V .r,
_ 1.9m. _ .fl?
‘ r... .34..

Z
t
A. s
U

5

a.

. .awh q .

. $7.13.:-
if:

fitiflhfl‘ é.

n 1 w..."

5:}... 5

2.. r: .-.u .n

17. .. J .
~14 III

.6. A
.. ,mfiafnm s
, . ‘IV

I. I.

I 1.. $vati|a 5.
Av. I71 39:2(1!
c. t '7):ny

.1.

.. lair
3:5. vu‘ is?
‘ yr). 1 1 w

 

 

 

an r! 2:13:31 :534. 5..
Eh‘IRICII. 3|..- ‘

 

N

“‘3' A I ;smu

2
20041

[)(JS‘S'NS’L

 

LIBRARY
Michigan State
U hive rSltY This is to certify that the

 

 

 

dissertation entitled

The Impact of Maize Technologies on Inca-e
Distribution in Marginal and High Potential
' Regions of Kenya

presented by

Daniel David Karanja

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Agricultural Economics

 

 

  

Eric H. Crawford

Major professor

 

Date June 24, 2003

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

DATE DUE DATE DUE DATE DUE

 

6/01 c:/CIRCIDaleDuo.p85—p. 15

, N, ,

THE IMPACT OF MAIZE TECHNOLOGIES ON INCOME DISTRIBUTION IN
MARGINAL AND HIGH POTENTIAL REGIONS OF KENYA

By

Daniel David Karanja

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Agricultural Economics

2003

ABSTRACT

THE IMPACT OF MAIZE TECHNOLOGIES ON NCOME DISTRIBUTION IN
MARGINAL AND HIGH POTENTIAL REGIONS OF KENYA

By

Daniel David Karanja

The need to feed a rapidly growing population on declining per capita arable land and
dwindling research and development (R&D) resources is a common reality in many
developing countries and has increased pressure for governments and donors to fund
priority R&D activities that promise the greatest welfare benefits. In addition, many R&D
institutions are struggling to identify and allocate scarce resources among competing
research agenda and target regions. This debate is crucial in Kenya, especially because
agriculture is the major source of food, income and livelihood for the majority of the

population, and it has been performing poorly lately.

The objectives of this study are: (1) to provide a comprehensive review of production and
technology of Kenya’s most important staple crop, maize; (2) to evaluate the differential
impacts of maize technologies diffusion on farm profits and income distribution for
different households and regions; and (3) to help policy makers and research managers
make informed decisions on investments in Kenyan maize R&D. To achieve these
objectives, this study uses a GIS-referenced farm- and village-level survey data collected
in 1999 from 426 farmers in 30 population clusters. This and other secondary data are
used to construct multi-market models that simulate differential impacts of maize

technologies on farm profits and income for various households and regions. Gini

coefficients are calculated to gauge income distribution effects of those technologies.
Simulating impacts of technological change through input and product markets reveals
great insight into the distributional implications of alternative technology diffusion

patterns.

The results of the simulations indicate several things. First, without technological change,
Kenya will suffer a large deficit in maize output, necessitating greater maize import to
meet consumer demand or suffer unsustainable increases in maize prices in future.
Second, maize technologies that have been developed for high potential regions will
continue to have more profound aggregate impacts on maize production, leading to
reduction in import demand (if maize prices are controlled) or reduction in maize prices

(if maize prices are flexible).

Third, diffusion of maize technologies in the high potential regions has substantially
greater positive impacts on aggregate real income and farm profits, with or without
accompanying diffusion in marginal regions. However, technology diffusion in the
marginal regions has better income distribution effects than when technology diffusion
occurs only in the high potential regions, or in both regions. Lastly, the way in which the
maize market clears has important ramifications for both the magnitude and distribution
of gains and losses from various technology adoption scenarios. In general, aggregate
incomes are greater when maize prices are controlled than when the prices are flexible. A
notable exception is in urban households, whose welfare improves when maize prices are

flexible but remains unchanged when maize prices are fixed and unchanged.

DEDICATION

First, this study is dedicated to the Almighty God and to my Lord and Savior, Jesus
Christ, whom I give all the glory, honor and praise! His divine mercy and strength has
brought me this far! I thank Him for the courage, perseverance and all the resources I
needed, and for miraculously healing my wife in Lansing, Michigan. Glory to God in the

highest! Amen.

Secondly, I dedicate this work to my dear wife, Dr. Lucy W. Karanja, and our three
lovely children: Janet (9 years old), Michael (6 years old) and Elyon (2 years old). Their
love and patience strengthened my resolve. And lastly, to my mother (Janet Wanjiru),
mother-in-law (Serah Ngendo) and father-in—law (John Muchina), who tirelessly

supported and prayed for my family and me. I love you all and may God richly bless you.

ACKNOWLEDGMENT

First, I acknowledge the Chairman of my dissertation committee, Dr. Eric W. Crawford
for his support and kindness to my family. Indeed, Dr. Crawford has been more than just
an advisor to me. I am grateful to my thesis supervisor, Dr. Mitch Renkow of North
Carolina State University. His insight and experience with multi-market modeling made
my work a whole lot easier. I appreciate comments and support fi'om my committee
members: Drs. Tom Reardon, Scott Swinton, John Hoehn and Carl Liedholm, all of
Michigan State University. I also feel indebted to Dr. Carl K. Eicher, my major professor
during my Masters degree program at Michigan State University in 1988-90, for

encouraging me along this journey.

I am also grateful for the goodwill of Drs. Cyrus Ndiritu, Joseph Wanj ama and Adiel
Mbabu, former directors at the Kenya Agricultural Research Institute, respectively; the
support towards my tuition and research from the Rockefeller Foundation (New York);
and partial funding for my field research in Kenya from CIMMYT’s Economic Program
(Mexico). I wish also to thank all my Christian brothers and sisters who believed God
with us through the years, including the African Christian Fellowship (Lansing Chapter)
the Fellowship of Christian Internationals and Friends (Trinity Church, Lansing) and “At

His Feet" Fellowship, which met at our home on Fridays. God keeps all His promises!

