Front Cover
 Title Page
 Table of Contents
 List of Tables
 List of Illustrations
 Research and policy issues in adoption...
 Analytical approach and method...
 Agriculture in eastern provinc...
 Characterization and determinants...
 Labor allocation patterns
 Intrahousehold decisionmaking
 Food consumption and nutrient...
 Effects on health and nutritional...
 Conclusions and policy implica...
 Back Cover

Group Title: Research report
Title: Adoption of hybrid maize in Zambia
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00085468/00001
 Material Information
Title: Adoption of hybrid maize in Zambia effects on gender roles, food consumption, and nutrition
Series Title: Research report
Physical Description: viii, 126 p. : ill., map ; 26 cm.
Language: English
Creator: Kumar, Shubh K
Publisher: International Food Policy Research Institute,
International Food Policy Research Institute
Place of Publication: Washington D.C
Publication Date: c1994
Copyright Date: 1994
Subject: Corn industry -- Zambia -- Eastern Province   ( lcsh )
Hybrid corn -- Economic aspects -- Zambia -- Eastern Province   ( lcsh )
Food consumption -- Zambia -- Eastern Province   ( lcsh )
Nutrition -- Zambia -- Eastern Province   ( lcsh )
Agricultural laborers -- Zambia -- Eastern Province   ( lcsh )
Women agricultural laborers -- Zambia -- Eastern Province   ( lcsh )
Sex role in the work environment -- Zambia -- Eastern Province   ( lcsh )
Maïs -- Industrie -- Zambie -- Eastern   ( rvm )
Maïs hybride -- Aspect économique -- Zambie -- Eastern   ( rvm )
Aliments -- Consommation -- Zambie -- Eastern   ( rvm )
Nutrition -- Zambie -- Eastern   ( rvm )
Travailleurs agricoles -- Zambie -- Eastern   ( rvm )
Travailleuses agricoles -- Zambie -- Eastern   ( rvm )
Rôle selon le sexe en milieu de travail -- Zambie -- Eastern   ( rvm )
Genre: bibliography   ( marcgt )
statistics   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Zambia
Bibliography: Includes bibliographical references (p. 116-126).
Statement of Responsibility: Shubh K. Kumar.
 Record Information
Bibliographic ID: UF00085468
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 31753910
lccn - 94048103
isbn - 0896291030

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Title Page
        Page i
        Page ii
    Table of Contents
        Page iii
    List of Tables
        Page iv
        Page v
    List of Illustrations
        Page vi
        Page vii
        Page viii
        Page 1
        Page 2
        Page 3
        Page 4
    Research and policy issues in adoption of hybrid maize in Zambia
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
    Analytical approach and methodology
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
    Agriculture in eastern province
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
    Characterization and determinants of hybrid maize adoption
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
    Labor allocation patterns
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
    Intrahousehold decisionmaking
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
    Food consumption and nutrient intakes
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
    Effects on health and nutritional status
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
    Conclusions and policy implications
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
    Back Cover
        Page 127
        Page 128
Full Text
Cq c// ^ o. 7 /y/ y73< /^9



Shnhh K Kumar

IFPRI Research Reports
Publications Review Committee and Procedures

Christopher Delgado.
Ousmane Badiane
Romeo Bautista
Law rence Haddad

Keijiro Otsuka
Mark Rosegrant
Barbara Rose (ex ofic io)

All manuscripts submitted for publication as IFPRI Research Reports undi rgo extensive
review. Prior to submission to the Publications Re\ ie\% Committee, each manuscript is
circulated infomiall among the author's colleagues. presented in a forma seminar. and
reviewed by two IFPRI reviewers. Follow ing submission of the manuscript to the Com-
mittee, three additional reviewers-at least t'o external to IFPRI and one from the
Committee-are selected to review the manuscript. Re\ ieers are chosen for their
expertise in the manuscript's subject matter and methodology and. when applicable. their
familiarity \\ ith the country setting TheCommittee provides the author its ea.tion to the
reviewers' comments. After re\ rising as necessary, the author resubmits the manuscript to
the Committee ith a written response to the reviewers' and Commitee's comments. The
Committee then makes its recommendation on publication of the man iscript to the
Director General of IFPRI. With the Director General's approval, the manus.ri t becomes
part of the IFPRI Research Report series.


IFPRI Board of Trustees

Gerry Helleiner
Chairman. Canada
Sjarifuddin Baharsjali
David E. Bell
Henri Carsalade
GCi.dfrc\ (;uniaillelke
Sri Lanka
Ibrahim ,aad Ahmed Hagrass
Yujiro lla,.ami
Uwe I tlitz
Federal Republc ofGermany

James Charles Ingram

Nora I.usLti

HFarris 1utio Mule -

I. G. Patel

Martin Piieiro

.Abdo'ulav3e SaS adoE'
Cl te d'Ivoire

Per Pinstrup- \ndrrsen
Director General
L\ t-ffil iu. Denmark


Shubh K. Kumar

Research Report 100
International Food Policy Research Institute
Washington, D.C.

Copyright 1994 International Food Policy
Research Institute.
All rights reserved. Sections of this report may
be reproduced without the express permission of
but with acknowledgment to the International
Food Policy Research Institute.

Library of Congress Cataloging-
in-Publication Data
Kumar, Shubh K.
Adoption of hybrid maize in Zambia : effects
on gender roles, food consumption, and nutrition /
Shubh K. Kumar.
p. cm. (Research report; 100)
Includes bibliographical references.
ISBN 0-89629-103-0
1. Corn industry-Zambia-Eastern Province.
2. Hybrid corn-Economic aspects-Zambia-
Eastern Province. 3. Food consumption-Zambia-
Eastern Province. 4. Nutrition-Zambia-Eastern
Province. 5. Agricultural laborers-Zambia-
Eastern Province. 6. Women agricultural laborers-
Zambia-Eastern Province. 7. Sex role in the
work environment-Zambia-Eastern Province.
I. Title. II. Series: Research report (International
Food Policy Research Institute); 100.

HD9049.C8Z3648 1994



Foreword vii
1. Summary 1
2. Research and Policy Issues in Adoption of Hybrid Maize
in Zambia 5
3. Analytical Approach and Methodology 15
4. Agriculture in Eastern Province 30
5. Characteristics and Determinants of Hybrid Maize Adoption 38
6. Labor Allocation Patterns 53
7. Intrahousehold Decisionmaking 69
8. Food Consumption and Nutrient Intakes 78
9. Effects on Health and Nutritional Status 96
10. Conclusions and Policy Implications 111
Bibliography 116


1. Per capital consumption of basic staples and total calorie intake
in rural areas, by province, 1980 6
2. Growth of marketed maize production, by province, 1973-89 10
3. Population distribution in urban and rural areas, by province 30
4. Production of major crops by agro-ecological zones, Zambia,
1979-80 32
5. Crop area and share, by site 35
6. Crop area and share, by region 36
7. Crop area and share, by hybrid maize adoption 37
8. Hybrid maize adoption by farm size, plateau and valley 39
9. Oxen use by farm size 40
10. Oxen use by hybrid maize adoption 40
11. Farm size and hybrid maize adoption by head of household 42
12. Determinants of hybrid maize adoption and its income effects 44
13. Women's crop ownership and share 46
14. Women's crop ownership and share, by household head 48
15. Women's crop ownership and share, by hybrid maize adoption,
plateau 50
16. Factors in women's management role 51
17. Labor use per hectare, by crop 54
18. Intrahousehold labor, by crop 55
19. Total labor use, by farm size 56
20. Total labor use by farm size and cultivation method 57
21. Analysis of household labor allocation 66
22. Percent of females making agricultural decisions 72
23. Average amount received from crop sales by gender, by type
of crop ownership, all crops combined 73
24. Frequency of spending from proceeds of crop sales on various
items, by sex 74
25. Mean expenditure from the proceeds of crop sales, by sex 75
26. Percent of food expenditure decisions made by females in
households growing hybrid versus local maize 76

27. Annual average of daily per capital consumption of calories, by
food groups, Eastern Province 81
28. Annual average of daily per capital consumption of nutrients
Eastern Province 82
29. Daily per capital consumption of cereals, by farm size, in
valley and plateau regions and high and low adoption areas 83
30. Annual average of daily per capital nutrient intakes, by valley
and plateau regions 84
31. Food diversity, by month, region, and level of adoption of
hybrid maize 85
32. Mean daily per capital nutrient intake, by household use of
hybrid maize 86
33. Mean daily per capital nutrient intake, by hybrid maize adop-
tion and farm size 87
34. Daily per capital food consumption, by hybrid maize adoption
and farm size 88
35. Daily per capital nutrient intakes, by season 89
36. Daily per capital nutrient intakes, by season, for high- and
low-adoption areas in the plateau region 90
37. Mean daily per capital nutrient intake by household use of
hybrid maize, by seasons 92
38. Regression summary of consumption and food diversity 94
39. Illness rates 97
40. Disease episodes and duration per person by region and high-
and low-adoption areas 99
41. Disease episodes and duration per person by region and hybrid
maize adoption by households 101
42. Water and sanitation, by hybrid maize adoption 102
43. Malnutrition in children, Eastern Province, 1986 103
44. Malnutrition in children, by adoption of hybrid maize, Eastern
Province, February 1986 104
45. Anthropometric regression summary: Random effects model,
10 years and under 108


1. Map of Zambia's agro-ecological zones 33
2. Average hours of labor, by activity, adult males, by deciles of
total area farmed 58
3. Average hours of labor, by activity, adult females, by deciles
of total area farmed 58
4. Average hours of family labor spent in cropping activities, by
hybrid maize adoption 59
5. Average hours of male labor spent in cropping activities, by
hybrid maize adoption 60
6. Average hours of female labor spent in cropping activities, by
hybrid maize adoption 61
7. Average hours of nonfamily labor spent in cropping activities,
by hybrid maize adoption 61
8. Average hours of family labor spent in household mainte-
nance, by hybrid maize adoption 63
9. Days ill with diarrhea, malaria, and other infections 98
10. Percent of children who are below -2 Z-scores of weight-for-
age in areas of high and low adoption of hybrid maize 105
11. Percent of children who are below -2 Z-scores of weight-for-
age by household level of hybrid maize adoption and farm size 106


Enhanced agricultural productivity in Sub-Saharan Africa is critical to promote
economic growth and poverty alleviation and to avoid increasing food scarcities in
the region. The impact of commercialization and intensification of agriculture on the
well-being of the rural poor depends on how they are carried out. Past research by
IFPRI and collaborating institutions on commercialization of small-scale farming in
about a dozen countries provided new knowledge about the relationships between
commercialization and rural well-being as measured by incomes, consumption, and
nutrition. These links were shown to depend greatly on household behavior, which in
turn is influenced by intrahousehold processes. A better understanding of these
processes is likely to identify policy measures that will be effective in achieving both
productivity and household welfare goals.
This report contributes to improved understanding by examining household and
intrahousehold processes influencing the welfare effects of the adoption of hybrid
maize among farmers in a region of Zambia. The report identifies a number of key
policy options likely to be central to achieving higher agricultural productivity and
improved rural welfare simultaneously.
While extending previous research on agricultural commercialization and techno-
logical change in several countries, the research reported here is a part of a larger
collaborative research project undertaken with the University of Zambia, Rural Devel-
opment Studies Bureau, and the Zambian National Food and Nutrition Commission.
Several other IFPRI reports are available from this project, including an occasional
paper, Adopting Improved Farm Technology: A Study of Smallholder Farms in East-
ern Province, Zambia, edited by Rafael Celis, John T. Milimo, and Sudhir Wanmali,
and Research Report 94, Fertilizer Use on Smallholder Farms in Eastern Province,
Zambia, by Dayanatha Jha and Behjat Hojjati. Past research on commercialization of
small-scale agriculture is synthesized in a book published for IFPRI by the Johns
Hopkins University Press, Agricultural Commercialization, Economic Development,
and Nutrition, edited by Joachim von Braun and Eileen Kennedy.

Per Pinstrup-Andersen
Director General


I am deeply grateful to Catherine Siandwazi and colleagues at the National Food
and Nutrition Commission in Zambia for their partnership. They ensured that a
nutrition and intrahousehold component was included in the study of improving
agricultural technology from which this report is derived. The research in Zambia
benefited from the institutional base provided by the Rural Development Studies
Bureau, University of Zambia, and in particular from the committed and insightful
support of John Milimo and his colleagues.
I wish to acknowledge Saroj Bhattarai for his research assistance and careful
documentation of data analysis. Emelia Elson and Lyndon Naverra also assisted in
key portions of the data processing. Others who were instrumental in the whole
intricate process of managing the data from this comprehensive survey include Neal
Bliven and Cindy Holleman, and most importantly, the late Emmanual Shula and
Rafael Celis, who supervised the whole survey. Others who were key in starting and
supporting the overall collaborative IFPRI/RDSB/NFNC study on Growth and Eq-
uity Effects of Improved Farm Technology in Zambia on which this research is based
are John Milimo, Peter Hazell, and Marco Ferroni.
Many of my colleagues at IFPRI generously contributed ideas and other input to
this work. In particular, I would like to thank Francesco Goletti, Lawrence Haddad,
Chris Delgado, Tesfaye Teklu, Yair Mundlak, and Steven Vosti for their insights and
suggestions. Others who read all or parts of the manuscript and provided valuable
input are Dunstan Spencer, Paul Heisey, lan Hopwood, and Barry Reily.
I am especially grateful to the reviewers, Pauline Peters, Bob Evenson, Sara
Scherr, and Maarten Immink. They helped strengthen and tighten the analysis and
presentation, and I hope the final work does justice to their patience.



This report examines the role of hybrid maize adoption in Eastern Province,
Zambia, in improving the welfare of the population. Improving agricultural produc-
tivity of farmers in Zambia is important for the success of the country's new
economic growth strategy, and past investment in hybrid maize research has devel-
oped a potential for increased productivity that needs to be fully utilized.
Maize is the single most important food in the Zambian diet, and its primacy has
grown steadily as the result of past government policies that encouraged the produc-
tion of maize in all parts of the country, including areas where it may not be
economically efficient to grow maize. Given current market liberalization efforts, it is
likely that maize production in general and marketed maize production in particular
will remain viable only in areas near the major population centers because of transport
costs. Such a contraction in maize area would increase the need for improved tech-
nologies to raise agricultural productivity in outlying areas in order to maintain their
current level of income. Since most of the previous agricultural research in Zambia has
been on maize, this crop offers more options for increased productivity than other
crops. However, even if other crops are promoted, the experiences in optimizing
growth and welfare outcomes with hybrid maize should be useful.
Until the late 1980s, aggregate increases in maize production were limited,
despite a substantial expansion of hybrid maize adoption and fertilizer use. Since
then, a wide range of improved maize varieties suitable for small farms has been
released by agricultural research stations in Zambia. The potential thus exists for
rapid improvements in productivity, income, and welfare.
This report examines farm household-level factors that influence the adoption of
hybrid maize in Eastern Province and the implications of adoption for improvement
in household income, food consumption, and nutrition and health of the rural popu-
lation. The characteristics of adoption, such as who adopts and what other changes
are associated with it, in particular its implications for household labor allocation and
intrahousehold access to resources, are expected to influence food consumption and
the nutritional status of the population.
The analytical approach is geared to trace the distributional and welfare conse-
quences of hybrid maize production. An instrumental variable approach is used to
make predictions on the effects of hybrid maize adoption. Previous IFPRI work has
generally shown that incomes rise with the adoption of improved agricultural tech-
nologies, but child nutrition does not necessarily improve. In this report, the implica-
tions of a wide range of resource allocation decisions that are associated with
adoption and that influence the distribution of welfare improvements are examined.
These include changes in women's access to resources and decisionmaking, labor
allocation decisions, and characteristics of cash flow and allocation of income. Area-
level characteristics such as access to infrastructure and markets and geographical
variation in adoption rates are also considered. Although the analysis identifies

adoption of hybrid maize production as a key element of technological change in
agriculture, adoption is nearly always accompanied by increased use of chemical
fertilizers and an expansion of cultivated area associated with a shift from hoe to
ox-plow use.
This report is based on a collaborative study in Eastern Province conducted in
1986 by the International Food Policy Research Institute, together with the Univer-
sity of Zambia's Rural Development Studies Bureau and the Zambian National
Food and Nutrition Commission, to examine the growth and equity effects of
technological change in agriculture. Results of this study were presented to the
government of Zambia between 1987 and 1990, and this report presents a detailed
analysis of those results.
Eastern Province is one of the major agricultural regions of Zambia; it consis-
tently produces large maize surpluses. Its predominantly rural population depends on
agriculture for nearly 80 percent of its income. Agriculture is mainly smallholder,
with an average farm size of 2-3 hectares. It has some of the best agricultural land in
the country, but, as in many other parts of Africa, it has a single rainy season, thus
providing only one main growing season for farmers.
Study sites, located in each of the districts, were selected to provide a repre-
sentative sample of households from the province and its two main ecological
zones-plateau and valley. During 1986, 330 households, drawn from a stratified
random sampling, were visited monthly and interviewed on agricultural production
practices, labor allocation, off-farm income sources, food and nonfood consumption,
morbidity, and intrahousehold decisionmaking. In addition, weights and heights of
each household member were recorded four times during the year to determine
anthropometric status and hence nutritional status.
Among the 10 percent of farmers with the largest farms in Eastern Province,
nearly all with more than 5 hectares adopted hybrid maize. However, adoption was
also substantial among the smaller farms, with about 50 percent of those in the
3-to-5-hectare category, 37 percent in the 2-to-3-hectare category, and 25 percent in
the 1-to-2-hectare category adopting hybrid maize. Data indicate, however, that
hybrid maize production is more profitable for smaller farms. Marginal improve-
ments in income deteriorate beyond 4 hectares under hybrid maize. This implies that
policies directed to adoption by larger farmers may be contributing to lower produc-
tivity gains from this technology.
Because it is harder to process and store hybrid maize, farmers also grow local
maize for home consumption and sell most of the hybrid maize. Where labor supplies
are short, farmers are likely to devote more attention and resources to local maize.
Policy measures to improve local storage and processing options could further
improve hybrid maize productivity because these measures would shift its place in
the cropping system from cash crop to food crop, so that farmers would give it
priority in timing of planting and other operations. The market liberalization now
under way should provide an incentive for investments in low-cost rural storage
improvements, for which technologies already exists. Improvement in rural infra-
structure will also be critical.
Female-headed households have a lower adoption rate for hybrid maize (22
percent) than male-headed households (34 percent). However, the pattern varies
across farm sizes. Female-headed households of less than 3 hectares have a lower
adoption rate than larger farms headed by females, indicating that once women are

able to overcome resource constraints, they are just as likely or even more likely to
become technological innovators.
Women play an important role in agriculture in both female- and male-headed
households. Overall, about half the cultivated area is either independently or jointly
managed by women. This share is highest for local maize and traditional cereals (60
and 70 percent, respectively). Women have less involvement in hybrid maize than in
any other crop, with only 25 percent of area being independently or jointly managed
by women. Moreover, adoption of hybrid maize by a household tends to reduce
women's share in crop management and agricultural decisionmaking, independent of
farm size. This may be because women have less access to resources such as credit,
inputs, and human resource improvements, which are essential for producing the new
crop varieties, or it may reflect men's desire to control income from cash crops.
Overall, women provide nearly 60 percent of family labor in agriculture, but with
adoption of hybrid maize, men tend to shift from nonagricultural activities to agricul-
ture, thereby increasing their share of labor input. Although the amount of time that
women spend in farm work is reduced with hybrid maize adoption, the time women
spend on household maintenance activities increases.
The distribution of crop income within households reflects the extent of house-
hold members' participation in crop management. Therefore, women's share of
income and the value of their time relative to men's declines with adoption of hybrid
maize. The failure to use women farmers effectively, both in female-headed house-
holds and those headed by men, contributes to productivity losses by shifting
women's labor away from farming activities.
Policies that support participation of women in decisionmaking and production
of improved grain varieties not only could improve efficiency but could also improve
household food consumption and children's nutritional status. Women's share of
income and the time they spend in household maintenance activities are significantly
positive factors in improving overall household dietary intake, but only women's
income share is significant for improving child nutrition. This suggests that patterns
of child care available are compatible with women's work in rural Zambia. The
trade-off observed between women's work at home and household food consumption
is very small in absolute terms and could be reduced with better access to improved
technologies for household maintenance activities, such as hammer mills for grind-
ing grain. With a larger share of income, women are better able to obtain access to
such household maintenance improvements. For example, women themselves now
pay for the majority of household food processing costs. Viewing women's mainte-
nance and home production roles as simple trade-offs in family and especially child
welfare is therefore not justified, given all the dynamics involved.
In examining household food intake, this study uses a modified food expenditure
record to compute calories and protein consumed. Micronutrients analyzed include
iron, calcium, and vitamins B,, B2, and B3, all of which are important in energy
metabolism. Diet diversity is also measured. Results indicate that areas that have a
high level of adoption of hybrid maize also have a higher level of food intake than
areas of low adoption. Looking at the household level, however, with adoption of
hybrid maize, intakes improved only for the smaller farmers. The larger farmers who
adopted hybrid maize actually had lower consumption of nutrients. This finding is
consistent with the limited profitability found beyond 4 hectares of hybrid maize
planted. This adoption pattern helps reduce income inequality between small and

large households, while increasing income inequality within large farm households.
Decisions on household food consumption and income are closely interrelated, which
is plausible in a farming system facing pronounced seasonal labor shortages, where
farmers cannot provide the necessary labor if the supply of food available is insuffi-
cient to maintain their energy. If they sell their food crops to increase income, their
food supply falls. Similarly, if they denote more labor to the cash crop, the food crop
will suffer. Measures to improve food consumption are therefore likely to be as
effective as measures to improve income in making sustainable changes in welfare.
Analysis of the nutritional status of children shows both household income and
women's income to be significantly positive for the longer-term nutrition indicators,
such as height-for-age, but women's time spent on household maintenance activities
is not significant and contributes more to short-term nutrition improvement. Whether
a male or a female headed a household was not a significant factor in improving child
nutrition: who manages the crop and therefore allocates the income from it is the
important factor. Better access to health services and improved sanitation facilities
are significantly associated with improved child nutrition. Some catch-up growth in
height between the ages of 5 and 10 years is indicated, and there is no difference in
the nutritional status of boys and girls.
Efforts to promote productivity gains through better access to inputs by the
smaller farmers, and to ensure access to physical and human resources by women
including those in male-headed households, will be important if the full potential of
new technologies for improving food production and welfare of the population is to
be realized. The progress of new hybrids and composite varieties of maize in the
farming system also needs to be monitored to ensure that they are being grown not
only as cash crops but also as food crops and, therefore, receiving the same degree of
priority as local maize. This will be facilitated by the effects of market liberalization
on incentives for better on-farm storage for maize. Spread of small-scale food
processing facilities should also have a favorable impact on the integration of
improved maize varieties into the farming system. Other areas where policy attention
is needed are the reduction in diet diversity and micronutrient intakes observed in
areas with higher levels of adoption and the increase in welfare inequalities between
high- and low-adopting areas.



Agricultural growth in Zambia is increasingly recognized as central for sustained
improvement in economic growth of the nation and food security and nutrition of the
population. In the past, there was a pronounced urban bias in Zambia's development
strategy, and this was reflected in lower levels of income and nutritional status in
rural areas than in urban areas, a declining rate of agricultural growth, and high rates
of rural-urban migration. This was accompanied by unsustainable growth of urban
food subsidies and public-sector expansion. However, since the mid-1980s, Zambia
has undertaken the difficult process of structural adjustment, with accompanying
efforts to reduce public-sector control of agricultural prices and markets for both
inputs and outputs. This process has raised the prospect of better incentives for
agricultural production growth.
The need for technological change in Zambian agriculture is likely to be critical
during this period of transition for several reasons. First, traditional agriculture relies
primarily on area expansion for achieving growth, and this alone has limited potential
for sustained growth. Second, the dismantling of public-sector control of pricing is
likely to limit, geographically, the areas with a comparative advantage for producing
marketed maize surpluses. Increased agricultural productivity will therefore be es-
sential both for maintaining food supplies for the large urban populations and for
expanding agricultural improvements and incomes across the country.
This report examines the nature and effects of technological change in maize
production in Eastern Province, Zambia. It is primarily a study of impacts, and in the
process, it also uncovers characteristics of adoption of new technology that may help
clarify why adoption may have had a limited welfare benefit for households. In
particular, the report focuses on changes in the intrahousehold dimensions of farm
families-changes in women's role in crop management and their relative position in
resource allocation of both money and time. In this process, a clearer picture of the
intrahousehold changes that take place with adoption of new technology is drawn,
and this is linked to both the policy environment and to welfare outcomes in terms of
nutritional status.
It has been postulated that since women continue to play an important role in the
production of household food crops in many parts of Zambia, agricultural growth
programs and strategies need to ensure that women have access to resources and
inputs. Absence of such measures may lead to limited success of growth measures
and also failure to gain nutritional benefits from such growth. Earlier IFPRI studies
in Kenya, The Gambia, and Rwanda on technological change and commercialization
in agriculture have shown limited nutritional benefits from these efforts, even where
income benefits for the household were noted (Kennedy 1989; von Braun, Puetz, and
Webb 1989; von Braun, de Haen, and Blanken 1991). In this report, these associa-

tions are further examined to investigate the consequences of intrahousehold dynam-
ics that reduce the relative decisionmaking role of women in the process of agricul-
tural production while increasing the demand for women's time, especially in house-
hold maintenance activities. These dynamics are commonly observed during
agricultural commercialization and can be traced to limited improvements in real
household food security and nutritional status.
The challenge in proposing policy remedies for such dynamics is to examine both
policy and cultural issues. Even though the effects of policies occur within the existing
cultural context, policies and their institutional structures have a large responsibility in
shaping the effects. In practice, the solutions need to emerge from within the commu-
nities affected. The choice for action, however, is with policymakers.

Importance of Maize Production

Although maize is only one of the many grain and root crop staples consumed by
the Zambian population, it is overwhelmingly the favored staple food in urban areas.
In rural areas, it is the main staple food in the central, southern, and eastern parts of
the country (Table 1). Since these are the most densely populated parts of the country,
maize emerges as the single most important food item in the Zambian diet. Although
maize production has been encouraged in other parts of the country through a variety
of public policy measures during the past six decades, other grains, such as finger
millet, pearl millet, sorghum, and cassava, are still the predominant staples in the
northern, western, and northwestern parts of the country.
Several historical factors have contributed to the spread of maize in Zambia.
Diffusion of maize followed its introduction into the Congo basin around the six-
teenth century. Maize, along with cassava, which was introduced about the same
time, first appeared in the western and northwestern parts of the country and gradu-
ally spread eastward. However, in the early twentieth century maize received a big
production boost with the opening of the interior of the country (then Northern

Table 1-Per capital consumption of basic staples and total calorie intake in
rural areas, by province, 1980

Sorghum Wheat Total Calories
Province Maize Cassava and Millet and Rice per Day
Central and Lusaka 171.1 4.0 19.2 17.0 2,103.4
Copperbelt 108.0 3.0 53.0 19.0 1,783.7
Eastern 143.0 0.6 6.5 2.2 1,524.2
Luapula 32.5 74.0 12.5 4.9 1,184.9
Northern 33.6 50.0 54.0 2.8 1,343.0
Northwestern 46.8 34.0 46.0 1.2 1,234.9
Southern 155.0 0.3 14.0 4.4 1,733.3
Western 90.3 28.2 13.0 2.6 1,365.2
Average 102.2 23.8 24.4 5.4 1,517.0

Source: Food and Agriculture Organization of the United Nations, Zambia: Comprehensive Agricultural Devel-
opment and Food Security Programme (Rome: FAO, 1991), 21.

