, Kenya for years has
sought to maintain a price level that would produce self-
sufficiency in maize, but would not result in a large surplus
which would have to be exported at a loss. The first
guaranteed price emerged from farmer pressure in the
1930's when price declines accompanying the world wide
depression led to a government guarantee. This required
a guaranteed price, since swings in the market price due to
variations in rainfall would be too great to assure adequate
production if farmers could not be certain of covering their
costs. Underlying this assumption, in the colonial period,
was the belief that peasant farmers would not respond to
.price incentives, and European farmers had to be relied
upon to produce adequate supplies to feed the African
labor force. The Food Shortage Commission of Inquiry
of 1943 made the point:44
Since the Colony cannot safely depend on native grown maize to
satisfy its internal requirements, it must look to the European
farmer to produce a certain amount of maize. The European farmer
cannot afford to produce maize for sale unless he obtains a return at
least equal to that obtainable from other farming activities.
In fact, deliveries to the Maize Board from large farms
have varied pretty consistently with variations in small
farm deliveries, both depending primarily on the weather
(Figure 2) The only difference may be that, whereas
large-scale farmers may vary the size of their plantings
according to the price level, small-scale farmers may vary
the amount of surplus maize delivered to the official market-
ing channels rather than the amount actually planted.4s

We have already noted how a guaranteed price was
introduced temporarily during the Depression and again,
permanently, in 1942. To keep the large-scale producer
from being undercut by lower-cost producers in the African
areas, however a controlled market had to be created as
well. Since export losses were deducted from the payout to
farmers the Europeans desired that this cost be shared
across the crop as a whole. As a result, a system of tight
controls was introduced which has persisted to the present.
Without a special movement permit, it is illegal to move
more than ten bags of maize within a district, or two
bags of maize across district boundaries. Since the price
differential between a maize-surplus and a maize-deficit
district can be enormous (from 20/- to 100/- per bag),
the possession of a movement permit enables the holder
to make substantial profits. Otherwise, all sales must be
made at the official price through the Maize and Produce
Board. The recent Select Committee report stated that,

There were many complaints both about pressure exerted on the
Maize and Produce Board to issue permits to particular people,
and about the illegal issue of maize movement permits by authori-
ties other than the Board. Given their value, these movement
permits provide potential sources of patronage and corruption on
a large scale...the uncontrolled issue of movement permits obvious-
ly plays havoc with the controlled maize market; it is grossly
inequitable, and it encourages production inefficiencies.46

To enforce the regulations requires frequent searches of
trucks and bus traffic by police, which is also liable to
corruption. Because the Maize and Produce Board sells
most of its maize to millers and other large-scale users,
the controlled market can create large price differentials
even in the same town. In July 1973, we found maize
selling in local markets for as much as 60/- a bag while
maize and Produce Board depots not 300 yards away were
overflowing with maize purchased at 35/-. Until 1967, in
fact, the Board would only sell in quantities of 140 bags or

Fig. 2. Sales to maize board by large and small farm sector, 1955-1965. Source: Massell et a. "Maize
Policy in Kenya." Institute for Development Studies, University of Nairobi, Discussion Paper, 1965, p. 8.
(Data were not kept on this basis after 1965).













\ ,~ \

SLarge Farms

'.Sm.all Farms

*6b 1*61 '6 '6) '64 '66

# '57 '58 '59

4o 3n0

0 70

o 0
30 6o.
5. 0

C 20


r 40


Large Farm Maize Acreage
(thousand hectares)

/---~ Producer Price
/ (Shillings per bag)

'71 '72

Fig. 3. Producer price and acreage planted to maize on large farms, 1959-1972.
Sources: Select Committee Report, pp. 13 and 15; Economic Survey, 1973, p. 58.

more, but since has sold in quantities of five and, recent-
ly, one bag.
There have been many proposals that the Maize and
Produce Board should operates as a buyer of last resort,
setting a floor price at which it would take all maize
offered and offering to sell at a ceiling price to hold
price fluctuations within a given range. Within that range
farmers would be able to sell to anyone they pleased and
traders could move maize easily and quickly between
surplus and deficit areas, preventing the large differences
in price which now prevail. Given the excellent system of
highways and abundance of transport in Kenya, price dif-
ferentials would vary only slightly, reflecting the costs
of transport between producer and consumer. Trading in
maize would itself be an asset in the development of
entrepreneurial activity in the whole economy. There would
also be a moderation in the sharp price variations over the
course of the year which now prevail (Figure 4), since
it would pay farmers to hold the maize on their farms
longer after the harvest. At present, farmers have no
incentive to hold back surplus maize since it may spoil
and since the Board offers no differential in price accord-
ing to time of delivery. In fact, farmers who deliver late
may find, as many did in 1973, that the Maize and Produce
Board stores are full.47 Efficiency of production would
also be improved by freeing the internal market, since at


present the high price of maize in marginal (deficit) areas
encourages farmers to grow maize there and surpluses in
the optimal growing areas discourage maize production
where it can be produced best. Under a free internal maize
market, the Board would, of course, continue to maintain
strategic storage reserves and would handle all exports and
The main argument advanced against a free internal
maize market has been that it would reduce the quantity
of maize handled by the Board and that "overhead costs
would be distributed over a much smaller turnover." 48
Given the profits Kenya can presently make on maize
exports, however, this worry certainly should not continue
to be a problem. Greater obstacles seem to be that those
who now benefit from possessing movement permits would
lose their monopoly positions, plus a general bureaucratic
fear in government circles of any relaxation of controls.
Because of the political outcry whenever imports have
been required, civil servants are hypercautious about allow-
ing any exports to take place. An analysis of Maize and
Produce Board purchases and sales over the past 14 years
(Figure 5) shows an average surplus of purchases over sales
of 0.95 million bags (1.27 million over the past seven years)
with the peaks getting successively higher. The greatest
deficit in that period was only 600,000 bags. With present
storage capacity of almost 3 million bags including a stra-

1970 maize

1971 maize

I 3 I X I

> E n I 0 z I p J '< I i 0 1 Z C4 0 I I
S rr19 *

Fig. 4. Maize price fluctuation in Vihiga, Kakamega district. Source: Peter R. Moock, "The Vihiga SRDP
Farm-Level Survey" Institute for Development Studies, University of Nairobi, Discussion Paper, 1971
p. 214.

tegic long-term reserve of more than a million, the country
has ample capacity to avoid future domestic shortages. The
increasing adoption of hybrid varieties means that Kenya
has almost certainly entered a long-term surplus position.
Moreover, the management of the Maize and Produce
Board since the 1971 crisis has been greatly improved along
with its capacity to plan imports and exports rationally.
The deficits of the 1960s in fact were due more to mis-
management of stocks than to actual shortfalls of national
production. In both 1965 and 1971 the decision to
import maize came so late that the harvests were starting
before the imports arrived. When they did come they
were either reexported or sold for cattle feed.
Although it is clear that a guaranteed price has been an
incentive to maize production in Kenya, the success of the
maize industry has outgrown the rigidly restricted market-
ing system that continues. The recent increase in world

prices has removed whatever rationale existed for an internal-
ly controlled market. Proposed bulk handling facilities
would significantly reduce transport costs, increasing ex-
ports profitability margins still further. The widespread
adoption of hybrid maize may yet bring about a freeing
of the internal market. With maize in abundance in the
country, the incentive to maintain monopolistic movement
permits may itself be reduced to the point that a free
internal market becomes a reality.
Bernd Schubert has suggested four criteria for evaluat-
ting agricultural marketing systems: guaranteeing of urban
food supplies; stabilization of prices, both seasonally and
between localities; technical and allocative efficiency; and
promotion of agricultural development.49 Although Kenyan
marketing policy has had a reasonable success in achieving
the first and last of these objectives, it has performed
poorly on the others. Given the increased adoption of

per Bag



H 1 f' 4 >t> 'fl C'- O O4 0 -

so o SN o o s0o r o O 0'
Fig. 5 Kenya maize and pr e b p s C1

Fig. 5. Kenya maize and produce board, purchases and sales of maize, 1961-1973.

hybrid maize, the construction of improved handling faci-
lities, higher world prices, new management in the Maize
and Produce Board, and, hopefully, a relaxation of out-


1. L.H. Brown, A National Cash Crops Policy for Kenya (Nairobi,
1963) p. 79.

2. M.N. Harrison, "Maize Improvement in East Africa", in
C.L.A. Leakey, Crop Improvement in East Africa, 1970. p. 21-59;
Festus Ogada, "A Statement of the Status of the Maize Crop in
Kenya," mimeo. (n.d.); S.A. Eberhart, M.N. Harrison, and F.
Ogada, "A Comprehensive Breeding System," Zuchter (1967):
169-174; "Development and production of Hybrid Maize in
Kenya," momeo, Maize Research Section (n.d.); B.D. Dowker,
"Breeding of Maize for Low Rainfall Areas of Kenya," Journal
of Agricultural Science (1971): 523-530; and L.D. Smith, "The
Role of Maize Research in Kenya Agricultural Development,"
mimeo (n.d.).


moded internal marketing restrictions, however, there is
no reason why maize cannot continue its preeminent role
as the most important crop in th Kenya economy.

3. M.N. Harrison, "Maize Improvement in East Africa", in
C.L.A. Leakey, Crop Improvement in East Africa, 1970. p. 37.

4. The heterosis between these two groups has been so good,
in fact, that the plants easily reach 17 to 18 feet in height and
suffer damage from lodging (falling down) in storms. A special
program has had to be introduced to lower the height of the
plants by reducing the distance between leaves on the stalk.

5. Peter R. Moock, "Managerial Ability in Small-Farm Pro-
duction: An Analysis of Maize Yields in the Vihiga Division of
Kenya" (Ph.D. dissertation, Columbia University, 1973), p. 135.

6. Olaf Hesselmark, Kenya Maize and Produce Board, personal
communication, February 20, 1974.

7. M.N. Harrison, "Maize Improvement in East Africa" in
C.L.A. Loakey, Crop Improvement in East Africa, 1970. p. 45
"Maize Diamonds," Kuenyii Farmnr (January 1968); A.Y. Allan,
"Some Effects of Time of Planting on Maize in Western Kenya," mi-
meo (n.d.) and A.Y. Allan, "The Influence of Agronomic Factors
on Maize Yields in Western Kenya," op. cit.

8. E.R. Watts, "The Adoption of Maize Growing Practices in
Embu District, Kenya," East African Agricultural Economics
Society Conference, 1969, pp. 2,5.

9. John C. de Wilde, Experiences with Agricultural Development
in Tropical Africa, Vol. II (Baltimore, 1967), pp. 38-39.

10. D.G.R. Belshaw and Malcolm Hall, "The Analysis and Use of
Agricultural Experimental Data in Tropical Africa, "East African
Journal of Rural Development 5 (1972): 39-72.

11. For an excellent recent summary of knowledge, see George
H. Axinn and Sudhakar Thorat, Modernizing World Agriculture:
A Comparative Study of Agricultural Extension Education Systems
(New York, 1973). On Kenya, see J.R. Weir, Report of the
Agricultural Education Commission (Nairobi, 1967), and John C.
De Wilde, Experiences with Agricultural Development in Tropical
Africa, Vol. II (Baltimore, 1967) pp. 157-197.

12. Hugh Fearn, An African Economy: A Study of the Economic
Development of the Nyanza Province of Kenya, 1903-1953 (London
1961), p. 84.

13. J.R. Ascroft et al., "The Tetu Extension Pilot Project. A
Special Rural Development Report," paper read at the workshop
on Strategies for Improving Rural Welfare, University of Nairobi,
1971, p. 27. "Progressiveness" was defined from an index of
adoption of eight key cash crops.

14. David K. Leonard, "The Social Structure of the Agricultural
Extension Services in the Western Privince of Kenya," Institute
for Development Studies, University of Nairobi, Discussion Paper,
January, 1972, p. 4. "Progressiveness" was defined as growing
hybrid maize and certain cash crops.

15. See Everett M. Rogers, Communication of Innovations
(New York, 1968).

16. See J.R. Ascroft et at., The Tetu Extension Pilot Project. A
Special Rural Development Report". Paper read at the Workshop
on Strategies for Improving Rural Welfare, University of Nairobi,
1971: David K. Leonard, "The Social Structure of the Agricultural
Extension Service inthe Western Province of Kenya", Institute of
Development Studies, University of Nairobi, Discussion Paper,
January, 1972; and International Labour Office, Employment,
Incomes, and Equality (Geneva, 1972), p. 153-154.

17. International Labour Office, Employment, Incomes and
Equality (Geneva, 1972), p. 154-246.

18. David K. Leonard et al., "The Work Performance of Junior
Agricultural Extension Staff in Western Province," Institute for
Development Studies, University of Nairobi, Discussion Paper,
June 1971, pp. 1,3,25.

19. "An Interim Report on the Evaluation of Agricultural
Extension," Rural Development Research Committee, University
of Dar es Salaam, 1968, p. 4.

20. Peter R. Moock, "The Vihiga SRDP Farm-Level Survey,"
Institute for Development Studies, University of Nairobi, Discussion
Paper, 1971. Joseph Ascroft, Carolyn Barnes, Ronald Garst,
"The Kissi SRDP Survey," IDS Discussion Paper, 1971.

21. Kenya Seed Company, mimeo, March, 1973. Also interviews
with W.H. Verburght, A.Y. Allan, E.J.H. Hazelden, Festus Ogada,
February August 1973.

22. Republic of Kenya, Report of the Seelct Committee on the
Maize Industry (Nairibo, 1973), p. 4.

23. Agricultural Extension Service, Kitale, "Economic Aspects
of Seed Maize Production in Trans Nzoia District" (April 1973),
pp. A1, A4.

24. Republic of Kenya, Report of the Select Committee on the
Maize Industry (Nairobi, 1973), p. 5.

25. See for example M.N. Harrison, "Maize Improvement in
East Africa". In C.L.A. Leakey, Crop Improvement in East Africa,
1970, pp. 52.

25. J.M. Colyer, "A Background to the Development of Fertilizer
Utilization in East Africa," East African Agricultural Economics
Society Conference Paper, 1969, p. 2.

27. K. Zschernitz, "Marketing of Fertilizers through Local
Stockists in Kenya," FAO/DANIDA Regional Seminar on Fertilizer
Use Development (Nairobi, 1972). Zschernitz concludes that the
stockist system is the "most successful" method for a country at
Kenya's level of development.

28. See for example, Fred E. Chege and Joseph Ascroft,
"Marketing Farm Supplies in Rural Areas: A Study of Farm
Inputs Availability in Tetu Division," IDS, University of Nairobi,
1972, and Republic of Kenya, Report of the Select Committee on
the Maize Industry Nairobi, 1973, p. 4.

29. International Labour Office, Employment, Incomes, and
Equality (Geneva, 1972), pp. 432-433; Republic of Kenya; Report
of the Select Committee on the Maize Industry (Nairobi, 1973)
p. 5; and Republic of Kenya, Report of the Working Party on
Agricultural Inputs (Havelock Report) (Nairobi, 1971), pp. 19-20),
pp. 19-20. The Havelock Report forecast a "conservative" increase
of 100% in nitrogenous fertilizer demand by 1975 (over 1969) and
a 47% increase in demand for phosphates.

30. D. Outram, Twiga Chemical Industries, "The Role of Private
Industry for the Small Farmer Market," Conference on the Dis-
semination of Research Findings and Technology in Kenya, IDS
University of Nairobi, 1973.

31. G.F. Donaldson and J.D. von Pischke, A Survey on Farm
Credit in Kenya (Washington, D.C., 1973).

32. Josef Vasthoff, Small Farm Credit and Development (Munich,
1968), p. 23.

33. Donaldson and von Pischke, Farm Credit, pp. 4 and 7a.

34. Meyer et al., Rural Development, p. 55

35. Department of Cooperative Development, Development Plan
1971-1974, pp. 35-39.

36. Donaldson and von Pischke, Farm Credit, p. 120.

37. See Niels Roling, Fred Chege, and Joe Ascroft, "Rapid Develop-
ment for Kenya's Small Farms," IDS, University of Nairobi, 1973:
Peter Weisel, "The Vihiga Maize Credit Program," AID Spring Review
of Small Farmer Credit, 1973; and "Minutes of Meeting to discuss
Tetu and Vihiga SRDP Extension, credit, and Training Programmes,
"Ministry of Agriculture, mimeo, July 1973.

38. Peter Weisel, "An Analysis of Maize Credit Loan Repayments,
1972," mimeo, July 1973.

39. Donaldson and von Pischke, Farm Credit, pp. 155-162. Also
interviews with Ministry of Agriculture and Provincial Planning
staff, August 1973.

40. See especially, in order of their appearance, V.G. Matthews,
"Report on Kenya Maize Industry," Ministry of Agriculture

mimeo, 1963; Republic of Kenya, Report of the Maize Commission
of Inquiry (Nairobi, 1966); W. David Hopper, "Marketing of
Grains in Kenya, "Ministry of Agriculture, mimeo, 1968; "Final
Report of the Working Party Studying the Maize and Produce
Board in Relation to the Expected Crop Production in the 1970's,
and in Particular with Regard to the Major Cereal Crops," mimeo,
1969; and Republic of Kenya, Report of the Select Committee on
the Maize Industry (Nairobi, 1973). For a good discussion of
some of the earlier reports and marketing in general, see William
O. Jones, Marketing Staple Food Crops in Tropical Africa (Ithaca,
1972), pp. 198-232.

41. East African Standard, January 11, 1974, p. 1. A favorite
activity for officials during the 1971 maize shortage was to visit
traders' stores during the middle of the night to find the "missing"
maize; no mention was ever made officially of the serious drought
then occurring as a possible explanation of the shortage.

42. East African Standard, January 12, 1974, p. 1.

43. For a good discussion of underlying factors see lan Livingstone
"Production, Price and Marketing Policy for Staple Foodstuffs in
Tanzania," in V.F. Amann (ed.), Agricultural Policy Issues in East
Africa (Kampala, 1973), pp. 208-242, and B.F. Massell, J. Heyer,
and H. Karani, "Maize Policy in Kenya," IDS, Discussion Paper,
University of Nairobi, 1965. Both of these were written, however,
when profitable exports from East Africa were not yet considered a
realistic possibility because of lower world prices.

44. Food Shortage Commission of Inquiry Report, 1943, p. 46.

45. William O. Jones, Marketing Staple Food Crops in Tropical
Africa, (Ithaca, 1972), p. 205, quoting Michael Harrison. It is


While other sources of data related to the adoption of
maize technology exist in Kenya, it was decided that the
larger purposes of this study could best be served by
basing it on newly collected primary data. These data
emerged from a sample survey of 360 maize farmers conduct-
ed in June and July of 1973.
Data from the survey have been analyzed using multi-
variate probit analysis, and method designed specifically
to deal with a dichotomous dependent variable, which in
this case is adoption and non-adoption of hybrid maize.
Ordinary least squares regression has been used where the
dependent variable is a continuous one, and chi-square tests
have been used with simple cross tabulations. In addition,
the study makes use of extensive documentary evidence
from such Kenyan sources as the Maize and Produce Board,
the Kenya Seed Company, and the Maize Research Section,
earlier farm level surveys in the region, interviews with
persons involved in the Kenyan maize industry, and pu-
blished sources.

even questionable how much large farm acreages follow price
changes. Except for the large drop in acreage when farms were
changing ownership in 1963-65, large farm acreages have remained
fairly stable over the past 15 years (Figure 3). Moreover, price
changes are often announced too late to affect planting, and
sometimes even after harvests have been partially delivered.

46. Republic of Kenya, Report of the Select Committee on the
Maize Industry (Nairobi, 1973), p. 21.

47. The drop in acreage of farmers applying for seasonal credit
(GMR) in 1973 was apparently due entirely to the fact that many
farmers were unable to deliver their maize to Maize and Produce
Board stores and hence could not clear their 1972 accounts before
planting time. As late as August 1973, when the 1973 crop was
already beginning to come in, stacks of maize from the 1972 crop
were piled under tarpaulins outside many MPB reception centers.

48. Republic of Kenya, Report of the Maize Commission of
Inquiry (Nairobi, 1966), p. 23. The Report also claims that it
would be impossible to control the price of maize meal to the
consumer without permitting millers to make excessive profits. Al
distrust of millers and traders, many of whom are Asians, seems to
have been a strong motivation for control among both Europeans
and Africans from the earliest introduction of controls, even
though Africans in the maize-deficit districts are the main victims
of the present policy and African transporters would be major
beneficiaries of the proposed changes.

49. Bernd Schubert, "Some Considerations on Methods for Evalua-
ting Marketing Systems for Agricultural Products," Eastern Africa
Journal of Rural Development 6 (1973): 39-54.

The farmers were drawn from the population of maize
growers west of the Rift Valley. This area is one of the
most densely populated in Kenya, produces the bulk of the
country's maize and has had the longest history of support
from the maize industry. Map 2 shows the distribution of
Kenya's population and outlines the area of the study.

Delineating Agro-climatic Zones:

The general potential of the area west of Rift can perhaps
best be summarized with the reference to the major agro-
climatic zones used in this study, as shownon Map 3. These
were determined with the advice of agricultural scientists
at the National Agricultural Research Station and the
criterion used for establishing these zones was that they
should be internally homogenous with respect to the suita-
bility of the new maize technology. Zone 1 is defined as
that portion of the region receiving more than 60 inches of

Map 2 Kenya, showing population distribution and area of study.