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES viii
LIST OF FIGURES x
KEY TO SYMBOLS AND ABBREVIATIONS xi
Chapter 1
‘l' a. .I A.‘ 1
Problem Statement and Study Objectives 1
Methodology and Organization of the Study 8
Sampling Strategy and Data 8
Organization of the Study 11
Chapter 2
Characterizing the Maize Production 2.. .' ‘ 12
Characterizing Maize Farming 16
Current Maize T ‘ ' 9' 19
Changes in Maize Technology Adoption, 1992-1998 74
Agricultural Inputs, Markets and Prices 7 7
Land 77
Labor 1 1
F ertilizer ‘15
Maize Marketing and C rim. “#7
Chapter 3
Impact of Technological Change 43
Partial Equilibrium Analysis of Technological Change 47
Indirect Effects through Commodity Markets 47
Indirect Effects through Labor Markets 51
The Role of Policies and Institutions 55
The Multi-Market Model for Technological Change 59
The Analytical Framework 59
Closing the Model and Simulation Scenario: 6%
Parameter Selection and Share Computations 68
Production and Consumption Shares 68
Input Demand, Input Supply and Population Shares ..................................... 71
Income, Profit and Expenditure Shares 72
Maize Technology Shifler 75
Elasticities and Model Structure 81

 

vi

Chapter 4

 

 

 

 

 

 

 

 

 

 

 

Results and Discussions from Simulations 8‘5
Fixed-Price Model Simulations RR
F lexible-Price Model Simulations 101
Aggregate Income and Farm Prnfit 112
Income Distribution Fff‘eotq 115
Sensitivity Analysis 1 17
Chapter 5
Conclusions and Implications for Research and Policy 121
From Descriptive Analysis 121
From Multi-Market Model Simulations 127
Study I ' " " 132
APPENDICES 135
Appendix A: 1999 F arm-Level Survey Questionnaire .................................. 136
Appendix B: 1999 Village-Level Survey Questionnaire ............................... 147
BIBLIOGRAPHY 153

 

vii

LIST OF TABLES

 

 

Zone

 

 

Size and 7on9

 

 

Table 1: Distribution of Farmers by District and 70119 10
Table 2: Basic Agro-Climatic, Demographic and Maize Production
Characteristics by 7 one 14
Table 3: Maize Farm and Farmer Characteristics by Farm Size and
17
Table 4: Maize Varietal Releases between 1964-1999 70
Table 5: Maize Technology Adoption and Yield Potential by Farm
7’)
Table 6: Percentage Change in Maize Technology Adoption and
Institutional Support, 1992-98 75
28

Table 7: Trends in Per Capita Agricultural Land Use, 1960-2000 ......................

Table 8: Maize Labor and Wage Rate by Farm Size and Zone ...........................

Table 9: Per Capita Maize Production and Consumption, and Marketing
by Farm Size and Zone

 

Table 10: Multi-Market Model Characteristics

Table 1 1: Production, Consumption, Input Demand and Population Shares

Used in the Simulations

Table 12: Profit, Income and Expenditure Shares Used in Simulations ..............

Table 13: Net Yield Gains and Input Demand Associated with Zone-
Specific T L ‘ C;

Table 14: Selected Elasticities Used in Model Simulations

Table 14 (Contd.)

Table 15a: Fixed-Price Model Results: Baseline Scenario (% Change) ..............

Table 15b: Fixed-Price Model Results: Technological Change in
Marginal Regions (% Change)

viii

34

41

66

69

73

an

R?

R?

89

90

Table 15c: F ixed-Price Model Results: Technological Change in

 

 

 

Highland Regions (% Change) 93
Table 15d: Fixed-Price Model Results: Technological Change in

All Regions (% Change) 94
Table 15e: Fixed-Price Model Results: No Technological Change,

Maize Price Increase by 20% (% Change) 96
Table 15f: Fixed-Price Model Results: Technological Change in

All Regions, Maize Price Increase by 20% (% Change) ............................... 98
Table 16a: Flexible-Price Model Results: Baseline Scenario (% Change) ................ 102

Table 16b: F lexible-Price Model Results: Technological Change in
Marginal Regions (% Change) 104

Table 16c: Flexible-Price Model Results: Technological Change in
Highland Regions (% Change) 105

 

Table 16d: Flexible-Price Model Results:
Technological Change in All Regions (% Change) ....................................... 107

Table l6e: F lexible-Price Model Results: No Technological Change,
Doubled Imports (% Change) 109

 

Table 16f: Flexible-Price Model Results: Technological Change in
In All Regions, Doubled Imports (% Change) 110

 

Table 17: Aggregate Farm Profits and Income Effects, and Gini
Coefficient by Technological Change Scenarios l 13

 

Table 18a: Sensitivity Analysis of Selected Elasticities and Shares
for Fixed-Price Model 118

 

Table 18b: Sensitivity Analysis of Selected Elasticities and Shares
for Flexible-Price Model 1 19

LIST OF FIGURES

Figure 1: Kenyan Maize Output and Area, 1961-99

Figure 2: Kenyan Maize Agro-Climatic Zones

 

Figure 3: Fixed-Price Model of Technological Change

 

Figure 4: Flexible-Price Model of Technological Change
Figure 5: Technological Impact on Adopting and Non-Adopting Households .........
Figure 6: Impact of Technological Change on Labor Markets

Figure 7: Modeling the Probability Distribution of Research Outcomes ..................

 

Figure 8: Maize Technology Adoption Profile
Figure 9: Multi-Market Model Structure
Figure 10: Kenyan Maize Price Variation, Jan 1996 - Dec 1998 ..............................
Figure 1 1: World Market Maize Price Variation, Jan 1996 - Dec 1998 ....................

Figure 12: Calculation of the Gini Coefficient

 

48

50

52

54

76

79

84

86

87

116

CBS

DMA

DRSRS

DT

HT

KMDP

KMIS

LT

MMA

MT

NASSEP III

NCPB

R&D

T&V

WMS

KEY TO ABBREVIATIONS

Central Bureau of Statistics

Dry Midaltitude zone

Department of Remote Sensing and Resource Surveys
Dry Transition zone

Highlands zone

International Monetary Fund

Kenya Agricultural Research Institute

Kenya Maize Database Project

Kenya Maize Impact Study

Lowlands zone

Moist Midaltitude zone

Moist Transition zone

National Sample Surveys and Evaluation Program
National Cereals and Produce Board

Research and Development

Training and Visit

Welfare Monitoring Survey

xi

CHAPTER 1

1.1 Introduction

The need to feed a rapidly growing population on declining per capita arable land and
dwindling research and development (R&D) resources is a common reality in many
developing countries and has increased pressure for governments and donors to ftmd
priority R&D activities that promise the greatest welfare benefits (Renkow, 1993). In
addition, many R&D institutions are struggling to identify and allocating scarce resources
among competing research agenda and target regions. The quest to do this assumes
greater significance in Sub-Saharan Africa (SSA), which is the only region in the world
that is experiencing declining per capita food production, rapid population growth and

serious economic stagnation (World Bank, 2001).

Conventional wisdom suggests that in order to sufficiently improve agricultural
productivity, investments in R&D should be made in high potential regions rather than in
marginal regions. The argument is that greater productivity in the high potential regions
will generate faster economic growth, greater employment, higher wages and lower food
prices that will benefit the country more, especially the poor. In additional, there will be
less pressure to cultivate fragile marginal lands thereby reducing environmental
degradation. Moreover, investments in marginal areas historically have been low because
of poor returns, and diverting research resources away from the high potential regions

may do more harm than good (Coxhead and Warr, 1991; Renkow, 2000).