Rhodesia) to mining and settler interests. The earliest documented agricultural policy
is that of encouraging European settler farmers to grow maize to supply food to mine
workers (Dodge 1976). Gradually, some of these incentives were extended to the
African farming communities close to the mining centers and along the original "line
of rail." This line of rail has tremendous historical significance in the settlement
pattern of Zambia. It is the train route that connects the mining areas in north central
Zambia with ports in South Africa. Although a new line of rail has emerged with the
opening of the Tazara line to Tanzania, the areas that lie along the original line of rail
are still the best served infrastructurally.
Of the other staple food crops traditionally grown-sorghum, pearl millet, finger
millet, and cassava-none has received the policy support given to maize. This is
primarily due to the force of the urban demand for maize. By the early 1980s, nearly
half of the country's population resided in urban and periurban centers, and maize
dominated agricultural research and extension programs as well as the agricultural
pricing and marketing policies of the government of Zambia. As a consequence, its
importance in both agricultural production and in food consumption has grown
steadily over time.

Technological Change in Maize Production

Growth in maize production during the past two decades is primarily due to area
expansion. According to World Bank (1992) estimates, maize yields have declined
by 2 percent annually, while area has expanded by about 11 percent, for a production
growth rate of 8.5 percent during 1974-89. Much of the decline in maize yields is the
result of the rapid expansion into relatively marginal areas encouraged by the agri-
cultural production and extension policies.
Area expansion was fueled by a high growth rate of population (3.7 percent
annually) and by some reverse migration during the period of structural adjustment.
In addition to population growth, use of mechanical land preparation technologies
such as ox-plow and tractor cultivation in the 1970s also encouraged expansion of
area under crops. An abundant supply of land has facilitated the area expansion and,
according to a World Bank analysis, "it has tended to encourage smallholders, who
face a labor constraint, to substitute land for labor by adopting suboptimal crop
husbandry practices (for example, single weeding under high fertilizer application)"
(World Bank 1992, 31). Improved land preparation technology facilitates area expan-
sion, only to impose a labor constraint on subsequent farm operations that are critical
to raising yields. To some extent, farmers compensate for the labor shortage by
increased use of improved seeds and fertilizers, but with a lower output response. To
the extent that the availability of mechanical traction allows more land to be culti-
vated and makes it more likely that farmers will adopt the improved seed-fertilizer
package, they can increase their net profitability despite the ensuing labor constraint
(CIMMYT 1990, 34; Pingali, Bigot, and Binswanger 1987).
The yield-increasing technological changes available for maize production in-
clude improved varieties and modern inputs. Hybrid maize varieties have been
available in Zambia since the 1960s, and were introduced to the smallholder sector
around 1970. Reports available on growth in hybrid maize and other high-yielding
varieties (HYVs) of maize are mixed. Aggregate yield increases, however, have been

limited by expansion of maize area into ecological zones not considered suitable for
maize production and by reduction of fallow in the more densely populated but
suitable plateau zones.'

Hybrids and Other Improved Varieties
The most widely used hybrid maize is SR52, which was first introduced in
Zambia for smallholder production in the late 1960s. Although other hybrids have
been developed since then, none has matched the yields of SR52. Other hybrids that
have been released include ZH1, SR11, ZCA, and SR13. Production of seed has,
however, been sustained only for SR52, which accounted for 90 percent of commer-
cial maize seed produced in 1980/81, and ZH1, which accounted for the remaining
10 percent in that year (Zambia Central Statistical Office 1981). Both of these
varieties offer substantial yield advantages over the local varieties, especially in their
response to fertilizer, but they are both long-duration varieties requiring 170 days to
mature, which makes it critical to plant them early in the season.
More recently, Centro Internacional de Mejoramiento de Maiz y Trigo (CIMMYT)
has supported the development of high-yielding, open-pollinated composite maize
varieties that are expected to reduce the annual cost and availability risks to farmers
that are inherent in hybrid seed. Field trials on new composite varieties in the 1980s
showed that they were not able to improve on the performance of the SR52. However,
given that farmers do not have to buy seeds each year, they are still likely to be well
received. Short-duration maize varieties are supposedly available but difficult to
obtain in practice.
The progress in adoption is likely to be constrained by the seed production and
distribution mechanisms available in Zambia. The Zambia Seed Company (ZAM-
SEED), a state-owned parastatal, is the only producer of improved seed, and it
generally has not kept pace with the growth in demand for improved seed. During the
1970s, when growth in adoption was very rapid, seed availability increased only
threefold (Zambia, Central Statistical Office 1981). Distribution of seed to producers
is primarily through the National Agricultural Marketing Board (NAMBOARD), and
in some provinces this function has been transferred by NAMBOARD to provincial
cooperatives. In addition to the availability of improved HYVs of maize, agricultural
extension and availability of credit and fertilizer are important factors in their
adoption. As mentioned earlier, animal traction is a factor inasmuch as it allows area
expansion, which, for various reasons, makes adoption of hybrid maize easier.

Progress in Adoption of Improved Maize Varieties
There is little reliable information on the adoption of improved crop varieties or
other agricultural technologies in Zambia. Here, as elsewhere, use of improved maize
germ plasm is difficult to estimate precisely because one cannot easily distinguish
between improved and unimproved materials. As a consequence, tracing sources of
seeds replaces visual inspection (CIMMYT 1990). According to CIMMYT sources,
hybrids and other improved, open-pollinated varieties accounted for 64 percent of

'Yield reductions with continuous maize cultivation are especially pronounced in highly weathered soils
such as those in northern Zambia, and occur despite fertilizer applications and soil pH control.
Micronutrient depletion of the soil is a factor in this (SPRP 1987, 38).

maize area in 1985/86, but only 46 percent by 1988/89. Hybrids alone accounted for
53 percent and 45 percent of maize area, respectively, in those years (CIMMYT
1987, 1990). World Bank sources, on the other hand, indicate the share of maize area
planted to hybrids increased from 47 percent in 1984/85 to 60 percent in 1988/89
(World Bank 1992). Given uncertainties such as these and lack of good national
statistics on area planted to improved varieties, perhaps some idea about trends can
be obtained from small-scale farm- and household-level surveys.
In the smallholder sector, hybrid maize was introduced in the late 1960s and began
to take hold in the farming system very quickly. Its earliest adopters were, not
surprisingly, the larger farmers in the traditional sector. A survey by Harvey (1973) in
the Kalichero area of Chipata District, Eastern Province, in 1972 showed that about 7
percent of farmers grew hybrid maize and this accounted for less than 5 percent of
maize area. All of the adopters were among the largest 15 percent of farmers.
Hybrid maize production grew rapidly during the 1970s. Fieldwork conducted by
the author for the International Food Policy Research Institute in collaboration with
the National Food and Nutrition Commission and the Rural Development Studies
Bureau, University of Zambia, in 1981/82 showed that in the same part of the district
surveyed by Harvey, 55 percent of the maize area was now sown in hybrids. At the
same time, however, for Chipata District as a whole (including valley sites), the rate
of adoption of hybrid maize (33 percent) was lower than that for Kalichero. The same
study found a substantial decline in hybrid maize use in 1982-down to about 42
percent in share of maize area in the Kalichero sites.
In the present IFPRI survey in Eastern Province, the rate of adoption in Chipata
District was about 23 percent of maize area in 1986, suggesting that there may have
been a downward trend in HYV use there during the 1980s. This is consistent with
the CIMMYT estimates and also with reports of declining maize yields from the
World Bank.
Even though the study sites in the different surveys are not identical, the results
from farm-level studies over the past two decades suggest that adoption of hybrids
grew rapidly during the 1970s but stagnated or even declined during the 1980s.
The rate of adoption may be different in other parts of the country. Reports from
provinces along the line of rail suggest that adoption grew more rapidly during the
1970s, especially in Central and Southern provinces as compared with Eastern
Province (CIMMYT/GRZ 1978). There is also some indirect evidence that there has
been rapid growth in adoption of hybrid maize as a rural cash crop in nonstaple maize
areas. Primary emphasis on maize production, especially as a commercial crop, has
been the focal point of agricultural extension and development efforts in all parts of
the country, including some areas not considered suitable for maize cultivation
(Wetterhall 1981; Evans 1981; Keller and Mbewe 1988). The most rapid growth in
marketed maize production has been in areas where maize is not an important staple
food and nearly the entire increment in production is likely to be marketed, as seen
in Table 2 for Luapula, Northern, Northwestern, and Western provinces.
Factors Conducive to Adoption of Hybrid Maize
Agro-ecological Conditions. Most of the literature on Zambian agriculture
clearly states that maize cultivation is not suitable in either the northern high rainfall
or the western sandy areas. It is also in these areas that maize is not a primary staple
food. Because agricultural price policy and agricultural extension emphasis has been

Table 2-Growth of marketed maize production, by province, 1973-89

Marketed Production Percent Change
Harvest Year Harvest Year Range
Province 1973 1977 1981 1985 1989 1973-81 1981-89 1973-89

Central and Lusaka 2,510 3,278 3,001 2,500 4,961 20 65 98
Copperbelt 132 70 37 242 484 -72 1,208 267
Eastern 501 942 1,184 1,781 2,471 136 109 393
Luapula 16 32 30 59 386 88 1,187 2,313
Northern 59 212 328 738 1,446 456 341 2,351
Northwestern 26 40 42 746 160 62 281 515
Southern 1,172 3,077 3,039 1,583 3,358 159 11 187
Western 15 86 43 92 285 187 563 1,800

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

on promoting maize production throughout the country,2 however, adoption of hybrid
maize has grown in all parts of the country.
Seed and Fertilizer Availability. Information on the seed industry in Zambia is
scarce. ZAMSEED, a parastatal, provides very little information on the production and
distribution of different varieties of maize seed. Available statistics indicate that SR52
is still the main variety. SR52 is a long-duration variety, requiring 170 days to mature,
which may explain why its yields have been substantially lower in the smallholder
sector. Because it requires early planting, it is well suited to commercial farms but not
to traditional farms. Since farmers prefer to plant the plots that supply their food before
planting hybrids, late planting of hybrid varieties is one of the main constraints to its
productivity (CIMMYT/GRZ 1978). Varieties that are suitable for later planting,
MM502, for example, give a similar yield and fertilizer response and may be more
suitable for small-farm adoption (Central Province 1984). Demand for the short-
duration varieties is strong, but their availability is poor (Keller and Mbewe 1988).
The distribution of improved seed and fertilizer is through NAMBOARD and the
provincial cooperatives. The cooperatives primarily make inputs available to mem-
bers who also have access to the cooperative's credit program. Nonmembers theoreti-
cally have access, but whether they receive inputs is likely to be a function of first
meeting demand from members, the input supply situation, and cash availability.
Although fertilizer use in Zambia appears to have grown steadily during the
1970s and 1980s (Zambia, Ministry of Agriculture and Water 1983; World Bank
1992), the numerous problems with its distribution system have limited effective
yield responses. Zambia imports the major part of its fertilizer supply3 and channels
the input through the parastatals (NAMBOARD and the provincial cooperatives).
Problems with imports, the domestic distribution mechanism, and poor rural roads
and storage have made timely availability of fertilizers a chronic problem.

2For example, a program of the 1980s called the Lima Crop Extension Program was aimed at improving
the productivity of smaller farmers by assisting them in the efficient use of chemical fertilizer.
30ver two-thirds of nitrogenous fertilizers and all nonnitrogenous fertilizers are imports (FAO 1991).

Most analysts agree that the monopolistic nature of fertilizer marketing and
distribution has contributed to the inefficiency of the system. During the 1992 harvest
year, some licenses were being made available to private traders to market fertilizer,
but mandated price subsidies may deter any substantial private involvement (FAO
1991). Seed distribution closely mirrors that for fertilizer. In this case, importation
problems do not enter the picture; the main bottlenecks are inadequate production of
improved varieties and the inefficient distribution mechanism. Available information
suggests that seed production has lagged behind the development of short-maturing,
open-pollinated varieties that are well suited to smallholder adoption (Keller and
Mbewe 1988; CIMMYT 1990).
Improved Means of Cultivation. The traditional, low-tillage hoe cultivation
method is still widespread in Zambia. Use of ox-driven plow cultivation, relatively
recent, has been promoted in the country's agricultural growth strategy. The primary
advantage of ox cultivation is that it makes it easier to plant a larger area (in a
labor-constrained agriculture), and therefore it is considered essential for promoting
the surplus production essential for commercialized agriculture.
There are still large parts of the country where ox cultivation is neither practiced
nor feasible. The high-rainfall, northern areas raise cattle but, until recently, not
oxen. In areas, including parts of Eastern Province, where tsetse flies carry the deadly
trypanosome parasite, it is infeasible to maintain oxen or cattle. Hence, farmers in
those areas only have access to hoe cultivation. In the 1970s, when the country's
foreign exchange situation was relatively comfortable, there was an effort to promote
tractor cultivation by local agricultural camps that would hire out tractors. By the
1980s, these had all fallen into disrepair, and the farmers in the traditional sector in
tsetse areas again had no recourse but to use hoe cultivation.
To the extent that farmers in the smallholder sector plant HYV maize as a cash
crop and depend on local maize or other staples for their food, use of ox-plow
cultivation provides a means of expanding HYV production through area expansion.
Numerous studies in Zambia and other countries in Sub-Saharan Africa have shown
that increasing farm size allows farmers to expand cash crop production, while
retaining as much acreage (or even more) for their own crop production. This is
confirmed by studies in Zambia that show ox-plow users to be four-to-five times
more likely to adopt HYVs than hoe cultivators, and they also plant a much larger
area to the new varieties (ARPT/Eastern Province 1988).

Rural Infrastructure
In many parts of the world, improvement in rural infrastructure has been a major
factor in providing farmer incentives for investing in improved agricultural technolo-
gies, for stimulating growth in off-farm employment opportunities, and encouraging
nonfarm growth linkages. Evidence of these effects is most clear-cut in Asia, where
there is a longer history of improvement in rural infrastructure (Hazell and Roell
1983; Ahmed and Hossain 1990).
Indirect benefits of improved infrastructure include better access to and use of
social- and consumer-oriented services that help improve standards of living. In
addition to improved health and education, improvement in rural physical infrastruc-
ture components, such as access to roads and markets, affects diets in a variety of
ways. It opens up many new income-earning options for farm families; it may lower
consumer prices and increase the variety of foods consumed by improving access to

markets; and it often alters dietary preferences and eases women's work burdens by
improving food-processing opportunities. More explicit work is needed on the effects
of rural infrastructure development in Zambia.
By all indications, spread of improved agricultural technologies in Zambia would
be facilitated with investment in rural infrastructure through improved access to input
and output markets, better functioning of labor markets, and better access to agricul-
tural services. In addition, it is also likely to improve the acceptance of new maize
varieties as part of the diet through the expansion of local processing facilities such
as hammer mills, which make it easier to process hybrid maize. The home premilling
process that is used for local maize is unsuitable for hybrids.

Policy Environment for Hybrid Maize Adoption
Price Policies Favor Maize over Other Crops
At independence in 1964, Zambia inherited an agricultural price policy frame-
work that was primarily geared toward commercial production of maize on the
large-scale estate farms that provided maize to urban mine workers at subsidized
prices in order to ensure an elastic supply of workers for the mines. Marketing of
maize from producers to consumers was highly regulated and carried out by parasta-
tals. Producer prices v.ere fixed annually on the basis of numerous criteria, including
cost of production and fair return to producers.4
In the postindependence period, this basic price policy framework has remained
intact. However, a major additional objective was added in the early 1970s: produc-
ers in the traditional sector were to be subsidized at the cost of the large-scale estate
sector in order to reduce reliance on the estates for the marketed supply of maize.
This was achieved through a panterritorial pricing policy that, by establishing a
uniform producer price in all parts of the country, subsidized farmers in remote rural
areas and taxed the large-scale commercial enterprises established near the major
town, road, and rail networks.
The price policy for agricultural products favors maize over other crops. This is
reflected in the guaranteed producer prices paid over the past 15 years for maize
relative to other crops. Figures available indicate that producer prices for maize
increased 1,100 percent between 1975 and 1986, compared with 849 percent for
soybeans, 773 percent for groundnuts, and 713 percent for sorghum. Increases for
tobacco, wheat, sunflower, and cotton--other important commercial crops-were
lower (Jansen 1986). Since 1986, the emphasis on maize prices has increased even

Promotion of Hybrid Maize as a Cash Crop
Consistent with the factors that were driving the emphasis on maize production
was the focus on marketed production. Until recently, even production statistics
interchangeably used the concepts of increase in production and marketed produc-
tion. The technology package promoted for maize also matched this policy with the

41n the ex post facto analysis, however, these prices remained between the import parity (high) and export
parity (low) prices in the years for which this information was examined (Kumar 1987b).

linked hybrid seed-fertilizer-credit package made available primarily to the emergent
farmers in the smallholder sector.5
There were other characteristics inherent in the management of the production
policy that contributed to production of hybrid maize as a cash crop. One of these was
the unpredictability of when payment would be received for the maize harvest sold
to the parastatal. Liquidity and management problems often led to a wait of several
months before farmers got paid. Only the larger, more economically secure farmers
could withstand this delay, because it meant that sufficient food stocks had to be
retained to tide them over the wait.
Physical characteristics of the hybrid maize grain also fit in with the cash crop
role. The hybrid maize varieties that have commonly been available belong to the soft
(dent) grain type. They are difficult to process in the manner preferred traditionally,
that is, wet milled-milled after being soaked, pounded, and sifted to remove the
pericarp. This partially processed grain is then ground in the local hammer mill or, if
none exists, at home. The refined meal is preferred for taste, storage, and cooking
qualities. At the village level, the soft grain varieties are taken to the hammer mill to
be ground into whole, unrefined flour, but this form is less preferred. The consumer
subsidies for maize in Zambia have usually made it much more attractive to sell
hybrid maize and buy back the refined form of subsidized maize meal if necessary.

Price Policy Environment Favors Sale versus Storage
Past maize pricing and marketing policies that have discouraged rural storage in
favor of sale after harvest are also likely to have encouraged production of hybrid
maize, at least in the short run. In the long run, lack of improvements in rural storage
and increasing problems with parastatal handling of the marketed maize may have
contributed to a plateau or even reduction in hybrid use, as is indicated by CIMMYT
and farm-level statistics reported earlier (CIMMYT 1987; 1990).
One defect of the maize pricing policies has been the absence of a seasonal price
increase for maize. The main contributor to this effect has been the large consumer
price subsidy for maize meal. To the extent that this subsidy is being eliminated, the
incentive for farm storage will improve. As a consequence of the lack of a seasonal
price increase, virtually all maize marketing have been completed in the postharvest
period when the guaranteed producer price can be obtained by selling to the grain
marketing parastatals. Hybrid maize constitutes the main component of marketed
maize production, while local maize is mostly stored for home consumption. Pur-
chases, if required, have generally been of the preferred refined commercial maize
meal product (breakfast meal) which, until the mid-1980s, actually had a higher
subsidy than the coarser maize meal product (roller meal). Rural areas where these
products are readily available have a flat seasonal producer price for maize, com-
pared with a 17 percent seasonal increase in the price of groundnuts, a 23 percent
increase in the price of beans, and a 14 percent increase in the price of sorghum
between the postharvest months of June-July and November-December of 1981 in
plateau areas of Chipata District (Kumar 1984).

5"Emergent farmers" is a classification of farm size used in Zambia to denote farmers in the traditional
sector farming 10-20 hectares of land, whereas the majority of farmers cultivate less than 10 hectares.
Therefore, these are the larger farmers and are likely to plant more hectares to commercial crops.

The extent to which hybrid maize has fit into the farming system of the small-
holder Zambian farmer primarily as a cash crop may also have contributed to its
failure to achieve its yield potential and, consequently, its acceptance. Since farmers'
first priority is to ensure food security, they protect that by preferentially allocating
area to local maize. This is indicated by the amount of area allocated to local maize
production, which remains much the same on a per capital basis across farm size and
level of hybrid maize adoption (as seen later). Farmers' priorities are also reflected
in their planting of local maize at the onset of rains and hybrid maize later. As a
result, the planting of hybrid maize is often delayed past its last recommended date,
which is in mid-December (Harvey 1973, 20). This late planting of hybrid maize
appears to be an important constraint in achieving its yield potential.

Input Subsidies
Subsidizing of both improved seed and fertilizer has been a steady feature of
Zambian agricultural policy. However, there has been little careful analysis of the
impact of these policies on agricultural productivity and input use patterns. To a large
extent, political economy has dictated these policies, especially the panterritorial
pricing policy for both inputs and outputs. Despite progress in structural adjustment,
available information indicates that these subsidies have been hard to eliminate
(Graham 1994, Ch. 6). As a consequence, the production of inputs locally also has to
be subsidized and is likely to be hampered. For example, ZAMSEED has found it
difficult to maintain a supply of seed from commercial seed producers, and thus to
meet demand, especially from widely dispersed smallholders. Some efforts have
been made, beginning in the late 1980s, to price locally produced fertilizer (primarily
urea) at import parity levels, and production levels for that are reported to have
improved as a consequence.

Access to Improved Technologies by Women
By the late 1980s, Zambian policymakers were beginning to recognize the
importance of integrating women into sectoral development policies-to identify
women farmers as a target group in agricultural sector strategies. Evidence was
growing that women's role in agriculture was significant and that previous public
sector programs had failed to reach them. Studies have shown that women farmers
are less likely to receive agricultural credit, and when they do, the amounts are
significantly lower than those for men (J. T. Milimo 1989). Women are poorly
represented in agricultural extension and training programs (Eklund 1985; Che-
noweth 1987; Bliven 1991), and they are also likely to face obstacles to stable land
tenure not encountered by men (J. T. Milimo 1989).
However, improving women's access to improved technologies and resources
requires more than political posturing and pronouncements. There is a wide gap
between so-called policy statements and effective action (Keller and Mbewe 1988;
Keller 1990). The problems are compounded by cultural practices that constrain
women's rights and access to resources. Some of the major actions required include
an improved awareness for both men and women at all levels-farmers, service
providers in agricultural and other institutions, and policymakers-of the need for
providing production opportunities and of the costs of inaction.



Nutrition Situation in Zambia

Protein-calorie malnutrition is a widespread and serious public health problem in
Zambia. Zambia is not one of the most severely affected countries in Africa, but
levels of protein-calorie malnutrition are above average (OAU/UNICEF 1992). A
national survey in 1990 found 25 percent of children between the ages of 6 and 60
months to be undernourished (Cogill and Zaza 1990).6 Malnutrition occurs more
often in areas where infrastructure is poor, especially provinces off the line of rail and
in urban squatter settlements. Eastern Province, where this research was carried out,
has some of the best agricultural areas in Zambia, but it is an outlying province and
therefore levels of protein-calorie malnutrition tend to be higher there (Ekberg and
Mwale 1988).
There is also evidence of varying degrees of Vitamin A, iron, and iodine deficien-
cies in different parts of Zambia. Vitamin A deficiency occurs primarily in the
northern areas, with sporadic cases in other parts of the country. Anemia, on the other
hand, is a serious public health problem throughout the country. In 1970/71, it was
present in about 70 percent of children under five years of age, 45 percent of men, 17
percent of all women, and 22 percent of pregnant and lactating women. Iron defi-
ciency is only part of the etiology of anemia, with parasitic infections such as
malaria, hookworm, and bilharzia being major causes (UNDP/FAO 1974). Goiter,
resulting from primary or secondary iodine deficiency, affects people living primar-
ily in the western and northern regions, where soils are poorer and heavily leached
due to high rainfall and soil erosion, and where people rely heavily on cassava in the
diet (OAU/UNICEF 1992).7
The nutrition of rural populations is inextricably linked with agriculture as the
central source of food and livelihoods. Even though there have been high rates of
rural outmigration in many parts of Zambia, remittances are, in general, an insuffi-
cient source of support for rural households with migrant members. Also, off-farm
income sources such as wage employment or self-employment are extremely limited
in areas outside the line of rail, where agricultural commercialization is not wide-
spread. A number of area, household, and intrahousehold characteristics of agricul-

6Below two standard deviations from median weight-for-age of accepted international reference
standards (U.S. National Center for Health Statistics [NCHS], as adopted by the World Health
7High levels of cassava consumption have been noted to be goitrogenic, especially when the cassava is
not adequately processed to remove the toxic compounds present in most varieties grown in Africa.

tural production influence levels of nutrition. Area factors include the natural re-
source base and agro-ecological characteristics, agricultural technology, cropping
patterns, and the extent of commercialization of agriculture; household factors in-
clude farm size, productivity, off-farm income, fluctuations in income, education,
and access to food; and intrahousehold factors include women's resource access and
work patterns. These and other characteristics of agriculture affect the supply of
foods and their prices and the effective demand for food and other services. Together
they are among the major determinants of nutrition of population groups.

Role of Technological Change in Agriculture in Nutrition
Since the early days of the Green Revolution in the 1970s, there has been an
ongoing debate about the effects on human welfare of technological change in
agriculture. Generally, arguments have taken two opposing views. The proponents of
each could be classed as the optimists and the sceptics. The optimistic view is that
labor-intensive technological change in agriculture improves land and labor produc-
tivity, raises employment and incomes of the poorest households, and leads to large
incremental improvement in aggregate food production and, hence, in food prices
and affordability (Mellor 1966; Pinstrup-Andersen and Jaramillo 1991). In addition
to the direct effects expected of these technologies, which are essentially embodied
in better seed and fertilizer use, they generate rapid agricultural growth with linkage
effects that stimulate rural investment in the off-farm sector (Hazell and R6ell 1983).
These linkage effects are expected to be especially favorable for the distribution of
welfare benefits, particularly improvements in food security and nutrition, to the
neediest groups in the population, including the landless (Ahmed and Hossain 1990;
Kumar 1992).
On the other hand, the sceptics point to potential problems such as excessive
concentration of wealth, unsustainability, environmental problems with improper use
of inputs, loss of plant genetic diversity in the low-income countries, and exploitation
by seed-producing multinationals (Brown 1970; Messer and Heywood 1988). The
works of authors such as Lipton and Longhurst (1989) and Hart (1989) have been
useful in putting the potential of these technologies in perspective and in pointing out
that the results are not always as favorable as expected. Others have been concerned
about the effects of localized improvements in the agricultural sector on the growth
of poverty in areas where such improvements have not occurred (Pradhan 1993).
In addition to the issues of the level and distribution of household welfare, the
intrahousehold dimension also entered into the debate during the 1980s. This was
stimulated primarily by work in Africa, and was associated with efforts to bring new
agricultural technologies to farming populations. Work done by several authors
showed that although women generally had distinct responsibilities in agricultural
production decisions, they were seldom able to obtain access to improved agricul-
tural inputs.8 This limited access by women was observed to be a factor in the limited
success of these programs in raising agricultural productivity (Dey 1992). Other side
effects of adoption of improved agricultural technologies by smallholder farmers that
have been cited in other studies include a reduction in the share of household income

See Kumar 1987a for a review of this work.

received by women and an increase in their share of household labor, often including
women's labor devoted to agriculture. These intrahousehold dynamics have gener-
ally been viewed as a negative factor in producing improvements in child nutrition,
since both a reduction in share of household income earned by women (Garcia and
Pinstrup-Andersen 1987) and an increase in their workload (Kumar 1978; McGuire
and Popkin 1989) decreases the quality of child care.
Recent reports that have analyzed the effects of household adoption of improved
agricultural technologies have observed that in many cases, technological change and
commercialization of agriculture are virtually synonymous (von Braun and Kennedy
1994). There is, however, wide diversity in the types of effects observed. A number
of studies have documented this diversity in household and intrahousehold changes
with technological change or commercialization in smallholder agriculture. Overall,
household diets improved in terms of dietary caloric availability and intake, and this
led to some improvement in child nutrition. However, the magnitude of the improve-
ment was often small; in Kenya, for example, a doubling of household income led to
a 7 percent improvement in child nutritional status.9 Similar improvements in house-
hold income produced somewhat better nutritional gains in Malawi and Rwanda
(Kennedy 1993).