Map 3. Western Kenya, agroclimatic zones.

rain per year. Ths includes virtually all of Kisii and Kaka-
mega districts, about half of Nandi and Kericho districts,
and the eastern third of Busia district. This is the zone
with the highest agricultural potential, with generally good
soils, and is consequently densely populated, especially in
Kisii and parts of Kakamega. The major cash crops grown
are tea, pyrethrum, potatoes, and dairy cattle in the-higher
elevations and coffee, sugarcane, and bananas at the lower
levels. Maize is the dominant food crop everywhere. Zone
2 consists of that portion of the region found above 1500
meters which receives less than 60 inches of rain per year.
This high-altitude, moderate rainfall zone includes half of
Kericho and Nandi districts plus all of Uasin Gishu, and
Trans Nzoia districts and almost all of Bungoma.' This
zone includes most of the large-scale farms (in Uasin Gishu
and Trans Nzoia) as well as settlement schemes. Farm
sizes tend to be quite large by Kenyan standards (15
acres average) and there is considerable commercial farm-
ing, including especially maize (throughout), wheat (in
Uasin Gishu), dairy cattle, and some tea. Zone 3 consists
of that portion of land found below 1500 meters in altitude
which receives less than 60 inches of rain per year. This
zone includes part of Busia district, and all of Siaya, Kisu-
mu, and South Nyanza districts. It is a moderate potential

area with some cash farming of sugar, cotton and ground-
nuts, but mostly subsistence production of maize, millets,
and sorghum. Farm sizes are relatively small and farming is
hindered by special problems including flooding, impeded
drainage, and tsetse fly infestation in some areas.

Stratification of the Sample.

In order to ensure a sample size adequate for making
comparisons between agro-climatic zones, and farm size
groups, the sample was stratified in four categories. Three
of these are the zones described earlier.
Because it was believed that small-scale, primarily sub-
sistence farmers would behave differently from large-scale
commercial farmers, a special sample of some 50 large-
scale, African-owned farms was taken in the Uasin Gishu
and Trans Nzoia districts. These were formerly European-
owned enterprises which have been or are being purchased
by African owners and which have received extensive
government attention and assistance in recent years. Al-
though all of the farms in this category are located in zone

2 the large farm sample is referred to collectively and in
the tables as Set 4. A few farms of over 50 acres from
zones 1 and 3 were also included in Set 4.

Sampling Procedure

Within each of the three agro-climatic zones, a three stage
sampling process was used to select approximately one
hundred farms. First, grid squares of 100 square kilo-
meters were numbered sequentially in each zone using
1 to 250,000 scale maps. Within each zone the populations
of the various districts and parts thereof were summed and
sampling segments were assigned to each district in pro-
portion of its percentage of the total population in that
zone. This step adjusted the sample for population density
between districts though not within districts. Using a
random number table, 34 sampling segments were selected
in each zone.
Second, using 1 to 50,000 scale maps, each 100 square
kilometer grid was further divided into 100 one square
kilometer segments. A random number table was again
used to select one square kilometer sampling segments
within the larger grid. On this scale map, physical features
including roads and even huts are identifiable and some
adjustments were made, throwing out segments that fell in
the middle of a lacke, game reserve, or town.
Third, within each segment, the interviewer made a
list of all households by name before selecting any farms
for interviews.2 If there were not ten households in that
segment, the inverviewer took additional households from
neighboring segments to make a minimum total of ten.
Interviewers took a maximum of twenty households with-
in any one segment. After preparing the list of house-
holds, the interviewer used a random number card to select
three names for interview. This procedure was followed to
prevent the interviewer from being led to "better" or
more prestigious farmers. Random field checks of the
lists of names and the farmers interviewed indicated that
interviewers kept closely to the sampling procedure. If the
farmer grew no maize whatsoever, the interviewer was
instructed to conduct the interview but also to take another
farm which did grow maize in the same segment. Only
four of the 361 farmers interviewed (about 1 percent) were
growing no maize at the time of interview.
The large farm sample of some 60 interviews was taken
from lists of some 700 large-scale commercial farms provid-
ed by the Ministry of Agriculture. With the help of the
District Agricultural Officers in Trans Nzoia and Uasin
Gishu and the German Extension Team Leader in Trans
Nzoia, farms which had been effectively subdivided or
which were managed by the Agricultural Development
Corporation were removed from the list and a random
selection was made among the others. These large-scale
farms are referred to as set 4 in the tables and discussion.

The Interview

The chief instrument of the survey was a structured inter-
view containing 81 primary questions, many with a number
of different parts. The interview focused on the maize
growing practices of the farm but included basic informa-
tion on the farmer, his knowledge of recommended practi-
ces, communication, channels used by the farmer, extension
contacts, and other factors.
The questionnaire was prepared in consultation with
scientists in the Maize Programme of the National Agricul-
tural Research Station at Kitale. It draws on the experience
of similar interviews both in Kenya and in other developing
The findings of the.survey show a high internal consist-
ency. The only area of significant question involves acreage
estimates which are notoriously difficult in areas where
farms are not surveyed and fields are neither large nor
necessarily rectangular. Whereas farmers whose land has
been adjudicated are likely to know the total size of their
farm, their estimates of field size vary widely.
Fortunately determining the nature of the maize tech-
nology being used on the farm did not depend (as a yield
survey would) on accurate land measurements. However,
wherever field size does enter into the analysis (as in
fertilizer applications per acre), the limitations of the data
should be kept in mind.
Farm interviews were carried out by ten trained and
experienced interviewers of Research Bureau Limited of
Nairobi, a market research firm with considerable experien-
ce in both urban and rural surveys in Kenya. Interviewers
were given an intensive orientation seminar on the survey
by the author and Dr. A.Y. Allan, Chief Agronomist in
the Maize Research Section at Kitale. The interviews
were carried out either in the vernacular or in Swahili.

The Pattern of Hybrid Seed Adoption in 1973.

Table 7 shows the pattern of hybrid seed adoption in
western Kenya in June 1973 by agroclimatic zone.
The zones used and the survey methodology was des-
cribed earlier in this chapter. For hybrids, "Adoption" was
defined as use of hybrid seed for more than half of the
farmer's acreage. However, only a handful of farmers were
found growing both hybrid and local varieties on the same
farm. Although farmers no doubt "tried out" hybrid on a
portion of their maize acreage in earlier years, the innova-
tion is now well enough known that farmers are apparent-
ly willing to adopt it on all their acreage or not at all
depending on its reputation in their immediate area.
For other inputs examined, "adoption" was defined
as use of the input on maize. Two additional modern
inputs were examined, fertilizer and insecticide.
Farmers were asked two different questions relating to
each of their maize fields, namely "What kind of maize

seed was planted in this field? and "Was the seed planted
this year new seed from a bag or did you use your own seed
from last year's crop? Table 8, gives the responses for
specific types of maize seed. The distribution of these
types is determined largely by availability since only recom-
mended varieties for a given area are sold in that area
The second question was asked in order to ascertain whether
farmers were planting second generation hybrid seed and
still calling it "hybrid". Only three of 360 farmers ex-
plicitly said, when first asked, that they were using second
generation seed and an additional six said they were using
"hybrid" but also said it was "seed from last year's crop."
The basic figures on hybrid use, therefore, are those res-
ponses to the second question (i.e. second generation seed
is treated as a "local" variety). As discussed in Chapter II,
some agricultural experts argue that hybrid seed is not
practical for developing countries because farmers have to
buy seed every year and if they do not, second generation
hybrid seed gives a sharply reduced yield. The fact that
only 2.5 percent of the farmers interviewed in the 1973
survey appeared to be planting second generation seed
indicates that, in Kenya at least, the farmers' willingness to
buy seed every year has been considerable.
The most striking feature of the adoption pattern is the
wide disparity in adoption between Zones 1 and 2 and
Zone 3. Much of the analysis of this thesis seeks to
explain the discrepancy. According to Ogada,3 a regres-
sion analysis of the 1971 Late Maturity Maize Variety
Trials indicates that the yield performance of hybrid (in
this case primarily Hybrid 632, which is recommended for
Zone) indicates that the differential yield between hybrids
and local varieties increases significantly with improvements
in the environment. Using data from controlled experiments
at a total of 20 different sites, Ogada compared the mean
yield of each variety at each site on an environmental index.
The index is simply "the difference between the mean
yield of all varieties in an environment and the mean
yield of all varieties at all environments or grand mean,"
and runs from roughly -30 to +30. Ogada went on to say,
"In this particular case I can state without doubt that the
Zone 3 environments would be represented by the range of
-10 to -25 and the high altitude areas by +10 to +25."
Figure 6 shows Ogada's findings and the declining advantage
of Hybrid 632 over local varieties (in this case approximated
by Kitale II Composite) as the environmental index declines.
The analysis of risk perception in the next chapter
further supports this hypothesis as does certain additional
evidence provided by the farmers themselves. After t'-
examination of the farmer's maize fields, those farmers
who did not plant hybrid maize were asked why they did
not. Their responses are presented in Table 9. Responses
total more than the number of farmers planting local
maize since some farmers gave more than one answer. The
interesting part of this question is that, whereas 100 per-
cent of non-hybrid planters in Zones 1 and 2 gave "expense"
as their reason, almost half the responses in Zone 3 gave
other answers. I have grouped these answers under five

main headings: cost, availability, information, performance,
and "congruence" with traditional maize planting practices.
In doing this, I also included the eight farmers in Zone 3
who said they had never heard of hybrid maize.
The coding of the questionnaire may have. obscured
a distinction between those farmers who said hybrid was
"too expensive" (i.e. not worth the investment) which
would have been a judgement of its performance and those
who felt it was "to expensive" in that, although they would
have liked to use it, they just did not have the money. The
fact that a few farmers gave "has no money" as a separate
answer may indicate that this was the case. The fact that
several farmers said they did not plant hybrid because
they "planted too late" may indicate that they are aware

Table 7. Percent of farmers adopting hybrid maize in
western Kenya by zone, 1973.

Zone 1 Zone 2 Zone 3 Set 4
(N- 96) (N= 93) (N= 95) (N= 76)

94.8 89.2 15.8 100

Table 8. Types of maize seed used, 1973, by zone (All
answers in percentage)

Type Zone 1 Zone 2 Zone 3 Set 4
511 0 0 3.2 0
512 0 1.1 0 0
611 0 4.3 0 17.4
612 12.5 12.9 0 4.3
613 53.1 54.8 0 71.8
622 5.2 4.3 7.4 2.2
632 12.5 2.2 2.1 0
Other 1.0 3.2 3.2 4.3
Local 3.1 7.5 78.9 0
Don't Know 12.5 9.7 5.3 0

Table 9. Responses of farmers not planting hybrid maize
to question, "why did you not plant hybrid maize? "

Answers in %
Zone 1 Zone 2 Zone 3
Number of Responses 9 14 102

1. Cost: Too expensive/Has no money
2. Availability: Not available/Distance
to Stockist/Must buy every year
3. Information: Never heard of it/
doesn't know how to use it/ No
experience of it
4. Performance: Yields less/Less certain
Doesn't do well
5. Congruence: Too much work/
Planted late/Miscellaneous

100 100

0 0 6

0 0 14

0 0 20

0 0 7

Note: Includes 8 farmers in Zone 3 who had never heard of
hybrid maize. Total responses exceed the number of non-adopters
since some farmers gave more than one answer.

Table 10. Adoption and farm size.

Zone 1 Zone 2 Zone 3
Smaller half 95.7% 83.7% 14.9%
Larger half 95.8% 95.1% 17.4%
Chi-square value 0.21 1.91 0.01
Significance level NS NS NS

of the importance of early planting for hybrid yields and
though that with late planting it would not be profitable.
In any case, we can conclude from the responses that,
whereas hybrid is universally known and well regarded
in Zones 1 and 2, a substantial number of farmers in Zone
3 have doubts about its efficacy or are still ignorant of its
use. When asked "Have you ever planted hybrid maize? ,
24 of the 90 farmers not planting hybrid in 1973 said yes,
18 of whom (75 percent) were in Zone 3. This indicates
that hybrid has received some considerable trial in that
area and, for various reasons, has been found undesirable.
Looking now at the impact of farm size on adoption of
hybrids, the data of Table 10 shows that larger farmers
consistently lead smaller farmers, even when zones are
held constant. The difference between larger and smaller
farmers shown in the table are not great. It must be
remembered, however, that all farms over 50 acres -most;
of them in Zones 1 and 2- have been eliminated from the
data on which the table was constructed. These farms had
an adoption rate of 100 percent. Were they reintroduced,
the percentages of adoption in all zones would increase
slightly. The resulting comparison of farm sizes within
zones would show virtually no difference between smaller
and larger farms in Zone 1, a difference of some 14 per-
centage points in Zone 2, and a difference of roughly
eight percentage points in Zone 3.
Looking now at fertilizer use by agro-climatic zone
and farm size the impact of both factors are reflected in
the data of Table 11 except in Zone 3, where rates of
adoption are quite low. Even here, where differences
between farm size groups within Zones 1 and 2 are large,
the differences between agro-climatic zones are still more

Table 12 Farmers first hearing about
zone (All figures are percentages).

Table 11. Farm size and the use of other inputs.

Zone 1 Zone 2 Zone 3
Smaller half 47.8% 71.4% 2.1%
Larger half 75.0% 92.5% 6.4%
Chi-square value 6.23 5.03 0.25
Significance level 0.05 0.10 NS
Smaller half 2.2% 21.3% 7.1%
Larger half 10.6% 43.9% 2.2%
Chi-square value 1.53 4.17 0.33
Significance level NS 0.10 NS

notable. For insecticides the relative impact of agro-clima-
tic zones is even more evident.
The results of these two tables manifest clear support
of the hypotheses that agro-climatic zone and farm size
play a role in the adoption of new inputs. As differences
among zones exceed differences within zones for each of
the inputs, the tables tend to support the idea that agro-
climatic factors play a dominant role in shaping adoption
With respect to hybrid seed especially, the influence
of farm size is negligible as compared with that of the
climate. However, because the use of commercial fer-
tilizers and insecticides are significantly related to size,
we cannot conclude that yields (and therefore profitability)
of hybrids are neutral to scale. Moreover, since mechanized
plowing is conducive to early planting and has genuine
economies of scale, it is reasonable to assume that larger
farms plant earlier and consequently get higher yields.
Whether these advantages are offset by more intensive
weeding or more timely harvesting on smaller plots is
possible, but can only be determined by careful (and
expensive) field sampling. In any case, we can conclude that
small farmers in Kenya have not been excluded from
enjoying the benefits of hybrid maize and have adopted
it with alacrity where environmental circumstances war-
rant it.

and first using hybrid maize, by year and

Zone 1 Zone 2 Zone 3 Set 4
Year Hearing Using Hearing Using Hearing Using Hearing Using
1963 or earlier 2.4 0 3.5 1.1 4.5 0 19.5 8.7
1964 12.3 2.1 20.2 6.5 2.2 1.1 19.6 13.0
1965 11.1 6.3 17.8 16.1 1.1 0 26.1 23.9
1966 9.9 4.2 10.7 3.2 1.1 1.1 6.5 15.2
1967 12.3 10.4 7.1 9.7 2.2 1.1 8.7 8.7
1968 24.7 16.7 9.5 7.5 10.3 2.1 4.3 6.5
1969 7.4 11.5 9.5 6.5 13.7 5.3 6.5 8.7
1970 6.2 16.7 9.5 11.8 19.5 2.1 8.7 4.3
1971 8.6 7.3 4.7 9.7 19.5 11.6 0 10.9
1972 3.7 11.5 5.9 11.8 11.4 5.3 0 0
1973 1.2 1.0 0 3.2 4.5 3.2 0 0
None/Don't know 0 12.5 1.1 12.9 9.1 67.4 0 0

The Rate of Adoption Over Time.

Table 12 presents the percentage of farmers first hearing
about hybrid maize and first planting it by year and by
agroclimatic zone. Figure 7 presents the data on first
planting in cumulative form. Both sets of information are
based on the farmers' recollection of events as of June,
1973. For the reasons given in Chapter III, it is believed
that this information is quite reliable in the case of first
use of hybrid seed, although the farmers' ability to recall
when they first heard about hybrid is probably less reliable.
The mean time period between first hearing about and first

planting hybrid was approximately 1.5 years for all zones,
which compares with a mean time lag of five years for
Iowa farmers.4 The modal frequency of first hearing
about hybrid varied considerably from 1964 (one year
after hybrid was introduced) in Zone 2 to 1970 (seven
years later) in Zone 3. In Zone 1 the modal frequency was
in 1968. The mean year for hearing of hybrid varied in the
same pattern as follows: Set 4 -196.5, Zone 2-1966,
Zone 1- 1967, and Zone 3-1969. These means vary
directly with the distance of the respective zones from the
National Agricultural Research Station at Kitale (in the
large farm area of Trans Nzoia) but also reflect the relative

Fig. 6. Response of varieties to environments. Source: Late maturity
1971; Festus Ogada, National Agricultural Research Station.

) -20

maize variety trials,

/ H.611

SSR52 x EC573



- H.511
- KIT.II.Co.

-ib 6 +Ib



importance of maize as a cash crop among the different
zones, and, hence, the general level of commercialized
The modal frequency for first use of hybrid seed varied
from two years after its introduction in Zones 2 and 4 to
eight years in Zone 3. For Iowa corn farmers the modal
frequency of adoption was ten years after its introduction
and for Punjabi wheat farmers three years.5 The mean
frequency for first use varied from three years after first
introduction for the large farm areas to five years in Zones
1 and 2 and seven years in Zone 3. It is apparent from
these data that both large- and small-scale farmers in the
high rainfall portions of western Kenya adopted hybrid
maize at a rate faster than American farmers in Iowa in the
1920's and 30's. It should be remembered, however, that
the relative yield advantage of Kenyan hybrids over local
varieties was much greater at the time of its introduction in
1963 than was the comparable American case.6 Neverthe-
less, it would be difficult to conclude from the rapid adop-
tion rates and short time lags between first hearing and
first use that African farmers are in any way inextricably
bound by tradition or unopen to change.
A closer look at the adoption pattern in Figure 7
shows clearly the differences in adoption rates and ceilings
between zones. The former European-owned farms (Set 4)

reached almost 80 percent usage within five years. Hybrid
use on the smaller, but still relatively large, commercial,
maize-growing farms in the neighboring areas also spread
quickly but was soon matched and even slightly overtaken
by the smaller subsistence farmers in the high rainfall areas
of Kakamega, Kisii, and Kericho. Hybrid use in the lakeside
districts of Nyanza Province was much slower to develop
and may actually be falling since the percentage of farm-
ers who reported ever planting hybrid is now considerably
higher (32.9 percent) than the percent who reported plant-
ing in 1973 (15.8 percent). These data are presented visual-
ly in Map 4.
Interestingly enough, the rate of adoption in the high
rainfall zones has been so steady that they hardly show
the traditional S-shaped curve characteristic of most in-
novations (in which potential adopters at first hesitate,
then adopt quickly, and finally slow down again as "con-
servative" late adopters a lag behind).7 Griliches reported
that the S-shaped or "normal" curve typical of hybrid
adoption in the U.S. was also found to exist for the
adoption of combines, corn-pickers, and field forage harvest-
ers, as well as new types of prescription drugs.8 Griliches
also pointed out that the equilibrium level of adoption in
all parts of the U.S. never reached 100 percent. The
western parts of Nebraska, South Dakota, and Kansas

Fig. 7. Percentage of farmers first planting hybrid maize by year and by zone. (Figures are cumula-
tive and therefore may exceed the total percentage of farmers planting hybrid in a given year).





ZONE 1 ---
./ "ZONE3

10. '"


reached an equilibrium level after 30 years of 30 to 60
percent hybrid use. These are areas of highly variable rain-
fall where the use of hybrid was profitable only for those
farmers on particularly good land or able to employ irriga-
tion.9 This may eventually prove to be the case in Zone 3.
Unless new varieties specially suited to that environment are
developed, hybrid use may stabilize only among farmers
with certain, as yet unspecified, characteristics.

The Adoption of Associated Maize Technology.

Although the discussion so far has focused primarily on
the use of hybrid maize seed, this has never been consider-

ed in Kenya to be the sum total of hybrid maize technolo-
gy. From the outset it has been recognized that a complete
package of associated practices, including both physical
inputs and improved husbandry, would produce the highest
yields. Allan's definitive work on agronomic factors,'
already discussed in Chapter II, and Moock's study of small
farm maize yields in Vihiga both indicate the importance
of early planting, plant population, fertilizer use, and weed-
ing for achieving higher yields.
According to Hazelden,12 the conscious policy of the
seed company and the extension service from the start
was to use the new seed as a catalyst for achieving a
general improvement in maize husbandry and yields. "Hy-
brid" was considered a completely new crop requiring

Map 4. Diffusion of hybrid maize in Western Kenya, 1964-1973. Source: 1973 Kitale/CIMMYT hybrid
maize survey.