An alternative hypothesis suggests that increased public investments in “less-favored”,
marginal regions can generate competitive or greater agricultural grth than comparable
additional investments in high potential regions. The argument cites the fact that past
investments in agricultural development tended to focus on irrigated agriculture and high
potential regions and never resolved increasing poverty, hunger and food insecurity
problems in the marginal regions. Coupled with increasing evidence of stagnation in
agricultural productivity growth in high potential areas, this strategy proposes that more
investments in marginal regions’ agriculture may yield higher aggregate social returns,
given that most of the poor are located in these regions, compared to investments in high
potential regions (Fan and Hazell, 2001). Also, if the bulk of past agricultural research
investments were made in high potential regions, the incremental rate of return to
investment may decline to the point where it competes with, or is surpassed by,

investments in marginal regions (Byerlee and Morris, 1993).

This debate is critical in Kenya and many other African countries that face a severe food,
agricultural or economic crisis. Since agriculture is the dominant sector of the economy
and provides food, employment and income to 70-80% of poor people who live in rural
areas, the best way to achieve poverty reduction and welfare gains is to increase
agricultural productivity, improve access to food and markets, and invest in supporting
infrastructure, institutions and policies. Increases in agricultural productivity will, in turn,
provide better access to non-tradable and semi-tradable foods, and improve rural and

urban employment and wages.

This study contributes to the debate by using cross—sectional, farm— and village-level data,
as well as other secondary data, to evaluate the impact of maize technologies on income
distribution in different households and regions, while providing information on the
status of maize production and technology in a way that aids decision-making concerning
investments in maize R&D in Kenya. A multi—market model framework is used to
provide insight on how the technology works through output and input markets to
distribute benefits and losses to different farm households (those adopting and non-
adopting; net producers and net consumers; small and large scale farms; and urban
consumers). Such analysis provides valuable information for Kenyan policy makers and
research managers to make objective decisions about specific maize technological

investments.

1.2 Problem Statement and Study Objectives

Globally, maize ranks second to wheat in terms of production output, but in Africa it
ranks first. Maize is the most widely grown cereal crop, with seventy countries, each
planting more than 100,000 hectares (Ha) of maize. It is grown in diverse regions,
elevations and production cycles (Dowswell, Paliwal and Cantrell, 1996). Five hundred
million metric tons (MT) of maize are produced on 130 million Ha annually in the world.
Twenty of the largest maize-producing countries account for 91% of the global output
and 80% of the global maize area. Kenya’s share in the global maize production and area

are 0.6 and 1.2%, respectively (FAOSTAT, 2001).

Despite a small global share, maize is Kenya’s staple food. The crop, introduced into
Kenya about a century ago, is grown in almost all agro-ecological zones on nearly two-
thirds of the total food crop area (Hassan and Karanj a, 1997). The crop’s popularity has
surpassed that of other traditional cereals, such sorghum and millets, due to ease of
cultivation, storage and processing, resistance to pests and diseases, and utilization in
many forms. Maize supplies large shares of proteins and calories for the majority of poor
people, and dominates food policy decisions in Kenya, to the extent that insufficient
domestic supplies of maize easily translate into major food shortages and, often, famine

(Blackie, 1990; Karanja, 1990).

In general, Kenya produces sufficient quantities of maize in most years, except in adverse
weather. Figure 1 shows the trends of area and production of maize in Kenya between
1961-99. The trend reveals increasing maize production and area between 1963-77, a
“peak” on both area and production between 1991-94, and a decline followed aflerwards.
The increase between 1963-77 is attributed to increased availability and use of new maize
varieties while the latter stagnation reflects a slower uptake of new technologies or few
newer technologies available to farmers, a limited capacity for area expansion, and

adverse weather and poor policy environment effects (Hassan and Karanja, 1997).

The remarkable past progress in maize production in Kenya was a product of consistent
development and injection of new streams of maize technologies from the public-funded
agricultural research institutes, now coordinated under the Kenya Agricultural Research

Institute (KARI). This research system has over the years developed more than 25 maize

ENE.» 7 flags... Zine 0.5.... 3... >3... 33-09

 

 

 

wmooooo

woooooo

Nmooooo

moooooo

 

Amooooo,

Maize Output (MT), Area (Ha)

 

 

doooooo.

 

mooooo. .........................

 

 

 

 

damn down Ammu .mwo dwuw doum amum

<34

noun

_ 1.! J

imam

.

. _ .

Ammm

_ .

Amos

_ a _

Amen

meow

 

Tags 9%... u .. sea >3» ...... .3? Ages >3». 1 u r .6? Eng 05...:

_

 

 

 

varieties targeted for different agro-climatic zones across the country. The adoption of
these varieties and other complementary technologies, such as fertilizer and intercropping
techniques, contributed immensely to increases in maize production in the past, which
partly was responsible for relative low consumer prices of maize (Karanja, 1990). After a
decade of dismal performance in the 1980s, a series of maize input and product market
reforms were carried out in the 19905 but have, seemingly, not yielded the intended
increase in input use and maize productivity, the reasons ranging from high input cost to
poor infrastructure and lack of market incentives for farmers (Karanja, Jayne and

Strasberg, 1999; Renkow, Hallstrom and Karanja, 2003).

Kenya faces a serious maize and food problem now and in the future if productivity
growth fails to match a rapidly increasing food demand. This study and debate are timely
and critical for several reasons. First, the economy is heavily dependent on agriculture,
with the sector accounting for 26-30% of the total Gross Domestic Product (GDP), 80%
of total employment, 60% of total export earnings and 45% of total government revenue,
and supporting livelihood for 85% of the population (Karanj a, 1990). Secondly, the
country has one of the fastest growing populations in the world, which puts considerable
pressure on limited arable land and raises food demand beyond domestic supplies.
Thirdly, the high demographic pressure on land has caused agricultural production to
shift gradually into fragile marginal environments posing a serious degradation threat
there. Finally, Kenya’s economy has been performing poorly in the past decade, thus

constraining public investments in agriculture. A decade of suspended donor funding,

three years (1999-2001) of the worst drought in 42 years and stress from transition

politics combine to create a potential political and economic crisis.