Welfare Effects of Technological Change and
Commercialization in Agriculture in Zambia
Few detailed analyses have considered the effects on nutrition and health of
agricultural change in Zambia. Useful insights have, however, been obtained from
studies associated with development programs, such as the Integrated Rural Devel-
opment Program (IRDP), in different parts of the country. In cross-section compari-
sons, households with a higher degree of commercialization of crop production were
generally found to have a higher incidence of child malnutrition than subsistence
producers (FAO 1984; IRDP 1986). Anthropological research in one area suggests
that there was a reduction in cropping and dietary diversity, especially in ingredients
used in relishes or sauces served with the staple food, and a greater reliance on basic
staples. In addition to a reduction in dietary diversity, an increase in the workload of
women was also found to be a factor in the lower levels of child nutritional status in
areas where agricultural development programs were being promoted (Keller and
Mbewe 1988). In the present report, these and other factors associated with advance-
ments in agriculture will be analyzed in detail.

Analytical Approach and Linkages

Technological change in agriculture has long been accepted as a necessary
condition for accelerating growth in food production in Sub-Saharan African coun-
tries. The decline in per capital food production and availability in these countries
over the past two decades bears testimony to the pressing need for addressing this
problem with speed and clarity.

9Z-score for weight-for-height.

Technological change in agriculture to promote increasing yields of food crops is
being emphasized for its expected effects on both supply and demand for food.
Though the precise factors underlying adoption of new technology may be relatively
complicated to analyze, tracing the distributional and welfare consequences of adop-
tion is even more problematic. Adopters are more likely to experience income benefits
at the household level, compared with nonadopters. However, translation of additional
income into consumption and nutritional benefits is mediated not only by the profitability
of hybrid maize production, but also by a wide range of resource allocation decisions.
In addition, even at the local level, the effects of adoption may be linked to secondary
income and employment generation that will benefit nonadopter households.
To trace the effects of hybrid maize production, the report looks at area, household,
and intrahousehold conditions. It is, however, beyond the scope of this work to consider
urban and regional benefits or disincentives that could be derived from the increase in
marketed surplus of foodgrains in food-deficit areas outside the local area of adoption.
The basic analytical approach in this report is similar to other IFPRI studies on
the effects of commercialization and technological change on household food secu-
rity and nutrition, but with greater emphasis on intrahousehold dynamics. Adoption
of improved technologies is expected to influence household-level characteristics
such as disposable income and consumption expenditures that influence food con-
sumption and nutritional adequacy. At the intrahousehold level, changes in income
source, control of income, and women's workloads can influence the income and
other resources that determine income allocation for food consumption, dietary
adequacy, and levels of child nutrition among households but also among household
members. These effects are predicted in both the commonly used neoclassical and
bargaining-type models (Senauer, Garcia, and Jacinto 1988). In addition to the
household and intrahousehold effects on resource access and allocation, technological
change can also influence food demand and availability at the local level, affecting
those who are not hybrid maize adopters as well as those who do adopt. For example,
crop diversity, demand for wage labor, and demand for other local goods and services
may all be affected.

Adoption of Hybrid Maize Cultivation Technology
Farmers in Africa face different and probably greater constraints to adopting
improved agricultural technologies than farmers in Asia. These differences stem
from three major conditions in Africa: low population densities, low average produc-
tivity of the resource base, and seasonal labor bottlenecks. All of these factors make
capital accumulation and adoption of technological change more difficult (Delgado
and Ranade 1987). These problems are compounded by poor infrastructure develop-
ment, which increases riskiness of production and reduces access to local off-farm
income sources that could assist in capital accumulation to sustain technological
progress in agriculture.
For the individual farmer, adoption of improved agricultural technologies in-
volves a complex set of considerations. Three aspects of the adoption process that are
discussed in the literature are (1) risk and production uncertainty, (2) profitability and
price incentives, and (3) the technological package and components selected for
adoption and intensity of use. According to CIMMYT (1990), up to 40 percent of
maize area in Sub-Saharan Africa faces occasional drought and production uncer-
tainty. Farmer strategies are, therefore, geared to reducing risk to ensure at least

enough production to meet minimum household needs. The adoption process itself is
a reflection of that: those who do adopt have a better capacity to deal with risk and
Characteristics of adoption of hybrid maize reflect underlying conditions of risk-
coping behavior and infrastructure constraints. For example, adoption is associated
with increased farm size and with capital accumulation, but at the same time adequate
area is still assigned to food crops to meet minimum household needs. Two other
technologies are nearly always associated with hybrid maize adoption in Zambia and
other countries in the region: application of a basal or top dressing of fertilizer (or
both) and use of mechanical traction. According to World Bank analysis, the produc-
tivity of this technology package is often constrained by the unavailability of inputs
when needed, and especially by a seasonal labor constraint that is accentuated on the
large farms. The efficiency of this practice of households with the ability to expand
area under cultivation in effect substituting land (with improved technologies) for
labor in a "high input-low output system" has been questioned (World Bank 1992, 32).
Characteristics of adoption, such as which components of the package are ac-
cepted at any point in time, even though they may have a sequential component, are
jointly determined by household and area-level factors. These characteristics include
the decision to use a particular input-hybrid seed, fertilizer (plow cultivation), or
pest control-as well as the intensity of use and timing and manner of application.
Studies of the adoption of HYVs of maize in Zambia and other countries in the region
indicate that it is jointly a function of adoption of planting technologies-mechanical
traction (by oxen or tractor) and basal fertilizer (Jha and Hojjati 1993; Rauniyar and
Goode 1992; Birch-Thomsen 1990). Less is known about the level of use of inputs,
and use intensity is mostly a function of the ability of households to gain access to
the limited supplies of subsidized inputs available.
Use of oxen or other mechanical traction for land cultivation has widely been
associated with increased adoption of improved germ plasm technologies in Africa.
There are several likely reasons for that, including the ability to spread out labor input
(Delgado and Mclntire 1982) and the ability to farm a larger parcel of land, which
households desire because they wish to continue to grow local varieties for their own
consumption. Very often, new seed varieties are not compatible with the existing
storage and processing facilities that are available or local food tastes and are then
treated as a cash crop. This has been widely observed in adoption of hybrid maize in
Zambia, where in addition to the local storage and processing constraints, maize
pricing and marketing policies did not encourage on-farm storage of surplus maize
production (Kumar 1987b). These factors have led farmers to keep their original land
under local maize varieties and to plant incremental area to cash crops like hybrid
maize. As local maize milling facilities improve, farmers become more willing to
keep hybrid grain for home consumption, but only enough to last until the start of the
rainy season. The existing storage facilities do not enable the softer hybrid grain to
withstand pest attacks with the onset of the rains. To encourage improvements in
on-farm storage, there has to be a shift in pricing and marketing policies as well.
Fertilizer use, encouraged by heavy price subsidies, has been growing in Zambia.
An IFPRI survey in Chipata District in 1981/82 showed that fertilizer use was limited
primarily to hybrid maize and cotton production. A repeat survey in Eastern Province
in 1986 found fertilizer use much more widespread, including its use on local maize
(Jha, Hojjati, and Vosti 1991). There has, however, been a limited yield response to

this additional fertilizer use (World Bank 1992). This can be attributed partly to poor
timing and quality of fertilizer application, insufficient weeding, and declining soil
quality due to continuous cultivation of maize. The use of fertilizer on local maize,
which is a recent phenomenon, is largely a response to declining soil fertility.10
Insufficient labor for crop management activities such as land preparation, plant-
ing, and weeding is widely acknowledged to be an important constraint in obtaining the
potential benefits from use of improved technologies. Factors that could accentuate a
labor constraint, particularly related to hybrid maize, are (1) area expansion associated
with hybrid maize production increases total household labor demand, (2) use of
fertilizers increases the need for timely and adequate weeding, and (3) intrahousehold
control of the crop by men could influence the extent of labor input.
The central focus for this analysis will therefore be on adoption of the improved
maize seed and fertilizer technology, with labor use and oxen cultivation as addi-
tional factors that determine overall farm and hybrid maize productivity.
At the household level, the effects of adoption consist primarily of(1) changes in
household income and consumption expenditure, (2) changes in who controls income
within the household and the effect on women's income, and (3) changes in labor
allocation patterns. These effects can, in turn, influence calorie requirements and
adequacy and the ability of women and other household members to care for children.
All of these primary consequences of adoption are therefore expected to influence
food consumption and child nutrition.

Consequences for Household Income and Expenditure
Jha and Hojjati (1993) find that income, especially from nonfarm sources in-
itially, is likely to facilitate the ability to grow hybrid maize. Once a household has
successfully adopted, the income advantage over local maize is usually demon-
strated. The extent to which hybrid maize adoption improves household income and
consumption expenditure will of course depend on the productivity and profitability
of the improved variety on a particular farm. As indicated earlier, the amount of labor
available and the timeliness with which procedures are undertaken is even more
important than fertilizer application in the improvement of agricultural productivity
with adoption of high-yielding varieties (World Bank 1992). This analysis takes this
issue further by distinguishing the income effect of hybrid maize adoption from the
effect of increasing area under hybrid maize.

Consequences for Women's Income and Decisionmaking
There is, by now, a voluminous literature on the effects of modernization of
agriculture on the economic role and welfare of women. Two main types of effects
can be related to nutrition outcomes, especially for children: the effects on intra-
household income and the effects on time allocation of women. It is well accepted by
now that men and women are likely to allocate income differently, and that the
pooled income of different household members is a poor predictor of nutrition
outcomes of individual members (Haddad 1992). Similarly, women's work in in-

'lDeclining soil fertility is attributed to villagization programs introduced during the 1970s that aimed to
bring dispersed homesteads together into clearly identifiable villages, which contributed to a shift from
extensive to intensive and continuous cultivation.

come-generating activities may conflict with their activities in providing food for the
household and in child rearing, especially feeding and caring activities that contrib-
ute to improvement in children's nutritional status (Leslie 1989).
In rural Zambia, women's production of agricultural products and access to
improved production technologies will largely influence their income and time
allocation. It is expected that women in male-headed households, who have relatively
low access to improved technologies, will have access to a smaller share of house-
hold income, and that demand on women's time for both income generation and
production of household consumption goods and other Z-goods" will increase with
an increase in household income.
In households headed by women, a low level of improved technology use would
mean that women, in general, have low access, and that this would be reflected in
lower income. It is hypothesized that reduced women's income will be detrimental to
household food consumption and child nutrition, but that reducing demand for
women's labor in agricultural production will benefit household food consumption
and child nutrition by improving their ability to allocate sufficient time to performing
these services.

Consequences for Household Labor Allocation
Hybrid maize adoption is associated with cultivation of a larger area so that there
is no decline in the area sown in local maize. This was found to be true in surveys by
the International Food Policy Research Institute in Eastern Province in 1981-82, and
again in 1986. In neighboring Malawi, where maize is also the staple food, nearly all
farmers that adopt hybrids also plant local maize for their own consumption (Smale
1991). In parts of Zambia-Central Province, for example-new varieties were
widely adopted by the late 1970s, but the majority of farmers were still planting
finger millet, the preferred staple in that area (CIMMYT/GRZ 1978). Because of this
and other characteristics associated with adoption of hybrid maize, together with
underlying seasonal labor constraints, household members are likely to face difficult
labor allocation choices.
Hybrid maize production requires a higher labor input per hectare than local
maize to reach potential yields. However, most field observations suggest that actual
labor input is lower on hybrid maize than on local maize (Jha and Hojjati 1993;
ARPT/Eastern Province 1988). This is consistent with the substitution of inputs for
labor suggested earlier and the difficulty of obtaining hired labor.
The extent to which household labor input, and, in particular, intrahousehold labor
allocation is influenced by hybrid maize adoption determines its consumption effects.
The availability and use of nonhousehold labor is a factor in this labor response, but
also intrahousehold dynamics of labor supply and demand. If additional household
labor is directed to hybrid maize, it could have two possible effects. First, it could
increase the demand for calories simply by virtue of the higher workload of members,
and, second, if this increase is accompanied by an increase in women's labor, then

"According to neoclassical economic theory, the composite of household utility is made up of Z-goods,
which combine purchased items with time of household members into final products that enter into their
utility function (Becker 1965).

there could be adverse household food consumption and child nutrition implications
because women's time for those activities might be curtailed or less effective.
When wage labor use in agriculture is widespread, as in Asian countries, and
competitive nonagricultural opportunities exist, household labor input in agriculture
is a decreasing function of total area cultivated by the household, especially the
women's labor component. However, in areas where population density is low, as in
most of Zambia, farmers are less likely to hire wage labor. Instead, household labor
in agriculture is expected to increase with farm size and to be positively affected by
increasing household size and the number of working-age members. The extent to
which this increase in labor input occurs with adoption of hybrid maize and increas-
ing farm size will depend on the opportunity cost of alternative uses of time by each
household member. This could differ for males and females if males are predomi-
nantly engaged in market-oriented nonagricultural work that is easily substitutable
for agriculture, whereas women are primarily engaged in household maintenance
activities for which increased income from agriculture may not easily substitute.
The household labor response for agricultural production is expected to be
determined simultaneously by exogenous factors that also influence the household
income effect. Both of these are expected to influence the supply (through the
substitution effect) of and demand (through the income effect) for household labor in
nonmarket activities, including leisure. This nonmarket labor includes household
maintenance activities such as food processing, cooking, collection of fuel and water,
and house repairs. These activities are hypothesized to be important in determining
the beneficial effects of agricultural technology adoption on household food con-
sumption and child nutrition. These associations will be examined in the analysis.

Effects on Household Food Consumption
In order to trace the effects of hybrid maize adoption on facets of household food
consumption, the approach adopted is to first trace the consequences of adoption on
the determinants of consumption, and then, from the strength of the different causal
factors, to draw implications for consumption and nutrition status. This approach is
preferable to using adoption per se as an explanatory factor, since the technology
itself creates a potential for improving the income and food consumption situation
but is translated into inter- and intrahousehold effects via a complex set of social and
economic conditions, which may be amenable to policy intervention.
In explaining differences in dietary intake, the main predictors that are influ-
enced by hybrid maize adoption are income- and time-related variables. Changes in
availability of foods, usually reflected by prices in local markets, are also important.
However, because of producer and consumer price controls in effect, a large part of
food transactions take place informally, and prices are hard to measure. Household
income can be influenced by adoption of agricultural technology through a combina-
tion of direct effects and by indirect area-level effects on employment and food
availability. In addition to household-level income effects, intrahousehold distribu-
tion of income is also tested for its impact on household diets. Time allocation,
especially by women in consumption-related activities, is also expected to be a factor
in consumption effects of increased income. The allocation of time is hypothesized
to be influenced by improved technology adoption through a combination of labor
allocation decisions and income effects.

Dietary characteristics examined are per capital calorie intake, calorie intake at
the household level in relation to household composition and workload, per capital
protein intake, and a diet diversity indicator for overall diets, as well as home-pro-
duced and purchased components. The protein and diet diversity measures are
included to give an indication of improvement in diet quality, which is often more
important for child nutrition than improvement in dietary calories alone.
Effects on Child Nutrition
The direct determinants of good child nutrition are adequate diet and child care
and absence of disease. However, there are several problems with tracing the effects
of hybrid maize adoption on child nutrition. First, the direct determinants of child
nutrition are the result of a complex mix of area-, household-, and intrahousehold-
level effects of agricultural change. Second, many child nutrition measures, for
example, weight-for-age and height-for-age, are the result of cumulative effects on
the child that begin even before its birth. This often makes it difficult to account for
a large part of the variation in these measures. Third, it is difficult to disentangle the
simultaneity between the various factors that contribute to nutritional status. For
example, time allocation of women is a contributory factor in food intake and child
health, both of which contribute to a child's nutritional status. However, in addition
to influencing these two factors, there may be an additional effect of women's time
allocation not captured by these variables, that is, the quality of care given. Similarly,
while women's control of income could influence the allocation of more household
resources directed to food, it could also influence allocation of food within the
household so that child nutrition is affected.
In order to resolve these problems, particularly the problem caused by the
endogeneity of causal factors such as household food intake and child morbidity, an
instrumental variables approach is used. Estimated values of household dietary
parameters and predictors of child morbidity are used-water, sanitation, and access
to health services. In addition to these and women's time allocation, other variables
are included that could influence intrahousehold resource allocation, such as who
heads the household and who controls income.

Theory and Analytical Model
The main issues of interest in the present analysis are the effects of agricultural
production, in particular adoption of hybrid maize cultivation, on household food
consumption and child nutrition. According to current analytical practice, these out-
comes can be directly modeled as part of the household's utility function (Behrman
and Deolalikar 1988). The basic model for an agricultural household must deal with
the problem that household decisions affecting production and consumption are likely
to be made simultaneously, with each affecting the other. It is, however, possible to
model these recursively, that is, with production decisions in one period affecting
consumption outcomes in the second period, which could then affect production
decisions and outcomes in the third period, and so on. In the present data set, the
information was available for only one production cycle (one calendar year), therefore
the problems of simultaneity in making these associations need to be resolved.
Since food consumption, nutrition, and health outcomes are part of the house-
holds' utility function, they are given by choices made in the allocation of both

income and time to these activities directly, and are subject to a combination of
budgetary and time constraints. In the household model, these are all endogenous
variables, but a set of reduced-form demand functions can be specified, in which both
the production- and consumption-related outcomes are left-hand-side variables and
are given by variables that are exogenous to the household.
At the base of the household model is the concept of full income in which both
income flows and time contribute. As initially proposed by Becker (1965) and
developed further in the human capital investment literature,12 households derive
utility from Z-goods--or final products that are primarily a combination of com-
modities and time of household members. These Z-goods include, for example,
health, nutrition, and food consumption outcomes, in addition to others such as social
and educational outcomes. Tastes, as well as individual and household endowments,
can influence the outcomes of this process. Thus,
Z, = Z(x,, t,, eu), (1)

where xi is a vector of market goods, ti is a vector of time used in producing Z,, and
ei. is a matrix of individual and household endowments.
It is these Z-goods, the health and nutrition of its members, that contribute to a
household's utility. Therefore, maize in the store does not provide much consumption
utility, but after it has been processed, cooked, and served, its utility goes up. Time
allocation data from rural households as well as field observations show us that much
of the time spent by rural households on non-income-earning activities is in the
production of such Z-goods. Household production functions for income contribute
to both the disposable income and allocation of time.
The income-generation function reflects a maximizing choice given the sum of
individual and labor market characteristics. Thus, while disposable income (Y) helps
in the acquisition of market goods, it is also a reflection of time allocation decisions
of household members between alternative sources of income. Simply put,
Y = Ex, + S = L,w, + E, (2)
Ex = total consumption expenditure,
S = savings,
Li = labor allocation to alternative income generation activities,
w, = the wage rate in the ith income-earning activity, and
E = nonlabor income.

The net income effect of hybrid maize adoption will be conditional on the improved
returns to labor in agriculture and shifts in labor allocation between different sectors.
Taking the two main sectors in which labor can be allocated to income-earning
activities to be farm and off-farm, w, the shadow wage rate in agriculture, and w2 the
wage rate in off-farm employment, then, first, w1 will be a function of profits, that is,
derived from a combination of farm production technology and prices of inputs and
outputs. In addition, when farm production is primarily for own consumption and

12See, for example, the discussion of household production of health and nutrition by Behrman and
Deolalikar (1988).

agriculture labor markets are virtually nonexistent, as was the case in Eastern Prov-
ince, then w, will depend on both the production technology and household prefer-
ences (Strauss 1986).
Second, L, and L2 are the labor of household members allocated to farm and
off-farm work, respectively, and each is given by w, and w2 as well as by prices of
inputs and outputs and the utility function. It can be shown that for farm households
with different utilities for farm and nonfarm work, structural equations for labor
supply to the farm and off-farm sectors can be derived with

Li = L,(T, q, w, Z), (3)
T = the farm production technology,
q = a vector of input and output prices that
influence the returns to farm labor,
w2 = the off-farm wage rate, and
Z = a vector of Z-goods demanded by the
household (Lopez 1986).

The food consumption and nutrition outcomes of hybrid maize adoption will be
the result of the effective demand for Z-goods, given the net income and time
allocation effects and given that

Z, = Z(x, t, e,) = f(Y). (4)
Thus, while disposable income contributes to the demand for purchased commodi-
ties such as foods, it also affects the demand for t, that goes into the production of Z,.
Based on the outline presented above, the analytical model consists of equations
that first predict the income and time allocation effects of hybrid maize adoption, and
then use these predicted values in explaining food consumption and nutritional
outcomes. The model consists of the following recursively estimated equations:

1. A predicting equation for the adoption of hybrid maize as the main indicator
for the level of farm production technology used. This is estimated using a
two-step or Heckman approach, which first estimates the probability of
hybrid maize adoption and then, conditional on adoption, estimates equation
(6) by ordinary least squares, correcting for truncation bias:

and HM= f(E,), (5)
HMA = f(HM*, E2), (6)
HM = observed hybrid maize adoption,
HM* = probability of hybrid maize adoption,
HMA = area under hybrid maize conditional on
adoption, and
El, E2 = vectors of exogenous variables.
Examples of exogenous variables are household and area characteristics,
fixed assets, and nonlabor income.

2. Since agricultural technology and the option of improvement in productivity
will influence both labor allocation decisions and disposable income out-
and Y= f(HMA*, E3), (7)
Li, L2i = f(HMA*, E), (8)
HMA = predicted area under hybrid maize production,
E3, E4 = vectors of exogenous variables,
L1, = household farm labor by males and females
estimated separately, and
L2, = household off-farm labor by males and females.
The allocation of household labor for farm and off-farm work is estimated
separately for males and females living in the household. Predictors include
farm technology adoption, human capital endowments of household mem-
bers, an indicator of intrahousehold differences in preferences in farm and
off-farm labor allocation, and exogenous variables. Since prices for both
inputs and outputs were fixed by policy measures, variations in input use and
marketed output are likely to be a function of other variables, such as being
located in a well-functioning cooperative area and membership in it. There-
fore, price variations per se are not likely to be a factor in farm productivity.

3. The next set of equations estimates the time spent on nonlabor activities such
as those involved in consumption support activities, T,. This will be the
compensated effect of the shadow wage rate changes with adoption of im-
proved farm technology, and will be through a combination of the well-
known income and substitution effects:

where T= f(Y*, HM*, HMA*, E5), (9)
Tj = time spent in household maintenance activities by
males and females,
Y* = predicted value of household disposable income
derived from equation (7),
E5 = vector of exogenous variables.
The other terms are defined as before.

4. Food consumption is next estimated as a function of disposable income (Y),
its intrahousehold control, time spent by household members in household
maintenance activities (T,), and exogenous variables:

where = f(Y*, Fs, T,*, A, E6), (10)
Cj = household caloric, protein, and diet diversity
Fs = an indicator of women's share of income,
T,* = predicted values for male and female time spent
in household maintenance activities,

A = area-level factors that influence food prices and
availability, and
E, = vector of exogenous variables.

5. Child nutritional status is estimated using the predictors of demand for food
and health care, which are the main direct factors in child nutrition, time spent
in household maintenance activities by males and females, indicators of the
health environment and access to health services, as well as other child- and
household-specific characteristics:

where NSk = f(Y*, Fs, T,*, H, C,, E), (11)
NSik = the nutritional indicator for the child in season k,
H, = indicators of the health environment and access,
C, = child characteristics such as age and sex, and
E7 = vector of household variables.
A seasonal dimension is added with seasonal intercept variables and a random
effects estimation model.

In the above system of equations, the vectors El through E, are such that the
equations are fully identified. Variable details and results will be presented later in
the report.

Data Sources

Selection of Study Sites
Ten representative sites were chosen from Eastern Province including sites in all
administrative districts. These locations had been originally selected by the World
Bank-funded Eastern Province Agricultural Development Project (EPADP) to moni-
tor their activities in the province. The branch was the local administrative unit in
each of the selected sites. Each branch consisted of about 10 villages, with each
village having an average of 25 households. Though sites were located in each of the
districts, they were not chosen to be representative of each district; rather, collec-
tively they represent the provincial rural population.

Selection of Households
Within each branch, four stratifying criteria were selected. These were (1) use or
nonuse of hybrid maize seed, (2) use of oxen or hoe cultivation, (3) male- or
female-headed household, and (4) contact farmer status.1 A census of the total
population in the sampled branches was carried out, which recorded information on
each of the stratifying criteria. Households were grouped into all possible combina-
tions of the stratifying criteria, and households were randomly drawn from each
group, selecting the Nh household, where

13The training and visit (T&V) program, which was a central part of EPADP's activities, used contact
farmers as a focal point for spreading the extension message.

N = B- (12)

where B, is the number of households in the branch and 33 is the number of
households to be selected in each branch. This size of sample represented about 15
percent of the branch population-a total of 330 households.


Frequency of Visits
Households were visited monthly beginning in December 1985 until December
1986 and interviewed to obtain information on the main questionnaire. Additional
modules were incorporated for information not requested each month.

Types of Measurements
Measurements were based on interviews with selected household members for
most items in the study. Actual measurements were made for some items, such as size
of farmed area and output, and for anthropometric indicators of nutritional status of
all household members. Area farmed and output were measured for the 1986 harvest
year. Anthropometric measures (weight and height) were taken during four of the
monthly rounds to capture seasonal variations in nutritional status.

Modules for Primary Focus in This Analysis
Labor Allocation. This information was obtained for each individual working
during the previous month on an activity in a set of five different modules: agricultural
work by plot, different postharvest activities, different livestock activities, nonfarming
activities of a self-employed nature, and wage employment activities. With the exception
of agricultural work, which was recorded in days worked, all other activities were
recorded in a way that could be converted directly into hours of work. Since information
was obtained according to the individual engaged in the activities, this was converted
later into labor allocation by age, sex, or other categorization.
Food Consumption. This information was obtained by using a modified food
frequency/expenditure recall for the past week. To do this, first a detailed list of all
possible food items that could be obtained was compiled. In conducting the interview,
the enumerator was instructed to interview the female who was primarily responsible
for food preparation. They first went through the list and marked those items that were
present in the household diet during the previous week. Then they obtained the number
of days an item was consumed and the typical amount used each day, in a meal, and
during the week, or, for items consumed in small amounts, the total time elapse for
completing a given amount. Quantities were obtained through a combination of a set
of standard units that were provided to the enumerators and a series of standardized
local units. The approach was intended to be as flexible as possible to fit into the
pattern of food acquisition employed by the household for each food item it consumed
during the previous week. Additional information was obtained on the source of the
item and, if purchased, the frequency, amount, and price or expenditure entailed.
Recording the status of the food quantity, whether in edible portions or "as pur-
chased," was also important in converting the quantities into nutrient consumption.