UI: Percoatae of Farmrs who Reported Navisg Planted Hybrid Ma i*, by District
Lss tb an 201 or an, 50_ or mor 0 o

new techniques of production. The Kenya Seed Company
painted "Hybrid" in large letters on its company Volks-
wagens and even today their publicity refers only to
"hybrid" not "hybrid maize." Seed was never subsidized
and farmers were taught that, if it was worth paying for,
it was worth using properly. The early demonstration
plots were few and well run and the farmers got high
yields from the total hybrid package. The fact that some
Europeans felt that African farmers were not "ready" for
hybrid only served to heighten the demand for it. Farmers
who laughed in 1963 at the idea of 20 bags an acre saw a
farm manager get 47 bags in a yield contest in 1972. How
well, though, did the strategy succeed in bringing about
the desired changes in husbandry that accompany the use
of hybrid seed? Have farmers accepted the new technology
in toto or have they merely bought the seed while revert-
ing to traditional methods of husbandry? And what has
been the difference in acceptance between commercial,
physical inputs and behavioral, cultural practices? The
1973 survey gives some interesting indications.
To begin with, Table 13 shows the percentage of farmers
in 1973 using various recommended maize growing practices
by zone. The use of manure (a beneficial but not actively
recommended practice) is also included. Although the
measurement of some of the practices may be fairly rough
(e.g. the timing of weedings may be more important.than
the number of weedings and the meaning of planting
"before the rains" instead of "with the rains" may vary
from place to place), nevertheless, these figures give a
reasonable picture of differential adoption of maize tech-
nology practices in western Kenya.
The pattern of associated technology use varies marked-
ly both by practice and by zone. The use of commercial
inputs varies from 100 percent (for hybrid seed in Set 4)
to 4 percent (for insecticide and fertilizer in Zone 3).
Cultural practices display a similar range, from a high of
100 percent planting in rows in Set 4 to allow of 10
percent planting "before the rains" in Zone 3. On the
basis of the evidence no clear distinctions are noted be-
tween the adoption of "physical" and "cultural" practices.

Each component of the recommended technology appears
to have been adopted or rejected according to its own
characteristics, including cost, profitability, risk, complex-
ity, and congruence with traditional diets or labor patterns.
Looking specifically at the commercial inputs, the weighted
average of farmers using inputs falls from 67 percent (seed)
to 49 percent (some fertilizer) to 46 percent (insecticide on
stored maize) to 13 percent (insecticide in the field). This
is in part a function of cost, since recommended levels of
fertilizer cost about five times as much per acre as hybrid
seed, and may also reflect availability. In the more com-
mercialized maize-growing areas, of Zone 2 the use of
commercial inputs of all types remains quite high whereas
it drops off sharply in the smaller, more subsistence-oriented
farms in Zone 1 and is uniformly low in Zone 3.
Among cultural practices, it appears that row planting
has been overwhelmingly adopted by hybrid growers. Inter-
planting, however, continues in importance in all the sub-
sistence areas. Moock found13 in Vihiga that interplanting,
controlling for plant population, was positively related to
maize yield and speculated that legumes may release nitrates
into the soil. He also noted that farmers claim the tall
maize protects the interplanted crop from hail damage.
Other reasons advanced for interplanting include improved
soil cover to preserve moisture and reduce erosion, control
of soil temperature, saving on weed control, and staggered
growing periods which affect both labor demands and
food requirements. As long ago as 1934, L.S.B. Leakey,4 '
explaining the interplanting habits of the Kikuyu, wrote:
The habit of regarding African methods of agriculture or of any
other activities as inherently bad because they are different from
our own is most unwise. I do not suggest that the methods used
by different native tribes are all perfect. Doubtless the methods
of agriculture employed by the Kikuyu could almost certainly
be improved in many details, but this could only be done if
European methods of research were employed in trying to develop
the African method of cultivation, which is very different thing
from trying to substitute European methods of planting for those
which have been evolved out of reach by trial and error.

As a result of the durability of interplanting as a practice,
the National Agricultural Research Station in 1973 began

Table 13. Percent of farmers using recommended maize technology by zone.

Zone 1 Zone 2 Zone 3 Zone 4 Weighted Average
A. Physical Inputs
1. Used Hybrid Seed in 1973 94.8 89.2 15.8 100 67.1
2. Used Some Commercial
Fertilizer 61.5 80.6 4.2 97.8 49.0
3. Used Insecticide on Stored
Maize 47.9 74.2 17.0 100 45.8
4. Used Insecticide in Field 6.3 30.1 4.2 50.0 13.2
5. Used Manure on Maize 25.0 20.4 15.8 10.9 22.8
B. Cultural Practices
6. Planted in Rows 96.9 93.5 32.6 100 75.1
7. Planted a Pure Stand 64.6 61.3 41.1 91.3 57.4
8. Weeded More Than Once 83.3 51.6 46.3 69.9 63.3
9. Thinned Maize 31.3 44.1 64.2 -43.5 45.2
10. Planted Early 31.3 26.9 10.5 47.8 23.0

Table 14. Husbandry
users, 1973.

practices of hybrid and non-hybrid

Hybrid Non-Hybrid
Users Users
Practice (N- 189) (N=93) X2 Value
Planted in rows 97.2% 29.0% **47.12
Used some commercial
fertilizer 68.2 3.1 "32.73
Used some insecticide
on crop 16.5 1.2 3.04
Thinned maize 35.7 66.6 **6.72
Weeded more than once 74.4 38.9 *12.37
Planted pure stands 67.0 36.4 6.57
Used manure 24.9 12.5 1.21
Planted before the rains 32.6 7.6 6.08
Used insecticide on
stored maize 61.1 14.3 "17.03

Sold some maize 51.5 22.0 5.79

* = 5% level of significance.
" = 1% level of significance.

trial studies on interplanting to discover what physical
interactions, if any, do exist.
Early planting, unfortunately, is difficult to measure
since the beginning of the low rains varies considerably
from place to place. For this reason, the question posed
in the 1973 survey asked if planting began "before the
rains started," "when the rains started," or "after the
rains started." Even this phrasing presents problems, how-
ever, since there is not necessarily consensus on when the
rains have actually "started" and since planting can be
suspended if the rains appear to halt. Farmers, therefore,
were also asked, "How many days passed between starting
and finishing seeding in this field? and the responses
indicate some variance (from an average of 4.1 days in Zone
1 to 6.9 days in Zone 3). This appears to indicate greater
caution among Zone 3 farmers. In any case, planting before
the rains had the lowest acceptance of any cultural prac-
tice and was, once again, lowest in Zone 3. This is con-
sistent with our risk hypothesis, namely, that farmers will
forego the higher yields which result from early planting
in favor of a greater certainty that the rains have actually
started. Another factor affecting early planting, however,
is that plowing (whether by oxen or by hand) is more
difficult when the soil is hard (and coincidentally when the
oxen are at their weakest due to the long dry period). It is
not surprising, then, that early planting is highest (48 per-
cent) in Set 4, where 100 percent of the farmers plowed by
tractor, as opposed, for example, to 7 percent in Zone 3
and 25 percent in Zone 1.
Interestingly enough, the practice of thinning out weak-
er plants is highest in Zone 3 (65 percent) and lowest in
Zone 1 (32 percent). This may be because farmers who
have purchases hybrid seed are reluctant to pull up even a
few stalks, even though it would improve their overall
yield. Weeding, on the other hand, is highest in Zone 1
(where 84 percent of the farmers weeded more than once)

although this is most likely a function of farm size rather
than a newly adopted practice. The use of manure was also
greatest in Zone 1.
Unfortunately, we do not really know the incidence
of husbandry practices at earlier periods and are, hence,
unable to assess precisely how much change is due to the
introduction of the hybrid technology package. The evi-
dence from earlier sample surveys, however, shows very
low levels of fertilizer use in 1965, with increasing use in
1968 and again in 1970, but still lower than in 1973. Such
information as does exist on row planting also shows a
gradual increase and it is asserted that row planting was
virtually unknown before the introduction of hybrid..
An indication of the "hybrid influence" can perhaps
best be gained by comparing the husbandry practices of
hybrid seed users with those of non-adopters. if we assume
that husbandry improvements are a "secular" phenomenon,
there should be no significant difference between hybrid
and non-hybrid users. To measure the strength of what-
ever association does exist, we have used a simple chi-
square test which measures the difference between the
actual and the predicted observation. Any chi-square great-
er than 3.84 with one degree of freedom indicates a less
than 5 percent chance that such a distribution would have
occurred randomly; 6.63 or greater indicates a less than 1
percent probability.
Table 14 shows the practices of hybrid and non-hybrid
adopters for the weighted average of the entire sample. In
every case except two there was a greater than 95 percent
probability that the distributions were not random. One of
these was insecticide use which has a 90 percent probability
and the other was manure use, which although not specifi-
cally recommended by the extension service, is nevertheless
used by more than twice as large a percentage of hybrid than
of non-hybrid growers.
Row planting and fertilizer use were especially strong-

ly related to hybrid use, while interplanting and thinning,
as noted above, were positively related to non-hybrid use.
Use of insecticide on stored maize and sales of maize off the
farm are strongly associated with hybrid use.
From the evidence of the survey, we can conclude that
the adoption of associated technology is strongly related
to hybrid seed adoption regardless of the overall levels of
adoption of each portion-of the new technology. In other
words the "package approach" seems to be justified even
if it is not 100 percent adopted in each case. The one cu-


1. A small portion of western Bungoma and northern Busia
districts which does not fall clearly in any of the three zones was
excluded from the survey area, as were the forest areas on Mt.
Elgon and in Kakamega and Nandi districts. The zones used
here correspond roughly to the Star Grass, Kikuyu Grass, and
Savannah zones mentioned in the literature.

2. "Households" were used in making this list to avoid a possible
bias of being directed to only the better "farmers," a frequent
survey problem.

3. Letter from Festus Ogada, Director, Kenya National Maize
Breeding Programme, February 12, 1974.

4. Bryce Ryan and Neal C. Gross, "The Diffusion of Hybrid Seed
Corn in Two Iowa Communities," Rural Sociology 8 (March 1943);

5. Zvi Griliches "Hybrid Corn and the Economics of Innovation."
Science 132 (July, 1960): 277-279, and Max K. Lowdermilk,
"Diffusion of Dwarf Wheat Production Technology in Pakistan's
Punjab. A Summary Report Condensed from a Doctoral Thesis",
Cornell University, 1972, p. 17.

rious exception is the thinning of young plants, which is
more prevalent among non-hybrid users. The general rela-
tionship is true both of physical inputs and of cultural prac-
tices, although the level of adoption of each practice varies
with its cost, complexity, and congruence with traditional
practice. Where the adoption .of an innovation such as
early planting may increase the riskiness of the food crop,
however, its acceptance has been much more limited. In
the following chapter we will look more closely at factors
affecting differential adoption levels among farmers.

6. Zvi Griliches, "Hybrid Corn and the Economic of Innovation,"
Science, 132 (July 1960): 278. Griliches estimated that hybrid out-
yielded open-pollinated varieties by 15-20 percent at that time.

7. Everett M. Rogers, Communication of Innovations (New
York, 1971), pp. 176-179.

8. Griliches, "Hybrid Corn," p. 275

9. Ibid., p. 279.

10. A.Y. Allan, "The Influence of Agronomic Factors on Maize
Yields in Western Kenya with Special Reference to Time of
Planting" (Ph.D. Dissertation, University of East Africa, 1971).

11. Peter Moock, "Managerial Ability in Small-Farm Production"
(Ph.D. dissertation, Columbia University, 1973).

12. Interview with E.J.R. Hazelden, Kenya Seed Company,
February, 1973.

13. Moock, "Managerial Ability in Small-Farm Production," p. 259.

14. L.S.B. Leakey, "Science and the African," 1934. Given the
author by T.M. Wormer.


In this chapter we look more closely at the characteristics
of adopters and non-adopters of the new maize technology
and at the characteristics of "early" adopters. In particular,
we consider the relationships between the adoption of
hybrid seed in 1973 and such factors as agroclimatic zone,
perceived risk, ability to "tolerate" or withstand risk, and
knowledge, as well as availability, of the new technology.
The purpose of this analysis is twofold. First, it is
important to see what, if anything, characterizes non-
adopters in order to be able to modify research, extension,
or input supply systems in ways that will better reach this
segment of the population. Secondly, it is important to
distinguish differences between early and later adopters


so as to judge what effects the introduction of the new
varieties has had on income distribution and employment.
After comparing the factors related to early and later
adoption we will consider whether time lags in adoption
have caused serious problems for income distribution in
Both bivariate and multivariate methods have been used
in the analysis of adoption, including difference of means
tests, chi-square tests, Pearson correlations, probit analysis,
and multiple regression analysis. Of these, probit analysis
merits special mention since it is as yet infrequently used
in social science applications. The most readable explication
of multivariate probit analysis is still probably James Tobin's

1955 paper the"Application of Multivariate Probit Analysis
to Economic Survey Data."'

The Multivariate Model for Hybrid Adoption

Based on the established procedures for multivariate probit
analysis the equations for examining the factors influencing
adoption are

/ =a +bXi +cX2 +dX3 ... +nXn


Y =g(l)

where I is an index reflecting the combined effect of the
X factors which prevent and promote adoption. Y =1
(adoption) if I exceeds a threshold value; Y = 0 (non-
adoption) if I falls below a threshold value.

The following independent variables are used:

Interval-Scale Variables
X, = Age (in years)
X2 = Education (formal education in years)
X3 = Size (size of farm in acres)
X4 = Distance (distance to nearest source of inputs in miles).
X5 = Cash (imputed value of cash crops in thousand shillings per

Dummy Variables
X6 = Zone (0 = Zone 3)
(1 = Zones 1 and 2)
X7 = Risk (0 = "risky" -famine crops present)
(1 = "not risky" -famine crops not present)
X = Job (0 = no job off farm)
(1 = job off farm)
Xg = Experience (0 = Farmer has never worked off farm)
(1 = Farmer has worked off farm)
Xio = Large farm (0 = Farmer has never worked on a large
commercial farm)
(1 = Farmer has worked on a large com-
mercial farm)
XI I = Credit (0 = Farmer believes farmers in his area cannot
get credit)
(1 = Farmer believes farmers in his area can get
X 2 = Extension (0 = No extension visit in past year)
(1 = Extension visit in past year)
X3 = Demonstration (0 = Farmer has never attended maize
(1 = Farmer has attended maize de-
X14 = Training (0 = Farmer has never been to farmers Training
(1 = Farmer has been to Farmers Training

The variable X7 warrants additional explanation. Inter-
planting maize with other crops is a common practice
among Kenyan farmers (see Table 13). Oftimes, especially

in Zone III, maize is found intercropped with millets and
sorghums, both drought resistant crops. While there may
be explanations for this practice, given the physical proper-
ties of the drought-resistant crops and the taste preference
for maize evidenced by its rapid spread, it seems reasonable
to assume the presence of drought-resistant crops in the
farmers' crop mix represents a risk averting action. In the
analysis which follows, the variable X7 serves as a proxy
variable for risk aversion.
A computer program, "N-chotomous Multivariate Probit
Program" was used to make the necessary computations.2
This program produces sample means and variances for
each independent variable, maximum likelihood estimates
for the coefficient of each variable, standard errors, and a
t-test (MLE/SE). The log of the likelihood function and
-2 times the log of the likelihood ratio are given, the latter
of which is distributed like chi-square with m degrees of
freedom for large samples and can be used as a measure of
the explanatory power of the independent variables taken
in groups.3 Using the coefficients obtained by the maximum
likelihood method and substituting the values of the inde-
pendent variables in the equation, the equation is solved
and the resulting solution is converted to a probability of
adoption from the normal distribution table. By substitut-
ing the average values of the sample for all but one variable
at a time, an idea of the relative importance of each variable
for an "average" case (if such exists) can be obtained.
Tables 15 and 16 give the "probits" or probit runs
which are used in the analysis. The following sections
discuss each of the key independent variables roughly in
order of their importance. Arithmetic means (for continuous
variables) and frequency distributions (for dummy variables)
are presented for adopters and non-adopters. Where these
statistics differ between zones, the reasons for those dif-
ferences and their pos-ib'e implications for adoption are
To recapitulate briefly, it is hypothesized that adoption
will be positively correlated with education, farm size,
presence of cash crops, presence of an off farm income,
availability of credit, relevant work experience off the
farm, attendance at maize demonstrations, an extension
visit in the past year, and location of the farm in a high
rainfall zone. It is anticipated that adoption will be negati-
vely correlated with age, distance to a source of inputs,
presence of "famine crops" in the crop mixture, and loca-
tion of the farm in a moderate or variable rainfall zone
(Zone 3). Special importance is ascribed to those factors
(agroclimatic zone and the presence of "famine crops")
associated with risk and, where these factors are themsel-
ves important, it is anticipated that "risk tolerance" or
ability to absorb risk (as measured by outside income,
access to credit, farm size, and presence of cash crops) will
be positively related to adoption. Chi-square and difference
of means tests are applied, where applicable, to the bivariate
analysis, and simple correlations between variables are
given with appendix.

Factors Affecting the Adotpion of Hybrid Maize.

Using the model presented previously, probit an
carried out separately on the sample population
and on the sample in each of the three agroclimi
Analysis of Set 4 was not undertaken since 100
farmers were adopters. The results of the ar
given in Tables 15 and 16. Because of the \
number of non-adopters in Zones 1 and 2, no
ly significant coefficients were found in Zone 1
imputed value of cash crops was significantly
adoption in Zone 2 (at the 0.5 level). Most of
variate analysis, therefore, concentrates on tl
taken as a whole (including agroclimatic zone a
pendent variable) and on adoption and non-ac
Zone 3 (excluding agroclimatic zone as an e>
As Table 15 shows, the following factors sig
influenced the adoption of hybrid maize in weste
agroclimatic zone, perception of risk (as measul
planting of famine crops), level of formal educati
ledge of credit availability, and imputed value of (
As hypothesized in Chapter IV, zone, education, i
cash were positively related to adoption and
negatively related. Although the significance
education as an independent variable declined
introduction of partially collinear variables suct
and cash crops, it remained significant at the 0.10 I
presence of all the major variables (equation one

Table 15. Multivariate probit

the other variables, age was negatively related to adoption
and size of farm was positively related, as hypothesized,
although neither had large coefficients or t-values. Measures
of extension contact (extension visits and attendance at
maize demonstrations or Farmers Training Centers) were
positively related to adoption, though only one was signifi-
cant at more than a 0.05 level (equation three). Previous
off-farm work experience, whether in general or on a large,
commercial farm, was, surprisingly, negatively related to
adoption, and having a job off the farm at the timesof
the interview, though positive in sign, was not significantly
related. Distance to a source of inputs did not prove a
useful indicator, since non-adopters were generally unable
to answer the question correctly, and it therefore was
dropped from the analysis. In fact, distance from a source
of inputs, when included in the equations, turned out to be
positively (but not significantly) related to adoption, perhaps
because the few non-adopters who knew the distance to a
source of inputs were those who lived closest to such
a source. Finally, the chi-square values for all of the
equations indicate that the expalantory power of the inde-
pendent variables taken together was significantly different
from zero at the 0.001 level. The R2 or pseudo-R2 values
for the two main equations were encouraging (0.76 and
0.75), and the coefficients obtained predicted almost 92 per-
cent of the cases correctly.
As Table 16 shows, the following variables were signifi-
cantly related to adoption in Zone 3: risk, education, and
credit availability (at the 0.05 level) and imputed cash income

analysis of hybrid maize adoption in western Kenya-all

Independent Variables Equation 11,2 Equation 2 Equation 3
CONSTANT 0.60 -1.22
Agroclimatic zone 2.07 (8.06)+++ 2.09 (8.55)--++ 3
Age in years 0.01 (1.61)+ -
Education in years 0.06 (1.30)+ 0.07 (1.74)+-
Farm size in acres 0.01 (0.05) -
Risk (famine crops) 1.42 (3.89)+++ -1.41 (3.90)t++
Credit availability 0.98 (3.08)+++ 0.88 (2.94)4+4
Cash (000 shillings) 0.21 (2.15)++ 0.21 (2.11)++
Job off farm 0.28 (0.92)
Ever worked off farm -0.69 (2.66)4++
Worked on large, commercial farm -0.75 (2.34)4-
Extension visit in past year 0.29 (1.32)+
Attended maize demonstration 0.13 (0.69)
Attended Farmers Training Center 1.33 (2.60)+-+

-2 times log likelihood ratio 221t++ 218+t4
X2 distribution-degrees of ; freedom 7 5
Estimated R squared (R ) .76 .75
Percent of cases predicted correctly 91.9% 91.5%

1. The maximum likelihood estimate of the coefficient is given first, followed by t statistics in
parenthesis. 2. One, two, and three plus signs (+) indicate significance levels of 0.10, 0.05, and 0.01
respectively for a one-tailed test. 3. Equation three included the same variables as equation two with
each of the work experience and information variables added in turn. The coefficients of the
independent variables in equation two changed only very slightly with the addition of an additional
variable, as did the constant and the chi-square value.

(at the 0.10 level). Attendance at Farmers Training Centers
was positively and significantly related to adoption, and
previous work experience off the farm was significantly and
negatively related to adoption (equation three). Age and
size of farm were not significantly related to adoption, had
very small coefficients, and'negative signs. Again, the chi-
square values of the equations indicate that the explanato-
ry power of the independent variables was significantly
different from zero at the 0.001 level. The R2 values were
0.72 and 0.69 and 91.6 percent of the cases were predicted
correctly. In the following sections, each of the inde-
pendent variables is discussed in greater detail.