Because of limited capacity for expansion of arable land, future increases in agricultural
productivity will largely rely on yield improvement (Karanja, 1996; Byerlee and Eicher,
1997). Therefore, investments in agricultural innovations that increase yields through
genetic improvement, agronomic husbandry or technology diffusion will be crucial.
However, concerns arise about the differential impacts of resource allocation decisions
and investment in such agricultural technologies on the welfare of different households
and regions (Renkow, 1993). For Kenya, the concern is the differential effects on small
versus large farmers and high versus marginal potential regions, hence this analysis uses
these classifications of farmers and regions to shed light on this issue. The investment
and policy choices that maize research managers in Kenya make has important
implications on whether those strategies will reduce or aggravate current levels of hunger

and poverty through their impacts on the farm income and its distribution.

Currently, farmers achieve, on average, between 20-60% of the yield levels attained in
research centers (Hassan, et a1. 1998). This gap is attributed to various constraints that
include a low use of existing maize technologies and may point to a potential for
productivity growth waiting to be exploited by better targeting of those technologies and
facilitation of their adoption (Hassan and Karanj a, 1997; Matlon and Spencer, 1985).
This study evaluates the impact of such potential (‘on-the-shelf’) maize technologies on

farm income and its distribution, where these are used as measures of farm welfare, on

different households and regions. Since maize is the major staple and accounts for large

production and consumption shares for the majority of Kenyans, any policy choices and

potential changes in maize productivity and incomes will have far-reaching economic

welfare and poverty reduction impacts. Therefore, this study has three main objectives:

1. To provide a review of the status of maize production and technology in Kenya.

2. To evaluate the differential impact of the diffusion of maize technologies on farm
profit and income distribution for different households and regions; and

3. To help policy makers and research managers make informed decisions on

investments in maize R&D in Kenya.

1.4 Methodology and Organization of the Study

1.4.1 Sampling Strategy and Data

This study used a GIS-referenced multi-stage stratified random sampling frame that was
created by a KARI project, the Kenya Maize Database Project (KMDP) back in 1992
(Hassan, Lynam and Okoth, 1998). Agro-climatic characteristics, population density and
intensity of maize production were used to define homogenous maize production zones.
The sampling sites were randomly selected from the National Sampling Frame of the
National Sample Surveys and Evaluation Program (NASSEP HI), Central Bureau of
Statistics, which contains 1048 rural and 324 urban population clusters (Kenya, 1994).
Since these sites were well distributed across maize-growing regions, the current study
randomly selected 30 survey clusters and 426 farmers out of the 65 clusters and 1407
farmers used by the KMDP project. The distribution of farmers between different zones

was determined by the relative importance of maize in each zone, logistical

considerations and available research funds. Table 1 shows the final distribution of
selected farmers and survey sites by administrative district and agro-climatic zone. The

farmers were interviewed using farm- and village-level questionnaires (in the Appendix).

Secondary data, complementing the primary data collected in 1999, included:

1. Maize production data for 1992-98 from the Department of Resource Surveys and
Remote Sensing;

2. Commodity price data for 1995-99 and different spatial markets from the Market
Information Branch, Ministry of Agriculture;

3. Climate data for updating the GIS information from the Department of Meteorology.

4. Data on infrastructure to supplement village-level survey from the Ministry of
Transport Communications and Public Works;

5. Detailed data on welfare attributes of sampled clusters from the Welfare Monitoring
Survey of the Central Bureau of Statistics, Office of the Vice-President and Ministry
of Planning and National Development; and

6. Regional farm production data from the Department of Rural Planning, Office of the
Vice-President and Ministry of Planning and National Development.

These primary and secondary data are used to generate parameters needed to simulate

impacts of maize technologies on income and farm profits for different households and

regions using fixed-price and flexible-price multi-market models. The multi-market
models are preferred because they allow detailed household-level analysis, especially
when farmers are a mixture of producers, consumers and labor suppliers, a situation that

would present complications in typical economic surplus models.

dew—a H” gun—.353: e». Has—.5»; 3. Ewing 9.... N25

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

33m?»— weHgaeH Neuom EIHmw H.323... N38
Quinn 53. Sean DJ. 3%.: >= N38
Pei—3.3 739.58% 33:52:... Hue—BEE. 5.255.... Ema—23m
Wanna Hm Hm
Hweamoae H m H u
9.me H m H u
mac: H m H m we
9.2... H m H m
5.555%. Hm Hm
HAHHHm H m H m
HnéwHo mo mo
gmoerSm Hm Hm
3.38.: H u H m
7.55.9 H m H m
73.3%.» H m H m
7.3:. Hm Hm
3.3an H m H m
238.... Am Am
2585. H m H m
waeawam H m H m
Wmoreoewo H m H m
.363 28:. 3 AN
genera Hm Hm
C85 92... mo “3
Hon: ac ac Au 3 H on H H... So

 

 

lO

Disaggregating simulated impacts of technological change across different household
types and regions, as well as across input and product markets, reveals greater insight into
the distributional implications of alternative technology diffusion patterns. The model
analysis used is similar to Renkow (1993) but it considers six agroclimatic zones instead

of two and does not differentiate between low- and high-income urban populations.

1.4.2 Organization of the Study

This study is organized as follows. Chapter 2 provides a detailed descriptive analysis of
maize farming and technology in Kenya. This updates information on maize farming and
technology in Kenya, as differentiated by small and large farmers in marginal and high
potential regions. Maize policy decisions are considered on the basis of these categories.
The chapter also outlines information on maize input and product markets, and changes

that may have taken place after market liberalization in the early to mid-1990s.

Chapter 3 presents a literature review on the impact of technological change on farm
productivity, income and welfare. It also lays out the structural details of the basic
analytical framework or multi-market models used in this study and outlines how
different parameters for the models are computed. Chapter 4 presents the results and
discussions fi'om a simulation of a fixed-price and a flexible-price model of technological
change, including calculating Gini coefficients from the scenarios considered to
determine the income distribution effects of maize technology diffusion. Chapter 5

summarizes the results and draws the implications for research and policy.

CHAPTER 2

2.1 Characterizing the Maize Production Environment

Assessment of the impact of technological change on income and welfare in different
households and regions requires careful characterization of the production envirorunent,
farm technology and its users, measures of specific technology effects on income, and
understanding of how the market distributes output gains and losses. To characterize the
production environment, this study used the maize zone classification developed in 1992
by the Kenya Agricultural Research Institute.1 The advantage of the new classification
over the previous J aetzold and Schmidt (1983) is that it is digitally referenced using

Geographical Information Systems, making it easier to use, update and adjust.

Figure 2 is presented in color and shows the six classified maize production zones used in
this study: (1) the Lowlands zone; (2) the Dry Midaltitude zone; (3) the Moist
Midaltitude zone; (4) the Dry Transitional zone; (5) the Moist Transitional zone; and (6)
the Highland zone. Table 2 summarizes the agro-climatic, demographic and maize
production characteristics of these zones, which indicate significant zonal variations. The
March-August season is the major season for most of Kenya’s maize producing regions
and has greater rainfall amounts. The September-February season has less rainfall but is
more important for parts of the Moist Midaltitude, the Dry Midaltitude and Dry

Transition zones.