Often, food expenditure surveys record foods in the "as purchased" form, and if this
entire quantity is treated as "edible," the consumption figures may be substantially
inflated. Other information included in the food consumption module included atten-
dance at meals and meals provided to guests and workers.
IntrahouseholdDecisionmaking. This information was obtained in two modules,
each of which was implemented once. The first was aimed at cash expenditures made
in agricultural production, distribution of income from sale of produce to different
household members, and their patterns of allocation of income received. This part of
the survey was conducted in the postharvest month of September 1986. The second
module was aimed at all food and nonfood cash expenditures and was carried out
during December 1986.
An attempt was made to determine the involvement of household members in
different parts of the decisionmaking process, as described by Acharya and Bennett
(1981). For example, in the case of cash expenditures in agricultural production, the
sequence of decisionmaking was represented by who suggested the expenditure be
made, who arranged or negotiated for the item to be purchased, and who actually paid
for it. For sale of produce, the sequence was who suggested the sale, who arranged or
negotiated it, and how were the proceeds distributed within the household. For food
and nonfood expenditures, the sequence was who suggested it, who paid, and who
went to purchase the item. Dividing the decisionmaking process in this way provides
a more realistic description of intrahousehold decisionmaking and reduces ambiguity
and conflicting responses.
Health Status. Morbidity recall during the previous month was included in the
main monthly questionnaire. Illnesses were identified either by commonly known
names or by symptoms. The duration of the illness for each member who was
reported ill during the month and the nature of treatment were also recorded.
Nutritional Status. Assessment of individual nutritional status was made by
anthropometry four times during the year to capture seasonal variations. Weight and
height (or length for children under two years) was recorded for all household
members during visits in February, June, September, and November of 1986. These
periods represented heavy work with severe food scarcity (February), early harvest
(June), postharvest (September), and start of the next planting cycle (November).
Age assessments and verification were made, based on a combination of hospital and
birth records, local events calendars, and questioning the mothers of young children
on the season or month of birth and years completed.

Involvement of Local Institutions in Study

The design and implementation of the study was done jointly by research staff
from three divisions at IFPRI in cooperation with specialists from the Zambian
Nutrition Commission, the Rural Development Studies Bureau of the University of
Zambia, and the Eastern Province Agricultural Development Program of the Provin-
cial Planning Unit. It was a truly interdisciplinary collaborative project, which was
first presented to Zambian analysts and policymakers and was later edited for
publication (Celis, Milimo, and Wanmali 1991). This report gives a fuller analysis of
the nutritional and food consumption effects of hybrid maize adoption and discusses
the implications in light of current policy reforms in Zambia.


Characteristics of Eastern Province

Eastern Province is one of the major agricultural regions of the country, consis-
tently producing large grain surpluses required for the urban centers. It has, neverthe-
less, remained predominantly a rural province. It has the lowest percentage of urban
population of any Zambian province--only an estimated 14 percent in 1990 (Table 3).
This compares with an overall Zambian urban population of 49 percent, and is even
lower than the relatively underdeveloped agricultural provinces of the north and west.
It has a low population density, about 10 persons per square kilometer, and the
farming population is largely in the traditional sector, with farms averaging 2-3
hectares. In rural areas, agriculture provides nearly 90 percent of household income
through production and employment. Most of the households rely on own production
for the major part of their consumption, with the relatively deficit areas relying more
on food purchases. It is also interesting to note that income sources are more
diversified in the poorer agricultural areas (Honeybone and Marter 1979). This is also
consistent with the changes in labor allocation patterns observed in this report; it
suggests that rural areas in Zambia are in the initial stages of agricultural transforma-
tion, in which other income sources are sought to overcome the uncertain and low
returns from agriculture. This diversification is different from that observed in the
later stages of agricultural development, in which households invest surplus pro-
duced in agriculture in local nonagricultural enterprises and produce linkage effects
in the growth process.

Table 3-Population distribution in urban and rural areas, by province
1990a Percent
Province Total Rural Urban Rural Urban

Central 720 411 309 57 43
Copperbelt 1,751 135 1,616 8 92
Eastern 882 760 122 86 14
Luapula 526 408 118 78 22
Lusaka 1,108 163 945 15 85
Northern 867 647 220 75 25
Northwestern 380 319 61 84 16
Southern 937 626 311 67 33
Western 574 479 95 83 17
Total 7,745 3,950 3,795 51 49

Source: P. D. Ncube, "The Zambian Food Strategy-Aspects of Production," in Agricultural Baseline Data for
Planning, ed. P. D. Ncube (Lusaka: Zambia, National Commission for Development Planning and the
University of Zambia, 1983).
aProjections for 1990 based on 1980 census.

The high degree of reliance on agriculture seen in the Eastern Province of Zambia
is typical of rural areas in Africa where there is a combination of poor infrastructure
development and relatively good agricultural potential. The poor infrastructure de-
velopment generally precludes the rapid modernization of agriculture and growth of
nonfarm employment and income in the area (Haggblade, Hazell, and Brown 1989).
At the same time, the relatively good agricultural potential enables rural households
to meet basic needs without seasonal migration, which in areas such as the Sahel are
essential for survival of rural households (Reardon, Matlon, and Delgado 1988).
Ecology, Rainfall, and Climate
Eastern Province has some of the best agricultural lands in the country. It is
mostly situated on the Eastern Plateau, which is characterized by moderate rainfall of
800-1,000 millimeters a year. Zambia is commonly divided into four agro-ecological
zones (Figure 1). Of these, Zone 1 is the northernmost high-rainfall area, with an
annual average of more than 1,200 millimeters. This zone occupies 46 percent of land
area and is traditionally a cassava and finger millet producing area. Although it is
ecologically regarded as unsuitable for maize growing, it has over recent years
become an important maize-producing area due to favorable production incentives.
Zone 2 consists of the western semi-arid plains and has a low rainfall (less than 800
millimeters); main crops are cassava, bulrush millet, and sorghum, as well as some
maize. It has a large cattle population and is well suited for it. Zone 3, which consists
of (a) the Central and Southern plateaus and (b) the Eastern Plateau, constitutes only
12 percent of the land, but it produces most of the agricultural surpluses, especially
for maize. It has moderate rainfall and some of the best soils in the country. The main
crop is maize, but smaller amounts of groundnuts, sunflower, cotton, soybeans, and
tobacco are also grown. Most of the study sites are located in this zone. Zone 4 is
composed of the Luangwa-Zambezi rift valleys, which have good soils and irrigation
potential but are located away from population and infrastructure and have had little
improvement in agriculture. Rainfall is low (less than 800 millimeters) and erratic.
Consequently, these are generally considered food-deficit areas. Two of the 10 study
sites are located in the Luangwa Valley in this zone.
The rainfall is unimodal, with the rainy season lasting from November to March
and with 70 percent of rainfall occurring during the months of December through
February. The temperature peaks just prior to the start of the rains, at a mean monthly
temperature in October of about 27 degrees centigrade on the Eastern Plateau. The
coolest month is in the middle of the dry season in June, when the mean monthly
temperature on the Eastern Plateau is about 18 degrees centigrade.
Vegetation is predominantly moist savanna (long grass) with scattered wood-
land, with the valley areas characterized as dry savanna. Tsetse infestation is high in
the Luangwa Valley (Zone 4) and cattle rearing is, thus, infeasible there. Efforts to
curb its spread on the plateau are under way and have been largely successful.
However, areas bordering Mozambique have had a recurrent problem with this
infestation coming across the border.
Importance of Maize in Production and Consumption
Eastern Province is in one of the most fertile and productive agro-ecological zones
in Zambia. It has an altitude of about 920 millimeters and an average rainfall of about
900 millimeters. The rest of the province is in the Luangwa Valley, of which nearly
half takes the form of a rocky and uncultivable escarpment. Although the soils of the

Luangwa Valley are better than in other areas in the low rainfall zones, rainfall tends
to be unpredictable, and flooding is common. The valley also has tsetse infestation,
making it unsuitable for livestock. Thus it lacks draft power for crop cultivation.
The plateau areas of Eastern Province (Zone 3b) are among the main maize-
growing regions of the country (Table 4). Although there is somewhat more crop
diversity in the valleys (Zone 4), maize is still the main grain crop. Other grains also
grown in the valley are sorghum, finger millet, and rice (bulrush millet is not grown
to any significant degree in the Luangwa Valley). Cotton is an important cash crop.
Provincial grain consumption patterns largely mirror the production pattern of the
agro-ecological zones, with maize being the staple in the plateau areas, and with a larger
share of cassava and sorghum and millets in the diet in the high- and low-rainfall areas
(Figure 1). The main producing provinces of the plateau areas are Lusaka, Copperbelt,
Southern, and Eastern and these all have maize as the predominant staple. Cassava is the
main staple (in weight terms) for Luapula and Western provinces, and it is an important
second staple to sorghum and millets in the Northern and Northwestern provinces.
For the country as a whole, maize is by far the most important staple (Table 1).
Even though large parts of the country lie in areas where it is not the main crop, it is
grown and consumed in the provinces that tend to have more densely populated rural
areas. Maize is also the main staple food in the urban areas, which comprise nearly half
the population of the country. Altogether, maize is the staple for nearly 75 percent of
the population. Table 1 also shows that cereals and cassava contribute from about 1,200
calories per day in Luapula to over 2,000 calories in Lusaka and Central provinces.

Agricultural Development Institutions and Programs
Overall, investment in agriculture has remained a relatively small share of public
expenditures. In the first and second national development plans implemented be-
tween 1966-70 and 1972-76, the level of planned public-sector investment for agricul-

Table 4-Production of major crops by agro-ecological zones, Zambia, 1979-80
Crop Unit 1 2 3a 3b 4 Total
Maize 90-kilogram bag 730.8 94.3 4,629.5 2,209.0 102.1 7,765.7
Sorghum 90-kilogram bag 186.0 24.0 21.4 ... 39.2 270.6
Finger millet 90-kilogram bag 512.6 64.2 4.6 10.0 13.7 605.2
Bulrush millet 90-kilogram bag ... 63.9 ... ... 20.0 83.9
Cassava 60-kilogram bag 2,937.8 881.4 24.8 4.2 10.7 3,858.9
Beans 90-kilogram bag 34.8 0.6 1.6 ... .. 37.0
Soybeans 90-kilogram bag 4.3 ... 12.6 ... .. 16.9
Rice 80-kilogram bag 27.0 15.8 0.8 7.5 4.4 55.5
Wheat 90-kilogram bag 33.3 0.9 53.1 ... ... 87.3
Groundnuts 80-kilogram bag 13.8 3.0 31.6 134.9 2.2 185.5
Sunflower 50-kilogram bag 13.3 1.5 291.7 53.1 18.2 377.8
Cotton I kilogram 139.0 40.0 20,997.0 3,908.0 4,696.0 29,780.0

Source: Zambia, Food Strategy Study (Lusaka: Ministry of Agriculture and Water Development, 1984).
Notes: Zone 1 is the northern high-rainfall zone.
Zone 2 is the western semi-arid plains.
Zone 3a consists of the Central and Southern plateaus; Zone 3b is the Eastern Plateau.
Zone 4 is the Luangwa and Zambezi rift valleys.

Figure 1-Map of Zambia's agro-ecological zones

1 Northern high-rainfall zone
2 Western semi-arid plains
3 Central and Southern plateaus (a)
and Eastern Plateau (b)
4 Luangwa and Zambezi rift valleys

E Valley area of Eastern Province
Ij Plateau area of Eastern Province

Notes: These zones are adapted from the World Bank's classifications. The dashed lines indicate province borders;
the solid lines are zones.

ture was only 12.2 and 11.3 percent of total public-sector investment, respectively.
Between 1971 and 1978, the real value of total agricultural allocations was reduced by
more than half. Another feature of expenditures on agriculture has been the prominent
role played by subsidies, especially maize marketing subsidies to NAMBOARD,
which accounted for more than 40 percent of all government allocations during the
1970s (World Bank 1981). Between 1978 and 1985, the real value of agricultural
sector allocations continued to decline sharply, but since then they have shown an
upward trend (World Bank 1992).
The main activities for agricultural development are undertaken either by the Ministry
of Agriculture, which is responsible for agricultural research and extension, or the
parastatal organizations that are responsible for delivering inputs-NAMBOARD in
some provinces and the provincial cooperative unions in the rest. Other organizations feed
their products through these, including those providing credit, seed, and fertilizers. Some
commodities, for example, cotton and tobacco, have separate institutions that deal directly
with all the farmer's needs for inputs for producing the crop and with its marketing.

Over the years there has been an evolution in thinking about mechanisms for
promoting agriculture. Initially, the efforts were largely centered outside the tradi-
tional farming sector, and these included state farms in the immediate postindepen-
dence years, followed by "rural reconstruction centers," which were intended to be
run by youths recruited from urban areas. In the early 1970s, areas with promising
agricultural potential were targeted with the Intensive Development Zone (IDZ)
programs, which aimed to deliver an integrated package of inputs and services to
"emergent" farmers from the traditional sector. By the end of the 1970s, the IDZ
concept was expanded into the IRDP, which was intended to expand coverage of
agricultural development efforts to all rural areas. Other programs adopted during the
1980s include the Lima program, the Adaptive Research Planning Team (ARPT),
and, on an experimental basis, the training and visit (T&V) system. All these pro-
grams were in operation in Eastern Province during the present study.

Crops Planted in the Study Area
There is a fair amount of variation within the province in the importance of
different crops in the farming system. Overall, 83 percent of land was devoted to
maize production, with local maize at 60 percent and hybrid maize at 23 percent
(Table 5). Maize production was found to be most important in the Chadiza and
Katete sites, where it was more than 90 percent of total farmed area. In parts of
Lundazi, the maize area was also nearly 90 percent of area. Maize area was lowest in
Mambwe and Chama, the two valley sites, while secondary cereal crops (rice, finger
millet, and sorghum) were highest. Even though the study sites in each district were
not representative of the overall district characteristics, these differences are indica-
tive of the nature of variations within the province.
Nearly twice as much of the local maize area was intercropped as in sole stands. As
one would expect, more of the hybrid maize was in sole stands, but a sizable amount
(about one-third) was intercropped. This represents a change from the early years of
hybrid maize adoption, when it was grown almost exclusively in sole stands. This could
be the result of recent emphasis on intercropping by the agricultural research and
extension groups in Zambia to reduce the time required for weeding labor.
The valley sites are striking in that much smaller areas are farmed per household
(Table 6). However, the total share of area devoted to cereals is very similar in the
plateau (89 percent) and valley (83 percent) sites, with valley areas more likely to
grow secondary cereals, such as rice, sorghum, and finger millet, while the plateau
sites specialize almost exclusively in maize production. Hybrid maize, as mentioned
earlier, is also virtually absent from the valley sites. Groundnuts, beans, and cowpeas
covered only 10 percent of farmed area (they had additional area as intercrops in
maize fields), while cotton and sunflower area was only 3 percent. For both of these
groups of crops, the pattern in terms of share of land was similar in plateau and valley
sites, despite the much smaller farm sizes in the valley.

Cropping Pattern with Hybrid Maize Adoption
A comparison of the cropping pattern of hybrid maize adopters with that of the
nonadopters in the plateau shows that adopters sow an even larger area to local maize
than the nonadopters: 1.7 hectares, compared with 1.3 hectares for nonadopters

Table 5- Crop area and share, by site

Agriculture Districts
Lundazi, Lundazi, Chipata, Chipata,
Crop Area/Share Mambwe Chama Chipili Kasendeka North South Chadiza Katete Petauke Nyimba All


Local maize
Sole crop
Hybrid maize
Sole crop
Total maize
Other cereals
Groundnuts and legumes
Cotton and sunflower
Total area

33 30 33 33 29 33 33 26 32 32 314




0.36 0.46
48.00 21.50
0.02 0.74
2.67 34.58
0.38 1.20
50.67 56.07

0.06 0.34
8.00 15.89
0.00 0.07
0.00 3.27
0.06 0.41
8.00 19.16
0.44 1.61
58.67 75.23
0.18 0.26
24.00 12.15
0.12 0.13
16.00 6.07
0.00 0.14
0.00 6.54
0.02 0.01
2.67 0.47
0.75 2.14
100.00 100.00



0.11 0.35
10.19 18.32
0.69 0.62
63.89 32.46
0.79 0.97
73.15 50.79













Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and
Nutrition Commission agricultural household survey, Eastern Province, Zambia, 1985/86.

Table 6-Crop area and share, by region

Crop Area/Share Plateau Valley All

Number of households 251 63 314

Local maize
Sole crop
Hybrid maize
Sole crop
Total maize
Other cereals
Groundnuts and legumes
Cotton and sunflower








Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

(Table 7). On average, the adopting households were putting slightly more area under
local maize than under hybrid and, overall, had 88 percent of their land under maize,
compared with 83 percent for nonadopting plateau households and 60 percent for
valley households. These observations should be related to the significantly higher
farm size of adopting households: 3.7 hectares, compared with 1.6 hectares for
nonadopting plateau households and only 1 hectare for valley households.
This pattern of area allocation, indicating a preferential treatment of local maize
despite a presumably higher profitability for hybrid maize, is similar to acreage
allocations to cash crops observed in other countries in Sub-Saharan Africa.14 Similar
results are not found in situations with better market integration, the Philippines, for

14This was found for sugar-producing farmers in South Nyanza District of Kenya by Kennedy and Cogill
(1987) and for potato-producing farmers in Rwanda by von Braun, de Haen, and Blanken (1991).

Table 7-Crop area and share, by hybrid maize adoption

Crop Area/Share Nonadopters Adopters Valley All

Number of households 154 93 63 310
Maize (hectares/household)
Local maize
Sole crop 0.44 0.62 0.19 0.44
Percent 28.39 16.71 19.79 21.15
Intercrop 0.84 1.04 0.35 0.80
Percent 54.19 28.03 36.46 38.46
Total 1.28 1.66 0.54 1.25
Percent 82.58 44.74 56.25 60.10
Hybrid maize
Sole crop 0.00 1.07 0.03 0.33
Percent 0.00 28.84 3.13 15.87
Intercrop 0.00 0.55 0.00 0.17
Percent 0.00 14.82 0.00 8.17
Total 0.00 1.62 0.03 0.49
Percent 0.00 43.67 3.13 23.56
Total maize 1.28 3.28 0.57 1.74
Percent 82.58 88.41 59.38 83.65
Other cereals 0.04 0.11 0.23 0.10
Percent 2.58 2.96 23.96 4.81
Groundnuts and legumes 0.15 0.22 0.10 0.16
Percent 9.68 5.93 10.42 7.69
Cotton and sunflower 0.07 0.09 0.03 0.07
Percent 4.52 2.43 3.13 3.37
Miscellaneous 0.01 0.00 0.03 0.01
Percent 0.65 0.10 3.13 0.48
Total 1.55 3.71 0.96 2.08

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

example, where purchased staples are preferred (Bouis and Haddad 1990). A study
in Guatemala where export crop production was introduced found a slight reduction
in area farmed to food crops, but this was offset by a substantial increase in use of
improved inputs in food crop production and higher yields (von Braun, Hotchkiss,
and Immink 1989). In Zambia, rural households have several reasons for treating
hybrid maize predominantly like a cash crop, and as discussed in Chapter 2, this is
along the lines of national policy objectives of promoting surplus production for
urban areas. However, in the long run, these policies may have been self-defeating,
as they have not encouraged production of hybrid maize for home consumption-
which could have led to a much higher marketed production of maize than has
hitherto been possible.
Whereas hybrid maize-adopting households specialize more in maize than other
households, they also have a higher area under secondary cereals, which, on the
plateau, is predominantly finger millet. These households slightly reduce their share
of land under legumes and the secondary cash crops (cotton and sunflower), but in
absolute terms land devoted to these crops is still higher than that allocated by the
nonadopting households.



In this chapter, some agricultural production characteristics of hybrid maize-adopt-
ing households are examined and some conclusions drawn on household income effects
and possible changes in intrahousehold control of income with hybrid maize adoption.
In particular, the role of increasing farm size, especially with oxen cultivation (which
permits area expansion), in hybrid maize adoption is examined. Allocation of fertilizer
to different crops and fertilizer use strategies are also examined, as are differences in
intrahousehold control of crop production. In order to predict the effect of hybrid maize
adoption on household income, an estimation equation is fitted for predicting changes
in household per capital consumption expenditure with adoption of hybrid maize.

Hybrid Maize Adoption by Farm Size
In 1986, only 30 percent of the farmers in the smallholder sector in Eastern
Province had adopted hybrid maize. Among the nonadopters were virtually all the
farmers living in the valley areas and more than 60 percent of the farmers in the
plateau areas. In area planted, hybrid maize accounted for only 28 percent of maize
area. This is much lower than the 64 percent area under hybrid maize reported for the
country as a whole for that year (CIMMYT 1987).
Adoption of hybrid maize was found to be heavily concentrated among the 10
percent of farmers with the largest farms in Eastern Province-all of the plateau
farmers with more than 5 hectares grew the hybrid (Table 8). However, adoption was
substantial among the smaller-size farms as well. About 53 percent of the farmers in
the 3-to-5-hectare category, 43 percent in the 2-to-3-hectare category, and 24 percent
in the 1-to-2-hectare category grew the improved varieties. In the smallest farm size,
less than 1 hectare, adoption was minimal. In the valley, there was virtually no
adoption of the hybrid because of the small farm sizes and a relatively favorable
climate for other cash crops, such as cotton and sunflower.
In an earlier IFPRI survey in the area in 1981/82, use of hybrid maize in valley
areas that were part of the Intensive Development Zone (IDZ) Program of the 1970s
was noticeably higher, along with other commercial crops, especially cotton and
soybeans. By 1986, production of both hybrid maize and soybeans in the valley areas
had declined appreciably. This is attributed to the availability of tractors for field
preparation during the earlier period provided by the local agricultural offices in IDZ
areas. The disrepair of this equipment in subsequent years and the shift of policies
away from the IDZ concept probably made it difficult for the farmers in the valley to
enlarge farm size and plant the hybrid maize crop.15 Cotton production, on the other

15Trypanosomiasis, a parasitic livestock disease transmitted by the tsetse fly, is endemic in the valley.
This prevents households in that area from keeping livestock and engaging in ox-drawn plow cultivation.

Table 8-Hybrid maize adoption by farm size, plateau and valley

Farm Size in Hectares per Household
Area/ Less More Weighted
Adoption than 1 1-2 2 3 3 5 than 5 Total Average
(N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent)
Nonadopter 64 57.7 50 56.2 19 47.5 21 44.7 0 0.0 154 49.7
Adopter 7 6.3 21 23.6 17 42.5 25 53.2 23 100.0 93 30.0
Nonadopter 40 36.0 18 20.0 4 10.0 1 2.1 0 0.0 63 20.3
Total 111 100.0 89 100.0 40 100.0 47 100.0 23 100.0 310 100.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Notes: Numbers may not add to 100 due to rounding. N is sample size.

hand, is preferred by the smaller farmers for various reasons, including the prompt
payment by LINTCO, the cotton parastatal, to farmers for their crop.
Area expansion (and factors contributing to it) has generally been accepted as an
important contributor to hybrid maize adoption in Zambia. Jha and Hojjati (1993), in
their model of fertilizer use in Eastern Province, postulated a simultaneously deter-
mined area farmed and a hybrid maize use function. In another study, Jha, Hojjati,
and Vosti (1991) found area expansion to be the single most important determinant
of hybrid maize adoption. At the mean, expansion of area by 1.0 hectare meant the
probability of adopting hybrid maize was nearly 0.8. Two factors that could contrib-
ute to this finding are (1) food security considerations of households and (2) access
to inputs such as improved seed and chemical fertilizers. The limited types of
improved maize seed available16 as well as the maize pricing and marketing policies
in effect at the time are likely to be the main contributing factors to this pattern of
adoption. Limited access to or demand for improved inputs does not appear to be as
much of a limiting factor as the supply-side and distribution problems (Keller and
Mbewe 1988). Fertilizer application was much more widespread than was hybrid
maize seed. This was especially so in areas with a high level of hybrid maize adoption
(Jha and Hojjati 1993). To a large extent, the supply-side variations were a function
of the effectiveness of local chapters of the Eastern Province Cooperative Union in
obtaining and distributing the inputs.

Hybrid Maize Adoption and Oxen Use
That the availability of mechanical traction for cultivation is a factor in the ability
of farmers to increase their farm size, and thereby to grow hybrid maize, was con-
firmed in a recent analysis of Eastern Province by Jha and Hojjati (1993). They found
that oxen cultivators farm an additional 1.4 hectares and have a 0.9 probability that

16SR52 was the main type of hybrid available. Since it is a hybrid, it has to be purchased annually, and it
is a long-duration variety. It therefore competes with planting of local maize. It is also a "dent" variety,
that is, it has a soft kernel that is difficult to store and process with traditional technologies.

Table 9-Oxen use by farm size

Farm Size in Hectares per Household Share of
Less More Total
Oxen Use than 1 1-2 2-3 3-5 than 5 Total Farms
(N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent)
Use oxen 23 20.7 39 43.8 23 57.5 42 89.4 22 95.7 149 48.1
Do not use oxen 88 79.3 50 56.2 17 42.5 5 10.6 1 4.3 161 51.9
Total 111 100.0 89 100.0 40 100.0 47 100.0 23 100.0 310 100.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: N is sample size.

they will grow hybrid maize. Although in recent years the perception of a land frontier
has become more real in the minds of farmers interviewed, virtually all of them still
claim that they could expand the size of their farms if they chose. This is a reflection
of the availability of fallow land rather than their capacity to cultivate more land.
It is the ability to use oxen to expand area planted that apparently drives the
association with hybrid maize production, rather than oxen use per se (Tables 9 and
10). The share of farmers using oxen cultivation increases rapidly with farm size,
with 96 percent of the households with more than 5 hectares using oxen. At the other
end, only 21 percent of the smallest farms cultivate with oxen (Table 9). Among the
farmers not growing hybrid maize, oxen users nearly equal those not using oxen,
while among the adopters, 82 percent use oxen (Table 10). In order to expand
cultivated area, farmers also need more workers; in areas where the labor market is
thin, this means a larger family. Without the assured labor supply, farmers hesitate to
expand cultivated areas (Kumar 1988).
As might be expected, no oxen are used in the valley areas because of the
presence of the tsetse fly, which spreads trypanosomiasis. Hybrid maize production
in valley areas was more widespread in the early 1980s, when some IDZ areas were

Table 10-Oxen use by hybrid maize adoption

Area/Hybrid of Sample
Maize Use Use Oxen Do Not Use Oxen Total Households
(N) (percent) (N) (percent) (N) (percent)
Nonadopters 73 49.0 81 50.3 154 49.7
Adopters 76 51.0 17 10.6 93 30.8
Nonadopters 0 0.0 63 39.1 63 19.5
All 149 100.0 161 100.0 310 100.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: N is sample size.

located there.17 A previous IFPRI study found 13 percent of maize area in the
Jumbe-Chikowa area of the valley planted in hybrids in 1980/81 and 10.5 percent in
1981/82 (IFPRI/NFNC/RDSB 1985).