Adoption and Agroclimatic Zone

The percentage of farmers in each zone using hybrid maize
seed on more than half their maize acreage in 1973 has
already been reported as follows:


Zone 1
(N = 96)

Zone 2
(N = 93)

Zone 3
(N = 95)

(N = 74)

The importance of agroclimatic zone is indicated in equation
(1), Table 15. By substituting the mean values for the
other variables and solving the equation, we find that
location of the farm in Zones 1 or 2 increases the proba-

ability of adoption from 17.7 percent to 87 percent. The
importance of zone is also indicated when adoption and
zone are cross-tabulated, giving a chi-square value of 53.64,
which is significant at the 0.001 level. The simple correlation
between zone and adoption is 0.78 which is also significant
at the 0.001 level.
The basic puzzle in the study of hybrid maize adoption
in western Kenya is to explain this substantial difference
between the higher altitude, high rainfall zones (1 and 2)
and the medium altitude, moderate rainfall zone (3). In
the absence of reliable yield data for farms in both areas,
we must regard agroclimatic zone as a measure of the
general environment conditions in which the farmer makes
his decisions. This includes not only the relative profitabili-
ty of the new technology in each zone (se Chapter II) but
also the variability of rainfall among zones. It seems to be
a well-established principle of meteorology that areas with
lower average rainfall tend also to have higher variability
from year to year. Evidently this is the case in western
Kenya, as is indicated by the persistence of famine crops
in the moderate rainfall area (see the discussion oft"risk"
The best data on relative yields in western Kenya are
from the F.A.O. Fertilizer Program demonstration plots.4
These plots, laid out by the extension staff on farmers' fields
in highly visible locations, have three portions: a control
plot with hybrid seed and no fertilizer, a plot treated with
60 kg. phosphates (PiOs) at planting, and a plot with
phosphates and 60 kg. of nitrogen topdressing. Since no

Table 16. Multivariate probit analysis of hybrid maize adoption in western Kenya-zone
three only.

Independent Variables Equation 11,2 Equation 2 Equation 3
CONSTANT -0.96 -1.67
Age in years -0.01 (0.82) -3
Education in years 0.13 (1.57)+ 0.14 (1.81)++
Farm size in acres -0.02 (-0.53)
Risk (famine crops) -2.33 (3.06)++- -2.23 (3.07)+++
Credit availability 2.04 (3.56)+++ 1.86 (3.56)+++
Cash (000 shillings) 0.28 (1.46)+ 0.27 (1.39)+
Job off farm 0.46 (0.77)
Ever worked off farm -0.95 (1.76)-+
Worked on large, commercial farm -3.41 (0.10)
Extension visit in past year 0.60 (1.22)
Attended maize demonstration 0.47 (1.21)
Attended Farmers Training Center 1.28 (1.76)++

-times log likelihood ratio 41+-t- 39+++
distribution-degrees qf freedom 6 4
Estimated R squared (R ) 0.72 0.69
Percent of cases predicted correctly 91.6% 91.6%

1. The maximum likelihood estimate of the coefficient is given first, followed by the t statistic in
parenthesis. 2. One, two, and three plus signs (+) indicate significance levels of 0.10, 0.05, and 0.01
respectively for a one-tailed test. 3. Equation three included all the variables in equation one plus each
of the work experience and information variables added in turn. The coefficients of the independent
variables in equation one changed very slightly with the addition of the other additional variable
and have not been reproduced.

local seed varieties are used, the results of these trials
only show the relative performance of hybrids in the three
zones and not of the relative performance of hybrids
vis-.-vislocal varieties. We should note also that the level
of husbandry on these demonstration plots is certainly well
above that of the average farm. Nevertheless, the results
give at least some indication of the differing performance
of hybrids in the three zones.
By averaging the yields obtained in Busia, Siaya, Kisumu,
and South Nyanza districts (all basically in Zone 3) in 1969,
1970, 1971 and 1972, and comparing them with the other
higher rainfall districts in western Kenya (all of which
would be in either Zone 1 or 2) we get the results shown
in Table 17.
Using this measure, hybrid yields averaged 22 percent
higher over a four-year period in the districts of Zones 1
and 2 than in the districts of Zone 3. Because of the
hybrids' greater genetic potential, moreover, we know
(Chapter III) that their advantage over local varieties will
be greater under good conditions than under poor condi-
tions. Therefore, the differences in yields between hybrids
and local varieties should be even greater than the differen-
ces found between hybrids taken alone. In sum, there are
good reasons to believe that differential profitabilities
play an important part in explaining the substantial dif-
ferences in adoption between the high and moderate rain-
fall zones and, consequently, in the overwhelming import-
ance of agroclimatic zones as an explanatory variable in the
sample taken as a whole.
Unfortunately, the problem is not this simple because
the three zones are not homogenous with respect to
their history or their existing infrastructure. Hybrid maize
was quite understandably promoted first in those areas
where it was thought to do best. Similarly, inputs were
supplied first to those areas where demand was greatest.
Knowledge of the new technology, in spite of extension
efforts, varies more or less inversely to the distance from
the National Agricultural Research Station at Kitale. The
lakeshore areas of Zone 3, therefore, combine a number of
features disadvantageous to the adoption of the new tech-
nology. lower altitude and rainfall, greater variability in
rainfall, poorer soils and drainage, poorer roads and input
availability, lower levels of extension and farmer training,
greater distance from the key research center and from
input outlets, and a later introduction of hybrid varieties in
the first place.5

Table 17. Yields of hybrids in 1074 F.A.O. fertilizer
demonstration control plots, 1969-1972.

Zone Number of Plots Yield, kg/ha
1 and 2 655 3417
3 419 2796

Agroclimatic conditions, then, are confounded by dif-
fering levels of general development and services. More-
over, the cash crop opportunities in Zone 3 (cotton, ground-
nuts, sugar) are generally not as lucrative as those in Zones
1 and 2 (tea, coffee, pyrethrum, dairy, bananas). Cash
incomes from the farm are therefore much lower, and
the corresponding interest in farming as a way of earning
a living can be logically expected to be less. Another prob-
lem is the fact that the cash crops in Zone 3 are more
seasonal than those (except coffee) in Zones 1 and 2, which
means that they provide a less steady source of cash in-
come with which commercial inputs can be bought. Be-
cause of the lag between harvesting (from August to Octo-
ber) and planting (in February-March) of the long-rains
maize crop, ready cash is not easily available for purchasing
inputs for those farmers without access to credit.
Another factor closely related to agroclimatic zone is
ethnic composition or tribe. Tribal boundaries in Kenya,
with few exceptions, are well delineated and often conform
closely to geographical features. This is especially true
in the region around Lake Victoria in Kenya, where the
Kisii hills and Nandi escarpment mark a sharp boundary
between the Luo-speaking people of the lakeshore and the
Bantu and Kalenjin-speaking peoples of the highest eleva-
tions. Only on the boundary between the Luo and Luyia
tribes north of the lake is there anything like a gradual
transition. And only in the formerly European-occupied
districts of Uasin Gishu and Trans Nzoia, where hybrid
adoption is virtually 100 percent, are these substantial
numbers of farmers from a variety of tribal backgrounds.
The result of this situation is that tribe and agroclimatic
zone are inextricably confounded. For example, there
were only three Luo farmers in the sample outside Zone 3
and ten non-Luo farmers in Zone 3. These 13 farmers
behaved as the majority of farmers in their zone rather
than as farmers from their tribe in other zones. I have not,
therefore, found it possible to distinguish tribe as a separate
explanatory variable. While not in any sense ignoring the
importance of cultural and ethnic factors in determining
responses to change, I would defend this conclusion in this
particular case on the grounds that, whereas the Luo tribe
make up the overwhelming proportion of farmers in Zone 3,
a wide variety of tribes in Zones 1 and 2, including those
more closely related to the Nilotic Luo than to the Bantu-
speaking peoples, have responded to the favorable condi-
tions for adoption hybrid maize. Within Zones 1 and 2,
there were no significant differences in adoption by tribe.
In the colonial period, the lower level of agricultural
development in the lakeshore region was attributed primari-
ly to "people problems." Typical of the colonial view
were the writings of Elsepth Huxley, the historian of and
apologist for the white settlers in Kenya. In her book,
The New Earth, published in 1960, Huxley entitled a
chapter "The Reluctant Luo."6 "Their obstinacy is le-
gendary," she reported, "they are inclined to be sullen
and suspicious.... Faintly Teutonic in an African fashion

...I! J. It is not surprising that of all the major Kenya
peoples, leaving aside the Masai, the Luo have put up the
stiffest resistance to the forward march of Progress in
the agricultural field." Similar, though less extreme, views
were expressed by Fearn, de Wilde, Ruthenberg, and a
variety of government reports, most of which complained
of the reluctance of the Luo to adopt practices recommend-
ed by the agricultural services.7
These opinions have recently been soundly rebutted by
Margaret Jean Hay in an excellent study of the economic
history of Kowe. a sub-location in Central Nyanza (now in
Kisumu District), from 1890 to 1945.8 In looking at the
Luo responses to a wide variety of innovations (white
maize, cotton, cassava, farm implements, education, trad-
ing) she concludes that "far-reaching changes" took place
in the Luo economy during this period and that programs
which were rejected "did not represent the optimal al-
location of land and labor resources."9 In general, she
says, "the agricultural potential of the area has been exag-
gerated byoutsiders who hoped to develop it." A survey
of Annual Reports of the Department of Agriculture from
1906 to 1959 showed only 12 years in which droughts,
floods, or poorly distributed rains were not cited by
agricultural officers.10
Although colonial officials were important in introducing
crops to the area, they were less significant in winning
acceptance of them. Introduction often involved coercion,
forced labor on demonstration plots, low prices, and con-
flicting recommendations. The poor acceptance of cotton,
for example, introduced by John Ainsworth near Kisumu
in 1906-07, has been often cited as evidence of Luo back-
wardness. As Hay points out, however, cotton faced a 50
percent price drop between 1910 and 1914 and again after
it was reintroduced in the 1920s and 1930s. Moreover,
the heavy labor demands of cotton conflicted sharply with
the Luo food crop cycle. In spite of these problems, the
Director of Agriculture noted in 1930 that "the failure of
the natives to accustom themselves to market fluctuations"
had retarded production.1 I
Hay contrasts the response to cotton with that to maize,
which spread steadily after its introduction in 1917, proba-
bly by one H.H. Holden, a Luo-speaking West Indian em-
ployed by the Department of Agriculture. The new maize
was called Orobi (after Nairobi) or ababavu ("a great thing")
because it was larger than existing varieties. It required
little adjustment in techniques of planting or preparation,
although it did not really become widespread until after
the introduction of a complementary innovation, the hand
gristmill, about 1929-when it could readily replace the
easier to grind sorghum as a flour staple. White maize
was also reputed to give higher yields and made a preferred
white flour.12 Also readily accepted were cassava, especial-
ly after the famines of 1918-19 and the locust invasion of
1931-32, and groundnuts, introduced in 1913-14 and again
in the 1920s.
After 1930, Hay found, famines in 1931-32 and 1943-

44, the collapse of cotton prices during the depression,
the discovery of gold nearby, and eventually the labor
demands of World War II drew men off the land. Labor
migration and local trade became important as sources of
cash income. These plus increasing population pressure,
shorter fallow periods, and soil exhaustion led to a "gradual
de-emphasis of the agricultural sphere within the economy
as a whole."' 3 By 1945 labor migration was an accepted
pattern. Although the colonial administrators took the
failure of cotton campaigns and soil conservation measures
as proof of a retrograde traditionalism among the Luo,
Hay concludes that the de-emphasis of agriculture in favor
of out-migration to work and engagement in trading op-
portunities was a rational response to their conditions.
Moreover, the colonial view ignored the successful adop-
tion of other agricultural innovations, ones which either in-
creased economic security (such as groundnuts and cassava)
or reduced labor expenditure (such as gristmills and plows).
From the evidence of the 1973 survey we can conclude
that agroclimatic zone is the most important factor in
determining the adoption or non-adoption of hybridmaize
seed in western Kenya. Although hybrid maize has done
well on some farms in the lakeshore zone, evidence from
fertilizer control plots indicates that hybrid varieties yield
substantially less in the lakeshore districts than in the
higher rainfall zones. This conforms with other data from
variety trials which indicate that the relative advantage of
hybrids over local varieties also declines with an index of
environmental conditions (Chapter III).
Further suspicion about the relative performance of
hybrid versus local maize is aroused by the responses of
farmers in Zone 3 when asked why they did not plant
hybrids (Table 9) and by the large difference between
farmers in Zone 3 who had ever planted hybrid (33 per-
cent) and those who were planting it in 1973 (16 percent)
(Figure 6). Farmers in Zone 3 have also suffered from
lower levels of extension services and farmer training,
more difficult access to inputs, poorer communications,
less well-organized marketing facilities, and a later introduc-
tion of the new varieties.14 Moreover, Hay's historical
study shows that the Luos, the major ethnic group in Zone
3, have readily adopted other agricultural innovations which
either increased economic security or reduced labor ex-
penditures. The record she presents of periodic famines and
irregular rainfall in Luoland indicates that risk-aversion may
be important in the way the Luos observe the new maize.
Just how important we will see in the following section.

Adoption and Risk

The theoretical basis for including risk as an independent
variables has been presented. A dummy variable-whether
or not the farmer planted drought-resistant but otherwise
less desirable "famine" crops- was created as a proxy for
the farmer's perception of the risk of not getting a maize

crop. The "famine" crops used in the dummy variable
were sbrghum, bulrush millet, cassava, and groundnuts.
The hypothesis was that farmers who include "famine"
crops in their crop mix will be less willing to invest cash
resources in the purchase of hybrid seed and complementary
inputs. The dummy variable "risk" was negatively cor-
related with adoption (-0.54) which was significant at the
0.001 level. Crosstabulated with adoption, the risk proxy
gave a chi-square value of 24.82 which is also significant
at the 0.001 level.' 5


Famine crops present
Not present



"Risk" was also negatively correlated with adoption in
each of the three zones taken separately:

Simple r
Level of significance

Zone 1

Zone 2

Zone 3

Chi-square values for the crosstabulation of risk and adop-
tion were also significant (at the 0.01 level) for Zones 2 and
3, though not for Zone 1.
Because risk was also correlated with zone (r= -0.50),
however, it was included both with and without zone in the
probit analysis. When included with zone (equation 1), risk
had a coefficient of -1.41 and zone, 2.09. Without risk
in the equation, the coefficient of zone was 2.32. In spite
of this multicollinearity, risk added significantly to the
equation (the chi-square value increased from 185 to
203) and risk was therefore retained as a variable in most of
the equations. Because zone is so important as an explana-
tory variable, the alternative of running risk without zone
in the equations was discarded. (When this was tried, the
chi-square value dropped to 119).
To get a better idea of the importance of risk perception
as an influence on adoption, we can substitute representati-
ve values for the independent variables in equation 1.
Setting the independent variables at their mean, we find
that, if a farmer in Zones 1 or 2 grows famine crops, his
probability of adoption is reduced from 97 percent to 71
percent. If a farmer in Zone 3 (Table 16) grows famine
crops, his probability of adoption falls from 25.2 percent
to 0.2 percent.
From the analysis we can conclude that the farmer's
perception of risk, as measured by the planting of "famine"
crops, is an important negative influence on adoption.
What we do not know is whether this reluctance is due to
the fact that hybrids per se are less reliable than local
varieties, or whether, in the presence of considerable
doubt about whether he will get a maize crop, the farmer
is unwilling to pay cash for commercial inputs. A possible
genetic explanation for the former case might be that the

local varieties of maize stem originally from lowland tropical
varieties imported at an early stage by the Portuguese from
the Caribbean and Brazilian coasts whereas the hybrids de-
veloped at Kitale stem primarily from a cross of a settler-
selected plateau variety with an imported Latin American
highland variety. The theory here would be that local
varieties are more tolerant of heat and humidity than
the cooler highland cross. It is certainly true that the
higher the altitude, the longer growing season a variety
My own opinion, however, is that given a basic doubt
about getting a maize crop at all, farmers are reluctant to
pay cash for commercial inputs. Because of better selec-
tion, production, storage, dressing, and handling, hybrid
seed is, if anything, probably more reliable in its germina-
tion than local seed. The distinction is an important one.
In either case, the development of varieties specifically
suited to the lakeshore zone is probably essential if wide-
spread hybrid use is to be achieved. But if a reluctance to
purchase commercial inputs is the problem the develop-
ment of a synthetic or compositive variety (in which the
seeds can be used for planting year after year) may in fact
be necessary. And, finally, until and unless a reliable
lakeshore maize variety is developed, farmers should certain-
ly not be discouraged from planting drought-resistant crops.
Much greater research emphasis should be devoted to im-
proving the yields and the nutritional value of such crops as
sorghum, millet, and cassava.

Adoption and Farm Size

Most of the studies of the Green Revolution in Asia and
elsewhere have paid particular attention to the importance
of farm size in determining the adoption of new technolo-
gies. Gotsch, Falcon, and others have argued that the new
agricultural technologies have increased income inequalities
because larger, wealthier farmers have enjoyed economies
of scale in production and have had easier access to seeds,
fertilizers, and, particularly, complementary inputs of
tractors and irrigation.'6 More recently, Harrison and
others have argued that economies of scale are not signifi-
cant in LDC agriculture and are often offset by more inten-
sive input use on smaller farms. 7 Lowdermilk, in his
Pakistan study, claims that there is "no conclusive evidence
that farm size itself is of major importance" in the adop-
tion of high yielding varieties, and quotes several studies
from India which found size of little importance in de-
termining adoption.' 8 He quotes Dr. S.R. Sen on the sub-
ject as follows:

The new technology is basically biological and chemical rather
than mechanical in nature. Therefore, the new [High Yielding
Varieties] did not give rise to such an economy of scale as to
place the small farmer in a substantially disadvantageous position
as compared to big farmers.9
and contrasts this with the conventional wisdom:

No theory of farm practice adoption can ignore the farm operator
and the scale of his oapeation.... The importance of economic
stntius in adoption, sire of unit, gross returns from the unit are
more highly correlated with adoption than other variables.2

In the present study, farm size was not found to be
significantly related to adoption in any of the probit equa-
tions, either in individual zones or in the sample taken
as a whole. In equation 1, even a ten acre increase in
farm size (almost 100 percent of the average farm size)
would have increased the "average" farmer's probability of
adoption by only 3 percent. Although the mean farm size
of adopters was larger than that of non-adopters in each
zone, these differences were not statistically significant
except in the entire weighted sample where the generally
smaller farm sizes in Zone 3 made the size difference
significant at the 0.001 level.21

Adoption and Cash Crops

It was hypothesized that adoption of hybrid maize seed
would be positively associated with the presence of cash
crops on the farm. The theoretical basis for this hypothesis
is twofold: first, that the presence of a cash income would
offset the risk of experimenting with a new food crop, and,
second, that the availability of cash from the sale of cash
crops would facilitate the purchase of the necessary inputs.
In the analysis, a variable "cash" was created by multiplying
the acreage of each major cash crop (coffee, tea, pyrenthrum,
bananas, sugar, wheat, and cotton) by an "average" estimat-
ed gross margin for that crop.22 To this was added a figure
for dairy production (the number of grade cattle times an
average estimated net return per cow) and for marketed
maize (the number of bags sold times an estimated net of
25 shillings a bag), since maize itself is the most important
cash crop in Zone 2. Although there are obvious difficulti-
es in using an "average" return since variations in yield are
substantial between farms (the best farmers often getting
double the average yield), nevertheless, this variable gives
a rough measure of the cash income accruing to each farm
from sales of farm produce. Sales of minor cash crops
groundnutss, passion fruit, wattle) would constitute a small
fraction of total farm income. The values of the "cash"
variable for the different zones are shown in Table 18.
Although these are only imputed cash incomes and are not
based on actual marketed farm produce, they do reveal the
substantial difference between cash incomes in Zones 1 and
2 and those in Zone 3, where both the number and value of
cash crops present are more limited.
The sharp difference in the "cash" variable between
adopters and non-adopters of hybrid seed lends strong sup-
port to the hypothesis that income from cash crops is
associated with adoption (Table 29). In each case, the
imputed income from cash crops was very much higher for
adopters than for non-adopters and these differences were
significant except in Zone 1.

Table 18. Mean value of "cash" variable by zone.

Zone 1 Zone 2 Zone 3 Average
Mean 2001 2178 639 1588
deviation 3228 3289 2255 3037
Number 96 93 95

In the probit analysis, "cash" was significantly and
positively related to adoption in the sample taken as a
whole. For the hypothetical "average" farmer, an addition-
al thousand shillings of imputed cash income would in-
crease the probability of adoption from 76 percent to 82
percent (equation 1). In Zone 3, "cash" was significantly
related to adoption at a 0.10 level for a one-tailed test.
A thousand shillings of "cash" would increase the proba-
bility of adoption of an "average" farmer in Zone 3 not
growing famine crops from 21 percent to 31 percent. We
may conclude from the analysis that the presence of cash
crops on the farm, itself a measure of the farmer's involve-
ment in the monetary economy, greatly facilitates the
adoption of the new technology. Marvin Miracle has argued
persuasively that the term "subsistence agriculture," defined
by some given percentage of farm production which is
marketed, is not a particularly useful way of classifying
small-scale farmers.23 More appropriate measures, he
feels, relate to the decision-making environment of the
small farmer, including degree of isolation, level of living,
commitment to agriculture, security of tenure, labor availa-
bility, and other factors. While we may agree that "sub-
sistence" as a term has been used so broadly as to be virtual-
ly meaningless, nevertheless, at the bottom end of the
market-subsistence continuum, the presence of crops grown
specifically for sale is an important influence on farm be-

Adoption and Credit Availability

Like cash crops, availability of credit was hypothesized to
be positively related to adoption of the new technology.
The reasons for this are obvious. As Ronald McKinnon
has written,24

Table 19. Mean value of "cash" variables by adoption.