 

l Corbett (1998) has details of how these zones were created and classified.

12

Figure 2: Kenyan Maize Agro-Climatic Zones

 

    
  
  

ETHIOPIA

UGANDA

SOMALIA

 

  

TANZANIA

. Towns

Agroclimatlc zones
Lowland tropical
Dry midaltitude
Moist midallltude

- Dry transitional
- Moist transitional

- Highland tropical

, | < 0.5% maize '

- Lakes oo o 00 100 Kilometer-
. @

 

 

 

Source: Hassan (1998)

dew... N" wean .9 3.9.5.5? 52.... .8 So a...— HSeHu... $39.92.... 9.92.2013.
HS... Ha... H.223». 82.8

 

"m 3%.;

 

 

 

 

 

 

 

 

I. 33.8....»— No.8...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

v3.2.8.2. 53. 3:... OJ. H59... >= Noam
Pet—nu... gas—:3... 3.9.53.3 56.85.... Hangman Ema—2....

Zn»: 5.23.... Aim... MN... ma H b Humab Hag... Gum. nwoqu -
3...... .9...me— 9......
gfio...>=mem. a3... unw. H qmob u H «b no H .3 a3... -
ma...oSvS.H.aH.EmQ ugh Sm... uAH b m3 .u Maw... wwcb -
.H....~H Howub .39.. Hum H .u Homn... H3...» H comb -
3...... 5.3.6325. A00.”
HERBS: N~.m 3... Hm... Huh Hm... mH -
3832.... mob 8H um... E: N... nub -
33.... we: HS... Hr=<Hmon moanueHm >385 <33on 2:82... 2.8...on -
HS». 3. 0.. >m.......H....eH :eEoroEm”
m8»: men... H Hwooo .88.... 38.... Huwooo 03...... 3.3.... $38..
Ema men... ~38 H32... 38.. 8...... H39... $8.. 38....
Kano 399.032.. 58.3“
ZEN." 0.5.... as... “.38 ~32... .33.... 5%.... 3.95.. 33.... P338
7..qu >8» a... .3ch wu :8 H880 53.... .33.... u: Huoo H.388
game 52.. Again. can PE Hbm H.H~ Pt Nb... HQm
3.5. 38:02.... 95833-3 ax...
m8»: oboa Pow.» Poo; Peso 0. H3 c. H H. -
Ema ob... Poem oboe 98m obmo PH Ho -
H08. 9.... 93¢ 93.. 993 9.3.. 930

 

 

 

mafia... W2...» A H003. gm 288. 0023: $83.

14

Nitosols are the best soils for maize production and major soils in the high potential
regions and are the best for maize production. In contrast, the soil types found in the
marginal regions, such as the Vertisols, F erralsols and Acrisols, have less potential for
maize productivity compared to Nitosols.2 These and other agro-climatic characteristics
influence the choice of maize varieties, the production potential and crop seasonality, and
based on them and for purposes of this study, the Highlands and Moist Transition zones
are classified as “high potential” regions and the rest of the zones are considered as

“marginal potential” maize regions.

The high potential regions are the most important maize production regions and account
for half of the total area under maize and more than two-thirds of the total maize output.
Contrasting this, the Dry Midaltitude and Dry Transition zones account for about 30% of
the total maize area and 15% of the total maize output. The Lowland zone is the least
important maize producing area and yields 1.6% of the total maize output on 3% of total
maize area. The 1992-98 average maize area and production was 1.44 million hectares
(Ha) and 2.56 million metric tons (MT), respectively, giving an average national maize
yield of 1.78 metric tons per hectare (MT/Ha). Small farms produced about half of the

maize in 1996-98 and accounted for 88% of all rural agricultural households in Kenya.3

 

2 Soil names used here refer to FAQ-UNESCO (1974) classification units.

3 This study uses 10 acres as cut-off size for differentiating small and large farms.

2.2 Characterizing Maize Farming

Hassan (1998) presents a detailed description of maize farming in Kenya based on data
collected in 1992. Since then, significant changes in maize policy and institutions have
taken place. This study updates that information and provides a comprehensive
characterization of maize farming and technology in Kenya. Table 3 presents a summary
of maize farm and farmer characteristics among small and large farms in the six maize
zones. The mean farm size among small and large farms was 1.32 Ha and 109.74 Ha,
respectively, and for marginal and high potential regions 2.3 Ha and 37 Ha, respectively.

The overall mean farm size was 19.90 Ha while the median farm size was 1.27 Ha.

The average proportion of female farmers was 45% for the entire sample, with significant
variance observed by household group and zone. Fifly percent of the small farmers were
women, compared to only 18% among large farmers. The proportion of women farmers
was noticeably higher among small farmers in the marginal zones, the Dry Midaltitude
and Dry Transitional zones, than in the other zones. This may reflect a bias against
women on land distribution and/or access, or a reflection a higher likelihood of men in

these zones to seek off-farm employment to supplement farm income.

Seventy-two percent of all the farmers had some formal education.4 This compares to
Kenya’s average of 83% (Kenya, 1998). Farmers in the Lowland zone were the least

educated whereas those in the Moist Transition and Highlands were the most educated.

 

" Education refers to “years of formal schooling”.

16

Hue—o u” 35% 5.3: 9.:— wunaon Grununnniumou cm 5...? 2% 2:— No.5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32. a». wan—Es— 888 m. .— 5325... N058
5.2.332. OJ. 33% 9Q 333 >= 82.8
rot—nan 3.9.52.3 33522:? H3359: H.335: E 5.5%
3o»: 5:: mmuo any
was: menu #3 #3 3: #8 ram ram 35
Ema mead. #2 Sum Hum - 3.?“ 3a.; 39.3
.x. m3»: madam AAA $5” 3 we mw So «A .2 mm
.x. mans—n 3383”
was: 3 mo 3 mm 3 mm mo
Ema we 3 . o - 3 E 3
.x. HERBS mean?
ma»: man: 3 um 3 Nu 8 mu 8
Fwfio man: .3 um um - 5 o B.
>ma 0». means. A<omamvn
m5»: 3 3 Au .3 2 .3 3
560 .3 mm 3 - mo 3 .3
.x. 3:: A033 >3» C32.
33%” mo 83 on 33 Am 8.: mo 33 um 33 3 33 mm 3:
was: mm 38 G 38 No 38 - 3 33 NM 83 mo 33
ream."
33$ 3855 32.9”
was: 33>? Oo<2o< moc3? MQEOS 33>? 33>? 3E5»?
Came 33>»? On. ““33? - 38.5? 3mg»? 33;?
33% 324835 30:3“
wan: FEE“ v.33? .3555 ~63? >=Qmov 2358 .3555
gnu €525 mun—U32. u=<>=m - 2358 2358 2358

 

 

l7

Overall, 71% of all the small farmers and 77% of all the large farmers had formal
education, and further comparison revealed that twice as many large farmers as small
farmers had higher levels of education. The overall mean age of the farmers was 44
years, indicating a relatively aged farming community. Large farmers were relatively

older (49 years) than small farmers (43 years).5

There was a notable differential allocation of land to maize by farm size and zone. On
average, small farmers allocated twice (68%) as much of their total farm area to maize
compared their large-scale counterparts (30%). Although most small and large farmers
allocated nearly 70% of their total crop area to maize, this was untrue for large farmers in
the Moist Midaltitude (49%) and the Highland (3 8%) zones. Notably, farmers in drier

environments allocated more cropland (75%) to maize than the average.