Hybrid Maize Adoption by Male
or Female Household Heads

Female-headed households accounted for 30 percent of all households included
in the study. This figure is in line with the 32 percent of farm households legally
headed by women in Eastern Province in a 1982/83 farm survey of Zambia (Safilios-
Rothschild 1985). Due to limited male-only outmigration, de facto female heads are
not present, except on a seasonal basis. Since the majority of the rural female-headed
households are primarily engaged in agriculture, from an agricultural policy perspec-
tive, they should be considered important in design and implementation of develop-
ment activities.
Evaluations of agricultural development programs in Zambia have generally
concluded that female-headed households were underrepresented as beneficiaries of
these programs (Due and White 1986; Bliven 1991). More recently, in some prov-
inces, such as Eastern and Northern, funds for credit were set aside for women and
made available to members of women's clubs in the Lima agricultural program.
Although these amounts were insufficient relative to the proportion of women
involved in agriculture (either as female heads of households or as wives), they still
represent an improvement over standard practices. Nevertheless, only 12 percent of
female heads in these provinces participated in the Lima program (IRDP 1983).
For the sample as a whole, in 1986 female-headed households had a lower rate
for adoption of hybrid maize (22 percent) than male-headed households (34 percent).
However, the pattern is not the same across farm sizes (Table 11). Except for the
smallest farm category, in which adoption is minimal in both male- and female-
headed households, the pattern for the next two groups is different from that of the
two largest farm size groups. In the 1-to-3-hectare sizes, female-headed households
have a much lower rate of adoption than the male-headed households. This is
consistent with the overall pattern and is also reported in most of the technological
change and commercialization literature. Surprisingly, however, female-headed
households in the 3-to-5-hectare category have a higher adoption rate than male-
headed households, and all households over 5 hectares are adopters. This difference
by farm size implies that once women are able to overcome resource constraints, they
are just as likely or even more likely to become technological innovators. However,
when they are faced with resource constraints, women are less likely to adopt new
technology, either because they tend to be more risk averse or because they face
greater hurdles in obtaining technological inputs or other requirements for adopting
improved technologies.

17Areas that were designated to be part of the IDZ program included some districts in the Luangwa
Valley, where farmers received access to tractors for field preparation in addition to seeds, fertilizer, and

Table 11- Farm size and hybrid maize adoption by head of household

Farm Size in Hectares per Household of Sample
Head of Less More House-
Household than 1 1-2 2-3 3-5 than 5 Total holds
(N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent) (N) (percent)
Nonadopter 69 62.2 41 46.1 16 40.0 17 36.2 0 0.0 143 46.1
Adopter 5 4.5 20 22.5 16 40.0 18 38.3 16 69.6 75 24.2
Nonadopter 35 31.5 26 29.2 6 15.0 5 10.6 0 0.0 72 23.2
Adopter 2 1.8 2 2.2 2 5.0 7 14.9 7 30.4 20 6.5
All 111 100.0 89 100.0 40 100.0 47 100.0 23 100.0 310 100.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: N is sample size.

Demographic characteristics of female-headed households by farm size offer
some possible explanations. In general, for both male- and female-headed house-
holds, as farm size increases, so does household size. While male-headed households
have a significantly larger number of male adults, female-headed households have a
larger number of female adults. As farm size increases, however, there is a statisti-
cally significant increase in male adults, female adults, and children in male-headed
households, but for female-headed households, the increase in adult males (15 years
old or more) is especially marked. The only other demographic difference is that the
age of female heads is significantly lower than male heads in the larger farm
households. Female heads cultivating 3 or more hectares are, on average, seven years
younger than male household heads cultivating 3 hectares or more.

Analysis of Hybrid Maize Adoption
and Its Income Effects

Analysis of factors contributing to the adoption of improved agricultural tech-
nologies for this sample of households in Eastern Province was also conducted by Jha
and Hojjati (1993) and by Jha, Hojjati, and Vosti (1991). In addition to results
confirming that area expansion is conducive to adoption of hybrid maize and facili-
tated by use of animal traction and increase in family size, they found increased
length of residence in the area also to be a factor.18 Other factors contributing to
hybrid maize adoption included membership in the cooperative Eastern Province
Cooperative Union (EPCU), age of head of household (younger households were
more likely to be adopters), and gender of head of household (male heads were more
likely to be adopters). Access to markets and infrastructure was not a factor.

18For background information on land tenure in Eastern Province, see Milimo (1991). Conversations with
farmers during the study suggest that most had access to fallow land.

In contrast to hybrid maize adoption, fertilizer use was more widespread and was
adopted by twice as many farmers as was hybrid maize. Jha and Hojjati (1993) did
not find an increase in farm size to be a predictor for fertilizer use. However, intensity
of fertilizer use was clearly associated with hybrid maize adoption. This is reflected
in the relative frequency of fertilizer use on hybrid versus local maize fields as well
as in the higher rate of application on hybrid maize. Fertilizer was applied on
virtually all hybrid maize fields but only on 47 percent of local maize fields. In terms
of rate of fertilizer application, however, hybrids received only 50 percent more
fertilizer (nutrients) per fertilized hectare than local maize. According to field trials
conducted by the Eastern Province Agricultural Research Station, this incremental
fertilizer application on hybrid maize is consistent with the relative yield response
and value-cost ratio for its application on hybrid versus local maize. No other crops
besides maize received any inorganic fertilizer application in the study year.19
In order to derive the income effects of hybrid maize adoption, the problem of
endogeneity of the measured adoption behavior needed in this report has to be
resolved. That data from only one crop season are available compounds the problem.
Exogenous variables that are not influenced by adoption are therefore used as
predictors of adoption behavior. The results of the Heckman two-stage model,
estimating the acreage planted to hybrid maize, as described in Chapter 3, are
presented in Table 12. Predicted values for adoption are then used to explain vari-
ations in household income. In estimating the effects of hybrid maize adoption on
income, both the predicted values for its adoption per se and area planted to it are
used in addition to other explanatory variables. Total consumption expenditure is
used as a proxy for disposable income.20
Results of the predicting equations for hybrid maize adoption show the same
signs and significance as work done earlier by Jha, Hojjati, and Vosti (1991). The
next set of results examines the income effects of hybrid maize adoption at different
farm sizes.
The analysis of the impact of hybrid maize on income and consumption expendi-
tures is complicated by the fact that (1) its adoption is collinear with increase in farm
size, and (2) its effect on income can be decomposed into two components: the effect
of adoption and the effect of increasing area under the hybrid. In order to incorporate
all of these factors, three cultivated area variables are modeled on the right-hand side
of the income equation:
Size of total area farmed (TOTHA);
Predicted area planted to hybrid maize in hectares (FAREAHMZ); and
Square of predicted area planted to hybrid maize (FAREAHMZ x FAREAHMZ
Other variables are household demographics, education, and ecological zones.
Per capital and total household consumption expenditure are used as a proxy for
disposable income.

19Application of organic manure is limited to a small fraction of plots on which animals are "corralled"
(fenced in for extended periods of time). Only six fields were corralled during the study period, three of
which were planted in local maize, two in legumes, and one in hybrid maize.
20Consumption expenditure is commonly used as an indicator of household welfare and is more likely to
predict permanent income than short-term income measures (Glewwe 1990).

Table 12- Determinants of hybrid maize adoption and its income effects

Probability of Area Under Hybrid Maize Per Capita Consumption Total Consumption
Hybrid Maize Adoption, (Conditional on Adoption), Expenditure Per Month, Expenditure Per Month,
Independent Mean = 0.29 Mean = 1.7 Mean = 3.9 Mean = 5.6
Variable Mean Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio

AGEHEAD 43.6 -0.0208 -2.55* -0.0322 -2.93* -0.0016 -1.01 -0.0028 -1.60
FAMILSZ2 5.9 0.0115 0.28 0.0035 0.08 -0.1101 -13.37* 0.0547 6.29*
HEADHH 0.7 0.6448 2.38* 1.0042 3.14* 0.0347 0.71 0.0781 1.50
DEPRATIO 1.1 ... .. ... -0.0773 -3.08* -0.0328 -1.24
EDHHLD 3.6 0.0483 1.39 -0.0124 -0.28 0.0147 2.01 0.0141 1.82
COOPMEM 0.2 0.1664 0.61 0.7418 2.69* .
DUMPLAT1 0.4 0.4978 1.20 -0.4725 -0.54 ...
DUMPLAT2 0.4 1.4351 3.41* 0.4084 0.43 ...
OXEN 0.5 0.0039 0.02 0.0832 0.24
VALLPLAT 0.2 ... ... ... ... 0.3376 6.17* 0.3711 6.43*
THa 2.1 0.5251 5.86* 0.6299 8.80* 0.0310 1.45 0.0426 1.89
TOTINOTH 115.1 -0.0001 -0.31 0.00003 0.12 ......
AREADINF 1.3 0.1651 1.49 0.0353 0.25 ......
MILLSI 0.7 ...... 0.8871 2.18*
FAREAHMZ 0.5 ...... ...... 0.1511 3.31* 0.1668 3.47*
PAREASQ 1.6 ... ... ... ... -0.0153 -2.94* -0.0223 -4.07*
Constant -2.70 -4.41* -1.26 -0.92 4.4818 49.80* 5.1063 53.82*
R2 (adjusted) 0.39 0.67 0.49 0.35
F 18.3 15.7 33.0 19.0
Sample size 304 88 304 304

Notes: Variable definitions:

AGEHEAD = Age of household head
FAMILSZ2 = Family size
HEADHH = Sex of the household head: 1 = male, 0 = female
DEPRATIO = Dependency ratio: +60/14-60
EDHHLD = Education of the household head: grades completed
COOPMEM = Membership in cooperatives: 1 = yes, 0 = no
DUMPLATI = Dummy area level low adoption: 1 = low
DUMPLAT2 = Dummy area level high adoption: 1 = high
OXEN = Cultivation method dummy: 1 = oxen use
The ellipses indicate a nil or negligible amount.
*Significant at the .05 level.

VALLPLAT = Valley or plateau: 1 = valley, 0 = plateau
THa = Total area farmed: hectares per household
TOTINOTH = Income from other sources kwachaa per household per year)
AREADINF = Area level infrastructure index based on distance to 12
elements (the higher the index, the poorer the infrastructure)
MILLS1 = Inverse mills ratio
FAREAHMZ = Predicted area under hybrid maize: hectares per household
PAREASQ = FAREAHMZ x FAREAHMZ (the square of predicted area
under hybrid maize)

The results of this analysis indicate that increasing both farm size and area under
the hybrid have a positive effect on household income. Increasing area under hybrid
maize has a much higher effect, as is to be expected. However, the variable for hybrid
maize area squared (PAREASQ) has a statistically significant negative sign, indicat-
ing a higher potential for income increments at the smaller farm sizes but diminishing
returns beyond a certain point of expansion of area under hybrid maize. In order to
derive the inflection point, the point at which rl = 0 (rl = elasticity) is calculated. When

then y = a + Px + yx2 + u, (13)
T = (Py + 2yx.y)(x/y), (14)
at rT = 0, x = -P/2y.21 The inflection point for both per capital and household income
equations is 4 hectares of hybrid maize, indicating that planting larger acreages is
There are several possible explanations for the negative quadratic sign on hybrid
maize area. First, hybrid maize may change intrahousehold interactions, so that even
though the return to households in per capital terms declines, gains to individuals who
control the crop may be higher. This may be an important factor in low productivity
improvements overall and related to the deterioration in women's access to resources
that sometimes occurs with hybrid maize adoption.22 This aspect is explored further
in the next section. Other explanations that have been proposed for this kind of result
are that subsidized inputs available to larger farmers, including credit and improved
seeds and fertilizers, distort the resource allocation patterns and, in effect, support
farmers who may be less productive than others.23
These results are mirrored later in the food consumption and child nutrition
results, where large farms adopting hybrid maize have a lower level of food consump-
tion and child nutrition than large, nonadopting farms, while adoption is positively
related to these effects on small farms. (The cutoff point for each farm size group is
the median for the sample of adopting and nonadopting farms.)

Small: 1.46 hectares
Adopters Large: 4.39 hectares
Farm Size -
Nonadopters Small: 1.22 hectares
Large: 3.01 hectares

21The following adjustments were made in the values of the regression coefficients for hybrid maize area
and hybrid maize area squared to accommodate the log value of the dependent variable. When the
equation log y = ct + 1*x + y*x squared is estimated, 13 and y can be calculated at the mean value of the
dependent variable as follows:
1 = antilog (logy + P*) antilogy,
y = antilog (log y + y*) antilog y,
and at l = 0, x = (-P)/2y.
22Milimo (1989) also refers to women's insecure access to land as a source of low agricultural
productivity in Zambia.
23Yadav, Otsuka, and David (1992) have observed this in Nepal.

Other variables contributing to increased disposable income are education of
head of household and having a male head of household, Valley households have a
significantly higher level of consumption expenditure than plateau households. In-
creasing household size benefits total household income but reduces per capital
income. A higher dependency ratio lowers both per capital and household income.

Intrahousehold Dimensions of Agriculture
As part of the crop production study, information on the person who owned or
managed each plot of land was obtained. This makes it possible to examine the extent
of women's responsibility and ownership of different crops. For the Eastern Province
sample as a whole, the results in Table 13 show the extent to which women are
responsible, either on their own or jointly with others, for different crops. For local
maize and other cereals, about one-fifth of the land is independently managed by
women, and an additional one-third is jointly managed with men. Overall, nearly 57
percent of local maize and 70 percent of other cereals are independently or jointly
managed by women.

Table 13- Women's crop ownership and share

Area Farmed by Women
Total Area Independently
Crop/Sharea Farmed Independently Jointly and Jointly

Maize (hectares/household)
Local maize
Sole crop 0.44 0.08 0.14 0.22
Share ... 0.18 0.32 0.50
Intercrop 0.79 0.19 0.30 0.49
Share . 0.24 0.38 0.62
Total 1.23 0.26 0.44 0.70
Share ... 0.21 0.36 0.57
Hybrid maize
Sole crop 0.32 0.01 0.08 0.09
Share ... 0.03 0.25 0.28
Intercrop 0.16 0.02 0.01 0.03
Share ... 0.13 0.06 0.19
Total 0.48 0.03 0.09 0.12
Share ... 0.06 0.19 0.25
Total maize 1.71 0.30 0.54 0.84
Share ... 0.18 0.32 0.50
Other cereals 0.10 0.02 0.03 0.05
Share ... 0.20 0.30 0.70
Groundnuts and legumes 0.16 0.02 0.06 0.03
Share .. 0.13 0.38 0.51
Cotton and sunflower 0.07 0.00 0.02 0.02
Share ... 0.00 0.29 0.29
All crops 2.04 0.34 0.65 0.99
Share . 0.17 0.32 0.49

Source: International Food Policy Research Institute, Rural Development Stdifes Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
aCrops are in hectares of area farmed and shares are the percentages of the total area farmed by women.

Hybrid maize is similar to the other cash crops, cotton and sunflower, in that
much less is under women's control. Only 25 percent of hybrid maize area is
independently or jointly managed by women. About 51 percent of the land in
groundnuts and other legumes, is managed by women. Of the total area farmed, 17
percent is farmed independently and 32 percent jointly by women for a total share of
49 percent.
When the pattern of crop ownership is considered by head of household, there are
differences between male- and female-headed households (Table 14). As might be
expected, in female-headed households, women farm a substantially larger share of
land--63 percent of the land independently and 22 percent jointly with men.
Women in male-headed households cultivate an insignificant amount of land
independently: only 4.5 percent. The share for local maize is only slightly higher than
the overall average. It is significant to note that the secondary (other) cereals (which
are also the traditionally grown cereals, such as finger millet, sorghum, and rice) have
the highest share of independent cultivation by women, about 19 percent in male-
headed households. A substantial share of land, 43.5 percent, in male-headed house-
holds is jointly farmed by women. Of all the crops, hybrid maize has the smallest
share of management by women, but it is still a substantial 31 percent. In female-
headed households, women manage (independently or jointly) 47 percent of the area
under hybrid maize.
The pattern of intrahousehold crop ownership indicates that women have little
independent ownership except in female-headed households. Secondary cereal crops
that are likely to have been traditionally grown by women (or have little commercial
appeal) are most likely to be managed independently by them in male-headed
households. However, 48 percent of the land in male-headed households and 85
percent of the land in female-headed households is independently or jointly managed
by women. The crop with the smallest share of women's management is hybrid
maize, but even in that, 31 percent of land in male-headed and 47 percent in
female-headed households are independently or jointly managed by them.

Women's Involvement in Crop Management
With Hybrid Maize Adoption
The concept of separate crop ownership for different plots of land farmed by a
household is well known and accepted in Zambia's Eastern Province, where the
traditional land tenure system predominates. This is similar to usufruct rights to land
and is distinct from land ownership. Milimo (1991) observes that the concept of
individual land ownership does not exist in the traditional land tenure system in this
area. Since both patrilineal and matrilineal systems coexist in Eastern Province, land
that has been assigned to the extended family is passed on to the next generation
either through the father or the mother.
Within the household itself, farming responsibility is divided between members,
with output produced in the household's fields the predominant food source. This is
generally local maize, which is usually jointly owned by husbands and wives. Other
fields may be cropped with primary responsibility for output resting with an individ-
ual. This assignment may depend on the perceived role of the individual; for example,
women traditionally have the responsibility for providing the "relish" ingredients for
the meal, which could include groundnuts, leafy vegetables, fish, or meat. Output from

Table 14--Women's crop ownership and share, by household head

Male-Headed Householda Female-Headed Household
Percent Percent
Area of Area Percent Area of Area Percent
Number Farmed Farmed Area of Area Farmed Farmed Area of Area
in Total Total Percent Indepen- Indepen- Farmed Farmed Total Percent Indepen- Indepen- Farmed Farmed
Crop Sample Area of Area dently dently Jointly Jointly Area of Area dently dently Jointly Jointly
(hectares) (hectares) (hectares) (hectares) (hectares) (hectares)
Local maize
Sole crop 123 1.10 56.0 0.08 6.7 0.32 39.1 1.14 68.3 0.54 55.4 0.49 32.3
(92) (92) (92) (92) (92) (92) (31) (31) (31) (31) (31) (31)
Intercrop 197 1.20 66.0 0.07 5.7 0.58 50.9 1.40 75.0 0.75 70.6 0.27 14.3
(131) (131) (131) (131) (131) (131) (66) (66) (66) (66) (66) (66)
Total 299 1.30 66.3 0.08 6.1 0.51 46.0 1.40 77.6 0.73 64.5 0.36 21.2
(208) (208) (208) (208) (208) (208) (91) (91) (91) (91) (91) (91)
Hybrid maize
Sole crop 59 1.70 44.0 0.08 2.7 0.42 34.9 1.60 27.7 0.06 20.0 0.44 20.0
(49) (49) (49) (49) (49) (49) (10) (10) (10) (10) (10) (10)
Intercrop 35 1.50 45.3 0.02 3.7 0.17 13.8 1.40 35.4 0.72 50.0 0.00 0.0
(27) (27) (27) (27) (27) (27) (8) (8) (8) (8) (8) (8)
Total 88 1.80 47.7 0.06 2.0 0.36 29.0 1.60 33.0 0.38 35.3 0.26 11.8
(71) (71) (71) (71) (71) (71) (17) (17) (17) (17) (17) (17)
Total maize 307 1.80 79.6 0.10 4.1 0.61 44.6 1.70 83.8 0.80 64.3 0.41 21.1
(216) (216) (216) (216) (216) (216) (91) (91) (91) (91) (91) (91)
Other cereals 60 0.51 30.8 0.06 18.8 0.15 38.7 0.49 31.0 0.14 31.3 0.14 18.8
(44) (44) (44) (44) (44) (44) (16) (16) (16) (16) (16) (16)
Groundnuts and legumes 183 0.29 17.7 0.01 6.4 0.11 41.3 0.23 15.0 0.13 71.4 0.07 22.4
(134) (134) (134) (134) (134) (134) (49) (49) (49) (49) (49) (49)
Cotton and sunflowers 53 0.38 18.9 0.01 2.8 0.12 38.9 0.41 15.7 0.06 25.9 0.13 32.9
(36) (36) (36) (36) (36) (36) (17) (17) (17) (17) (17) (17)
Miscellaneous 27 0.11 8.9 0.0004 2.2 0.03 34.3 0.27 18.3 0.26 75.0 0.01 25.0
(23) (23) (23) (23) (23) (23) (4) (4) (4) (4) (4) (4)
Total 310 2.10 100.0 0.12 4.5 0.72 43.5 1.90 100.0 0.90 63.1 0.49 21.7
(218) (218) (218) (218) (218) (218) (92) (92) (92) (92) (92) (92)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and
Nutrition Commission agricultural household survey, Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are sample sizes.
aThese are crops managed by women in households headed by males.

groundnut fields is therefore predominantly reported to be owned by women. Alterna-
tively, individuals perceived to have a comparative advantage or familiarity with a
crop, especially a new crop, may decide to cultivate that independently. This could be
a factor in the predominance of hybrid maize fields independently owned by men.
Crop ownership by an individual for a particular field does not mean that labor is
provided only by that person. He or she may request, demand, or cajole labor from
other household members or kin, and this ability is very likely to be an important
influence on the crop outcome. It does, however, give that individual a high degree
of control of allocation of the output from that field. It also implies that the individual
is primarily responsible for management of the crop on it.
Since hybrid maize is primarily raised on the plateau, a comparison of changes in
women's involvement with its adoption is made only for the plateau sites. The total
area farmed independently by women is the same in adopter and nonadopter house-
holds, but the area farmed jointly increases by about 50 percent with adoption (Table
15). This shows that women take on somewhat more responsibility for cultivation in
absolute terms in adopter households. When these differences are seen in relation to
the 2.5 times larger farm sizes cultivated in the adopting households, it becomes clear
that the relative share of women's crop area increases less than proportionately with
hybrid maize adoption. Thus, in nonadopting households, women have primary
responsibility for about 31 percent of farm land and for an additional 38 percent of
land that is farmed jointly with other household members, for a total of more than 69
percent. In contrast, in adopting households, women have primary responsibility for
production on only about 12 percent of land and joint responsibility for an additional
30 percent, for a total of 42 percent. In the process, they reduce the area they farm
independently in local maize and substitute that for an additional share of hybrid
maize under joint production with men.
In order to test whether the reduction in women's share of crop ownership that is
observed to be associated with hybrid maize adoption is a function of adoption or of
other characteristics associated with it such as increased farm size, the following
multivariate relationship is analyzed24:

Fs = f(HM*, THa, Head, EG), (15)
Fs = share of land under independent and
joint production by females,
HM* = predicted hybrid maize adoption,
THa = total farm size,
Head = sex of head of household, and
EG = patrilineal or matrilineal ethnic group.

The results presented in Table 16 confirm that the effect of hybrid maize adoption is
independent of the effect of increasing farm size, and adoption has a large impact in
that it decreases the share of area farmed that is owned and managed by women,
either independently or jointly. This impact is similar in size and direction to that of
shifting from a female to a male head of household. However, holding other factors

24This analysis is limited to the plateau sites only.

Table 15-Women's crop ownership and share, by hybrid maize adoption, plateau

Hybrid Maize Adopted Hybrid Maize Not Adopted
Area Area Area Area
Crop/Share/ Total Farmed Farmed Total Farmed Farmed
Number of Sample Area Independently Jointly Area Independently Jointly

Maize (hectares/household)
Local maize
Sole crop 1.75 0.17 0.61 1.20 0.28 0.33
Percent 41.0 12.0 33.3 73.2 26.6 31.0
N (33) (33) (33) (58) (58) (58)
Intercrop 1.60 0.32 0.41 1.20 0.35 0.56
Percent 48.2 20.2 29.5 80.4 35.1 42.1
N (61) (61) (61) (106) (106) (106)
Total 1.80 0.29 0.52 1.30 0.35 0.51
Percent 49.9 17.7 31.4 83.4 31.1 38.8
N (86) (86) (86) (153) (153) (153)
Hybrid maize
Sole crop 1.74 0.08 0.44 0.00 0.00 0.00
Percent 40.8 5.8 33.5 00.0 00.0 00.0
N (57) (57) (57) (0) (0) (0)
Intercrop 1.50 0.18 0.13 0.00 0.00 0.00
Percent 43.0 14.3 10.6 00.0 00.0 00.0
N (35) (35) (35) (0) (0) (0)
Total 1.80 0.13 0.34 0.00 0.00 0.00
Percent 44.5 8.6 26.3 00.0 00.0 00.0
N (86) (86) (86) (0) (0) (0)
Total 3.30 0.39 0.80 1.30 0.35 0.51
Percent 87.4 13 29.6 83.4 31.1 38.8
N (93) (93) (93) (153) (153) (153)
Other cereals 0.73 0.02 0.11 0.46 0.17 0.07
Percent 18.7 7.1 28.6 19.7 38.5 15.4
N (14) (14) (14) (13) (13) (13)
Groundnuts and legumes 0.36 0.04 0.14 0.29 0.06 0.11
Percent 11.5 17.5 36.8 21.7 33.3 33.0
N (57) (57) (57) (80) (80) (80)
Cotton and sunflower 0.46 0.02 0.07 0.39 0.02 0.16
Percent 12.8 10.5 15.8 20.9 11.5 42.4
N (19) (19) (19) (26) (26) (26)
Miscellaneous 0.10 0.00 0.00 0.25 0.15 0.00
Percent 4.3 00.0 00.0 14.5 42.8 00.0
N (3) (3) (3) (7) (7) (7)
Total 3.70 0.42 0.92 1.50 0.41 0.60
Percent 100.0 12.3 29.7 100.0 30.6 37.7
N (93) (93) (93) (153) (153) (153)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The numbers (N) in parentheses are sample size.

constant, a greater share of women's crop area was found to be in patrilineal ethnic
areas. The main ethnic groups in Eastern Province are the predominantly matrilineal
Chewa and Nsenga in the southern and western districts of Petauke, Nyimba, Katete,
Chadiza, (and the valley districts of Mambwe and Chama), and the patrilineal Ngoni
and Tumbuka of the northern districts of Chipata North, Chipata South, and Lundazi.
The higher share of women's crop area found in the patrilineal sites was surprising
but may be related to a greater responsibility for production by the marriage partner
who moves into the other's kinship land (Davison 1994; Crehan 1994). This may not

Table 16- Factors in women's management role

Independent Variable Mean Coefficient t-Ratio

HEAD 0.69 -0.3276 -5.71*
THa 2.40 -0.0443 -0.29
HM* 0.29 -0.3203 -4.17*
EG 0.51 0.1663 3.19*
Constant ... 0.8219 13.90*
R2 (adjusted) 0.26
F 224.1*
N 242

Notes: The dependent variable is the proportion of area farmed by females independently and jointly. The variables
are as follows:
HEAD = Sex of household head: 1 = male, 0 = female;
THa = Total area farmed: hectare per household;
HM* = Predicted hybrid maize adoption; and
EG = Ethnic group: 1 = patrilineal, 0 = matrilineal.
*Significant at the .05 level.

necessarily mean that women in patrilineal areas have better access to resources than
those in matrilineal areas.

Implications for Agricultural Policy
in Eastern Prpvince
The analysis of adoption patterns of hybrid maize production and its effects on
household income are generally positive, but the intrahousehold income benefits
appear to be unevenly distributed between men and women-either as female heads
or producers in male-headed households. The effects on interhousehold income
distribution is mixed in that, while the smaller farmers had less access to adoption of
hybrid maize, it was the smaller adopters that had the most income benefit from its
adoption. Larger farms had a smaller income benefit from adoption of hybrid maize,
and this is likely to be associated with labor shortages, inefficient use of inputs, and
increasing intrahousehold inequity.
Overall, smaller farms and those headed by women are less likely to adopt the
hybrid maize. This is because of their limited access to inputs, especially credit, seed,
and fertilizer, which were available primarily through cooperative membership, and
also their poorer access to extension and training services. Farmers who adopted
hybrid maize experienced an overall improvement in their level of income in com-
parison with those who did not adopt, holding farm size constant. However, for farms
cultivating more than 4.0 hectares of hybrid maize, the income gains from incre-
mental acreage in hybrid maize were declining. One possible explanation for this
counterintuitive result is the declining share of women's involvement in these farms
(except on female-headed households) and, hence, their lower access to income
gains. Since the intrahousehold control of hybrid maize income is expected to accrue
primarily to the male members, especially the male heads of households, the reduc-
tion in disposable income gain for the household as a whole could occur even as area
planted in hybrids is increasing.