Zone 1 Zone 2 Zone 3 Average
Adopters 2056 2424 2163 2180
Non-adopters 1244 135 355 386
T-Value 0.49 5.95 2.93 2.79
Level of
significance NS 0.001 0.01 0.01

Table 20. Availability and sources of credit (All answers in

Zone 1 Zone 2 Zone 3 Set 4
Farmers reporting credit
available 27.7 40.9 17.9 63.5
Sources of credit cited:
Kenya Farmers
Association 9 25 0 24
Friends and relatives 9 0 25 0
Agricultural Finance
Corporation 63 63 50 63
Cooperatives 18 12 25 12

Poverty and the inability to borrow to finance discrete increases
in expenditures can be formidable barriers to the adoption of even
the simplest and most productive innovations.
In spite of the high divisibilityy" of hybrid seed itself, the
recommended package of seed, fertilizer, and insecticides
can represent a substantial cash outlay for a farmer with a
small net income. This is no doubt part of the reason why
the use of fertilizer (49 percent) and insecticide (13 per-
cent) lagged substantially behind the use of hybrid seed
(67 percent) in the sample as a whole and particularly in
the small farm areas (Zones 1 and 3). The fact that fertili-
zer is less profitable than hybrid seed is obviously also
In the 1973 survey, a farmers were asked whether or not
credit was available in their area, but for fear of biasing
the responses were not asked directly whether or not they
used outside credit sources themselves. The percentages of
farmers reporting that credit was available, and the sources
of credit reported are shown in Table 20. Interestingly, no
farmer cited either stockists or commercial banks as sources
of credit, testimony to the poorly developed nature of
private lending institutions in the rural areas.
Although we can safely assume that farmers who were
not aware of credit did not avail themselves of it, we cannot
be certain that farmers who felt it was available either
used it or required it. Nevertheless, knowledge of credit
availability was highly significant (at greater than the 0.01
level) in the probit analysis, even when included with
variables such as education, cash, and farm size, with
which it might be expected to be correlated. For the hypo-
thetical "average" farmer in the sample taken as a whole,
for example, knowledge of credit availability increased the
probability of adoption from 77 percent to 96 percent.
When crosstabulated with adoption, knowledge of credit
availability was significantly related to adoption in all zones
except Zone 1 (Table 21).
The peculiar response in Zone 1 is due to the presence
of only four non-adopters and is not significant. Although
it is tempting to draw grandiose conclusions from the
high level of significance in Zone 3, namely, that lack
of credit is a major barrier to adoption, my view is that

Table 21. Availability of credit and adoption (Farmers
reporting credit available (percent)).

Zone 1 Zone 2 Zone 3 Average
Adopters 27.0 45.1 60.0 35.5
Non-adopters 50.0 10.0 10.3 12.4
Chi-square values 0.18 3.20 17.64 4.40
Level of significance NS 0.10 0.001 0.05

what is being measured here is something less. Since
private sources of credit are virtually non-existent and
since government loans for maize were limited (prior to
1974) to farms over 15 acres, lack of credit cannot have
been a formidable barrier to the many small-scale adopters
in Zones 1 and 2. What I suspect is actually being measured
here is the farmer's knowledge of and interest in commercial
maize growing per se. In this case we would certainly
expect adopters to be aware of the available sources of
credit to a degree unmatched by non-adopters. In fact,
GMR credit is given only to those maize farmers who grow
hybrid maize. Some support for this explanation may be
found in Moock's study of maize yields in Vihiga.25 He
found that farmers receiving credit were invariably richer
and better farmers than average and that credit per se was
not essential to the operations of those who received it,
nor was it significantly related in a multivariate analysis
to obtaining higher yields. This, of course, does not say
anything about the possibility that lack of credit is in
fact a very real constraint to farmers at the bottom end of
the scale. It is merely evidence of the fact that farmers
who can get credit generally do not need it.

Adoption and Off-Farm Work

The influence of labor migration on the adoption of farm
practices has long interested social scientists. Rogers re-
ports that innovators are more "cosmopolite" in that they
travel widely beyond the boundaries of their local system.
He records 132 studies (including the classic Ryan and
Gross sutdy of Iowa farmers) which support a relation-
ship between "cosmopoliteness" and innovativeness and
only 42 which do not support such a relationship.26
Miracle points out that the nature of off-farm activities is
Farmers who work elsewhere in agriculture-especially those working
on farms much like their own except that a higher level of produc-
tivity has been achieved-are more likely to adopt innovations
than are those whose wage experience is in mining or manufactur-
ing. The spread of plows and improved varieties of maize among
small-scale African farmers in Rhodesia, Zambia, Kenya and the
Malagasy Republic seems clearly to have been encouraged by the
work experiences on farms operated by white settlers.2

Table 22 Off-farm employment experience.

Zone 1 Zone 2 Zone 3 Zone 4
Presently has job off-farm
("job") 33% 22% 15% 22%
Ever worked off-farm
("work") 54 48 58 38
Worked on large commercial
farm ("large farm") 12 23 9 15

De Wilde points out that the Luos, unlike the Kikuyu, did
not find European farming areas in which they could work
nearby and that, although many Luos have worked on tea
plantations in Kericho, this employment had little relevance
to the agriculture of Central Nyanza. While reporting the
conventional view that "progressive farmers are often those
whose receptivity to change has been enhanced by their
work experience," he also points out wisely that "the
extent to which people return to their farms for the
purpose of earning a living, rather than of simply retiring,
tends to depend on the relative attractiveness of agricul-
tural opportunities in the district, on the one hand, and of
job opportunities outside the district, on the other."28
Off-farm work experience is generally thought to be
positively related to technological change because (1) it
provides an outside source of capital to finance necessary
inputs, (2) it widens the horizons of the farmer and in-
creases his desire for a better standard of living, and/or (3)
it provides the farmer with work experience directly re-
levant to his own farming situation. The percentages of
farmers with present or former off-farm employment is
seen in Table 22. When crosstabulated with adoption,
however, neither "job" or "work" had significant chi-
square values for any zone. In the probit analysis, "job"
was positively but not significantly related to adoption.
Having previous off-farm work experience, however, was
negatively (and significantly) associated with adoption in
both Zone 3 and the sample as a whole. Even more sur-
prising, having worked on a large, commercial farm was
negatively associated with adoption (also at significant
levels in the sample taken as a whole.) This is partly ex-
plained in the case of Zone 3 by the fact that 61 percent of
these farmers had worked on plantations where sisal, sugar
cane, or tea rather than maize was the major crop. In
Zones 1 and 2, however, almost half had worked on
farms where maize itself was the most important crop.
In spite of the intuitive reasoning (and in some cases
historical evidence) for the importance of work exper-
ience, the 1973 hybrid maize study found no statistical
evidence to support a positive relationship between previous
work experience and adoption per se. The reasons for this
are probably twofold. First, farm owners presently away
from their farms have neither the time nor the interest to

devote to farming because farming in general does not pay
better than modern sector employment. Moreover, in
these cases the farm decision-making is often not clearly
allocated between husband and wife; the husband usually
retains the "spending" authority while the wife actually
does the work. In these circumstances it is difficult to
make timely purchases and application of inputs.29 Se-
condly, long-term residence outside the area (such as is
typical of many farmers in Zones 1 and 3 who have work-
ed in the police, railways, and the like) is often in Moock's
terms, "a liability in the acquisition of relevant agricultural
knowledge." Joyce Moock found in Maragoli that farm
heads employed off the farm are less likely to make invest-
ments at home in spite of their higher incomes. She
concluded that "the individual most likely to respond to
local opportunities...is one [like a teacher] who is em-
ployed and who, at the same time, can live at home."30
We can conclude that the farmer's attitude toward farming
and the profitability of farming in a given area are probab-
ly better indicators of progressiveness than work experience
per se. That this is not always the case, however, will be
seen in our analysis of early adoption.

Adoption and Information Variables

In this section we discuss the relationship of such "in-
formation" variables as formal education, extension visits,
attendance at maize demonstrations, and attendance at
Farmers Training Center courses to adoption. It is hypo-
thesized that all of these will be positively associated with
it. Here we can also conveniently discuss age, which we
hypothesized will be negatively associated with adoption,
most likely because of its negative correlation with educa-
tion (r = -0.34) as well as because of greater conservatism
in older, more traditional farmers. In his summary of studies
of early adopters, Rogers found that education (74 percent
of the studies), change agent contact (87 percent of the
studies), and exposure to mass media (69 percent) were
positively related to early adoption, while age had no clear
relationship.31 In his Vihiga study Moock found formal
education, extension visits, and attendance at demonstra-
tions to be positively related to maize yields although only
schooling was significant in all the equations at the 0.05
level. Differences among different types of extension
contact, however, are of interest as a measure of extension
effectiveness. Since one-to-one visits are of necessity
more time consuming and more expensive than group tech-
niques, it is more than of academic interest whether they
'show more concrete results.
The mean values of the "information" variables in the
survey are shown in Table 23. Farmers in Zone 3 on
average are older, have less formal education, and receive
markedly lower levels of extension contact than farmers
in the other zones. Farmers in the large-farm sample, on
the other hand, have considerably more formal education
and much higher levels of extension contact.

Table 23. Age, education, and extension contacts by zone.

Zone Zone Zone Zone Weighted
1 2 3 4 Average
Age (years) 39.3 37.2 44.5 39.8 40.6
Formal education (years) 4.1 3.5 2.9 6.1 3.6
Extension visit in past
year (%) 39.6 38.7 26.9 56.5 35.0
Ever attended maize
demonstration (%) 26.0 30.1 20.0 64.0 24.9
Ever attended Farmers
Training Center course (%) 24.2 26.4 8.1 48.6 19.3

A difference of means test for age and education avera-
ges of adopters produced the results given in Table 24. Non-
adopters were significantly older in Zones 1 and 2 and
in the sample as a whole, but not in Zone 3, whereas
adopters had significantly more formal education in Zone
3 and the sample as a whole but not in Zones 1 and 2. In
all cases, however, the direction of the relationship was as
hypothesized. 2
In the probit analysis, age was not significant in any
equation, either in Zone 3 or in the sample as a whole,
although it always had a negative sign. Its coefficients
were uniformly low (an additional ten years of age de-
creasing the "average" farmer's probability of adoption
by only about 2 percent) and it was left out of subsequent
equations. Formal education was significantly and positi-
vely related to adoption in most equations, though general-
ly not in the presence of "credit." Its coefficient, how-
ever, was generally low. For an "average" farmer in Zone
3, for example, ten years of education (a very large
amount in Kenya) would increase the probability of adop-
tion from roughly 2 percent to only 22 percent. We
may conclude then that age and education, although relat-
ed to adoption in ways consistent with the hypotheses, are
not decisive influences in their own right in the presence
of other, more specifically relevant variables.
Turning to the extension variables, Table 25 shows the
percentage of adopters and non-adopters in each zone
receiving different types of extension contact. All three
forms of extension contact were significantly different for
adopters in Zone 3 but were not in Zones 1 and 2 because

of the small numbers of non-adopters. Nevertheless, the
direction of the relationships was as hypothesized in all
Again, however, a word of caution must be advanced
about these findings and the cause and effect relationship
implied. Other studies (Chapter II) have found that exten-
sion workers are predisposed toward visiting the better
farmers and that actual farm visits are more often related
to veterinary problems or cash crops than to food crops.
Likewise, farmers selected for F.T.C. courses tend to be
better, richer farmers. Since participation fees are usually
charged, in fact, only the wealthier farmers are sometimes
able to attend. Maize demonstrations, on the other hand,
are open to all farmers in a given area and the farmer decides
whether or not he will attend. Since the demonstration
is aimed at imparting a particular message, it is logical to
expect it to have a more direct impact on adoption. The
fact that demonstration attendance was the only one of the
three variables that was significant in more than one zone
may give some support to this thesis. In the probit analy-
sis, all three extension variables were positively related to
adoption in all equations and at significant levels in some
(Tables 15 and 16). Again, however, it is unclear whether
the extension contacts are successful in promoting hybrid
adoption or whether extension agents are merely visiting
the more progressive farmers. Since the better farmers
obviously make more demands on the extension staff and
are probably more responsive to advice, it is not surprising
that they receive the most extension contact.

Adoption of Fertilizers

Efforts to relate fertilizer use per acre of maize to indepen-
dent variables were singularly unsuccessful. In models
containing all three zones only the dummy variable for
Zone 3 showed appreciable levels of statistical significance,
reflecting the differences between Zones 1 and 2 and
Zone 3 manifested in Table 11. Again as in Table 11 farm
size showed appreciable levels of significance.
In general these models explain so small a part of the
variation in fertilizer use that they offer little information
over that contained in Table 11. Because of this they are

Table 24. Age and education variables by adoption (years).

Zone 1 Zone 2 Zone 3 Average
Adopters 40.27 ++ 37.50 +++ 43.14 39.62++
Non-adopters 50.25 46.66 45.59 45.93
Adopters 4.20 3.53 4.40 +++ 4.01+-
Non-adopters 2.75 2.90 2.67 2.69
-+ = significant at 0.05 level for a one-tailed test.
I+ = significant at 0.01 level for a one-tailed test.

Table 25. Adoption and extension contacts.

Zone 1 Zone 2 Zone 3 Average
Extension visit in
past year (%)
Adopters 40.7 39.8 40.0+++ 40.3
Non-adopters 25.0 30.0 23.7 24.3
Attended maize
demonstration (%)
Adopters 26.7 35.0+ 40.0++ 30.4
Non-adopters 25.0 0.0 15.6 15.0
Attended farmers
Training Center
courses (%)
Adopters 25.3 29.5 26.7+++ .26.7++
Non-adopters 0.0 0.0 2.9 2.5

+ = chi-square value significant at 0.10 level. ++ = chi-square value
significant at 0.05 level. +++ = chi-square value significant at 0.01 level.

not reported here. It should be emphasized that the
absence of explanatory power might be a consequence of
inaccurate estimation of the dependent variable pounds of
nutrient per acre. Recall that this point was made earlier
when it was acknowledged that the area of fields is only
loosely known by most of the farmers of the sample.
This reduces the reliability of any figure purporting to
describe on a per acre basis.

Analysis of Early Adoption

In this section we look at the characteristics of "early"
adopters as opposed to those of adopters and non-adopters
ten years after the innovation was first introduced. Con-
centrating on early adopters has been a popular focus in the
diffusion study literature. Such a focus is, in fact, much
more common than studies of "mature" adoption levels
or ceilings for a variety of reasons. First, social scientists
are often interested in deviant behavior, whether progressive
or regressive, precisely because it is different. Second, the
growing realization of the importance of entrepreneurial
behavior in the development process has led us to seek
to identify the innovating or achievement-oriented indivi-
dual. Third, it has often been asserted that by working
through early adopters of innovations we can influence
and even accelerate the adoption process among the general
population. And, fourth, because the year of adoption
gives us a continuous dependent variable, social scientists
can avoid the methodological problems hitherto encounter-
ed with a simple dichotomous or binary dependent variable
such as adoption and non-adoption. (For the same reason,
the percentage of acreage planted to new varieties and an
index of dissimilar innovations are also frequently en-
countered as dependent variables). As we shall discuss
later, this preoccupation with early adoption per se has
some limitations, not the least of which is the emphasis it

puts on the characteristics of the innovator rather than
the innovation. It is useful first, however, to characterize
early (or earlier) adopters and to see whether this character-
ization corresponds to that of the adopter vis a vis the
Rogers33 has characterized early adopters of innova-
tions as having more education, higher social status, larger
units of operation, a commercial rather than a subsistence
orientation, more specialized farm operations, and a favora-
ble attitude towards credit. "Laggards," on the other
hand, have been characterized as being "localite" in their
outlook, past-oriented, suspicious, and "alineated." In
the bivariate analysis which follows, we have defined
"early" adopters arbitrarily as those who first used hybrid
seed more than one standard deviation in time before the
mean year of adoption. In a normal distribution, this would
be approximately the first 16 percent of adopters. The
mean year of adoption and standard deviation by zones
are reported in Table 26.
In the multivariate analysis, the same variables used for
the analysis of adoption in 1973 have been included,
except that the dependent variable is the year of adoption
subtracted from 1974 (so that 1973 equals 1,1972 equals
2, and so forth up to 1963, which equals 11). One qualify-
ing point needs to be made here. Because we asked the
farmer in what year did he first plant hybrid seed, we may
be violating our definition of adoption as "use of hybrid on

Table 26. Mean year of adoption by zone.

Zone 1 Zone 2 Zone 3 Average
Mean year of adoption 1968.8 1968.3 1970.3 1968.9
Standard deviation 2.2 2.9 2.2 2.5
"Early" adoption (before) 1967 1966 1968 1967

at least 50 percent of maize acreage." It is logical to
think that farmers first tried hybrid seed on less than
their total maize acreage, although in 1973 only a tiny
proportion of farmers were planting both hybrid and local
varieties on the same farm. We also know that some farmers
in Zone 3 tried hybrid seed in the past but were not using it
in 1973. Since we do not have any way of knowing what
percentage of maize acreage was planted to hybrids in the
various first years of use, our definition of "adoption" in
this section is not strictly comparable to that used when
discussing adoption and non-adoption as of 1973. We have
no a priori reason to believe, however, that farmers who
first used hybrid seed are significantly different from those
who first used hybrid seed on 50 percent or more of their
maize acreage.
Table 27 gives the characteristics of "early" adopters
(those adopting one standard deviation or more before
the mean year of adoption), as compared with later adop-
ters in the survey. Non-adopters have not been included in
the analysis because of our belief that the difference be-
tween non-adoption and adoption is of a different magni-
tude, if not generically different, from the numerical differ-
ence which would separate a non-adopter from a 1973

Table 27. Crosstabulation

of early adoption and selected

Zone 1 Zone 2 Zone 3
Completed 4 years of primary
education (%)
Early adopters 36.4 40.0 80.0
Later adopters 45.2 39.3 34.6
Grows famine crops (%)
Early adopters 4.5 0 15.4
Later adopters 6.5 1.8 40.0
Joboff-farm (%)
Early adopters 38.1 26.1 25.0
Later adopters 31.7 24.1 20.0
Previous work experience (%)
Early adopters 57.1 50.0 100
Later adopters 48.4 48.1 69.2
Worked on large farm (%)
Early adopters 26.7 58.3 0
Later adopters 20.0 42.3 21.1
Extension visit (%)
Early adopters 54.5 44.0 50.0
Later adopters 41.9 41.1 50.0
Attended maize demonstration (%)
Early adopters 40.9 30.4 20.0
Later adopters 25.8 32.7 20.8
Attended farmers training center (%)
Early adopters 36.4 30.4 40.0
Later adopters 24.1 30.8 16.7
Farm greater than 10 acres (%i
Early adopters 54.5+ 70.8 20.0
Later adopters 29.5 49.1 15.4
"Cash" variable greater than
500 shillings (%M
Early adopters 68.2 64.0 40.0
Later adopters 54.8 64.3 23.1

+ = chi-square value significant at 0.10 level. ++ = chi-square value
significant at 0.05 level.