Maize was planted at different times depending largely on the onset of rains. Thus, there
was little variation between small and large farmers. Two distinct planting seasons are
observed: February-April and October-November, with the majority of farmers planting
in the former. Farmers in the Lowland, Moist Midaltitude, Moist Transition and Highland
zones mostly planted their maize between F ebruary-April. On the contrary, most farmers
in the Dry Midaltitude and Dry Transiton zones planted their maize in the October-

November season.

 

5 Average life expectancy in Kenya is 58 years for men and 61 years for women (World
Bank, 2002).

Harvesting time depended on the length of the rainfall season and the maturity duration
of the maize varieties planted by the farmers. The majority of farmers in the Dry
Midaltitude and Dry Transition zones harvested their maize between February-March
while in the Lowland and Moist Midaltitude zones, the majority harvested between J uly-
August. Most farmers in the Moist Transition and the Highland zones harvested between
November-December, apart from small farmers in the former zone who harvested
between August-September because they planted shorter maturing varieties. Overall, the
majority of small farmers harvested their maize in July—August period and large farmers

in Nov-December period.

2.3 Current Maize Technologies

Karanja (1990) and Karanj a (1996) provide a detailed historical perspective and
evolution of maize research in Kenya, noting that past success of the maize research
program was due to: (1) expansion of the research program to develop maize varieties
suited to different agro-climatic regions; (2) the ability of the government to forge a
public-private sector partnership to ensure the dissemination of the hybrid maize; (3) an
aggressive agricultural extension program that planted field trials and taught farmers how
to plant the new hybrids; and (4) guaranteed maize prices and markets for nearly 50
years. As a result, Kenya’s maize research program is one of the most successful in
Afiica, and has developed over 25 maize varieties for the six agro-climatic zones, giving

a range of varietal choices to farmers (Table 4). Current research is attempting to provide

Table 4: Maize Varieties in Kenya, 1961-99

 

 

 

Variety Year Maturity Elevation Yieldl
Released (MT/Ha)

Kitale Synthetic II 1961 Late High 3 .4
Katumani Synthetic II 1963 Early Medium 2.0
H61 1 1964 Late High 4.5
H621 1964 Late High 4.1
H631 1964 Late High 4.5
H622 1965 Medium High 5.2
H632 1965 Medium High 4.5
H612C 1966 Late High 5.9
Katumani Composite A 1966 Early Medium 2.3
Katumani Composite B 1968 Early Medium 2.8
1451 1 1968 Medium Medium 3 .6
H512 1970 Medium Medium 4.1
H6HC 1971 Late High 5.9
H613C 1972 Late High 6.0
Coast Composite 1974 Medium Low 3 .3
H614C 1976 Late High 6.3
H625 1981 Late High 6.8
H612D 1986 Late High 6.4
H613D 1986 Late High 6.0
H614D 1986 Late High 6.6
H626 1989 Late High 6.8
Dryland Composite I 1989 Early Medium 2.9
Pwani Hybrid I 1989 Early Low 3.8
H627 1989 Late High 6.9
H628 1999 Late High 7.1
H629 1999 Late High 7 .1
I Research yield potential

Source: Karanja (1990); Ochieng (1999)

20

regional specific agronomic recommendations using an eco-regional approach (Ochieng,
1999). Hassan and Karanja (1997) summarize the status of maize technology, adoption
and productivity in 1992, before the government implemented drastic agricultural input
and product market reforms. Mills (1998) went further and developed a framework for
estimating ex ante impacts and priorities for maize technology in different zones using an

economic surplus methodology.

Table 5 shows various parameters on maize technology and adoption. Sixty-two percent
of all the farmers (61% of small farmers and 70% of large farmers) planted hybrid maize
seed in 1998. This proportion varied by zone and household group. About 90% of small
farmers and 97% of large farmers in the high potential regions planted hybrid maize. In
marginal regions, less than one-fifth of large farmers and one-third of small farmers
planted hybrid maize. Overall, H614 was the most popular variety and was planted by
46% of all farmers; the second popular varieties were “local” maize varieties, which were

planted by 26% of all farmers.

Fifty-eighty percent of all farmers used basal fertilizer, with relatively more of the large
farmers (73%) adopting basal fertilizer compared to the small farmers (55%). However,
just like with hybrid seed, fertilizer adoption levels varied between different zones and

household types, with higher adoption taking place in the high potential regions and

21

.320 mu 3&3“ Han—:53 3

 

 

 

 

tease: 2:— 50..— wenoana a 5:3 mun." 2:— N25

 

 

  
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32. in. $32.2:— Neaon I. _. 5.8....»— Noun“
5.2—522. OJ. 33: UQ 3cm: >= Neaau
rot—nan 35-53% 3.9.53.3 56.552. 5.2552. Ema—saga
wove—E. $503. ax. Cmmum
$353.. H.08— AS 5:55: 0& Pena 33 mm: GE a: 3: la: 88 mo: 3:
was: menu Cunn— 28 “3553 Cd H.08— Cuv - ma: 33 mm: Cos =0: 38
E
o\.. CmEm mum». moan—E?
man: menu a N.» 3 nu ma 8 mm
Ema mun: o 3 mm - :5 3o 3
wean. H.635? has". Gama»?
m3»: mun: rm no; NNb “56 3b mob any
Sumo $5: ob 3b 3; - Sou. are Bub
.x. Cat—m H39? Hanna .
man: man:— o c o I No 3 8
Fauna 3:: o o o - So So 3
33.013298 ax. Enonnnouvomvu
man—.123 means—5.. 33 women 38 wow—8 8: W33 33 Wow—5 33 wogm 83 Wang 33
game 35: Gamma!» 33 woman 22: mama 33 - wanna Guy wowa $3 wanna Eur
335;. <55. Sow A3653“
was: man: 9% Pd row 9% fig ~.~m 3%
gm.“ 55: 98 0.3 3% PS w.~m NS
3358. <53. Son A553”.
m8»: 9% on: 35 Pm— gbo #3 _.me
Ema out. 9; ode oba PS mug PE
093:: Wong—d: <55 Non rua P3 #3 who P S -
235”“
9.930: ”338.”. 505 PM... Nae ch Pmo Maw 93 -
HE» “a
361. 59.3 ES. 52: a >530}. Denice—mo:

22

among large farmers; all large farmers and 80-84% of small farmers used fertilizer.
Except in Moist Midaltitude zone, less than one-third of farmers in marginal regions used

fertilizer.