In the past, the agricultural policy decisions in Zambia were made largely from the
perspective of increasing marketed surplus of maize. This focus probably led to the
observed patterns of adoption and effects. With the structural adjustment program and
market liberalization efforts currently under way, the need to promote policies that are
efficient in making productivity gains and that are sustainable becomes more crucial.
One of the initial effects of the move away from the panterritorial pricing policies of
the 1970s and 1980s will be a contraction of the geographical area where maize is
produced for the market, which will only be profitable in the line-of-rail provinces and
some adjoining areas. For the rest of the country, including Eastern Province and other
outlying areas, agricultural policies will need to focus on making real productivity
gains and in involving the private and nongovernmental sectors in making investments
in services and infrastructure that will promote development in these areas.
The focus in the outlying provinces such as Eastern Province will need to be less
on promoting larger farm sizes and more on increasing labor productivity and
efficient use of inputs. Seeds, inputs, and extension approaches more suitable for
dispersed populations of small farmers will need to be packaged according to the
ecological, farm system, and consumption patterns that exist in these areas. Improved
maize seed-fertilizer packages will remain integral to the agricultural growth needs
in areas where its yield potential is above average, as in Eastern Province.



Understanding changes in patterns of labor allocation is an integral part of examin-
ing the effectiveness of technological change in agriculture. For small farmers such as
those in Eastern Province, household labor is an important part of the production
process as well as the welfare and consumption function, and thus it is central to their
utility outcomes. Household production models and the theory of human capital have
allowed economics to go beyond the work or leisure dichotomy in explaining the
allocation of time (Becker 1965; Singh, Squire, and Strauss 1986). In these formula-
tions, a household member's contribution to "full income" includes both disposable
income and time. While income activities generate the primary resources, time spent on
other activities is essential for the production of household goods and services (such as
meals and child care) that directly enter into the household utility function. These
activities are primarily undertaken by women. It follows from this that, as incomes go
up, the demand for both commodities and time required for generating the final product
could also increase (provided there is no improvement in technologies used in produc-
ing the Z-goods). This process is usually reflected in a substitution of women's labor
from production to home consumption-related activities as incomes rise.
Technological change in agriculture can create competing demands for house-
hold labor. It may require additional time spent in agricultural production even while
it generates an increase in household income. In this process, households may be
adding to their food energy requirements as well as generating additional demand for
labor in consumption and welfare support activities, the demand for which increases
with incomes. The resulting allocation of household labor determines not only the
effectiveness of the production response, but also the effectiveness of the consump-
tion response. In this process, factors external to the household, such as the charac-
teristics of the rural labor market, and factors internal to the household, such as
intrahousehold decisionmaking and control of resources, can influence the outcomes.
In this chapter, the patterns of household labor allocation with adoption of new
maize varieties are examined for agricultural production, other income activities, and
household maintenance work. Labor allocation data were obtained between Decem-
ber 1985 and December 1986. During this period, the first crop cycle was completed
by June, when the harvest of maize was completed. Tabulations reflecting crop labor
use are therefore only made for this period. Use of nonhousehold labor in agriculture
is also examined to see the extent to which the agricultural labor market influences
production outcomes.

Labor Use by Crop
Among the crops, reported labor input per hectare is lowest for hybrid maize-50
percent lower than that for local maize (Table 17). As indicated earlier, only crop
labor up to harvest is included in the data (that is, December through June). Obvi-

Table 17- Labor use per hectare, by crop

Number Average Hours of Labor/Hectare/Seasona
Crop of Households Area Farmed Family Exchange Hired
Local maize
Sole crop 115 1.2 1,269.9 34.3 10.1
Intercrop 191 1.3 1,028.3 41.2 24.6
Total 291 1.3 1,142.1 33.7 17.7
Hybrid maize
Sole crop 57 1.8 658.6 17.6 49.5
Intercrop 34 1.5 610.5 113.1 60.6
Total 86 1.8 643.9 49.2 44.0
Total maize 301 1.8 1,128.4 34.5 39.2
Other cereals 53 0.6 1,894.6 47.2 24.4
Groundnuts and legumes 135 0.4 2,652.4 28.4 45.8
Cotton and sunflower 45 0.5 1,231.2 40.9 59.0
All crop average 308 2.08 1,558.2 38.8 42.4

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
aMonths covered were December 1985 to June 1986.

ously, hoe cultivators use a substantially higher labor input per hectare than oxen
users, controlling for type of crop, but the labor input for local maize is still twice as
high whether oxen or hoes are used. These figures for labor use are comparable with
Bessell's (1971) observations, made in detailed labor use investigations for Eastern
Province. They are also consistent with studies of labor productivity in Zambia,
which show a flattening of the yield-to-labor curve after 1,500 hours per hectare in
hoe-cultivated maize (Stromgaard 1984).
One of the most striking observations of this report is that hybrid maize has a
significantly lower labor input than local maize, but research on farming systems in
maize production in Eastern Province indicates that hybrid maize requires 30 percent
more labor than local maize for optimum yields (EPADP 1987).25 That farmers pay
much more attention to local maize than to hybrid maize is consistent with earlier
observations on priority given to early planting of local maize at the start of the
season. A large part of the additional labor in local maize appears to be spent in
weeding. Research in Eastern Province has found that up to 25 percent of local maize
fields may receive a third weeding, whereas the majority of hybrid fields are weeded
only once (Eastern Province Adaptive Research and Planning Team 1984).
Groundnuts and legumes receive the highest level of labor input, with other cereals
coming next. Hired labor use is highest for the cash crops, hybrid maize and cotton,
followed by groundnuts. Hired labor for other cereals and local maize is much lower at
only 24 and 18 hours per hectare. Exchange labor input is highest for hybrid maize and
other cereals. Overall, nonhousehold labor provides an insignificant share of crop labor,

25Estimated labor use for hybrid maize was reported at 130 workdays compared with 100 workdays for
local maize, both using oxen cultivation.

with the amount provided by exchange and hired labor nearly the same. Farmers are
most likely to use hired labor rather than exchange labor on sole-cropped hybrid maize,
while they are more likely to use exchange labor and rely less on hired labor for local
maize, intercropped hybrid maize, and other cereals. This higher wage labor input could
also be a response to the reduced input of women's labor on this crop.
Intrahousehold labor share by crop shows that women's and children's share of
labor is lowest for the most commercialized crops: hybrid maize, cotton, and sun-
flowers (Table 18). Women's share of labor for these crops is 44 and 38 percent,
respectively, while that for children is 11 and 9 percent. Women's share in sole crop
local maize is 52 percent and children's share is 11 percent; in other cereals, women's
is 54 percent and children's 15 percent; and, in groundnuts, women's is 53 percent
and children's 12 percent. At an aggregate level for all crops and households, women
provide nearly 49 percent of all household crop labor, men provide only 39 percent,
and children provide 13 percent. This sample includes female-headed households,
and shows the overall importance of women's farm labor in agricultural production.

Variations in Labor Use by Farm Size

In most labor-abundant and land-scarce rural areas, there is a clear pattern of
lower family farm labor input with increasing incomes and farm sizes. This is largely
due to greater use of wage labor and, to some extent, to a reduction in the intensity of
labor input per hectare on larger farms. The pattern of labor use in smallholder
agriculture where labor is scarce and land is abundant, as in this part of Zambia, is
generally expected to be different. There are, however, many similarities.

Table 18-Intrahousehold labor, by crop

Number of Area Male Labor Female Child Labor
Number of Area
Crop Households Farmed Hours Percent Hours Percent Hours Percent
Local maize
Sole crop 123 1.1 312.6 36.1 454.6 52.5 98.2 11.3
Intercrop 197 1.3 319.8 39.6 372.2 46.1 115.0 14.3
Total 299 1.3 345.3 38.2 441.0 48.8 117.4 13.0
Hybrid maize
Sole crop 58 1.7 260.1 46.4 243.7 43.5 56.6 10.1
Intercrop 35 1.5 243.2 44.4 237.1 43.3 67.2 12.3
Total 88 1.7 275.0 45.5 264.9 43.9 64.0 10.6
Total maize 307 1.8 432.7 39.6 521.3 47.7 139.5 12.0
Other cereals 60 0.5 174.2 30.9 306.5 54.4 83.1 14.7
Groundnuts 183 0.3 211.6 34.5 326.2 53.1 76.4 12.4
Cotton and sunflower 53 0.4 214.8 53.3 152.1 37.8 36.0 8.9
Miscellaneous 27 0.1 222.2 55.9 133.0 33.5 42.3 10.6
All crop average 310 2.1 703.8 39.0 874.1 48.5 225.9 12.5

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Months covered were December 1985 to June 1986.

Labor use per cultivated hectare of land has a pattern similar to that in areas with
high population density, and it declines as the area farmed increases (Table 19).26
This is found to be so for all types of family farm labor: men, women, and children.
Adult male labor on the smallest size farms (less than 1 hectare) is 734 hours per
hectare, and this is reduced by 84 percent to 120 hours per hectare in the more-than-
5-hectare group, about one-sixth the higher figure. The comparable reduction for
women is from 1,103 hours per hectare to 134 hours per hectare (an 88 percent
reduction), and, for children between 5-15 years, labor per hectare is reduced from
157 hours per hectare on the smallest farms to 31 hours in the largest-an 80 percent
reduction. It should be noted that women, despite a reduction in the overall intensity
of household labor input per hectare as farm size increases, still tend to put in more
hours of labor on crop production than men.
Intensity of nonhousehold labor input increases with increasing farm size, espe-
cially hired labor (Table 19). Though nonhousehold labor use increases with farm
size, it does not counteract the rapid reduction in labor intensity per hectare with
increasing farm size.
Use of technologies such as oxen-plow cultivation is generally expected to
reduce the labor requirement for farming. For the sample as a whole, this is found to
be true, with farmers using oxen reporting 934 labor hours per hectare compared with
1,655 hours per hectare for hoe-using households (Table 20). But this is partly due to
the larger farm size of oxen-plow cultivators. When the differences are examined for
farms of similar hectarage, then the labor-saving effect of oxen cultivation is evident
only for the larger farm sizes-those of more than 2 hectares. For these larger farm
sizes, the reduction in women's and children's labor is most pronounced for those in
the 2-to-5-hectare range.
Another question addressed in this section is how household labor allocation for
all farm and off-farm activities changes with an increase in farm size. For men, the

Table 19- Total labor use, by farm size

Share of Family Labor Total
Share of Family Labor Family Exchange Hired Total
Farm Size N Area Male Female Child Labor Labor Labor Labor
(hectares) (percent) (hours per hectare)
Less than 1 hectare 109 0.6 37 55 8 2,121.7 14.2 8.5 2,144.5
1 2 hectares 89 1.5 35 51 14 1,106.9 16.5 15.4 1,138.8
2 3 hectares 40 2.4 45 42 13 775.5 16.7 16.6 808.8
3 5 hectares 47 3.7 42 44 13 526.0 18.2 17.2 561.4
More than 5 hectares 23 7.5 42 47 11 285.6 19.2 34.1 338.9
All farm average 308 2.1 38 52 10 1,273.0 16.2 14.8 1,304.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Months covered were December 1985 to June 1986.

26The crop labor use tabulated in Tables 19 and 20 includes labor from December-June only.
Comparisons of labor between crop and other activities use annual information.

Table 20- Total labor use by farm size and cultivation method

Share of Family Laborchange Hired Total
Family Exchange Hired Total
Farm Size N Area Male Female Child Labor Labor Labor Labor
(hectares) (percent) (hours per hectare)
Less than 1 hectare
Hoe 85 0.6 36 56 8 2,128.1 13.8 7.9 2,150.0
Oxen 24 0.7 41 51 8 2,098.7 15.9 10.5 2,125.1
1 -2 hectares
Hoe 50 1.4 33 49 19 1,138.8 13.7 8.4 1,160.9
Oxen 39 1.6 40 54 6 1,065.9 20.2 24.3 1,110.5
2 -3 hectares
Hoe 17 2.3 42 45 14 944.7 15.1 11.2 971.0
Oxen 23 2.4 50 39 11 650.4 17.9 20.6 688.9
3 -5 hectares
Hoe 5 3.6 34 52 13 644.8 26.6 42.6 714.0
Oxen 42 3.7 44 43 13 511.9 17.2 14.2 543.2
More than 5 hectares
Hoe 1 6.4 55 33 12 611.8 0.0 a 611.8
Oxen 22 7.6 41 48 11 270.7 20.2 35.6 326.5
All farm total
Hoe 158 1.2 35 54 11 1,631.7 14.2 9.5 1,654.9
Oxen 150 3.0 42 49 9 895.7 18.3 20.4 934.4

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Notes: Months covered were December 1985 to June 1986.
aNo mean value was assigned since there was only one case.

main change with larger farm size is a slight increase in agricultural work but a
substantial decline in most other categories of labor use, especially in off-farm
employment and business activities (Figure 2). The net result is a decrease in men's
total labor per capital with larger farm size, from about 80 hours per capital per month
in the lowest farm-size decile to about 50 hours per capital per month in the highest
farm-size decile. The results for women's labor with respect to increasing farm size
are different in that there is no decline in total labor intensity (hours per capital) in
contrast to that seen for men (Figure 3). In fact, their work seems to go up in both
agriculture and household maintenance activities.
These observations on labor allocation by men and women as farm work in-
creases with farm size are consistent with observations made by Cleave (1974, 180).
In his words,

several of these calls (nonfarm activities) on the family's time may have some economic or
social value. The data available suggest, however, that, for men, at least, extra calls on time for
agricultural work are drawn mainly from this collection of activities rather than from recorded
resting time. For women, much of whose nonfarming time is devoted to a regular routine of
domestic duties, extra work on the farm may mean less leisure.

Variations in children's work by income and farm size show that the major
component of children's work is crop labor, which, on average, fluctuates between
10 and 12 hours per child per month. The second largest category of work in which
children help out is in household maintenance activities. For all activities combined,

Figure 2-Average hours of labor, by activity, adult males, by deciles of total
area farmed


1 2 3 4 5 6
M Cropping [ Self-employed g Off-farm employment

0 Household maintenance

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The largest farms are in the tenth decile.

Figure 3-Average hours of labor, by activity, adult females, by deciles of total
area farmed


B Cropping [ Self-employed j Off-farm employment

8 9
R Household maintenance

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The largest farms are in the tenth decile.

children's work averages about 15-20 hours per child per month. There is no clear
pattern of any change in children's work with increasing income or farm size.

Seasonality in Crop Labor Use

Seasonal variations in labor for cropping activities largely dictate the labor
allocation patterns of households. The peak labor input in agriculture is in January,
with a secondary peak in June during harvesting of hybrid maize (Figure 4). Harvest
of other crops is more spread out. August and September have virtually no agricul-
tural work reported, and with the onset of rains in October or November, the cropping
cycle begins again. From January, when the observations began, until the end of that
cropping cycle, households that grew hybrid maize had a significantly higher house-
hold labor use in most of the cultivating months and again during harvesting. In the
next crop year, the hybrid-adopting households seem to get off to a slow start. This
is because they use oxen for land preparation, and early in the crop season is
primarily when the labor-saving effect of this technology would occur.
As was seen earlier, the differences between adopting and nonadopting house-
holds in labor use is due to a combination of factors; for example, larger farm sizes
contribute to a decrease in labor intensity per hectare, but a slightly increased
intensity of labor use per person in agriculture. Oxen use, on the other hand, which is
usually associated with hybrid maize adoption, brings about a reduction in labor use
per hectare, if farm size is held constant. In other words, differences in seasonal

Figure 4-Average hours of family labor spent in cropping activities, by hybrid
maize adoption
Average Hours/Month
500 -e- Adopter
450 Nonadopter

Jan. Feb. March April May June July August Sept. Oct. Nov. Dec.

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The squares and dots indicate standard deviations.

patterns may reflect differences in labor use strategies. Labor strategy during harvest
is an example of this. Nonadopting households have an early labor peak that extends
for several months-between April and July-before tapering off in the postharvest
period. Although this is partly due to an early harvest in the valley areas, it primarily
reflects the much earlier beginning of harvest of local maize, as well as an extended
period of harvest. The longer harvest period may be an effort to conserve labor
energy following a long period of food scarcity. For hybrid-adopting households,
there is a sharp peak in harvest labor in June (family labor, especially women) and in
July (nonfamily labor).
The higher crop labor required in hybrid-adopting households is especially
marked for males (Figure 5), and, to a smaller extent, for females (Figure 6).27 It
should be noted, however, that the absolute amount of crop labor provided by females
is higher than that provided by males in both the adopting and nonadopting house-
holds. These findings are consistent with other research, which shows that with
commercialization of agriculture, there is an increase in men's involvement in
farming, but at the same time the demand for women's labor on crops also increases.
Seasonal variations in use of nonhousehold labor in crop production are similar
to those for household labor, except that differences in season are especially pro-
nounced for the hybrid maize adopters (Figure 7). January also has a pronounced
labor peak for nonhousehold labor, similar to that for household labor. However, the

Figure 5-Average hours of male labor spent in cropping activities, by hybrid
maize adoption
Average Hours/Month
220 -- Adopter
200- -.- Nonadopter
Jan. Feb. March April May June July August Sept. Oct. Nov. Dec.

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The squares and dots indicate standard deviations.

27This is partly because farm size of hybrid-adopting households is larger.

Figure 6-Average hours of female labor spent in cropping activities, by hybrid
maize adoption

Average Hours/Month

Jan. Feb. March April May June July August Sept. Oct. Nov. Dec.

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The squares and dots indicate standard deviations.

Figure 7-Average hours of nonfamily labor spent in cropping activities, by
hybrid maize adoption

Average Hours/Month

Jan. Feb. March April May June July August Sept. Oct. Nov. Dec.

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The squares and dots indicate standard deviations.

--- Adopter
-- Nonadopter

10 k



harvesting peak for nonhousehold labor comes later, in July instead of in June,
suggesting that individuals supplying nonhousehold labor are more likely to finish
their own harvesting before undertaking work for other households. This is in
contrast to the overlapping peaks in January, which indicate that individuals provid-
ing labor at that time are facing competing labor demands for own production or
wage labor, and they are opting at that time for wage labor. This is consistent with
assertions made by numerous authors that food scarcity for some households during
the planting/weeding season could be driving these labor choices (Kumar 1988). That
nonadopters have an early rise in household crop labor could be due to the predomi-
nance of hoe cultivation in that group versus ox cultivation among the adopters. The
higher labor required in land preparation and planting with hoe cultivation could
explain the differences between the groups at the start of the season. During the rest
of the season, however, hybrid maize adopters have a higher level of family labor use
in crop production.

Labor in Household Maintenance Activities

Besides crop labor, household maintenance activities are the most time intensive,
especially for women. These activities include fetching water and fuel, cutting thatch
or bamboo, and other activities for house or storage repair. The seasonal pattern for
these activities differs for adopting and nonadopting households. At the household
level, these activities are at a seasonal high during the peak agricultural work period
of January in the adopting households, while they are at a seasonally low point in the
nonadopting households. After a short peak for both in April, the harvest months of
June and July are also a high period for both types of households. Nonadopting
households reduce this activity sharply in August, in contrast to the adopters, who
maintain the previous level through September, after which the level in both tends to
decrease (Figure 8).
In 11 out of the 12 months recorded, adopting households have an equal or higher
level of household labor input in maintenance activities than nonadopters. The
difference is statistically significant in January, February, and August. A similar
observation is made about the share of women's labor in total household maintenance
labor. Women from adopting households have an equal or higher share of labor in
maintenance in 10 of the 12 months recorded, on average. Combining the results for
both the absolute amount and the share of women's labor in household maintenance,
it appears that for the adopting households, when the total amount of household
maintenance labor use goes up, the women's share also goes up. This suggests that
while the share of crop labor by men increases with hybrid maize adoption, their
share of household maintenance activities goes down. There are three months when
this does not occur: April, August, and November. In both April and August, total
labor goes up, while women's share goes down. As these are the months for repairing
storage bins and housing, men's traditional high involvement in these two activities
is reflected here. In contrast, in November, while the total labor in these activities is
stable at the previous level, women's share increases sharply, suggesting that other
household members are not available to help to the extent they were before. This is
consistent with the increased labor demand at this time for cropping activities. This
pattern in adopting households suggests that not only is there a higher demand for

Figure 8-Average hours of family labor spent in household maintenance, by
hybrid maize adoption
Average Hours/Month

80- -- Adopter






Jan. Feb. March April May June July August Sept. Oct. Nov. Dec.

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: The squares and dots indicate standard deviations.

labor in household maintenance activities in adopting households, but it is primarily
the women who are providing it, even when demand for their labor in cropping
activities is also high.
Differences in the patterns of labor allocation between males and females with
adoption of improved agricultural technology are expected to be a function of factors
that influence the relative profitability or returns to labor28 and factors that influence
the returns to household and individual income.29 The net effect of each type of
change on time allocation will be a function of the compensated effect of the
opposing wage and income effects. These will be examined in more detail next.

Analysis of Household Labor Allocation

Farm and Nonfarm Activities
As discussed in Chapter 3, the adoption of improved agricultural technology will
influence the allocation of labor in agriculture through the combined effects of
changing the "virtual" wage, which includes the value of time, and increasing

28This is expected to lead to an increase in their returns to labor and to an increased work incentive and
labor supply for agriculture.
29The income effect will generally decrease the supply of labor for agricultural work. This is due to
increased demand for consumption time (that is, time required for household maintenance for which
demand increases with income) and for leisure activities.

household income. Higher returns from adoption will shift labor away from other
activities into agriculture, until the virtual wage in agriculture equates with the wage
in nonfarm activities, increasing labor use in agriculture. Increments to income, at the
same time, will contribute to an increased demand for time in consumption-related
(and leisure) activity and put downward pressure on household labor use (supply).
The optimum combination of inputs (goods and time) in the production of Z-goods
(Z1) is determined by the marginal rate of substitution between income and time,30
which is given by the change in the virtual wage relative to the price of goods. The
marginal rate of substitution is equal to the input price ratio:

8z X-i /ti = w/P, (16)

where w is the shadow price of time, and pi is the price of goods.
Since adoption of hybrid maize affects the productivity of labor and, hence, both
the shadow price of time and household income, it will influence the allocation of
labor in alternative income-generating activities (farm and nonfarm), as well as in
consumption-related activities.
These two aspects of household labor allocation are analyzed here. First, the net
effect of the combined change in virtual wage and income effects on labor supply in
farm and nonfarm activities that occurs with the adoption of hybrid maize production
is examined. Results in Chapter 5 showed that the income benefits of adoption may
not be uniformly distributed across household members, and this could mean that
labor supply of males and females is affected differently. The estimation equation is

L 1,L2i = f(HM*, Ha, HM*Ha, Ox, Educ, DI, Head,
Intrahh, HLabor, Dep, Ecol), (17)
Ll; = family farm labor input by males and females;
L2, = nonfarm labor input by males and females;
HM* = predicted hybrid maize adoption;
Ha = total farm size;
HM*Ha = interaction of HM* and Ha;
Ox = use of ox-plow;
Educ = education of head of household;
DI = infrastructure index;
Head = sex of head of household;
Intrahh = proportion of area managed jointly or independently
by female;
HLabor = household labor availability;
Dep = dependency ratio; and
Ecol = ecological zone, plateau or valley.

30Zi, as discussed in Chapter 3, denotes products such as food consumption and nutrition and health that
directly contribute to the household's utility. Z, is a combination of the commodity, market goods xi, and
time inputs, ti.

The results suggest that patterns of labor supply do differ for males and females
as the result of changes in household characteristics, including hybrid maize adoption
(Table 21). On the other hand, area-level changes such as ecology and infrastructure
development show a similar pattern of labor response by both sexes. Thus, while both
show a similar optimization pattern with changes in exogenous variables, intrahouse-
hold effects of household-level changes are different.
Hybrid maize adoption reduces women's labor input in both farm and nonfarm
activities, when farm size is held constant, with the effect on nonfarm work being
statistically significant. Since this effect is consistent for both farm and nonfarm
work, it is likely to be a function of higher income together with a relative reduction
in returns to women's labor with hybrid adoption. Farm size, which is closely related
to household income, increases men's farm work but is neutral (insignificant) for
women's. Thus, there is no evidence of a reduction in work with larger farm sizes, as
is generally expected. The significant increase in agricultural work by men suggests
the relative improvement in their marginal wage rate (shadow price of time) with
increasing farm size.
Other household-level characteristics of interest in examining labor effects are
sex of the head of household and the proportion of area managed by females. Male
headship is associated with an increase in men's labor in farm and nonfarm work and
a reduction in women's labor in both, compared with females who are heads of
households. This seems to be primarily due to changes in the gender composition of
these two types of households. An increase in the share of area managed by females
has a striking effect on the reduction in women's work time in all types of activities
being analyzed here. This is an interesting result; there are several reasons that will
need to be researched further. The results are, however, consistent with an income
effect (that is, women's income is higher when they have a larger area under their
management). This has welfare implications for women and their families. The
education variable has the expected positive sign for nonfarm work, indicating the
higher wage rates available in the nonfarm sector with additional schooling.
The family composition variable with more adult members in a family has the
expected positive sign for both men's and women's farm labor input. Nonfarm work
for women also increases with the number of adult members, suggesting that larger
families are associated with increased nonfarm work by women. The dependency
ratio has an unexpectedly negative sign for men's farm labor. This could be an
indication of additional (unrecorded) activities required of men when the proportion
of children and elderly members increases. Alternatively, it could also result from the
substitution of children's labor for a portion of male labor. This could not be
ascertained in the present analysis.
Poor infrastructure access is associated with a higher level of labor input in
agriculture by both men and women. This is consistent with an increase in the price
of consumer goods relative to returns to labor in low infrastructure areas.31 This is
also equivalent to the converse of an income effect on labor supply. Women are more
likely to be engaged in nonfarm activities in areas where infrastructure is poor,
compared with men, which is consistent with the effect of negative income and a

31The policy of panterritorial pricing of agricultural products also contributes to this effect.