adopter in the dependent variable. Although these simple
crosstabulations are almost all consistent with the hypothe-
sized relationships, only farm size had a chi-square value
significant at the 0.05 level. This is due primarily to the
limited sample size of early adopters in each zone. In
the sample taken as a whole, all the relationships without
exception are in the direction anticipated. "Early" adop-
ters had larger farms, more cash crop income, more ex-
tension contact, more work experience, were less likely
to grow famine crops, and were more likely to have a
primary education than later adopters. Interestingly, al-
though work experience on a large farm was negatively
associated with final levels of adoption, it was positively
associated with early adoption (especially in Zone 2, where
hybrids were first introduced). This is consistent with the
view that hybrids were first used by many of those African
farmers who had seen them being used on European mixed
farms. The negative relation of large-farm work experience
to early adoption in Zone 3 may in turn be due to the
irrelevance of the plantation-type off-farm work exper-
ience (on sugar, tea, and sisal estates) of most of the farmers
from that region to their own farming situation.
In addition to the bivariate analysis of early adoption
and other factors, multiple regression analysis of early
adoption was carried out with year of adoption as the
dependent variable. Table 28 shows the results.
Farm size, agroclimatic zone, and imputed cash income
were significantly related to adoption at the 0.10 level.
In the absence of zone, farm size was significantly related
to adoption at the 0.01 level. None of the other variables
(extension contacts, work experience, etc.) were significant
when included with zone and/or farm size. When the
regression was limited to the three variables farm size, zone,
and cash, the coefficients were essentially unchanged and
all three were significant at the 0.05 level.
Regression analysis was also carried out separately on
each of the zones with interesting results. Table 29 shows
the results for Zone 1.
Once again, farm size is the most significant variable,
risk in negatively associated with early adoption, and age
and education, while positive, have very small coefficients
(ten years of education increasing the time of adoption
by less than a year). Cash, interestingly, was not significant
in the presence of farm size in this Zone. The only other
independent variable which proved significant (at a 0.05
level) was attendance at a maize demonstration, which
increased the earliness of adoption by more than a year
(coefficient 1.064). Extension visits and attendance at
Farmers Training Centers, while positive in sign, were not
significant and had much smaller coefficients.
In Zone 2 only two independent variables were significant-
ly related to adoption, both at the 0.10 level. These were
farm size (coefficient 0.04) and work experience on a
large, commercial farm (1.30). The latter is especially
interesting since it supports the view, discussed earlier,
that hybrids were first used by those who had seen them

on European farms. This is especially likely to be the
case in Zone 2, where the former European farms are
located and where both the mix of crops and general
ecological conditions would be similar between European
and African farms. The seemingly contradictory fact that
large-farm work experience was negatively related to final
levels of adoption in the probit analysis is due perhaps to
the fact that (a) the large-farm work experience of farmers
in Zone 3 was not closely related to their home farm
conditions and (b) large-farm work experience may have,
in many cases, actually proceeded the introduction of
hybrid maize. When the sample was restricted, then, to
adopters and to farmers in Zone 2, it makes more sense
that such experience would be positively related to early
adoption. (When a regression on early adoption was run
on the very large-scale farmers themselves [Set 4], work
experience on a large commercial farm was the only variable
that proved significant).
In the analysis of early adoption in Zone 3, the only
variable which proved significant at a 0.01 level was im-
puted cash income (coefficient 0.44, that is to say, a
thousand shillings of cash crop income would increase the
earliness of adoption by almost a half year). Farm sizes
were fairly uniform in this zone and were not, therefore,
a particularly good measure of variations in income or
ability to absorb risk. Moreover, since this sample included
adopters only, farmers growing famine crops were virtually
entirely excluded and "risk" as such was not as important a
factor in the earliness of adoption as it had been in deter-
mining adoption per se. The importance of cash crops,
whether as a protection against risk or simply as a source
of money with which to buy the necessary inputs, is
strongly reinforced by this result.
The only other variable besides imputed cash income
which proved significant in Zone 3 (though at only a 0.10
level) was large-farm work experience, which had a coef-
ficient of-1.61, that is such experience on average delay-
ed adoption by more than a year and a half. Again, this
may be due to the irrelevance of the plantation-type ex-
perience of most of the Zone 3 farmers to their own
farming conditions or to the fact that this experience took
place before hybrids were introduced.
In sum then, there are at least two noticeable dif-
ferences between factors affecting early adoption and
factors affecting adoption per se. One of these-work
experience on a large commercial farm-has already been
discussed and, on closer examination, has been found to
relate to the particular relevance of the work experience
to the farmer's own farm situation (strongly positive in
Zone 2 and strongly negative in Zone 3). The other
interesting difference is the role of farm size, which was
not significant in the analysis of adoption per se, but
proved to be the most significant single variable influencing
early adoption, even controlling for differences in farm
size between zones. In one or more equations, farm size
was significant at the 0.01 level in both Zones 1 and 2
and in the sample as a whole. Only in Zone 3, where farm

Table 28. Regression coefficients, standard errors, F-values,
and levels of significance of variables related to early
adoption of hybrid maize in western Kenya.

Independent Regression Standard F-Value Level of
Variables Coefficients Error (T ) Significance
Farm size
(acres) 0.038 0.024 2.39 0.10
zone 0.849 0.601 1.99 0.10
Cash (000
shillings) 0.120 0.080 2.00 0.10
Age (years) 0.019 0.019 1.02 NS
Education (years) 0.062 0.083 0.55 NS
Risk -0.204 0.709 0.08 NS
Constant 2.970

sizes are quite uniform, was it not the most important
explanatory variable and there another possible proxy for
general wealth (cash income) was the most important
Without stretching the analysis more than it will bear,
this differing significance of farm size may have an im-
portant (and, I think, hopeful) implication about the
adoption process. For if genuine economies of scale
were involved in determining which farmers adopt the new
technology, one would expect that farm size, as an inde-
pendent variable, would retain its importance with the
passage of time. Instead, we find that, although farm size
was significant in determining which farmers adopt first,
it was not significant in determining "mature" levels of
adoption. This supports the hypothesis that the ability to
absorb or tolerate risk is important in the adoption process.
Because they are risking a smaller portion of their total
assets on a given investment, the larger farmers (or those
with significant cash incomes) are able to try out the new
technology without endangering their overall security.
Where, as in Zones 1 and 2, the new technology proves not
to involve significant risks, it is then rapidly adopted by
farmers of all categories. Where, as in Zone 3, significant
doubts about the efficacy of the new technology remain,
risk perception and ability to absorb risk (as measured by

Table 29. Regression coefficients, standard errors, F-values,
and levels of significance of variables related to early
adoption of hybrid maize in Zone 1.

Independent Regression Standard F-yalue Level of
variables Coefficients Error (T ) Significance
Farm size
(acres) 0.068 0.027 6.40 0.01
Risk -2.050 1.295 2.50 0.10
Age (years) 0.037 0.025 2.17 0.10
(years) 0.096 0.092 1.08 NS
Constant 2.589


imputed cash income) continue to be important in de-
termining who adopts. Although work experience (in
Zone 2) and attendance at maize demonstrations (in
Zone 1) were both important, as we might expect in in-
fluencing early adoption, their importance declined as the
new technology became widely known. Other "informa-
tional" variables (knowledge of credit availability, exten-
sion visits, and attendance at Farmers Training Centers),
although positively related to adoption, may not be entire-
ly causally related for institutional reasons we have already
These conclusions are consistent with the historical
process of hybrid maize diffusion as we know it. The
new technology as introduced first on European farms in
Zone 2 and spread first to African farms in the same area,
presumably because many of their owners worked on the
European farms. Intensive maize demonstration campaigns
by the Ministry of Agriculture, especially in such Zone 1
districts as Kakamega, Kericho; and Kisii, quickly spread
the new varieties to these areas. Adoption in the lake-
shore districts of Zone 3 lagged behind because the new
technology was riskier, appeared to have a smaller relative
advantage over local varieties, because employment exper-
ience on large mixed farms (as opposed to plantations) was
less, and because levels of all types of extension contact
were lower. In the face of greater climatic insecurity,
drought-resistant famine crops continued to hold an im-
portant place in the farmers' mix of staple food crops.

The Importance of Early Adoption

Having looked both at early adoption and at "mature"
levels of adoption ten years after hybrid maize technology
was first introduced, we can now ask what important
effects early (as opposed to later) adoption may have had.
Although we are inevitably interested in the nature of the
innovators who adopted the new technology first, it can
be argued in this case at least, that early adoption patterns
are ultimately less important than "mature" adoption pat-
terns for those concerned about economic progress and
income distribution.3 5
Every society may be presumed to have some propor-
tion of relatively more innovative individuals, who, for
reasons of achievement motivation, intellectual satisfaction,
scientific predilection, or simply better access to new
information are willing to try out new ideas and technolo-
gies. The absolute number (or proportion) of these indivi-
duals may be of some importance depending on externali-
ties involved in the particular innovation, but in most
cases an adequate number will exist. It may even be
fortunate that a majority of society's members are not in
this category since innovations often prove unprofitable.
By bearing the risks of experimentation for a larger society,
the innovator may prevent widespread losses or may direct
attention to improvements in the innovation necessary

for its wider acceptance. Because we too often tend to
think of innovations as being uniformly advantageous,
this positive aspect of differential innovation is often
What is much more important than who adopts an
innovation first is who is ultimately unable to adopt it
and why. Unless early adoption of an innovation by a
minority precludes subsequent adoption by others, final
or mature ceilings of adoption will affect levels of welfare
and income distribution more profoundly than relatively
modest differences in the time of adoption. There are
obviously cases in which early adoption of an innovation
by one segment of the population results in long-term dif-
ferentials in income distribution or political power. Those
areas of Kenya which had planted coffee before the intro-
duction of coffee quotas, for example, have achieved a
significant long-term income advantage over the other
areas. We have already mentioned an Ethiopian case in
which adoption of high yielding wheat varieties has great-
ly increased land values resulting in the expulsion of an
estimated 5,000 tenant farmer families from the land by
landlords.36 There is also evidence from parts of Asia
that, where economies of scale in wheat production are
joined with an active land market, early adopters of new
grain varieties have used their profits to buy out smaller
landowners, thereby worsening income distribution.
Fortunately, however, for a variety of reasons, this
does not appear to be the case with hybrid maize in
Kenya. First, it is questionable whether there are signifi-
cant economies of scale at work in Kenyan maize produc-
tion. Although mechanized plowing permits earlier plant-
ing and consequently higher yields, this advantage is at
least partly offset by more intensive weeding and more
careful harvesting and drying on smaller holdings. Second-
ly, the technology itself is highly divisible and not parti-
cularly complex. Thirdly, sharecropping and tenant farm-
ing are virtually unknown in Kenya and tenure is over-
whelmingly (de facto if not de jure) under individual
title. Fourthly, the government has maintained a guarante-
ed minimum price for maize for any given year, which has
generally (though not always) protected the producer from
sudden drops in the market which would undercut continu-
ed expansion of the new varieties. Fifthly, although the
buying and selling of land is increasing in frequency in
Kenya, this is primarily in the formerly Scheduled (Euro-
pean-owned) areas. Land in the traditional African areas
does not change hands easily and it is unlikely that those
land purchases which are taking place today are being made
by early hybrid adopters from their profits on hybrid.
maize. Adoption in the high rainfall maize growing areas
has been exceedingly rapid (faster than among American
farmers in the 1930s) with the result that the advantages of
early adoption per se have been relatively short lived. In
fact, early adoption by some farmers has facilitated rather
than hindered subsequent adoption by others, both by
reducing the perceived risks of the new technology in the
high rainfall areas and by stimulating improved availability

of seeds and fertilizers. Although there were local problems
of oversupply in 1972 and 1973, this is likely to be a
temporary problem until export handling and storage facili-
ties are improved. Given periodic droughts in Africa and
elsewhere and rapid population growth, overproduction of
maize'is not a serious threat to continued hybrid expansion.
We can conclude from this discussion that differences in
time of adoption per se have not significantly worsened
income distribution in the hybrid growing areas of western
Kenya. If one wants to generalize about the effects of the
diffusion of innovations on income distribution, a useful
distinction can be made between those innovations which
we might call preclusivee" whose adoption by some indivi-
duals precludes either directly (through monopolies or
quotas) or indirectly (through market effects) subsequent
adoption by others and those innovations which we might
call facilitativee" whose adoption by some individuals faci-
litates adoption by others. If coffee planting under quota
conditions is an example of the former, hybrid maize is
fortunately an example of the latter. The very rapidity
of the adoption process in the high rainfall of western
Kenya is testimony to the fact that early adoption by some
has not prevented the benefits of hybrid maize from being
shared by many.
Far more serious problems are presented by those who
have not adopted hybrid technology ten years after its
introduction. Without implying that further increases in
hybrid use will not still take place, real doubts must never-
theless be raised about the suitability of present hybrid
technology for the lakeshore areas. Whether for reasons
of lower relative yield advantage, lower reliability, inabili-
ty to finance cash purchases, lower levels of services, or

movie likely, a combination of all these, a substantial number
of farmers in Zone 3 have tried but later discarded hybrid
seed. Given this fact and the generally low level of hybrid
use (15 percent) in that zone, this must be regarded as a
problem of conscious non-adoption rather than mere "lag-
gardness." As discussed earlier, further research and a-
daptation and possible introduction of composite or syn-
thetic varieties may be required to overcome this problem.
Special action may also be needed to reach the remain-
ing non-adopters in the high rainfall areas, though as we
have seen in Chapter II, achieving this through government
programs is not always easy.
In the end, the "non-adoption" problem is serious be-
cause it is essentially ecological and regional. As Griliches
found in the U.S., hybrid corn was never fully adopted in
those western fringes of the high plains where corn could
only reliably be grown under irrigation. This might also
prove the case in the more marginal rainfall areas of Kenya
where intrinsically drought-resistant grains will continue
to dominate. If high-yielding maize varieties are ultimate-
ly restricted to the high rainfall zones, this will obviously
worsen relative income distribution on a regional basis.
Even so, absolute levels of welfare will not be worsened by
hybrids since the non-maize growers would still benefit as
consumers from a lower maize price than would obtain
in the absence of hybrid expansion. But because ethnic
and ecological boundaries are closely related in Kenya, a
worsening income distribution on a regional basis is a
dangerous prospect. Among the dualities of developed
and developing nations, urban and rural sectors, and the
like, the duality between high rainfall and low rainfall
areas may yet prove the most difficult to solve.


1. James Tobin, "The Application of Multivariate Probit Analysis
to Economic Survey Data," Cowles Foundation Discussion, No. 1
1955. See also, D.J. Finney, Probit Analysis (Cambridge, 1971);
D.R. Cox, The Analysis of Binary Data (London, 1970); Arthur
S. Goldberger, Econometric Theory (New York, 1964), pp. 248-
251, and especially an excellent recent paper on probit analysis by
Richard D. McKelvey and Willian Zavonia, "A Statistical Model
for the Analysis of Ordinal Level Dependent Variables," University
of Rochester, May 1973.

2. Richard McKelvey, "N-chotomous Multivariate Probit Program,"
Rochester University, as modified by Jan Juran, Princeton Univer-
sity Computer Center, 1974.

3. McKelvey and Zavonia, "Analysis of Ordinal Level Dependent
Variables," pp. 29-31. The authors have also developed a pseudo-
R2 or R2 to measure goodness of fit. R can be interpreted much
the same as the corresponding R2 for regression analysis, but should
be used with some caution because its sampling distribution is not
known (p. 34). The N-chotomous Multivariate Probit Program
also calculates the percentage of cases predicted correctly by
substituting the coefficients obtained back into the equations with
the actual values for each case and then solving each case and
comparing the predicted with the actual value.

4. F.A.O. Fertilizer Programme, "Summary of West Kenya
Results," 1969, 1970, 1971, and K. Zschernitz and A.J. Ikalo,
"F.A.O. Fertilizer Programme, Report No. 4. Planting season
1972. Average Yield Data," 1972. [I am grateful to Mr. A.J. Okalo
of the Ministry of Agriculture, Nairobi, and Dr. A.Y. Allan of the
National Agricultural Research Station, Kitale, for these data.]

5. Indicative of the problems in Zone 3 is the fact that the low
incidence of hybrid usage was not fully appreciated before the
1973 survey. Following the preliminary results of the survey,
the extension service and the Kenya Seed Company discussed
revising their promotion strategies, the latter reportedly assigning
a key sales agent to work in the lakeside area. A.Y. Allan, personal
communication, February 11, 1974.

6. Elsepth Huxley, A New Earth (New York, 1960).

7. Hugh Fearn, An African Economy (London, 1961), pp. 116-
117. John C. DeWilde, Experiences with Agricultural Development
in Tropical Africa, Vol. II (Baltimore, 1967), p. 121. Hans Ruthen-
berg, African Agricultural Production Development Policy in Kenya
1952-1963 (Berlin, 1966), pp. 30-31.

8. Margaret Jean Hay, "Economic Change in Luoland: Kowe,

1890-1945" (Ph.D. thesis, University of Wisconsin, 1972).

9. Ibid., p. 3.

10. Ibid., p. 21.

11. Margaret Jean Hay, "Economic Change in Luoland: Kowe
1890-1945" (Ph.D. thesis, University of Wisconsin, 1972) pp. 134-

12. Marvin P. Miracle, Maize in Tropical Africa (Madison, 1966).
p. 140 cited by Margaret Jean Hay, "Economic Change in Luoland:
Kowe, 1890-1945" Ph.D. Dissertation, University of Wisconsin,

13. Margaret Jean Hay, "Economic Change in Luoland: Kowe,
1890-1945" Ph.D. Dissertation, University of Wisconsin, 1972, p.

14. John Harris, "Some Theory of Agricultural Innovation,"
East African Agricultural Economics Society Conference Paper,
June 1969, p. 8, talking of Nyanza (Zone 3), says that "hybrid"
maize is more sensitive to rainfall variation than are traditional
maize varieties" and also that early planting, which is important for
hybrid maize, "involves not only greater physical effort but also
requires abandonment of fishing in a period when catches are
highest." I have not been able to confirm this point. Hay report-
ed (personal communication, July 1974), that "when there were
problems of the amount or the distribution of rainfall, local maize
fared much better than the hybrid, and people were very quick to
point that out. The hybrid maize was also much more susceptible
to bird damage."

15. Another proxy variable for risk was tried and discarded in the
analysis, namely the actual acreage of famine crops planted. In
addition to the problem of measuring acreages (especially when
many of the same crops are interplanted with each other) it was
felt that farmers do not in fact plant proportionately more "famine"
crops the more worried they may be about getting a maize crop
but, if they are worried, plant a fixed amount as a precaution. This
in fact proved to be the case, e.g., farmers planting cassava planted
a mean of 0.68 acres with a standard deviation of 0.47.

16. Walter P. Falcon, "The Green Revolution: Generations of
problems," American Journal of Agricultural Economics (December
1970: 598-712. Carl Gotsch, "Technology, Prices, and Incomes
in West Pakistan Agriculture: Some Observations of the Green
Revolution," mimeo (January 1972).

17. James Q. Harrison, "Agricultural Modernization and Income
Distribution: An Economic Analysis of the Impact of New Seed
Varieties on the Crop Production of Large and Small Farms in
India" (Ph.D. thesis, Princeton University, 1972).

18. Max K. Lowdermilk, "Diffusion of Dwarf Wheat Production
Technology in Pakistan's Punjab (Ph.D. thesis, Cornell University
1972),p. 185.

19. S.R. Sen, "Agricultural Development and Poverty," in Some
Issues Emerging from Recent Breakthroughs in Food Production,
edited, by Kenneth L. Turk (Binghornton, 1971), p. 480. Quoted by
Max K. Lowdermilk, "Diffusion of Dwarf Wheat Production Tech-
nology in Pakistan's Punjab" (Ph.D. Thesis, Cornel University,
1972) p. 188.

20. James Copp, "Towards Generalization in Farm Practice Re-
search," Rural Sociology 23 (June 1958): 107. Quoted by Max


K. Lowdermilk, "Diffusion of Dwarf Wheat Production Technology
in Pakistan's Punjab" (Ph.D. thesis, Cornell University, 1972),
p. 187.

21. Moock found for the range of farm sizes in his Vihiga Study
that size was not significantly related to yields. These were all
very small farms, however, the largest having only six acres of
maize. Peter R. Moock, "Managerial Ability in Small-Farm Pro-
duction: An Analysis of Maize Yields in the Vihiga Division of
Kenya," (Ph.D. dissertation, Columbia University, 1973), p. 194.

22. I am grateful to the Farm Planning Unit of the Kenya Ministry
of Agriculture for these figures. They are, of course, only rough
estimates; they assume average yields, and crops such as tea and
coffee are taken at maturity levels of output, whereas in fact much
of the acreage would not yet be in full production. The gross margin
figures used were (in shillings per hectare): Bananas, 89/-; Sugar,
225/-; Wheat, 269/-; Tea, 1330/-; Coffee, 910/-; Cotton, 210/-;
Pyrethrum, 907/-.

23. Marvin P. Miracle, "Subsistence Agriculture: Analytical Pro-
blems and Alternative Concepts." American Journal of Agricultural
Economics 50 (May 1968), pp. 292-310.

24. Ronald I. McKinnon, Money and Capital in Economic Develop-
ment (Washington, 1973), p. 12.

25. Peter R. Moock, "Managerial Ability in Small-Farm Production:
An Analysis of Maize Yields in the Vihiga Division of Kenya,"
(Ph.D. dissertation, Columbia University, 1973), p. 199.

26. Everett M. Rogers, Communication of Innovations (New
York, 1971), p. 369.

27. Miracle, "Subsistence Agriculture", p. 304.

28. John C. deWilde, Experiences with Agricultural Development in
Tropical Africa, Vol. II (Baltimore, 1967), pp. 128-129.

29. This should not be taken to mean that female farm managers
are less able than their husbands, only that they are often not given
adequate authority to make the spending decisions necessary to
their work. Sex of farm manager was crosstabulated with adoption
and no significant differences in adoption were discovered. Exten-
sion agencies would do well to recognize the importance of female
farm managers in their efforts to bring about agricultural change.

30. Joyce Moock, quoted by Peter R. Moock, "Managerial Ability
in Small-Farm Production: An Analysis of Maize Yields in the
Vihiga Division of Kenya" (Ph.D. Dissertation, Columbia University
1973) p. 121.

31. Everett M. Rogers, Communication of Innovations (New
York, 1968) p. 352.

32. To test the proposition that what is important in education as
it relates to farming is not the total number of years but the acquisi-
tion of basic functional literacy, the education variable was recorded
as a dummy variable, whether or not the farmer had four years of
formal schooling, the level used by UNESCO and some other
agencies as a basic equivalent of literacy. When crosstabulated with
adoption, this dummy variable gave the same results as education
treated as a continuous variable.