The level of fertilizer application by farmers was lower than the recommended rate,
except among large farmers in the highland regions. The average rate of fertilizer
application for all small farmers was 48.7 Kg/Ha, compared to 105 Kg/Ha for large
farmers, and 40% of the recommended rate. The overall average fertilizer rate was 58.3

Kg/Ha, half of the recommended level.

Besides seed and fertilizer, another important technological factor affecting maize
production is the land preparation method. This affects farm productivity through the
quality of the seedbed and timeliness of planting. According to Table 5, about 55% of all
farmers used the hand-hoe to till their land, ofien delaying subsequent farm operations
and experiencing lower farm yields. 23% of the farmers used a tractor and 22% used
oxen plow. Only 13% of small farmers used a tractor for tillage compared to 67% for

large farmers.

The most important crop inter-planted with maize was dry beans. This was true in all
zones except the Lowland zone where sorghum was the most popular intercrop among
small farmers and cassava among large farmers. About 75% of all small farmers

intercropped their maize compared to 47% of all the large farmers. In almost all zones,

23

 

more than half of the farmers intercropped maize, except among large farmers in the

Moist Transition (45%) and the Highland zone (13%).

Table 5 also presents the maize research yields and farmers’ yields in 1998 and 1992. The
average yield in 1998 among small farmers was 1.38 MT/Ha while that of large farmers
was nearly double that at 2.62 MT/Ha, with an overall average of 1.59 Mt/Ha. Farm
yields varied from 0.64 MT/I-Ia among small farmers in the Lowland and Dry Transition
zones to 3.61 MT/I-Ia among large farmers in the Moist Transition zone. Comparisons
between yield levels in 1992 and 1998 show similar patterns and variations. The farm
yields are also relatively lower than research yields, revealing yield gaps that can be

exploited for productivity gains, even without developing new varieties.

2.4 Changes in Maize Technology Adoption, 1992-1998

Table 6 compares maize technology adoption between 1992 and 1998; over 90% of all
the households sampled in 1998 had been sampled in 1992.‘5 Overall, the number of
farmers using hybrid maize seed declined by 5-6% while basal fertilizer usage improved
by 9% among small farmers and 16% among large farmers. Significant increases also
occurred for large farmers (17%) in the Dry Midaltitude, small farmers (25%) in the

Moist Midaltitude zone and both small (35%) and large (37%) farmers in the Highland

 

6 Major maize policy reforms occurred between these two time periods and are discussed
in sections 2.5 and 2.6.

24

._ m: 2... 38.3.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hue—o a" men—62.3% Or»: a :— gsfio Hoar—53mm >925...— nun Ewan—:2:
32. En. $32.3: 82.8 E 3 ~32...»— 82.8
5.3532. OJ 33m» 63. 3%. >= N658
Pot—2:: 23:52:? 3:35:50 Aggie: H3332. In 5.5%
.x damn Egan moon
was: man: — I A: 5 a A3 Gv
gnu 1&3 a I 88 - A8 a A3
.x. Cmgm moan—Eon
was: 128 a e nm 2.: G mu 0
Smo mun: o 3 o - q 3 3
mamm— moasuon has... GAE—Aux
was: man: 5. 3.3 3 _ .8 ANN—.3 ASE Nah 38
gnmo mud: Po 2 .N Sm - who .86 uub
53:08. 5%. Sam AER»!
was: man: 98 SK 3.3V aid 8; C 93 8b:
ram“ 3:: 9: oz: 9% - 9; 8.5 a; s
.x. ”8959 mxnoammon >ainon
m3»: man: 3: 6.3 3: 33 G3 Gov $8
Fauna man: 08 23 G C - 38 $3 38
.x. ”8023 08%:
m5»: mun: a N o q E 9 o
rain mad. 0 o g - 38 23 28
.x. 22 mo=n~m om KEN...”
was: man: 5 5 3 Q; N; 3 3
gm... mun: A3 o NA - wu 3 Nb

 

 

 

 

 

 

 

 

 

 

mofino” gm :8va 555...? Sec
_ Emu—.8 3383: $8 .98— 835. Gen 33: A v Rana 8 names: 5??

25

zone but a significant reduction of 17% was noted among small farmers in the Dry

Transition zone.

The level of basal fertilizer shows a mixed picture with an overall reduction on the
average level of fertilizer use by about 29 kg/Ha among small farmers compared to an
average increase of about 35 kg/Ha among large farmers. Significant reductions in the
rate of basal fertilizer applications were observed among small farmers in the Moist
Midaltitude, Dry Transition and Moist Transition zones but there was a significant
increase in the rate in the Highland zone. Large farmers in the Dry Midaltitude, Moist
Transition and Highland zones recorded significant increases in the level of basal

fertilizer use.

Given major policy and institutional changes that took place between 1992 and 1998, the
observed changes in institutional access by farmers were profound. For instance, 43% of
small farmers and 46% of large farmers had lost access to extension service in 1998
compared to 1992. This reduction was observed in all the zones and household types,
with the worst reduction being in the most important maize-producing region, the Moist
Transition zone. Access to farm credit for maize production was less dramatic, with
access to small farmers remaining at 2% and for large farmers declining by 10 % between
1992 and 1998. However, there were significant increases in the number of farmers that
had access to maize markets, an increase of 15% for small farmers and 24% among large
farmers, confirming an increase in maize marketing afier liberalization (Karanja, Jayne

and Strasberg, 1999).

26

2.5 Agricultural Inputs, Markets and Prices

2.5.1 Land

Of Kenya’s 57 million hectares of land, only 19% is considered having high and medium
potential for agricultural use, and another 12% is good for livestock production (Kenya,
1998). The rest is arid and semi-arid land. Previously, agricultural production activities
occurred only in the high potential regions using highly mechanized technologies on
large farms averaging more than 50 hectares. But after Kenya’s political independence in
1963, many large farms were bought and subdivided into smaller land units as part of a

massive land re-distribution and settlement program.