Table 21-Analysis of household labor allocation

Total Male Labor Total Female Labor
Farm Nonfarm Household Maintenance Farm Nonfarm Household Maintenance
Independent (Mean = 1,540.6) (Mean = 336.4) (Mean = 48.6) (Mean = 1,749) (Mean = 823.1) (Mean = 670)
Variable Mean Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio Coefficient t-Ratio
(hours/year) (hours/year)
Head 0.8 470.89 2.18* 130.87 1.59 15.78 1.71 -680.17 -3.60* -337.34 -4.75* -263.42 -4.34*
Dep 1.1 -218.02 -2.12 -19.84 -0.50 -4.65 -0.82 -24.09 -0.27 20.28 0.60 48.13 1.28
Educ 3.8 -37.35 -1.30 23.66 2.16* -0.70 -0.56 -20.70 -0.82 9.53 1.01 -5.19 -0.63
Hlabor 3.9 288.00 4.82* 41.82 1.83 6.15 1.32 316.03 6.04* 125.09 6.36* 178.38 5.81*
Ox 0.5 292.77 1.29 -59.45 -0.69 -4.35 -0.46 -7.79 -0.04 123.18 1.65 89.39 1.44
Ecol 0.2 -806.37 -3.36* 50.66 0.55 57.88 4.18* -431.64 -2.05* 40.90 0.52 -134.37 -1.47
Ha 2.1 324.25 2.69* -1.82 -0.04 ... ... 35.56 0.34 12.26 0.31
DI 1.4 430.56 4.69* -44.75 -1.27 3.65 0.94 287.72 3.57* 185.66 6.14* 163.49 6.39*
Intrahh 0.6 -237.87 -1.14 16.59 0.21 -15.60 -1.75 -801.57 -4.38* -209.54 -3.05* -203.90 -3.48*
HM* 0.2 81.40 0.19 -19.04 -0.11 4.89 0.38 -389.88 -1.01 -304.11 -2.10* -146.00 -1.73
HM*Ha 1.1 -248.73 -1.84 -12.54 -0.54 ... ... 6.22 0.05 59.67 1.34
HMA* 0.5 ... ... .. ... -8.64 -1.55 ... ... .. ... 10.94 0.30
LogY* 4.0 ... ... ... ... -9.04 -0.34 ... ... ... ... 342.90 1.99*
Constant -432.14 -1.15 103.07 0.72 52.62 0.44 1,290.20 3.93* 300.45 2.44* -1,318.20 -1.68
R2 (adjusted) 0.33 0.03 0.23 0.28 0.37 0.39
F 10.4 1.6 6.7 8.5 12.5 13.1
Sample (N) 213 213 213 213 213 213
Notes: The variables are defined as follows:
Head = Sex of the household head: 1 = male, 0 = female;
Dep = Dependency ratio: + 60 / 14-60;
Educ = Education of the household head: grades completed;
Hlabor = Adult equivalents per household;
Ox = Cultivation method dummy: 1 = oxen use;
Ecol = Valley or plateau: 1 = valley, 0 = plateau;
Ha = Total area farmed: hectares per household;
DI = Area level infrastructure index based on distance to 12 elements; the higher the index, the poorer the infrastructure;
Intrahh = Proportion of area farmed by female, independently and jointly;
HM* = Predicted hybrid maize adoption;
HM*Ha = Predicted hybrid maize adoption (HM*) x total area farmed (Ha);
HMA* = Predicted area under hybrid maize: hectares per household; and
LogY* = Predicted value of household disposable income (log of per capital consumption expenditure).
*Significant at .05 level.

clustering of women in the low nonfarm wage sector. Valley sites have a significantly
lower farm labor input by both males and females, holding farm size constant. The
higher relative productivity of valley soils is likely to a factor in this result, combined
with little opportunity for area expansion.

Household Maintenance Labor
Labor input by men and women in activities such as collection of fuel and water,
food processing, home repair, and maintenance are included in this group of activi-
ties. These activities contribute to an improvement in household welfare as well as to
the production of Z-goods discussed above. While the demand for Z, increases with
income, the combination of goods (xi) and time (ti) are a function of the relative
increase in price of time relative to price of goods, as seen in the previous section.
The response of household labor allocation to these activities with the adoption of
hybrid maize is therefore a reflection of both changes in income and shadow price of
time of household members.
The estimation equation for the labor supply to household maintenance activities is

T, = f(Y* HM* HMA *, Ha, Ox, Educ, DI, Head, Intrahh, Hlabor, Dep, Ecol), (18)

T, = labor input in household maintenance activities by
men and women,
log Y* = predicted value of household disposable income,
HM* = predicted value of hybrid maize adoption, and
HMA = predicted value of area under hybrid maize (other
variables previously defined).

The results are presented in Table 21.
As discussed earlier, the effect of adoption of improved technologies on labor in
household maintenance activities is the result of both the income and shadow price
of time of household members. Income increments increase the demand for women's
household maintenance labor; this effect is positive and statistically significant. This
is a reflection of the labor-intensive nature of consumption-related technologies, such
as pounding of maize and collection of fuel and water, which have improved little,
and, finally, of the relatively lower contribution of women's productive labor to
income increments.32 As women's share of farm area increases-an indication that
their share of income is also increasing-the effects on women's maintenance work
are significantly negative. This is consistent with higher women's income and substi-
tution of goods for time. These combined results indicate that, while the demand for
women to spend time in household maintenance labor increases with household
income gains, it can be offset if women's share of income is also maintained.
Areas with poor access to infrastructure have an effect on household maintenance
time similar to a relative increase in price of goods relative to time (also shown in the

32This is not inherently so, but the result of women's limited access to improved production resources,
technologies, training, and information.

amount of labor devoted to farm and nonfarm work discussed earlier) and a signifi-
cant increase in time spent on all activities. This is especially so for women. As for
men's time spent in household maintenance activities, the only explanatory variable
of interest is the ecology variable, which indicates a significantly higher input by men
in valley areas. The almost exclusively matrilineal ethnic background of the valley
areas33 may help explain this result. The limited adoption of hybrid maize in valley
areas, which appears to rapidly raise the returns to men's agricultural labor relative
to women's, is also a related factor.
The results of the examination of labor allocation to farm, nonfarm, and household
maintenance activities are influenced by a mix of area, household, and intrahousehold
dynamics. When farm size or income is held constant, hybrid maize adoption leads to
lower labor input by women across all activities, but higher income is associated with
a higher use of women's labor in household maintenance. Men's farm labor rises with
farm size but is lower on farms of similar size with hybrid maize adoption.

33In plateau areas, there is a mix of matrilineal and patrilineal ethnic groups.


Issues in Intrahousehold Analysis

There have been at least three interrelated rationales for the increased attention
that is being paid to issues of intrahousehold access to resources and women's
decisionmaking roles within their households. This set of concerns includes physical
as well as human capital resources. These intrahousehold issues are relevant for
policy attention on the grounds of equity, efficiency, and welfare. They have emerged
roughly parallel with the larger policy emphasis on promoting the role of women in
development and have been facilitated by it.
The initial concern about equity grew in direct relation to the growing body of
information indicating that development accentuates patterns of discrimination
against women where such patterns already exist, and that, in general, women are less
likely to receive their share of the economic benefits generated by development
(Ahmad 1980; Savane 1986; Pala-Okeyo 1988; Baser 1988; Beneria and Sen 1986).
Gender-blind development is clearly not achieving gender-neutral effects.
Regarding efficiency, as evidence on the role of women in household production
grew, so did the realization that if the benefits of development programs, such as
improved agricultural technologies, methods, and practices, were developed and
promoted without the inclusion of women from farm households as direct benefici-
aries, then there would be costs in efficiency and in productivity gains (Dey 1983;
Jones 1986; Burfisher and Horenstein 1985). Independent analyses have shown that
the returns to providing women access to agricultural extension and training are
extremely favorable, often even more so than providing the same services to men.
This is especially so where traditionally women have been more involved in agricul-
tural production than men, as in many parts of Sub-Saharan Africa (Boserup 1970;
Moock 1976).
The third set of concerns addresses the welfare costs to households, children, and
women when insufficient attention is paid to issues of intrahousehold resource
access. These welfare losses are the most complex to document and analyze because
they can derive from several sources. First, at the household level, welfare costs are
a direct outcome of the efficiency losses already discussed. Welfare losses for women
and children also arise from the relative changes in the value of men's and women's
time arising from their differential access to development programs, resources, and
technologies. It is at this point that the current debate on whether neoclassical or
bargaining models better reflect household behavior becomes moot. When there is a
change in the relative economic productivity of men and women in favor of men,
both the neoclassical and bargaining models may predict similar outcomes in relative
income shares, patterns of time allocation, and even intrahousehold food distribu-

tion.34 However, the welfare and policy implications may be different. If the changes
are a result of optimization by the household with a single utility function, as is
assumed in the neoclassical model, then the case for intrahousehold policy interven-
tion is unnecessary. However, when the same changes occur as a result of the reduced
bargaining power of some individuals in the household, who in a sense "lose out,"
then policies also need to be designed for individuals within households.
Differences in intrahousehold utility, proposed in the bargaining models, have
another possible welfare outcome that can arise if some individuals, in this case women,
have a higher propensity than others to allocate income for basic household needs such
as food and child care. Several studies have shown that there is a higher marginal
propensity for improvements in food consumption and child nutrition to occur from
increments to women's income, compared with other sources of income (Tripp 1982;
Guyer 1980; Kumar 1978; Garcia 1991; Garcia and Pinstrup-Andersen 1987). How-
ever, that children are better off when women's value of time (or share of income)
increases is also predicted in the neoclassical model (Becker and Lewis 1973).
In this chapter, some facets of intrahousehold decisionmaking are explored to
examine the extent to which women's economic decisionmaking in agriculture and
share of income are affected by household adoption of hybrid maize. Also, differ-
ences between men's and women's expenditure patterns are analyzed.

Dynamics of Women's Role in Agriculture

The precise role of women in agriculture varies widely not only geographically,
but even within particular cross-sections of the population (Kumar 1987a). The
context for needs, constraints, and opportunities is, therefore, relatively specific for
local situations. To the extent to which the dynamics and costs of the marginalization
of women can be documented in different situations, policymakers and institutions
would be helped in assessing the importance of extending resources and benefits
specifically to women.
Women's role in agriculture, at least in terms of their labor input, has usually
increased dramatically with higher outmigration of labor from rural areas, as they
take on greater responsibility for farming in de facto or de jure female-headed
households (Kennedy and Rogers 1992). There is conflicting evidence on the forces
that contribute to urban migration, including structural adjustment. Evidence from
Zambia suggests that during the 1980s, when structural adjustment was initiated, the
rate of urban population growth slowed significantly, compared with the 1970s: it
declined from 5.8 percent per year to 3.7 percent per year in the 1980s. At the same
time, the growth rate of the rural population increased from 1.6 percent to 2.8 percent
per year, suggesting a dramatic slowdown in urban growth (Chiwele 1992).
Liberalization of agricultural prices and markets has been proceeding steadily in
Zambia since the mid-1980s. This has provided increased incentives for agriculture,

34Differential rates of return to market activity for men and women can, for example, be traced to patterns
of intrahousehold food allocation and nutrition and health outcomes that favor males over females in the
neoclassical model (Rosenzweig and Schultz 1982). The bargaining model predicts similar
intrahousehold allocation of welfare (Senauer, Garcia, and Jacinto 1988).

and combined with the reduction of urban food subsidies, it has improved the
prospects for the rural economy in general. These changes are favorable for the rapid
expansion of improved agricultural seed and fertilizer technologies (Byerlee 1992).
Women's access to productive resources within this context is both more important
because of the need to use the increased flow of resources to agriculture efficiently
and more problematic because of the increased competition for resources due to a
slowing of male labor migration to urban areas.
To summarize some relevant information presented in earlier chapters:
Women's management of hybrid maize plots-whether independent or joint-
is lower than that for other food crops (Chapter 5). This appears to be a function
of the cash and credit requirements associated with input purchases for hybrid
production as well as women's lower access to training and extension services
(Saito 1992).
Women's share of agricultural labor, 53 percent in households that do not
cultivate hybrid maize, is reduced to 47 percent with its adoption. Even though
the absolute number of person-days of agricultural work increases for women
with adoption of hybrid maize, the relative share of men's labor rises faster.
Women's labor input per hectare of hybrid maize is lower than that for other
food crops. The timing of hybrid maize activities and the degree of weeding
required by hybrid maize are found to be conducive to relatively low yields for
this crop relative to its potential.
The labor for household maintenance activities increases with hybrid maize
adoption, and this is consistent with the income effect associated with its use.
These findings are consistent with a low involvement of women in production of
hybrid maize, and, consequently, in limited access to the income improvements
derived from it. This aspect will be further explored on the basis of information on
intrahousehold decisionmaking that illustrates the extent of women's economic

Changes in Intrahousehold Decisionmaking

This section examines characteristics of women's economic decisionmaking in
Eastern Province in order to analyze how decisionmaking changes with technological
change in agriculture and hybrid maize adoption. Women's contributions to house-
hold food and nonfood expenditures are analyzed as well as the following facets of
women's decisionmaking role: (1) decisionmaking regarding use of improved inputs
and nonhousehold labor, (2) women's access to and allocation of crop proceeds, and
(3) women's contribution to household food and nonfood expenditures.
The methodology used in this part of the investigation was drawn from anthropo-
logical work on decisionmaking, in particular that used in the Status of Women in
Nepal series of case studies synthesized by Acharya and Bennett (1981). To elicit a
precise response and clarity in analysis, this methodology separates the different
facets of the decisionmaking process into several actions taken by the household. It
overcomes many of the subjective biases in responses to queries on decisionmaking
that are characteristic of such investigations. Accordingly, questions about three
facets of decisionmaking regarding use of improved inputs and hiring of nonhouse-
hold labor were asked: Who initiated the use of the inputs? who arranged for it? and

who paid for it? Regarding crop sales and access to income from sales, the queries
for each crop sold were, who suggested it? who negotiated it? and how were proceeds
distributed? For the allocation of income from crop sales, the query for each individ-
ual receiving income from crop sales was, what amounts were allocated to different
categories of expenses? For household expenditures for each food and nonfood
purchased during the previous month, the queries were, who initiated the purchase?
who paid for it? and who actually went to purchase it? These questions were asked in
September for the crop production and sale-related decisions (harvest was generally
completed in July) and in December for the household expenditure contributions.
Even though these questions were asked only once, they were designed to reflect
agricultural decisionmaking and income for the previous crop cycle, and consump-
tion expenditures at the start of the lean season.

Agricultural Decisionmaking and Income from Sales
Use ofInputs. There were 71 input decisions reported in households using hybrid
maize (HM) and 68 such decisions in the nonadopting households using local maize
(LM). The HM and LM terminology is used for the sake of convenience here, but in
actuality both groups of households grew local maize. A significantly greater share
of women participated in decisionmaking in the LM households (Table 22). Only 5
percent of input uses in HM households were suggested by women, compared with
27 percent in LM households. That women were much less involved, not even
"suggesting" the use of an input, implies a perception in HM households that
questions regarding new technology are men's decisions or ones that men know or
can manage better.
Use of Hired Labor. Compared with the use of improved inputs in Table 22, a
much larger share of decisions to hire labor were initiated and paid for by women in
the HM households. This greater involvement in hiring could be largely a function of

Table 22-Percent of females making agricultural decisions

Number of Agricultural Decisions Made by Females
Decision Type Decisions Reported Suggested Arranged Paid
Purchase of inputs
Hybrid maize 71 5 8 10
Local maize 68 27 21 24
Hiring labor for
Clearing and planting
Hybrid maize 28 15 15 19
Local maize 34 24 29 35
Weeding and fertilizing
Hybrid maize 50 10 10 14
Local maize 490 38 43 39
Harvesting and marketing
Hybrid maize 50 16 12 26
Local maize 65 38 38 37

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

the labor crunch faced by women, as seen earlier. Also, women may be more
knowledgeable in this area than in input purchase decisions for which they have not
received any training or advice. However, women from HM households did not
participate in labor-use decisions as much as women from LM households. The share
of labor hire decisions by women roughly parallels the share of independently farmed
land area reported in Chapter 5. In HM households, the share of land independently
farmed by women was about 12 percent, compared with 31 percent in nonadopting
households. The only category of labor decisions that stands out from the pattern is
the high 26 percent of payments made to hired labor for harvest by women from HM
households. This is consistent with the significant labor bottleneck during harvesting
in adopting households discussed in Chapter 6, which appears to be more serious for
women because they traditionally have a greater responsibility for harvesting than do
men in this region.
Income from Crop Sales. As with decisionmaking on agricultural transactions,
decisions on income from crop sales are expected to be a function of the degree of
crop ownership. As mentioned earlier, even though a plot may be under the inde-
pendent ownership of a man or a woman, labor input is still provided for it by other
household members. While the majority of cropped land under hybrid maize was
reported to be independently managed by men, women provided half or more of the
labor used on this crop. Similarly, men provided labor on crops independently owned
by women. The pattern of sharing income from crop sales is, therefore, analyzed in
relation to the pattern of crop ownership. There is little ambiguity in the responses to
this question, and no claims of a common purse or pooled income were obtained.
In looking at the share of income from crop sales received by males versus
females from plots under different types of ownership, it can be seen that men receive
88 percent of the proceeds from plots under the male head's ownership and women
receive 83 percent of the proceeds under the female head's ownership (Table 23). In
plots owned by a male spouse (in a female-headed household), men received 91
percent of the proceeds, while women received 87 percent of the proceeds from plots

Table 23-Average amount received from crop sales by gender, by type of crop
ownership, all crops combined

Males Females
Amount Share of Amount Share of
Principal Owner Received Proceeds Received Proceeds
kwachaa) (percent) kwachaa) (percent)
Crops owned individually
Male-headed household
Male owned 614 88 87 12
Female owned 30 13 206 87
Female-headed household
Female owned 164 17 787 83
Male owned 497 91 48 9
Crops owned jointly
Head of household and spouse 232 70 101 30

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

owned by a female spouse (in a male-headed household). These figures confirm that
the concept of independent intrahousehold crop ownership/management is a valid
one for this area, since this is associated with access to a large share of the crop
income. In plots that are jointly owned, only 30 percent of the proceeds went to
women and 70 percent to men. Since the bulk of plots under women's ownership are
in the joint ownership category, this division of crop income probably represents a
large part of their income from crop sales.
Data on sharing income from sale of crops suggest that the premium is on
management decisions and not on provision of labor and that management decisions
parallel the extent of crop ownership. Therefore, a decline in the share of the crop
owned by women will directly translate to a reduced share of income. Assuming that
the overall pattern of crop ownership could be translated into income shares along the
lines of the cash income shares, then women's income share is calculated to decline
from 40 percent in nonadopting households to 23 percent in adopting households.
The absolute amounts of women's income are, however, likely to be higher with
adoption, since it is also associated with larger farm sizes.

Intrahousehold Expenditure Patterns
Allocation of Income from Crop Sales. As mentioned earlier, income allocation
decisions from crop sales were documented in September, two-to-three months after
harvest. Since cash sales are lump sum receipts, their allocation is likely to be
different from allocation of periodic expenses. Only expenditures made up to the date
of the interview were thus captured, and amounts saved for later use were not
reflected in the numbers reported.
In terms of the frequency with which men and women spend on different items,
women spend most often on food items for the household, followed by personal
items, primarily clothing (Table 24). Men, on the other hand, are likely to spend most
often on repayment of agricultural loans and second on social expenses. The actual
amount of expenditure incurred by men each time is almost always higher than that

Table 24-Frequency of spending from proceeds of crop sales on various items,
by sex

Expenditure Item Males Females Total
kwachaa) (N) (percent) kwachaa) (N) (percent) (N) (percent)
Household dietary item 27.05 56 46.7 21.03 64 53.3 120 100.0
Food outside of home 30.04 52 57.8 8.33 38 42.2 90 100.0
Social expenditure
(including alcohol) 13.49 71 78.0 3.95 20 22.0 91 100.0
Gift 50.49 34 52.3 90.25 31 47.7 65 100.0
Personal item 79.55 116 47.2 70.63 130 52.8 246 100.0
Repayment of loan 404.07 26 83.9 71.80 5 16.1 31 100.0
Other 158.05 35 58.3 54.02 25 41.7 60 100.0
All items ... 157 49.4 ... 161 50.6 318 100.0

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: N is the number of expenditures.

spent by women, with the greatest difference in the case of loan repayments, foods
consumed outside the home, and social expenses. The only item on which women
spend more than men is gifts. It is possible that gift giving is a way for women to pay
for or derive services from their relatives and friends, and that they depend on this
network more than men do.
When average household expenditures from proceeds of crop sales are examined
(in the months immediately following the harvest period), the biggest item for men
is repayment of loans followed by personal items. For women, personal items is the
most important expenditure group. The average amount of expenditures made by
women on all items combined is about half that of men (Table 25). However, if loan
repayments are removed, then the differences are reduced substantially.
Household Consumption Expenditures. The pattern of intrahousehold expenditures
from the proceeds of crop sales in the postharvest period reflects expenditures at that
point in time. Intrahousehold consumption expenditures were obtained in December, at
the start of the lean season. For each expenditure item, information was collected on
who initiated the expenditure, who paid for it, and who actually went to purchase it.
Food items accounted for 21 percent and nonfood items for 79 percent of the total
cash expenditure budget for the month. Men made up 80 percent and women 20
percent of the monthly expenditures. In the men's budget, food items accounted for
19 percent and nonfood items for 81 percent. Women, on the other hand, spent 28
percent for food and 72 percent for nonfood (out of which maize milling was a major
item). Since women had a higher propensity to spend on food, their share of food
expenditures was nearly 30 percent. These figures show that even though the major
share of both food and nonfood cash expenditures were made by men, virtually all of
women's expenses were either for food or food processing.
In order to examine changes in intrahousehold decisionmaking with hybrid maize
adoption, the three facets of decisionmaking for food expenditures were examined
(Table 26). Women were most likely to initiate purchases of green leafy vegetables,
meals consumed outside the household, fats, oils, beans, and groundnuts. All of these
are items that women have the primary responsibility to provide. Women spend large
amounts of time collecting and drying leafy vegetables from the wild and from fields;
expenditures on this item largely reflect their concern for its provision in the diet.

Table 25-Mean expenditure from the proceeds of crop sales, by sex

Expenditure Item Males Females
Household dietary item 9.65 8.31
Foods outside of home 9.95 F.95
Social expenditure
(including alcohol) 6.10 0.49
Gift 10.93 17.27
Personal item 58.78 56.68
Repayment of loan 66.92 2.22
Other 35.23 8.34
Total expenditure 197.56 95.25

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

Table 26-Percent of food expenditure decisions made by females in households
growing hybrid versus local maize

Percent of Food Expenditure Decisions by Females
Food Initiated Paid Purchased

Hybrid maize households 40 27 41
Local maize households 50 33 58
Beans and nuts
Hybrid maize households 60 54 70
Local maize households 62 34 61
Leafy vegetables
Hybrid maize households 91 61 88
Local maize households 86 62 93
Meat and fish
Hybrid maize households 42 32 52
Local maize households 55 35 59
Milk and products
Hybrid maize households 42 42 42
Local maize households 43 57 60
Fats and oils
Hybrid maize households 67 46 54
Local maize households 84 41 64
Hybrid maize households 36 31 38
Local maize households 59 40 52
Meals outside of home
Hybrid maize households 75 52 65
Local maize households 80 50 64

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

Similarly, groundnuts and beans are the traditional relish items provided by women,
and fats and oils are also essential in the diet. Since provision of meals is the woman's
responsibility, it is not surprising that she often initiates an expenditure on a meal
consumed outside the home. In most cases (except milk and milk products), women
may initiate the expenditure, but men are more likely to actually pay for it.
For these foods-those for which women are most likely to initiate an expendi-
ture-their ability to actually incur an expenditure is either the same or better in
hybrid maize households. In contrast, in the case of food items for which women are
less likely to initiate an expenditure, such as cereals, meat and fish, sugar, and milk
products, women in hybrid maize households are less likely to incur an expenditure.
It is possible that women expect men to take care of expenditures for items for which
women generally do not feel responsible. This reduction in women's willingness to
incur expenses is plausible, given that their income share is reduced in hybrid maize
households, even though their absolute level of income may increase with adoption
of hybrid maize.

The study of intrahousehold decisionmaking shows that women are less likely to
take on economic transactions for agricultural production in hybrid maize-adopting

households. In all households, they are more easily involved in labor hire decisions
than in input use decisions, which is probably a direct reflection of their lack of access
to extension services. Access to income from crop sales within the household is clearly
indicated as a function of the crop ownership pattern, with the primary owners
receiving the bulk of proceeds. Where men and women jointly own the crop, women
receive only 30 percent of income from sales. This figure is more likely to reflect the
extent of women's management decisions than their labor input. To what extent a
bargaining process was involved in this division of proceeds could not be identified,
but it is likely. Intrahousehold expenditure patterns confirm that women's expenses
are virtually all for either purchases of food or for maize milling, indicating a high
propensity for food-related expenditures. The pattern shows that in most instances,
women in hybrid maize-adopting households are less likely to initiate food purchases
than those in nonadoptive households, which could mean that their bargaining power
is lower. In terms of actually paying for food items, women's share of expenses in
hybrid maize-adopting households is more likely to decrease or stay the same.



Issues and Consequences of Technological
Change in Agriculture

This chapter examines the patterns and distribution of food consumption and
nutrient intakes in Eastern Province and the effects of technological change in
agriculture on them. Growth in agricultural production and technological change in
agriculture influence dietary intakes in many ways. At the local level, the main
economic effects are through the improvement in returns to land and labor, including
the direct price and income effects and the indirect effects on agricultural incomes of
farm-nonfarm employment linkages (Pinstrup-Andersen 1979; Mellor and Lele 1973).
Several studies have shown that the level of food consumption improves with
technological change and commercialization of agriculture for households that partici-
pate in these changes (Kennedy 1989; von Braun, Puetz, and Webb 1989; von Braun,
Hotchkiss, and Immink 1989; Bouis and Haddad 1990). The results of the subset of
studies carried out in Sub-Saharan Africa are similar in this respect to those found in
other parts of the world. But the literature also recognizes the potential of new technol-
ogy to exacerbate income inequalities between households (Lipton and Longhurst
1989). Analysis of the effects on those not participating in the adoption of changing
agricultural practices is more difficult and the results are less clear-cut (Binswanger and
von Braun 1991). Income inequality has increased between geographical areas, be-
tween early and late adopters of new technology, and within households. It is important
to note that when these effects have occurred, they are generally not due to the nature
of the technology itself but to the underlying inequity in distribution of resources prior
to adoption (Pinstrup-Andersen 1979). Agricultural research and policies geared to
promote the spread of new technologies can, however, have a favorable impact on the
distribution of benefits, provided that awareness of the costs to society and economic
growth are incorporated into the decisionmaking process.
Dietary benefits can occur through income and employment growth and through
changes in intrahousehold income. Not only are different households affected in
different ways, but members of a household are also affected differently by marginal
improvements in returns to land and labor with technological change. The extent of
dietary improvement is generally greater when the inadequacies are greater, that is,
when the beneficiaries are among the lower income groups.
Marginal increments to caloric intake at very low income levels are expected to
occur rapidly and primarily through an increase in cereal consumption. Since cereals
are generally the least expensive calorie source, they predominate at the lowest

income levels. As incomes increase, foods with high income elasticity that played a
very small part in the diet before contribute an increasing share of calories. As this
process continues, household diets tend to move toward what is considered the
cultural norm or "ideal" diet.35 In general, this is desirable because the ideal diets in
most societies are nutritionally adequate for meeting average requirements, but this
does not necessarily mean that all individuals within a household will be adequately
nourished if nutrients are not equitably distributed. As incomes increase, not only is
the share of calories from high income-elasticity foods higher, but the number of food
items in the diet also increases. This contributes to greater diet diversity, which has
long been considered desirable from the standpoint of ensuring adequate access to the
full range of nutrients required by the body.36
More recently, attention has shifted to how intrahousehold utility functions and
income control affect allocation of household resources and outcomes, such as food
consumption levels. Both economic theory and empirical observations have generally
shown that increasing women's share of income (the value of women's time) results
in a higher marginal utility for household food consumption and other investment in
the quality of human capital than income from other sources (Senauer, Garcia, and
Jacinto 1988; Duncan 1992). What has been unclear is whether these allocations arise
from a single household utility function or are the result of intrahousehold bargaining.
Anthropological research in several rural societies, including some in Sub-Saharan
Africa, has questioned the concept of pooled household income, finding that women
have a separate purse and different expenditure patterns from men. The present study
verifies that this pattern also exists in Eastern Province, Zambia.
Analysis of the dietary effects of technological change is made more problematic
when the main source of income and consumption is own-farm production. This
problem of simultaneity is even greater where labor is scarce, as in Eastern Province.
In these situations, seasonal food availability can be a factor in labor allocation
decisions and in agricultural productivity (Kumar 1988). To the extent that food
consumption becomes a factor in income generation, the measured income will not
be exogenous, and the association is therefore more complex. This factor will be
taken into consideration in the multivariate analysis presented later in this chapter.