33. Everett M. Rogers, Communication of Innovations, (New
York, 1968) p. 195.

34. "Cash" is in some senses a measure of the quality rather than
the quantity of farm land since high value cash crops cannot be
grown in all locations.

35. I am grateful to Professor Don Kanel of the University of


This thesis has sought to look closely at the diffusion of
hybrid maize technology in western Kenya as a case study
of the spread of new agricultural technology among small-
scale African farmers. Its purposes have been (a) to de-
termine the pattern of that diffusion over the period 1964-
1973 and, particularly, as of June 1973; (b) to determine
the influence of environmental factors (such as agroclimatic
zone and risk perception) on adoption as well as the
influence of such other factors as farm size, cash crops, off-
farm income, work experience, credit availability, formal
education, and various types of extension contact; (c) to
characterize early, later, ard non-adopters of the new
maize technology in terms of these socio-economic factors;
(d) to determine which portions of the "package" of re-
commended practices associated with hybrid maize are
adopted 'and to what degree; (e) to describe the maize
industry in Kenya and to place the pattern of adoption in
the context of Kenya's rural economic services; (f) to
consider the costs and benefits to Kenya of developing the
new maize technology; and (g) to draw conclusions as to
why hybrid maize has been relatively successful in Kenya,
and how this development can be extended and improved.
In this chapter we will review our findings, first, in the
light of the 1973 Kitale/CIMMYT Maize Survey results,
and secondly, through our analysis of the Kenya maize
industry. Finally, we will comment briefly on the implica-
tions of this study for an overall view of small-scale farmers
and their responsiveness to change.

Summary of Main Survey Findings

In a period of only ten years, the use of hybrid maize
varieties in Kenya has increased from an initial 400 acres
to at least 800,000 acres, accounting for perhaps half
the country's total production. In the high and medium
potential areas west of the Rift Valley about two-thirds of
small-scale African farmers had adopted hybrid seed by
1973, including up to 90 percent of the farmers in the
higher altitude and rainfall zones. In the lower altitude,

Wisconsin for suggesting some of the ideas which follow.

36. For a recent discussion of the CADU case, see John M. Cohen
"Rural Change in Ethiopia: The Chilalo Agricultural Development
Unit," Economic Development and Cultural Change 22 (July
1974): 580-614.

lower rainfall areas around Lake Victoria, however, adop-
tion levels have been much lower, with twice as many farm-
ers sampled (32.5 percent) reporting ever having planted
hybrid maize as were planting it in 1973 (15.8 percent).
Aggcljratic zone, as described in Chapter I, was found
to be by far the most important variable in explaining
adoption within the farmer sample taken as a whole. Lo-
cation of the farm in a high rather than lower rainfall/altitu-
de zones increased the likelihood of adoption of an other-
wise "average" farmer from 18 to 87 percent.
The pattern of adoption over time, shown in Map 4 and
Figure 7, indicates that adoption began first in the large-
scale farming areas closest to the National Agricultural
Research Station at Kitale, but spread rapidly to the
densely populated high rainfall districts of Kakamega and
Kisii and then to the somewhat less densely populated but
also high rainfall districts of Kericho and Nandi. Evidence
from records of seed sales and from other, earlier farm-
level surveys in the region support the picture obtained
through farmers' recollections that the difffusion process
proceeded quite steadily, at a rate somewhat more rapid
than hybrid corn was adopted by American farmers thirty
years earlier. Within agroclimatic zones, farmers of all
ethnic groups and both sexes had adopted the new varieties
to a comparable degree.
The effect of farm size on adoption has changed marked-
ly over time. In the multivariate analysis, farm size was not
significantly related to adoption in 1973 in any of the
agroclimatic zones or in the sample considered as a whole.
In the bivariate analysis, farm size was also not significantly
related to adoption in any of the agroclimatic zones, al-
though-because of the larger farm sizes in Zone 2-it was
related in the sample taken as a whole. Size of farm, how-
ever, was significantly related to the earliness of adoption
in Zones 1 and 2 and in the entire sample, although the size
of the coefficients was not extremely large (ten acres in-
creasing the earliness of adoption by between 0.4 and 0.7
years). This relationship was though to be a function either
of greater capacity on the part of larger farmers to take
risks, and/or on the preferential treatment that large-scale

farmers receive from the extension services. Because of the
nature of the technology and the security of individual
land tenure in Kenya, however, early adoption by larger
farmers was probably more facilitative than preclusive of
subsequent adoption by smaller farmers.
Two other variables closely related to environmental
conditions were found to be significantly related to hybrid
maize adoption. The presence of drought-resistant crops in
the farmer's crop mix, used as a proxy variable for risk
perception, was strongly and negatively related to hybrid
maize adoption in both the multivariate and bivariate
analyses. Imputed income from cash crops, a rough measu-
re of the quality as well as quantity of the farmer's land,
was positively related to both earliness of adoption and to
adoption per se in 1973. The imputed value of cash crops
was particularly significant in the zone with the highest
risk perception, indicating perhaps that ability to with-
stand risk is an important determinant of adoption where
risk itself is perceived to be great. The presence of an off-
farm source of income, another possible measure of ability
to absorb risk, was positively but not significantly related
to adoption.
Other characteristics of the farmer and his farm environ-
ment which were found to be positively and significantly
related to adoption included formal education, knowledge
of credit, availability, and (in most cases) extension visits,
attendance at Farmers Training Center courses, and atten-
dance at maize demonstrations. The latter variable was
significantly related to early adoption in Zone 1, where
the greatest number of maize demonstrations took place in
the mid-1960s. Distance to a source of inputs had to be
rejected as an explanatory variable because very few non-
adopters were able to give the distance correctly. How-
ever, average distance to a source of inputs were significantly
greater in the zone with the lowest overall level of adop-
tion. Previous work experience off the farm proved to be
negatively and significantly related to adoption per se in
most zones and in the sample taken as a whole. However,
work experience on a large farm was positively and signifi-
cantly related to early adoption in Zone 2, the large-scale
farming area. Those with such experience adopted on
average 1.3 years earlier than others in their area, other
factors being equal. Farmers with large farm work exper-
rience in Zone 3, who had worked mainly on sisal, sugar,
and tea plantations, were, ceteris paribus, 1.6 years later in
adopting hybrid maize.
In addition to looking at the adoption of hybrid maize
seed, other components of the new technology were also
investigated. Adoption of related practices ranged from
4 percent to 100 percent depending on the practice and
zone involved. No clear differences were perceived between
adoption of cultural practices and the use of physical
inputs. In order of acceptance (in the weighted average of
all three zones) the most widely accepted recommended
practices were: planting in rows (75 percent), use of hybrid
seed (67 percent), weeding more than once (63 percent),

planting pure stands (57 percent), use of commercial fertili-
zer (49 percent), use of insecticide on stored maize (46
percent), thinning of plants (45 percent), planting "before
the rains" (23 percent), using manure (23 percent), and
using insecticide in the field (13 percent). Percentages varied
considerably, however, by agroclimatic zone, and the de-
finitions of what constituted "acceptance" of a recom-
mendation of course varied, as did the probable importance
of such inputs as insecticide and fertilizer in a given area.
In spite of the wide variation in levels of adoption of
individual recommendations, hybrid seed adopters had
significantly much higher levels of adoption of related
practices than non-adopters had. The only exception to
this principle was the thinning of young maize plants,
which was more common among non-adopters. The stren-
gth of the relationships between hybrid seed adoption and
the adoption of other practices, indicates that the "package"
approach of presenting hybrid as a new crop requiring new
practices has been relatively successful. The persistence of
such practices as interplanting and later planting, however,
indicates that closer study of farmers' labor patterns, food
preferences, and risk aversion behavior is required. Because
of the importance of early planting to obtaining higher
yields, and the farmers' strong aversion to planting in
advance of the rains, more work on early maturing varie-
ties and on small farm mechanization is needed.

Why Have Hybrids Been Successful in Kenya?

In describing the Kenyan maize industry and its history,
we have tried to discover what factors have accounted for
its relative success. It is of course difficult to say, on the
basis of one case, which of these factors are peculiar to
Kenya and which are generalizable to situations elsewhere.
It is hoped that readers familiar with circumstances else-
where will be able to discern for themselves which factors
are likely to be relevant to a particular situation. If
anything, this study shows how very site-specific agricultur-
al technology is, even within one region of one (admittedly
very. diverse) country. For this reason, I have refrained
from making broad generalizations about agricultural tech-
nology in developing countries. In this section, however, it
may be appropriate to speculate on some of the generaliza-
ble features of the Kenyan experience with hybrid maize.
To begin with, Kenya is fortunate in possessing certain
highland areas with good soils and reasonably reliable
rainfall. Although these high potential areas occupy a
very small part of the country's total area (7 percent in
one estimate), they are nevertheless large enough to sup-
port a thriving agricultural industry (Chapter I). Similar,
if not directly comparable, areas exist in most of the
countries of the Eastern African plateau, including Ethio-
pia, Ugada, Tanzania, Zambia, and Malawi, most of which,
incidentally, import hybrid maize seed from Kenya. These
high potential areas were largely responsible for the colonial

interest in Kenya, if not initially, at least once the Uganda
Railway was completed in the early 1900s. That interest,
in turn, was responsible for the construction of the rudi-
ments of an agricultural industry-including a transport
network, crop research, marketing boards, guaranteed pri-
ces, and an extension service-which have been greatly in-
creased and improved since Kenya's Independence in 1963.
The development of Kenya Flat White maize, improved
Pokot Rhodes grass, selected Kenya Boran Cattle, and
other indigenous products represents the kind of basic
agricultural research and development which can only take
place over time. Perhaps more important, the agricultural
success of the European farmers made farming appear profi-
table as an occupation worthy of investment and research.
Having based its independence movement on the rights of
the Kenyan people to the country's high potential land, it
would have been surprising if the Kenya government had
not made the continued development of its farming re-
sources a major priority. The political power of African
farmers, both large-scale and small-scale, ensures that agri-
culture continues to receive priority attention at the high-
est levels of government. This historical, psychological, and
political experience is perhaps the most difficult part of
the Kenyan case to duplicate elsewhere.
Secondly, Kenya possesses all or most of the com-
ponents deemed necessary to the successful development
of new cereal varieties. As Hill and Hardin have written,
"the principal barrier to increased agricultural production
in less developed countries is not the necessity of changing
farmers' traditions, attitudes and customs but a problem of
developing production technology capable of substantially
increasing yields and making it possible and profitable for
farmers to adopt it.... Adapted, high-yielding production tech-
nology; ready access to critical off-farm inputs; and product/
factor price relationships at the farm...are critical ingre-
dients in triggering an agricultural revolution."' That
Kenya possesses the basic production technology for the
high potential areas is beyond doubt. We have reviewed
the reasons for its development in Chapter II: an already
existing commercial maize sector that created the demand
for the product; a systematic breeding and agronomic
research program with close ties to the users of its product
and to the extension services; quality staff and remarkable
staff continuity; a locally based commercial production
company ready to put the product into widespread use;
foreign, governmental, and producer-supplied financial sup-
port; and contacts with international research institutions.
The Kenyan case is an excellent example of what Robert
Evenson calls "knowledge transfer" rather than direct
technology transfer: "the higher the level of indigenous
research activity the more productive is the regional re-
search...new varieties better suited to local regional condi-
tions [are] developed by crossing local varieties with inter-
national genetic material."2 It is not accidental that the
Kenya maize research program came up with a very high-
yielding hybrid variety. Rather, the existence of a well-

adapted local variety, a sound breeding program, and pre-
selected analogous varieties from an international collec-
tion of genetic material made the discovery of the Kenyan
hybrid varieties only a matter of time. There are few, if
any, shortcuts for developing countries seeking to "import"
the green revolution. As Griliches pointed out in the case
of American hybrids, hybrid maize is not a single invention
but a method for inventing corn varieties for specific locali-
ties. There is no alternative to investing in local research
capacity in individual countries in order to maximize the
returns on the investments that have been made in interna-
tional research institutions. In Kenya's own case, the need
now is to move toward still more localized research activity,
especially in agronomic work. More off-station research
trials are needed to determine the constraints to adoption
under smallholder farming conditions, including particular-
ly trials conducted with inter-planting of other crops.
Kenya has the core staff to direct such programs, but needs
many more well-trained subsidiary staff to carry them out.
The training and retention of well-qualified scientific staff
continues to be a matter of high priority.
The results of the present survey indicate the strong
influence of risk and of agrolcimatic zone on adoption.
More research on early-maturity, drought resistant varieties
is needed if the expansion of acreage under improved
varieties is to continue. Improved varieties of sorghum and
millet are also needed for the drier areas and work on high-
lysine maize needs to be accelerated. Because of its great
ecological diversity, Kenya cannot assume that the break-
through in high-yielding late-maturing varieties has solved
its food problems. Most Kenyans continue to grow all or
almost all of their own food. If productivity is not im-
proved in all farming areas, income inequalities between
high and low potential areas will worsen, thereby exacerbat-
ing ethnic differences.
After the production technology itself, the most im-
portant ingredient in the Kenyan hybrid success story has
probably been the commercial system of seed production
and distribution. As Guy Hunter and others have written,
different types of development institutions are necessary to
different stages and levels of development. It is unwise to
prescribe one type of institution for any and all develop-
ment situations. Although many persons have advocated
either state-run or cooperative-based input distribution
systems for Kenya, the available evidence indicates that
Kenya, at its present level of development, has been reason-
ably well served by the commercial sector both in seed
multiplication and in distribution. In the former case, the
most important factor has been the maintenance ofiquality
in the product, without which hybrid utilization would not
have continued to expand. Although the Kenya Seed
Company enjoys a monopoly on seed sales in Kenya, its
vested interest in maintaining the reputation of hybrid
maize and expanding its usage, plus government control
of seed prices and the government inspection service for
seeds have combined to prevent the kind of deterioration in

product which has taken place in many countries under
government run seed monopolies.
The maintenance of a system based on hybrid, rather
than synthetic, varieties, however, would not have been
possible without a distribution system that could deliver
new seed to the farmer each year. It is unlikely that a
public sector distribution system could do this as effective-
ly. By using established wholesalers and selected private
traders or stockists in the rural areas, the seed company
is sharing overheads of distribution with literally hundreds
of other inputs and consumer products. It has been estimat-
ed that while there is a health center for every 50,000 peo-
ple in Kenya, there is a small store or duka for every
500. As we have seen, the population per registered stock-
ist falls somewhere in between, ranging from 2,200 to
14,000 according to districts in western Kenya. Without
establishing an enormous distribution network of its own,
the government could not hope to distribute inputs except
through the private sector. Of course, cooperative socie-
ties are often used for distribution where they are active,
but functioning cooperatives are unevenly distributed around
the country. It is the view of some authorities, moreover,
that the slower uptake of improved maize varieties in central
Kenya is due in part to the fact that cooperative societies
play a larger role in input distribution in that area.
The rapid expansion in hybrid seed sales over the past
ten years would argue that the commercial distribution
system has been demonstrably successful. It may be, how-
ever that the cost of hybrids makes them too expensive
for farmers in the marginal areas to buy every year. For
this reason, the early maturing Katumani varieties used in
eastern Kenya have been produced as synthetics. Unless
improved hybrids can be developed for the lakeshore dis-
tricts of western Kenya, the development of a cheaper
synthetic for that area may also prove necessary. Belshaw
and Hall went so far as to recommend that a completely
revised package of free synthetic-type seed plus recommend-
ed interplanting with (supposedly) nitrogen-fixing legumes
be developed to avoid the necessity of purchasing any
inputs at all.3 With Kenya's reasonably good transport
system and a relatively well-developed retail system, how-
ever, it is not clear that the benefits of hybrid's higher
yields need be sacrified. Even if hybrids yield only 10
percent more than the best composites and synthetics, a
farmer (at the new price of 40/- a bag) need only get yields
over 5 bags an acre to make the 20/- price of seed worth-
while. It has also been proposed that the governmnet eli-
minate the KFA and other wholesalers as middlemen in
order to reduce the cost of seed to the farmer. It is
questionable whether the savings in seed cost would not
be more than offset by delays in distribution under such
a system. An incentive scheme giving discounts to stock-
ists who order and take delivery of inputs before a certain
date would probably be a better way of improving timely
distribution, allowing earlier planting, and thereby increas-
ing farmers' profits. A thorough study of fertilizer pricing,

subsides, and distribution is beyond the scope of this
thesis. We can only say that fertilizer usage is likely to
increase in importance over time as soil fertility is depleted.
The favorable increases of the late 1960s in product/factor
price relations have almost certainly been reversed in the
last two years by world fertilizer inflation. The deleterious
effects of this change on fertilizer utilization cannot yet
be determined.
A third important ingredient after technology develop-
ment and input distribution has been the extension and
training services of the Kenya Ministry of Agriculture. As
much maligned as it has been in recent years for its fairly
consistent bias toward large-scale farmers, the extension
service has nevertheless conducted literally thousands of
maize demonstrations, initially focusing on hybrid seed
and more recently on fertilizer utilization with hybrid
varieties. An agricultural journalist, visiting Kenya on
behalf of the Rockefeller Foundation in 1972, reported
that "Kenya has the most effective extension service for
reaching small farmers I have ever seen, and over the years
I have seen a good many."4 Although this view might sur-
prise some observers in Kenya, it may not be far from the
truth. In western Kenya, 35 percent of the farmers survey-
ed reported an extension visit in the past year, 25 percent
had personally attended a maize demonstration, and almost
as many reported first hearing of hybrid from an extension
agent as had heard about it through the usually far more
important interpersonal channels. In spite of these achie-
vements, however, there is still much room for improve-
ment in the extension services. Too many field staff are
largely untrained (in spite of serious efforts at in-service
training courses). Administrative back-up to field staff is
almost non-existent; promotion policies are poor; and em-
ployee morale is generally low. More use needs to be made
of group extension techniques and more contact is needed
between extension workers and research stations. More
extension effort needs to be directed toward women farm-
ers and toward farmers in the lakeshore districts which
receive considerably lower levels of agricultural services.
Contacts between research and extension staff need
to be further improved. In spite of the relatively short
distance across western Kenya, very few agricultural staff
interviewed below the district level had ever visited the
National Agricultural Research Station at Kitale. Farm-
ers Training Centers are generally poorly staffed, run at low
levels of capacity utilization, and are much too academic
in their approach to farmers. In one case a visitor found
farmers in a classroom being lectured on the dignity of
farming while prisoners from a nearby jail tended the
demonstration fields outside the classroom window. It is
probably safe to say that the extension service, no matter
how great its achievements, will always have considerable
room for improvement.
As for official sources of farm credit, because of minimum
acreage restrictions, they have played little or no part in the
diffusion of hybrid maize among small-scale farmers except

possibly through a demonstration effect in some key areas.
Kenya, however, has had the good sense to experiment with
various credit programs under its Special Rural Develop-
ment Program, and it is likely that improved smallholder
credit schemes will be developed drawing on the lessons
from that experience. At the very least, the stockist
credit scheme should help to improve input availabilities
in many areas and may result in the extension of some
private credit for maize inputs to local customers. The
problems of recovery of loans on a crop which can be sold
anywhere or consumed on the farm, however, make indivi-
dual credit for seasonal inputs difficult to extend on a
broader basis. Although credit has not played a major role
in the adoption of hybrid seed to date, it will become in-
creasingly important as fertilizer use increases over time.
The importance of cash crop income as an explanatory
variable in the analysis of adoption indicates that liquidity
can be a problem for small-scale farmers entering the market
economy for the first time.

Finally, the role of the marketing system in promoting
maize development must be recognized. Although it was
founded to serve the interests of large-scale European com-
mercial farmers, the guaranteed price system has been im-
portant in setting a floor to protect all farmers against wide
price fluctuations. The gradual improvement of storage
facilities, moreover, has given the country a strategic storage
capacity great enough to protect it against the shortages
which have periodically plagued the market in the past.
The recent increase in world prices to a level comfortably
above export parity has removed the problem of exporting
at a loss just at the time when hybrid production has placed
Kenya in a long-term surplus position. This is a prospect
that few African countries can envisage in the near future

and it should make maize an important export product for
Kenya over the next decade. Only the introduction of
bulk handling facilities to cut export costs and the ielaxa-
tion of controls on domestic marketing remain to be enacted
for maize to realize its full potential as Kenya's most
important agricultural product.
In conclusion, we have found that a modest investment
in the scientific breeding of maize has produced enormous
returns to the government and people of Kenya. Small-
scale as well as large-scale farmers in the high potential
areas of western Kenya and more recently in central Kenya
as well, have adopted hybrid maize technology at rates at
fast or faster than American farmers did in the 1930s and
1940s. No evidence has been found that "traditionalism,"
"custom," or "conservatism" has hindered the adoption
of a profitable innovation. Even in those areas where adop-
tion levels are low, historical and survey data indicate that
this has been a conscious process of non-adoption rather
than "laggard" traditionalism. Although farmers in the
drier areas suffer from lower levels of profitability, greater
risk, lower levels of agricultural services, and less income
from other cash crops, many of them have tried the new
technology but later given it up. As Griliches concluded
about American farmers, "taking account of uncertainty...
farmers have behaved in a fashion consistent with the idea
of profit maximization. Where the evidence appears to
indicate the contrary...a closer examination of the relevant
economic variables will show that the change was not as
profitable as it appeared to be." This appears to be the case
in western Kenya. It is hoped that a case study of the
willingness of African farmers to adopt profitable agricul-
tural innovations and the factors affecting adoption and
non-adoption will encourage those seeking to develop im-
proved technology for food production in tropical Africa.