Overall Diet and Nutrient Intake Levels
in Eastern Province

Estimates of food consumption were obtained through a modified food expendi-
ture record, to which adjustment was made for amounts actually consumed during the
previous week. Since estimates were based on food recall, some approximations were

35In a closed traditional society, it is relatively easy to determine the "ideal" diet. However, as markets,
new products, and advertising (or some of the more effective forms of nutrition education) come into the
picture, they produce new versions of the ideal. These factors, while most predominant in urban
situations, may also appear in some rural situations.
36The early use of Guttman scales in the 1960s was in acknowledgment of this need for diet diversity
(Sanjur and Romero 1975). Studies using the Guttman scale found this indicator to be highly associated
with child nutrition.

involved, but these were minimized by interviewing only the primary person in charge
of food preparation and obtaining quantities in local units, which were subsequently
standardized. The main advantage of obtaining a seven-day record is that a better
dietary profile is obtained, including foods consumed infrequently. Studies have
shown that food consumption estimates from 24-hour recall give lower figures for
intakes compared with other methods (Black et al. 1991). This was particularly
pronounced for higher-income groups in a study comparing 24-hour recall and food
expenditure methods (Bouis and Haddad 1990). Another recent study found the extent
of underreporting in 24-hour records to be about 18 percent (Mertz et al. 1991),
whereas recording recent expenditures alone tends to overestimate consumption for
households that make bulk purchases (likely to be the wealthier ones).
Maize is the main staple food in the average diet in Eastern Province. It contrib-
utes about 483 grams per capital per day out of a total cereal content of 518 grams
(Table 27). The maize consumed is predominantly out of local or household con-
sumption, with only 21 grams of it being purchased in the form of breakfast or roller
meal.37 Sorghum makes a slightly larger contribution to the diet than finger millet,
rice, or wheat. Overall, cereals contribute about 80 percent of calories in the average
diet. A comparison of the results of this survey with estimates from the 1970s shows
an increase in nearly all cereals in the diet, with the exception of finger millet, which
seems to have declined substantially (FAO 1976).
The average consumption picture culled from the survey shows a diverse and
fairly well balanced diet, with vegetable protein-rich foods providing about 45 grams
(on a dry weight basis) and animal foods providing about 60 grams (on a fresh weight
basis) of the daily diet. Vegetables and fruits constitute about 245 grams (on a fresh
weight basis) per day, and sweet potatoes, about 39 grams. Improvement of the diet
may require less dependence on cereals and a higher content of protein-rich foods and
fats and oils than currently available, especially in diets of children. Since both of
these food groups are expected to have a high income elasticity, income improve-
ments should improve the average diets consumed. If the comparison with consump-
tion figures from the 1970s is indicative of changes taking place, there does not
appear to be a clear-cut improvement. From the 1970s to 1986, the consumption of
cereals, legumes, and milk increased, while consumption of meat and fish and fats
and oils declined.
The average nutrient intake picture also looks relatively good, with the average
caloric and protein intake sufficient to meet the needs of the population. The per
capital caloric intake is 2,319 calories per day, with about 66 grams of protein
consumed (Table 28). Intakes of calcium, iron, and the B vitamins (thiamine, ribofla-
vin, and niacin) are adequate to marginally low. Although vitamin A content was not
calculated due to large gaps in the food composition tables for many vitamin A-rich
foods in the Zambian diet, the large amount of sweet potatoes, pumpkin, and leafy
vegetables in the diet indicates a good supply of this vitamin.
About 73 percent of calories are derived from carbohydrates. Proteins contribute
11.3 percent of calories, which is at the low end of the satisfactory range. Fats

37Roller and breakfast meals are two forms in which maize meal is marketed in Zambia. Roller meal has
an extraction rate of about 90 percent, while breakfast meal has a 65 percent extraction rate.

Table 27- Annual average of daily per capital consumption of calories, by food
groups, Eastern Province

Food Item

Breakfast and roller meala
Sorghum flour
Finger millet flour
Wheat flour
Fresh beans and peas
Dried beans and peas
Fresh groundnuts and groundpeas
Dried groundnuts and groundpeas
Sweet potatoes
Fresh cassava
Dried cassava
Roots, miscellaneous
Fresh vegetables and fruits
Pumpkins and gourds
Fresh leafy vegetables, cultivated
Fresh leafy vegetables, collected
Other fresh vegetables
Fruits, wild
Fruits, cultivated
Dried vegetables
Dried leafy vegetables, cultivated
Dried leafy vegetables, collected
Other dry vegetables
Fresh meat
Fresh meat, wild
Fresh meat, reared
Poultry, wild
Poultry, reared
Fresh fish
Dried meat, fish
Dried meat, wild
Dried meat, reared
Dried fish
Organ meats
Dried caterpillar
Fresh milk
Other milk products
Fats and oils
Other foods
Alcoholic beverages

Per Capita Calorie Consumption
Annual Standard
Average Deviation

20.48 72.01
462.44 266.70
16.41 49.30
5.01 22.23
7.61 25.58
6.17 18.32

















Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
aThese are two forms of purchased maize meal.

Table 28- Annual average of daily per capital consumption of nutrients,
Eastern Province

Average Daily
per Capita Standard
Nutrient Consumption Deviation

Energy (calories) 2,319.36 202.39
Protein (grams) 65.51 5.61
Fat (grams) 35.12 3.76
Carbohydrates (grams) 423.01 36.02
Calcium (milligrams) 445.93 37.87
Iron (milligrams) 22.39 1.72
Thiamine (milligrams) 1.05 0.10
Riboflavin (milligrams) 1.03 0.09
Niacin (milligrams) 14.01 1.45

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.

contribute only 13.6 percent, which is low compared with the common dietary
recommendation of about 20 percent of calories from fat for a normal population.
While the average consumption picture for the year gives some useful pointers
on the diet pattern in the area, it is of limited usefulness because of variations
between households, individuals in households, and seasonal consumption patterns.
Some indication of this variation can be seen in the high variances of the annual
consumption averages in Tables 27 and 28. Although individual consumption data
were not obtained for this study, some of the other dimensions of variations in food
intake in the area are examined in the next section.

Variations in Dietary Intakes by Ecology
and Area-level Technological Change

The two types of area-level variations in food consumption examined here are,
first, ecology or variations in natural resource endowments and, second, spread of
hybrid maize adoption. Within Eastern Province, there are two main ecological areas:
the plateau, where the majority of the population resides, which is better developed
infrastructurally, and the Luangwa Valley, which has good agricultural soils but
lower rainfall. The valley is heavily infested with the tsetse fly, carrier of the deadly
cattle disease, trypanosomiasis, which effectively eliminates oxen plowing. As a
result, farm sizes are much smaller in the valley, and hybrid maize adoption is
virtually absent. In addition to comparing consumption patterns in the plateau and the
valley, consumption at plateau sites that have a high degree of hybrid maize adoption
was compared with that at plateau sites with little adoption.
The main differences in the diets of the two ecological areas were found to be in
cereal composition and a higher intake of fish and wild meats in the valley. House-
holds in the valley had a higher content of sorghum, finger millet, and rice in their

diets, as well as a higher content of purchased maize meal. In order to examine the
pattern of cereal intake at the different sites in greater detail, the households were
further categorized by size into small and large farms, based on the median per capital
farm size for those areas. For example, households with more than 0.171 hectares per
capital in the valley were categorized as large, those in the low-adoption plateau areas
with more than 0.294 hectares per capital were categorized as large, and, in the
high-adoption plateau areas, households with more than 0.420 hectares per capital
were categorized as large. Both valley and low-adoption plateau areas have a higher
degree of dependence on purchased maize meal than plateau areas with higher
adoption rates. This clearly indicates that more maize is available in high-adoption
areas (Table 29).
Overall grain consumption is lowest among the small farmers in the valley areas
and highest among the large farmers in the high-adoption plateau areas. These
differences parallel the differences in farm sizes between the valley, low-adoption
plateau, and high-adoption plateau areas.
The differences in the annual average intake levels for the major nutrients, as
shown in Table 30, do not follow as clear-cut a pattern as the differences in quantity
of total cereal consumption. Valley areas tend to come out better than low-adoption
plateau areas in intakes of many nutrients that are indicators of diet quality, for
example, protein, calcium, and iron. The intake of calories and dietary components
that contribute to it, such as carbohydrates and fat, parallel the earlier observed
differences in farm size between the three areas. The most striking difference is that
the plateau sites have nearly twice as much fat content in their diets, whereas valley

Table 29-Daily per capital consumption of cereals, by farm size, in valley and
plateau regions and high and low adoption areas

Region Plateau
Plateau Valley High Adoption Low Adoption
Small Large Small Large Small Large Small Large
Cereals Farms Farms Farms Farms Farms Farms Farms Farms
Breakfast and roller meala 18 11 23 26 5 8 44 16
(64.9) (28.0) (44.8) (36.8) (17.5) (27.8) (104.5) (28.1)
Maize 453 584 244 302 489 633 383 488
(222.9) (295.6) (71.9) (133.8) (238.5) (320.3) (171.0) (212.3)
Sorghum flour ... ... 45 118 ...
(50.5) (92.2)
Finger millet flour 1 ... 28 20 1 ... ..
(5.8) (39.5) (49.2) (7.1)
Rice 2 2 40 23 2 2 1 ...
(4.6) (4.0) (52.6) (47.3) (5.5) (4.8) (1.7)
Wheat flour 6 6 1 4 7 10 6
(12.4) (12.1) . (2.5) (7.2) (12.3) (18.3) (11.9)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are standard deviations. The ellipses (. .) indicate a nil or negligible amount.
aThese are two forms of purchased maize meal.

Table 30--Annual average of daily per capital nutrient intakes, by valley and
plateau regions

Plateau Adoption Areas
Nutrient Plateau Valley High Low

Energy (calories) 2,367.28 1,819.60 2,491.65 1,908.78
(221.86) (168.19) (339.26) (83.71)
Protein (grams) 66.43 55.94 69.93 52.50
(6.20) (4.35) (8.79) (1.45)
Fat (grams) 36.81 17.50 39.14 25.55
(4.20) (1.23) (3.13) (1.76)
Carbohydrates (grams) 429.48 355.48 446.23 358.38
(39.58) (37.71) (68.36) (16.74)
Calcium (milligrams) 436.32 546.23 423.57 350.75
(44.14) (78.25) (51.04) (56.19)
Iron (milligrams) 21.09 35.87 21.42 16.85
(2.03) (9.65) (2.66) (1.20)
Thiamine (milligrams) 1.06 0.88 1.04 0.82
(0.12) (0.05) (0.10) (0.16)
Riboflavin (milligrams) 1.06 0.69 1.06 0.93
(0.09) (0.04) (0.14) (0.06)
Niacin (milligrams) 14.18 12.26 14.87 9.70
(1.68) (0.33) (1.62) (1.03)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Notes: Numbers in parentheses are standard deviations. Adoption of hybrid maize was negligible in the valley areas.

households consume more iron and calcium, largely because of the higher sorghum
and finger millet content of their diets.38
A comparison of the plateau sites by degree of hybrid maize adoption shows that
low-adoption plateau areas had a lower level of consumption than the high-adoption
areas. This difference was found across the board for all the nutrients, since the
composition of the diet on all plateau sites was essentially the same (in contrast with
the differences between the plateau and the valley diets).
To address the issue of diet diversity, a simple count of the number of food types
consumed in the past week was recorded during each monthly visit and compared for
the different areas (Table 31). (The total number of possible food types is listed in
Table 27.) The annual food count represents the total number of food types present
in the aggregated diet for the 12 months. During the lean months (January-February),
the diet was more diverse in the low-adoption plateau areas than in the high-adoption
areas, and this is attributed to greater diversity in both own-produced and purchased
foods. A larger number of food items are purchased in the low-adoption plateau areas
than in either the high-adoption plateau areas or in the valley areas. The diversity of

38Sorghum has 2.5 times the calcium and 3 times as much iron as maize, and finger millet has more than
30 times the calcium and more than 3 times as much iron as maize. Maize, however, has higher carotene
content. Most of these nutrients are present in the pericarp (bran and germ) of the grain and may be
biologically unavailable for absorption by the body in varying degrees, depending on the type of
processing and cooking used.

Table 31- Food diversity, by month, region, and level of adoption of hybrid

Plateau, Plateau, Valley,
Low Adoption Areas High-Adoption Areas Low-Adoption Areas
Own Own Own
Month Total Produced Purchased Total Produced Purchased Total Produced Purchased

January 6.56* 5.02* 1.86* 5.27 4.30 1.22 6.58 5.35 1.63
February 6.98* 5.14 2.18*,** 6.16 4.95 1.57 7.16 6.20** 1.48
March 6.97 5.22 1.96** 6.93 5.67 1.63 6.73 6.14** 0.79
April 7.33 5.69 2.02** 7.62 6.21 1.70 7.89 7.43** 0.68
May 8.18** 5.96 2.62*'** 7.62 6.36 1.57 7.22 6.63** 0.75
June 8.43 6.13 2.81*,** 8.05 6.35 2.13 7.69 6.70 1.39
July 8.77** 5.85 3.43*,** 8.34 6.66* 2.04 7.59 6.24 1.84
August 8.15 5.20 3.42** 7.76 5.70 2.49 7.59 5.71 2.63
September 7.83** 5.23** 3.09** 7.91 5.80 2.56 6.32 4.47 2.34
October 7.63** 5.16 3.13** 8.20 5.89* 2.90 6.47 4.95 2.21
November 7.18** 4.88 2.89** 7.48 5.46* 2.53 5.89 4.41 1.86
December 7.46** 5.83** 2.61** 7.44 5.52 2.40 5.86 4.53 1.56
Annual 15.64 11.41 7.93*,** 15.94 13.18* 6.05 18.56** 15.83** 6.32

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Notes: The measure of food diversity is derived by counting the number of food categories out of a total of 42
possible categories that were present in the diet during the period of the survey. One week's diet was recorded
each month and the results are based on that. Adoption of hybrid maize was negligible in the valley areas.
*Significantly higher at 0.05 level, comparing plateau low- and high-adopting areas.
**Significantly higher at 0.05 level, comparing plateau low-adopting areas and valley areas.

own-produced food items in the diet is greatest in the valley areas, and this is a
reflection of a more diversified production system and greater access to foods from
fishing, hunting, and collecting from the wild in these areas.
An examination of the geographical variation of diet in the valley and the low-
and high-adoption areas of the plateau indicates that caloric intakes are best in the
plateau areas with high adoption of hybrid maize, but that this also parallels the
variations in farm size in the three areas. Indicators of diet quality, such as protein,
calcium, and iron, and aggregate measures of diet diversity, however, suggest that
low-adoption valley and plateau areas do better. This is due to a combination of
factors that will be analyzed in the multivariate analysis.

Dietary Intake Variations by Household Level

In contrast to the earlier comparisons of hybrid maize adoption at the area level,
when consumption is compared according to hybrid maize adoption at the household
level, there are no clear-cut improvements evident. While the primary food compo-
nents-calories, proteins, carbohydrates, and fats-are slightly higher for the year in
the hybrid maize-adopting households, the micronutrients analyzed are all slightly
higher in the nonadopting households (Table 32). These observations suggest that

Table 32- Mean daily per capital nutrient intake, by household use of hybrid

Nutrient Intake
Nonadopters of Adopters of
Nutrient Hybrid Maize Hybrid maize

Energy (calories) 2,148.94 2,263.13
(1,057.30) (1,001.15)
Protein (grams) 61.53 63.44
(29.15) (26.99)
Fat (grams) 29.21 34.87*
(20.61) (21.73)
Carbohydrates (grams) 399.06 406.68
(193.22) (184.37)
Calcium (milligrams) 451.59* 367.17
(265.36) (167.14)
Iron (milligrams) 23.82* 19.54
(17.03) (8.67)
Thiamine (milligrams) 0.97 0.91
(0.56) (0.48)
Riboflavin (milligrams) 0.95 0.95
(0.55) (0.43)
Niacin (milligrams) 13.03 12.89
(7.68) (7.09)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are standard deviations.
*Significantly higher at the .05 level.

nonadopters in areas with a high degree of hybrid maize adoption are better off than
nonadopters in other areas. While this finding is consistent with the overall larger
farm sizes in the high adoption areas, it could also be the combined result of
improved returns to labor and improved food availability at the local level.
In order to examine the differences at the household level by hybrid maize
adoption, households are subdivided by farm size. Results show that small-farm
adopters have better diets than small-farm nonadopters, but large-farm adopters'
diets are not better than large-farm nonadopters' (Table 33). In fact, except for fat
consumption, which is higher among large farms that are adopters of hybrid maize,
all other diet constituents examined show that the large farms that have not adopted
hybrid maize actually have a higher level of nutrient consumption. The results
suggest that income elasticities of food consumption may be much lower for hybrid
maize adopters, compared with nonadopter households. These findings are consistent
with earlier results, which showed that hybrid maize adoption had a significantly
positive impact on household consumption expenditure, but this was not as evident
for larger farms.
To examine further the changes in the pattern of food consumption that occur
with hybrid maize adoption on small and large farms, the annual average intakes of
the main food groups are tabulated (Table 34). The main household dietary compo-
nents that are higher among small-farm adopters are cereals (up 11 percent), milk

Table 33- Mean daily per capital nutrient intake, by hybrid maize adoption
and farm size

Nutrient Intake
Nonadopters of Hybrid Maize Adopters of Hybrid Maize
Nutrient Small Farms Large Farms Small Large Farms

Energy (calories) 1,923.90 2,461.50* 2,165.53 2,306.01
(993.47) (1,069.08) (837.63) (1,068.23)
Protein (grams) 54.30 71.57* 59.48 65.18
(26.86) (29.37) (22.32) (28.78)
Fat (grams) 27.29 31.88 32.98 35.70
(21.36) (19.33) (22.04) (21.71)
Carbohydrates (grams) 353.91 461.76* 389.86 414.02
(176.15) (199.26) (141.46) (200.90)
Calcium (milligrams) 376.16 556.37* 360.40 370.14**
(218.14) (289.70) (172.01) (166.21)
Iron (milligrams) 20.72 28.12* 19.44 19.59**
(14.83) (18.94) (9.57) (8.33)
Thiamine (milligrams) 0.83 1.17* 0.88 0.92**
(0.54) (0.54) (0.46) (0.50)
Riboflavin (milligrams) 0.83 1.10* 0.92 0.96
(0.50) (0.57) (0.34) (0.47)
Niacin (milligrams) 11.72 14.85* 12.83 12.91
(8.06) (6.74) (7.02) (7.18)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are standard deviations.
*Significant at the .05 level, comparing small and large farm size groups.
**Significant at the .05 level, comparing nonhybrid maize producers and hybrid maize producers.

(up 180 percent), alcoholic drinks (up 71 percent),39 groundnuts (up 18 percent), and
pumpkin and gourds (up 23 percent). For the large-farm adopters, quantities of most
of the food items consumed are smaller than those consumed by large-farm
nonadopters, with the only items bucking the trend being fresh meat, sugar, and
alcohol, which are higher by 31 percent, 63 percent, and 19 percent, respectively. In
light of data on intrahousehold food expenditure patterns in Chapter 7, it may be
inferred that the food items whose consumption is higher among hybrid maize
adopters are likely to have been purchased by male household members. This could
be indicative of the changes taking place in the pattern of intrahousehold resource
availability as the result of hybrid maize adoption. In the smaller farm sizes, food
items identified as women's purchases are also higher to a smaller extent, but not
among the larger farm adopters.

39The increased consumption of alcoholic drinks in high-adoption areas is indicative of a high income
elasticity. Since most of the drink consumed is local brew, made and sold primarily by women, it may
represent a strategy for improving women's income share in high-adopting areas.

Table 34-Daily per capital food consumption, by hybrid maize adoption and
farm size

Daily per Capita Food Consumption
Nonadopters of Hybrid Maize Adopters of Hybrid Maize
Food Product Small Farms Large Farms Small Farms Large Farms
Roller and breakfast meala 22 19 7 7
Local 395 528 481 520
Other cereals 33 59 13 13
Sweet potatoes 36 43 36 42
Others 6 9 4 6
Groundnuts and beans
Fresh 16 20 15 12
Dry 38 39 46 48
Fresh leafy 44 64 43 48
Dried leafy 2 5 2 2
Pumpkin and gourds 107 125 132 123
Others 17 22 18 15
Fruits 23 46 26 21
Meat, fish, and poultry
Fresh 25 32 26 42
Dried 2 4 1 2
Milk (fresh) 5 12 14 14
Fats and oils 2 1 3 2
Sugar 8 8 10 13
Alcoholic drinks
(primarily local beer) 100 165 171 196

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
aThese are two forms of purchased maize meal.

Seasonal Variations in Dietary Intake

Seasonal variations in food consumption show the extent and severity of food
scarcity faced by households during the year. Monthly food consumption observa-
tions were grouped to get two-month moving averages between January 1986 and
December 1986. The results show that for the sample as a whole, the January-Febru-
ary period was the worst in terms of the macronutrients, with small improvements
occurring in March-April (Table 35). Beginning with the harvest months of May-
June, substantial improvements occur through July-August, which are sustained in
September-October and November-December. It is likely that consumption levels
will begin to decline sometime in the January-February period of the following year,
with depletion of grain and income from cash sales.
In contrast to the macronutrients, calcium and iron consumption is highest during
the periods when caloric intakes are the lowest. The precise reason for this is not
clear, and it is likely to be the result of the seasonally high consumption of leafy

Table 35-Daily per capital nutrient intakes, by season

January- March- May- July- September- November-
Nutrient February April June August October December

Energy (calories) 1,978.08 2,040.36 2,270.79 2,404.97 2,434.70 2,456.03
(170.53) (162.46) (208.48) (234.74) (240.29) (231.30)
Protein (grams) 55.47 60.55 66.88 67.53 68.07 67.58
(4.48) (4.73) (5.97) (6.49) (6.68) (6.41)
Fat (grams) 24.62 28.60 38.47 37.72 38.56 40.90
(2.06) (2.13) (3.65) (4.81) (5.52) (5.51)
Carbohydrates (grams) 390.89 396.33 401.72 428.21 427.73 428.95
(35.00) (32.90) (37.21) (39.83) (39.43) (37.27)
Calcium (milligrams) 506.60 550.16 500.59 391.02 361.79 354.27
(39.54) (42.15) (44.48) (39.08) (37.88) (32.30)
Iron (milligrams) 23.19 21.22 23.48 21.96 21.16 21.85
(1.55) (1.64) (1.98) (2.04) (1.98) (1.76)
Thiamine (milligrams) 0.88 1.04 1.16 1.09 1.05 1.06
(0.07) (0.07) (0.12) (0.12) (0.13) (0.13)
Riboflavin (milligrams) 0.73 0.80 0.95 1.07 1.20 1.22
(0.05) (0.05) (0.10) (0.11) (0.12) (0.12)
Niacin (milligrams) 10.10 13.19 15.78 15.21 14.31 14.91
(0.88) (1.02) (1.56) (1.76) (1.88) (1.93)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricul-
tural Development Project, and National Food and Nutrition Commission agricultural household survey,
Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are standard deviations.

vegetables, which are rich in these components at that time of the year. The B
vitamins (thiamine, riboflavin, and niacin) tend to follow a seasonal pattern similar
to that of the macronutrients, probably because their main source is the major cereals.
In comparing the differences in seasonal patterns between high- and low-adoption
areas, the broad pattern is similar, but some differences emerge. Consumption is lower
in low-adoption areas at all times of the year, especially around the "hungry" period
(February-March) (Table 36). Low-adoption areas come out of their low earlier (by
March-April), while the high-adoption areas do not increase consumption until May-
June. This is consistent with the staggered harvest in these areas, which is also reflected
in the labor allocation data, and may be due to the greater need in low-adoption areas to
harvest crops such as fresh maize, groundnuts, beans, and so forth, while still green, due
to the higher degree of seasonal food scarcity there.
High-adoption areas reach their peak consumption level in the postharvest period
of July-August; consumption declines slightly from then until the end of the year. In
contrast, the jump in consumption in the postharvest period is less perceptible in the
low-adoption areas, but consumption increases gradually between September-October
and the end of the year (Table 36), probably in response to the higher work require-
ments at the start of the new crop year.
In contrast to the area-level differences in amounts and seasonal patterns of nutrient
consumption, differences at the household level are not that great. This is partly due to
the much higher consumption variation between regions than between households
within each of the geographical regions. This suggests that the geographical dimensions
of consumption change are greater than differences between households in each of the
areas. Nonadopters of hybrid maize follow a seasonal pattern similar to that of the low

Table 36-Daily per capital nutrient intakes, by season, for high- and low-adoption areas in the plateau region

High-Adoption Areas Low-Adoption Areas
January- March- May- July- September- November- January- March- May- July- September- November-
Nutrient February April June August October December February April June August October December

Energy (calories) 2,199.83 2,187.12 2,576.54 2,746.93 2,652.23 2,577.14 1,288.37 1,657.42 1,597.06 1,704.24 1,922.97 2,141.69
(307.11) (269.28) (351.55) (390.05) (352.12) (374.39) (95.11) (63.22) (133.90) (27.32) (222.64) (344.66)
Protein (grams) 61.29 64.71 76.83 77.08 71.78 70.44 36.82 49.57 46.79 46.48 52.56 55.61
(7.94) (8.07) (9.06) (10.26) (9.48) (10.02) (3.69) (1.47) (4.74) (1.45) (4.00) (6.66)
Fat (grams) 28.03 27.35 48.89 45.05 44.06 44.23 18.79 29.08 24.80 21.58 23.00 27.61
(3.99) (1.71) (2.36) (4.84) (6.20) (6.30) (0.77) (0.78) (2.48) (2.35) (1.41) (2.18)
Carbohydrates (grams) 432.02 435.27 437.78 478.58 454.00 434.90 251.13 306.84 296.31 318.40 356.89 398.64
(66.35) (57.16) (72.90) (74.67) (66.09) (68.54) (23.49) (18.98) (29.37) (6.45) (43.39) (69.92)
Calcium (milligrams) 497.10 519.22 435.82 386.39 336.94 338.96 397.57 495.89 376.72 264.27 246.55 254.27
(64.90) (61.23) (58.96) (53.48) (47.98) (41.61) (47.63) (55.31) (65.55) (44.22) (55.96) (54.41)
Iron (milligrams) 21.77 24.34 22.67 21.86 20.00 18.88 12.72 15.00 16.73 15.73 16.25 17.03
(3.06) (2.56) (2.75) (2.92) (2.80) (2.80) (0.84) (1.01) (2.72) (1.80) (1.79) (2.16)
Thiamine (milligrams) 0.93 1.15 1.20 1.11 0.99 0.92 0.62 0.89 0.88 0.76 0.78 0.83
(0.09) (0.09) (0.11) (0.12) (0.15) (0.16) (0.14) (0.01) (0.24) (0.19) (0.19) (0.21)
Riboflavin (milligrams) 0.72 0.80 1.11 1.21 1.24 1.26 0.59 0.85 0.57 0.79 1.04 1.12
(0.06) (0.05) (0.13) (0.19) (0.20) (0.21) (0.02) (0.04) (0.04) (0.04) (0.15) (0.20)
Niacin (milligrams) 9.75 13.26 18.71 17.41 15.26 15.14 7.43 11.81 9.65 8.65 8.73 9.58
(1.09) (1.22) (1.85) (2.06) (2.32) (2.51) (1.75) (0.51) (1.62) (1.07) (0.86) (1.35)

Source: International Food Policy Research Institute, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and
Nutrition Commission agricultural household survey, Eastern Province, Zambia, 1985/86.
Note: Numbers in parentheses are standard deviations.

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