1. F.F. Hill and Lowell S. Hardin, "Crop Production Successes
and Emerging Problems in Developing Countries," paper presented
at a Workshop on Emerging Issues Accompanying Recent Break-
throughs in Food Production, Cornell University, 1970, pp. 9, 11.

2. Robert Evenson, "Science and the World Food Problem,"

mimeo, 1974, pp. 16, 19. Kindly made available by Mr. R. Mohan.

3. D.G.R. Belshaw and Malcolm Hall, "The Analysis and Use of
Agricultural and Experimental Data in Tropical Africa," East Afri-
can Journal of Rural Development 5, 1-1, (1972): 70.

4. Carrol P. Streeter, Untitled paper, mimeo, 1972, p. 3.

Appendix I. Correlation coefficients-adoption of hybrid maize and selected factors.

Edu- Farm Large Dxten- Demon-
Hybrid Zone Age cation size Risk Job Work farm Credit sion station FTC Cash

Zone 0.76
Age .24 -0.21
Education .19 .13 -0.34
Farm size .28 .30 .05 0.18
Risk .54 .50 .10 .06 -0.13
Job .16 .13 .15 .26 .07 -0.09
Work .08 .06 .08 .05 .09 .04 0.25
Large farm .05 .18 .18 .00 .12 .03 .07 -0.12
Credit .26 .17 .03 .13 .24 .10 .00 .00 -0.05
Extension .15 .11 .02 .15 .10 .07 .07 .00 .05 0.17
Demonstration .18 .09 .05 .14 .10 07 .03 .02 .15 .18 0.28
FTC .29 .20 .08 .13 .16 .13 .05 .01 .09 .15 .22 0.41
Cash .28 .22 .07 .19 .40 .17 .12 .09 .05 .18 .17 .21 0.22

Appendix II. The 1974 Kenya Maize Farmers Survey
by Olof Hesselmark,
Kenya Maize and Produce Board


In 1974 the Kenyan Maize and Produce Board did a follow-
up study on the 1973 Kitale-CIMMYT survey including
the higher rainfall areas of Central as well as Western
Kenya. This survey was jointly financed by the Maize
and Produce Board and the Kenya Seed Company in Kitale.
The suggestion to repeat the exercise came from the KSC,
which had a particular interest in adoption rate in Central
Province, where an almost explosive adoption-with annual
increases of 50 percent in seed sales-has taken place over
the last few years. It was therefore decided to extend the
survey into Central and parts of Eastern Province, i.e. into
the Central Highlands. Thus the KSC and the MPB had
joint interests in the survey-the MPB could use the results
for better predictions of the total maize production and
hence for better marketing and storage decisions; the KSC
would get a more solid basis for decisions on seed produc-
tion and marketing.
The planning of the survey was made together with
two statisticians at the Agricultural Statistics Section at
the East African Community, Messrs. Macandu and Dunbar.
Much valuable advice also came from Mr. Verburght of
KSC and Dr. A.I. Allan of the Maize Research Center at
Kitale. The interviews were carried out by Research Bureau
Limited as in the 1973 survey. In Western Province the
majority of the interviewers were the same as in the earlier
survey, and whenever possible the same interviewer went
back to the farmer. The RBL again performed very well
under sometimes difficult conditions.

Sampling Procedure

The sampling procedure was essentially that employed by
Dr. Gerhart in the 1973 survey. In Western Kenya it was
identical, insofar as the same farmers were visited. In
Central Kenya the same technique was used. The area was
divided into two zones, one at higher (above 1500 m) alti-
tudes. In each zone the total area comprised of a number
of 100 square kilometers segments of which a random
sample was drawn.

The Central Province

The area that constitutes zone 4 and 5, in the 1974 survey
is located in Kenya's Central and Eastern Provinces, and

includes the following districts: Kiambu, Murang'a, Nyeri,
Kirinyaga in Central Province, and Meru and Embu in
Eastern Province. This is a densely populated area with
almost exclusively small scale farms, inhabited by the Kiku-
yu, Meru and Embu tribes. In the southern part, the
country slopes eastwards from the Aberdare Mountains
from the forest zone at 2000 2500 m altitude down
towards the plains at 1100-1300 m in the eastern part of
Murang'a and Kiambu Districts. This land is intersected by
deep valleys, and the dominant landscape features are the
long almost parallel rivers and ridges running from north-
west to south-east. Further north there is a great valley
between the Aberdares and Mount Kenya, whose peaks
are about 80 km apart. This is the location of Nyeri
District. Around Mount Kenya on the southern and eastern
side are Kirinyaga, Embu and Meru Districts.
The rainfall pattern is fairly contrasting, with rainfall
generally increasing with altitude. The eastern slope of
Mount Kenya receives considerably more rainfall than the
western side; thus Meru receives 1800-2000 m compared with
900 mm per year in Nyeri. The natural forest boundary
is at about 1500 to 1700 m altitude, but population
pressure has led to considerable cutting down of forests,
and cultivation sometimes takes place as high as 2500 m
(in Limuru north of Nairobi).
In our sample, zone 4 consists of the land which is below
1500 m and zone 5 is that land which is above 1500 m.
Zone 5 then is between 20 and 30 km wide, and follows
the forest boundary along the two mountains. Zone 4 is
the valleys and plains below down to 1100-1300 altitude.
Below 1100 m the rainfall is usually not thought sufficient
for hybrid maize.

The Interviews

A main objective of the survey was to find out how hybrid
seed is actually being utilized by small farmers. This
is particularly important in those areas in Western Kenya
where we already knew that a vast majority uses hybrid
seed. The relative intensity of cultivation i.e. plant popula-
tions together with selected husbandry practices, would
indicate what potential increases in seed sales, land utilization
and maize production exist in that area. In the non-adopt-
ting zone 3, the survey perhaps could give more information
on the reasons for non-adoption in those areas. Since 1972,
in Central Kenya, the KSC has witnessed a spectacular in-
crease in seed sales, and it was therefore interesting to

establish the general level of adoption, in order to estimate
the level and the point in time at which seed sales would be
expected to level off. Compared with the 1973 question-
naire, the 1974 survey focused more on the question
"how many" of the farmers use hybrid rather than "why"
or "why not". Some of the questions used to find out the
reasons for adoption or non-adoption were therefore discard-
ed. This is perhaps unfortunate, since it is not possible to
treat the material in the same way as in the earlier study.
For Central and Eastern Provinces, however, the willingness
to adopt hybrid seed once it is available has been demonstra-
ted already, and the recent phenomenal adoption rate indica-
tes that full adoption will be reached in two or at most
three years.
The most important change in the questionnaire and in
the interviews was a precise plant count, from which the
number of plants per hectare and number of ears per hec-
tare was computed. The reason to include this was twofold.
First, a prceise measurement of plant populations would be
of value for KSC's marketing decisions. Secondly, this in-
formation can be used to estimate the total production
from hybrid seed in the country. Given the amount of seed
sold in various areas, the total number of maize plants and
ears can be computed. When the count is complemented
by a sample of mature, maize cobs, whose grain weights are
measured, the total hybrid maize production in each area
can be estimated-from the number of cobs and their
weight. The second sample has now been carried out, and
it is hoped that this will provide the first estimate of Kenya's
maize production that is not based entirely on guesses.


The sample. For zones 1, 2 and 3, a sample was drawn
from the 300 or so earlier questionnaires because finan-
cial reasons made it impossible to visit all the farmers from
the 1973 survey. The target was to have about 150
interviews and by tossing a coin one half of the segments
were selected. All in all, 117 successful respondents were
about 22 percent of the total sample. The discarded
interviews seem to be randomly distributed over the sample.
Zones 4 and 5 (in Central Kenya) had not been visited
in the 1973 survey, and a completely new sample was
drawn. After studying population maps it was concluded
that the population density was roughly equal in the two
zones, and it was decided that the number of segments, i.e.
the area of each zone would determine the sample size
for each of them. A total sample size of 120 was set as the
target, which would mean the selection of every second
segment in each area. The selection was randomized by
tossing a coin. The actual number of successful farm
interviews was 104, the remainder being non-maize farmers,
non-accessible segments and other field problems.
Adoption Rates. It appears that about two farmers out
of three have adopted hybrid maize in those areas of Kenya

where hybrids can be grown. If zone 3 is excluded, one
is left with practically all high-potential small-scale farming
areas in the country. The results in these areas only,
indicate an overall adoption rate of 74 percent. The
Central Province had about 55 percent adoption, with no
significant difference between zones 4 and 5.


Hybrid Users
Zone No. %
1 39 100
2 43 89
3 4 13
4 25 59
5 33 53
Total 144 65

No. %
0 0
5 11
26 87
17 41
29 47
77 35


Kenya's Hybrid Acreage. According to KSC and Ministry
of Agriculture recommendations, plant populations for pure
hybrid stands are between 45,000 and 53,000 plants per
hectare depending on variety. The achieved plant popula-
tions were measured by counting the number of maize
stalks in the field, and by measuring the distance between

(Weighted averages. Farms not practi-
cing row planting excluded).


Hybrid Users Non-users
37.5 32.5
22.5 20.0
36.6 35.8
32.2 28.9
34.4 28.R

KSC area estimates based on seed sales and an assumption
of 9 kgs. of seeds per acre should be increased by 1.24 (from
336,000 ha to 416,000) to give the true number of hectares
under hybrid maize in the country.
An important factor influencing plant population is of
course interplanting, a practice which is widely spread
throughout the small scale Kenyan Agriculture. The merits
of interplanting have never been studied or demonstrated
conclusively, although most authorities seem to agree that
the practice has a lot of advantages in the African small-scale






Interplanting, % Pure Stands, %
Zone Hybrid Non-Hybrid Hybrid Non-Hybrid Total
1 43 57 100
2 25 67 8 100
3 23 67 9 21 100
4 36 26 24 14 100
5 23 45 29 3 100
Total 26 27 38 8 100

Apart from the repeated pattern from the 1973 survey
for zones 1, 2 and 3, zones 4 and 5 show a remarkably
different picture from Western Kenya. First, interplanting
is one the whole more common than in Zones 1 and 2.
Second, interplanting does not seem to vary with hybrid
use as much as in zone 1, 2 and 3. In zone 1 and 2, only
35 percent of hybrid users interplanted, whereas 51 per-
cent of hybrid users in Central Province did. It is difficult
to explain this behaviour. In the 1973 survey it was
concluded that adoption of hybrid maize went along with a
whole series of agricultural practices, including planting in
pure stands. This problem will be examined later in great-
er detail, but it seems as interplanting may be a more deep-
ly rooted habit in Central Province.
The low plant populations per hectare mean that hybrid
maize is actually planted over a much larger area than has
previously been thought. Obviously, here lies the key to
increased maize production. If it is the case that low plant
populations are realized in pure stands as well as in inter-
planted fields, considerable increases in the number of
plants can be made.
Adoption and Husbandry Practices. In the 1973 survey
it was demonstrated that adoption of hybrid seeds tends
to go together with the adoption of a whole packet of
improved husbandry practices, such as use of fertilizers,
planting in rows and in pure stands, early planting, etc.
Through careful analysis of his material, Dr. Gerhart show-
ed that there actually was a relationship between seed
adoption and changed husbandry practices. It is not a
coincidence of two independent trends. Hybrid seed is
considered by the farmers as a new crop whose growing
is inherently different from the cultivation of the old
type of maize.
In the 1974 survey, the data for zones 1, 2 and 3 again
show the same pattern as in 1973. Zones 1 and 2 have
very high numbers using hybrid-related practices, significant-
ly higher than in zone 3. e.g. 93 and 94 percent of the farm-
ers in zone 1 and 2 planted in rows, vs. 47 percent in zone
3. The use of different practices is shown below:

1 2 3 4 5

Physical inputs
Hybrid seed
Insecticide in store
Insecticide in field


Cultural Practices
Row planting
Pure stand
Weeded more than once
Planted early
Nr farms

Central Province (Zone 4 and 5) showed a slightly dif-
ferent pattern. The high scores for row planting and for
fertilizer use should be noted, as well as the high percent-
age of farmers interplanting maize with other crops (62 and
68 percent). Farming Methods in the often very high and
hilly zones 4 and 5 are definitely different from those in
Western Kenya. This is of course a well known fact.
Particularly interesting to note is the use of manure, and
the fact that many farmers thinned their maize. In the
1973 survey thinning was shown to be negatively related
with adoption in zones 1, 2 and 3, in accord with the
hypothesis that planting a large number of seeds in each
hole was related with broadcasting and other husbandry
practices employed before hybrid was introduced. In Central
Province, however, it is common to pull out maize plants
during the growing season and feed them to cattle. Also,
planting many seeds in a hole reduces the number of plant-
ing stations and thereby labour requirements at planting
time. There is evidence that labour supply at planting time
is an important constraint on the area that can be planted
at the right time. In Central Province, where rainfall is less
abundant and it s variations are greater, than in the West,
late planting is related with severe yield reductions. There-
fore farmers should be expected to employ various tech-
niques to enable them to plant a large area within a short
time. In the absence of mechanical implements, planting
many seeds in a hole constitutes one such technique. Thus
many seeds in a hole may be done for other reasons than
to ensure germination from poor seed.

In the 1973 survey it was investigated whether there
was any difference between hybrid and non-hybrid users
with regard to other husbandry practices. The hypothesis
was that since hybrid has been introduced as a package of
husbandry practices rather than by simply encouraging
farmers to buy the new seed, it would be expected that
hybrid adopters are more likely to use for example fertilizers,
row planting, pure stands etc. This hypothesis was con-
firmed in the 1973 survey, and it has been repeated with
1974 data. The method employed was a simple chi-square
test where samples of non-adopters for one practice at a
time. For this analysis, a chi-square value of 2.71 or
higher makes the hypothesis accepted at a 90 percent
confidence level. The results are given in the table below.
The sample has been divided into two parts, Western Kenya
(Zones 1, 2 and 3) and Central and Eastern Provinces (Zone
4 and 5).

(Proportion of each group of respondents
that employed the practice in question).

Row planting
Insecticide on crop
Weeded more than once
Pure stands
Early planting

Hybrid users
n = 88

1, 2 and 3, 1974.
users and non-users-

n =34

2.5 ns
10.4 *
0.2 ns
0.4 ns


Row planting
Insecticide on crop
Weeded more than once
Pure stands
Used manure
Early planting

Hybrid users
n = 57

n =47

0.1 ns
0.1 ns
0.1 ns

moderate altitudes (3500 to 7000 feet). The requirements
of the climate were important all the way from the begin-
ning of the programme, which started in 1965 with a large
number of crosses from a widely based genetic material.
In 1967, the first results were Embu Composite I and
Embu Composite 2, both formed from coloured exotic
material. A systematic selection for whiteness was perform-
ed on these varieties. At the same time, crosses of late
maturing Kitale hybrids and Katumani was performed to
get a variety with the same maturity time as the most
common local maize. The local is called the Muratha Maize,
and has been used as a common denominator in most
comparisons throughout the programme.
In 1967, the first hybrid (H511) was produced. It had
the same maturity as Muratha, but this first version out-
yielded the local variety by 24 percent. The genetic ma-
terial was the following:

H621 Katumani IV

Embu 11

Cometico I Katumani IV

Embu 12


In both areas, row planting, fertilizer, insecticide on
crop, and pure planting were all highly significant, i.e.
adopters are more likely to use these practices than non-
adopters. Thinning the maize plants was significant in
Western Kenya but not in Central. The probable reasons
for this has been indicated above. Early planting was
significant in Central but not in Western, and likewise, this
has been discussed above. For weeding and use of manure
there were no significant differences between adopters and
The conclusions from this exercise as from the 1973
survey are that there are indeed significant differences
between adopters and non-adopters. When farmers start
using hybrid maize, they simultaneously take up a whole
series of husbandry practices. Hybrid maize is then re-
garded as a "new" crop, different from local not only in
seed, but also in farming methods. Increased production can
therefore not be attributed to the seed alone, but on the
other hand the seed was the vector that started the develop-
ment. Without the seed, it might have been much more
difficult to change the methods.

The Embu Medium Maturity Breeding Programme

What was responsible for the rapid increase in hybrid
use in Central Kenya after 1970? The following descrip-
tion is a concentrate of Annual Reports from the Maize
Research Institute of Embu 1966 to 1973.
The maize breeding programme at Embu focused on
the production of medium maturity varieties foy Central
and Eastern Provinces. The climate there is characterized
by two distinct rainy seasons of limited duration, and by

The variety was regarded as so successful, that it went on
sale commercially in 1968. It was also improved in the
next year, when it outyielded the local Muratha by 44
During the first years of the breeding programme, agro-
nomic trials were also conducted at Embu, and the find-
ings were similar to those at Kitale. It was concluded, that
early planting, a high plant population (53,000 plants/ha)
and good weeding increase yields significantly. Recommend-
ations for planting were issued in 1970, (then for H511).
In 1969, the next breakthrough in breeding was made
with the production of H512, being a cross of H511 with
another variety (SR52). H512 had very desirable character-
istics, and in 1970 it outyielded the local by 61 percent. It
was also released in that year.
By 1970, the immediate needs to produce an acceptable
hybrid had thus been covered. In the years since its
introduction, H512 has gained great popularity in Central
and Eastern Provinces, and well over half the maize acreage
in that area is expected to be planted with H512 in 1975.
The breeding programme has therefore focused more on
the longterm aspects of high protein and high lyseine maize,
andf some progress has been made over the last years.
Another aspect that has been of some interest is that
H512 outyields all other varieties under normal rainfall,
but in dry years the hybrids is outyielded by local maize
with the same maturity, and both are outyielded by Katuma-
ni. Efforts are therefore made to improve the overall
performance of H512 under dry conditions.
Like the maize research at Kitale, the Embu programme
must be judged as highly successful. In a very short time,
very tangible results have been achieved in terms of high-

yielding hybrids. The actual production of commercial
seed has been made by the Kenya Seed Company at Kitale,
and KSC has also been responsible for the distribution of
seed. The characteristics of H512 are very desirable in all of
Central Province, and farmers have been very quick to see
the benefits. There are indications that the rate of increase
in acceptance has been even faster than in Western Province,
and again the KSC has succeeded in distributing the seed.
A comparison between seed sales in Central and Western
Provinces is given below:


10-kg bags Increase, %
13,200 26
27,700 110
45,800 65
68,100 49
82,700 22
120,000 45
150,000* 25

10-kg bags Increase, %
64,000 78
80,000 25
127,000 59
135,000 6
185,000 37
205,000 11
218,000* 6

Compound interest growth
rate, 67/68-74/75

The growth rate in Central Province is a stunning 46 per-
cent a year since 1967/68, when H511 was first introduced,
and the projected 1974/75 sales are 14 times as large as in
the first year, and over 3 times as large as in 1970/71.
Without pressing the data too much, it appears that the
growth is even faster than in Western Province, where sales
now are levelling off because of saturation of the market, as
we have seen earlier. In Central Province, it should still
be possible to double the seed sales over the 1973/74
level, since only about 50 percent of the farmers indicate
that they are using hybrid seed. This suggests a saturation

level for Central Province of 220,000 to 240,000 packets,
or about the same as Western. As with Western Kenya, it
is the small farmers that have adopted hybrid seed. (In
Central Province, this can be asserted even without any
analysis, since there are hardly any large farmers in the
Province). The story of hybrid diffusion throughout Kenya
is thus a success story, where rapid adoption has been
achieved within the small-holder sector.


The 1974 survey has at least demonstrated two things.
One is that the results for 1974 in Western Kenya are
consistent with those of 1973. The other is that a similar
development of hybrid maize has taken place in Central
Kenya a few years later. The reason for the lateness in
Central Kenya is simply that a suitable hybrid variety
became available later there.
Perhaps the most valuable part of the whole survey
and the analysis has been to point out in what areas we
need further investigations. After the survey was completed,
work started immediately on the planning of a system for
objective yield surveys of maize in Kenya, which is being
conducted in 1975. The work is going to be carried out by
the Maize and Procedure Board together with the Central
Bureau of Statistics. The earlier surveys have been invalu-
able in the planning, because they provided information
necessary to determine sample sizes and error levels in the
yield surveys. For the objective yield survey, they have
acted as pilot studies, and we have been able to avoid
a large number of pitfalls and problems through what has
been learnt from them. It is through its indirect influence
on the more comprehensive study that the survey has been
of some value, not through its actual findings.



CIMMYT International Maize and Wheat Improvement Center

The CIMMYT International Maize and Wheat Improvement
Center holds all rights to the source text and shall be
considered the copyright holder for the text and images of
these publications.

The CIMMYT International Maize and Wheat Improvement
Center has made this publication available to the University of
Florida, for purposes of digitization and Internet distribution.

The CIMMYT International Maize and Wheat Improvement
Center reserves all rights to this publication. All uses,
excluding those made under "fair use" provisions of U.S.
Code, Title 17, Section 107 are restricted.

Contact the CIMMYT International Maize and Wheat
Improvement Center for additional information and

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