As population increased, demographic pressure and customary bequeathing of land led to
further subdivision of land. Consequently, per capita arable land has declined, especially
in high potential regions, causing an increase in migration and settlement in fi‘agile and
marginal land. This trend exacerbates land degradation and loss of agricultural
productivity. Table 7 shows this trend in per capita arable land, which has declined by
more than 60% between 1970-2000 as a result of a rapid increase in rural populations and
lack of sufficient arable land. Therefore, Kenya’s future increase in maize production will
have to rely more on higher yields through land-saving technologies rather than land area
expansion. This entails reliance on availability and use of new and better technologies

that raise productivity or reduce crop losses, while maintaining the natural resource base.

27

 

 

 

 

 

 

Table 7: Trends in Per Capita Agricultural Land Use, 1960-2000
Year Cultivate Land Agricultural Per Capita
(‘000 Ha) Population Agricultural Land

(‘000 People) (Ha/Person)

1960 3900 7321 0.53

1970 3945 9858 0.40

1980 4280 13674 0.31

1990 4500 18728 0.24

2000 4520 22683 0.20

 

 

 

 

 

 

Source: FAOSTAT (2001)

28

Because of limited irrigation capacity, complementary, low-cost technologies that
improve soil fertility and water retention or improve drought tolerance can greatly boost
agricultural productivity. Alternatively, the limited irrigation potential can be used for
high-valued crops to free up land in rain-fed regions for food crops such as maize while
generating money to import additional foods. However, transformation of land through
additional investments in irrigation may raise the value of land rental and purchase prices

(Renkow, 1991).

Kenya’s land market was very active after independence with large but differential
transfers of land by farming enterprises and region. But low savings and lack of long-
term credit opportunities has constrained land purchases. Ofien, land prices were
distorted upwards by government subsidized loan programs that were more accessible to
large farmers and people with higher social status. This hiked up land prices above the
market value and made it difficult for poor people to access land. Moreover, the demand
for land is depressed by reduced savings, which have to be accumulated over a long
period. Also, the supply of land tends to be constrained by customary land tenure (Migot-

Adholla, et al., 1991; Lyne, Roth and Troutt, 1997).

However, when compared in terms of costs to other land settlement programs in the
world, Kenya’s land transfer system was found to be cheaper and a better means of
achieving equitable distribution of land resources, job creation and increasing overall
demand for labor. Land exchange occurs in a variety of contractual modes ranging from

seasonal rental to permanent transfer of ownership. Payments can precede or lag the flow

29

of use of land, thus forming an inter-linked land and credit contract. Payments can be
fixed in advance to be dependent upon the output from the land, forming an inter-linked
land, labor and insurance contract. But none of these is a dominant practice in Kenya.
Most holdings are acquired through inheritance of family land. However, because of
limited cultivated land expansion, there has been an increase in incidences of renting
land, especially from non-resident farmers. The scarce data available on this suggest little
evidence on the relationship between land rental and land area but small farmers tend to

rent more land than large farmers.

Ofien, those who rent land are poorer than those who lease land. Since rental fees are
typically paid at the start of the contract, and a cash-flow problem is created by
cultivation of the rented land, it is normal for tenants to be in net receipts of credit.
Recent ethnic conflicts and land invasions in Kenya and Zimbabwe, and civil conflicts in
Rwanda, Somalia and Burundi have tended to distort land markets and raise the risks and
transaction costs of land ownership and farming. On the other hand, land adjudication
and the issuance of title deeds have positively affected land investments and, in many
cases, increased adoption of “lumpy” agricultural technologies and leading to improved

land productivity.

30

2.5.2 Labor

Agricultural labor markets in many developing countries are less structured and, unlike
industrial labor markets, selection of quality labor, supervision and investment in skill
accumulation are less applicable. In general, basic skills are developed within the
household or village. Quality variation between hired workers might be less significant
and better known by employers, the latter frequently being close neighbors. Also, poor
performance that may be instantly noticeable in an industry may be drowned by the
combination of long lags in agricultural production function and the large random
variations in productivity due to climate, diseases and pests. Moreover, specialization and
division of labor is generally less exploited, especially in smallholder agriculture, as
compared to industry. But most of all, poor enforcement of labor contracts, where such

exist, has kept agricultural labor inefficient and unorganized.

Since many large farms operations in Kenya tend to be capital intensive, save for non-
mechanized tasks such as harvesting maize, this section will focus on labor transactions
in smallholder farming. The most striking fact about labor utilization in smallholder
farming, even in the most commercialized commodities, is that hired labor forms only a
small fraction of the total labor use and much of it is used to meet seasonal labor
requirements rather than offset permanent differences in land/labor ratios between
households. Of the hired labor used during the year, it is estimated that two-thirds of it is
casual labor while a third is regular labor. Also, contrary to conventional wisdom, there is

virtually no difference in the mean daily earnings between the two groups of labor.

31

Cross-sectional data used in this study reveals little variation between wages in different

regions.

Hired labor in small farms is regionally differentiated is not surprising considering that
most of it comes from casual labor hired by the day from the immediate neighborhood.
Language barriers, lack of information and distance preclude long-distance labor
migrations. Thus, intra-rural regional labor migrations are confined to employment on
large plantations or estates, which are fewer now than years ago as a result of land sub-
division. This was observed and confirmed with the data collected in this study: there
were hardly any significant labor migration patterns between different villages, except for
regions close to plantation farming such as parts of Moist Transition and Moist

Midaltitude zones.

Labor sales data show evidence that, with dwindling farm sizes and increasing
agricultural populations, a larger proportion of hired labor is from neighboring small
farms rather than from rural landless laborers. Such labor, however, is merely a seasonal
exchange and not large enough to even out the disparity in factor proportions between
different farm categories. Moreover, the extent of small farm labor sales on non-small
rural labor market is still limited by fewer large farm operations so that the only viable
alternative for small farm labor is the informal “Jua Kali” rural enterprises, which in turn
are constrained by lack of credit, technology and market information. Access to jobs in
the formal sector in rural areas are limited and rationed by educational credentials while

access to self-employment opportunities commonly require skills and finances, both of

32

which are extremely scarce in rural areas. The only other alternative is employment in the
informal sector in urban areas, which has been the main source of employment for rural
youths escaping the consequences of rural unemployment. But urban wage rates have
declined due to a high rural-to-urban migration and the current economic slump (Kenya,

2001).

Table 8 presents family and hired maize labor use and cost by zone and household type.
On average, small farms are twice as labor intensive and use three times as much family
labor on a hectare of maize as large farms. On the other hand large farms use 1.5 times
more hired labor than small farms. There are, however, significant inter—zonal variations.
For instance, in the Lowlands and Moist Midaltitude zones, small farms only used 28%
more family labor than large farmers, whereas in the Moist Transition and Highland
zones, the proportion was 93% and 83%, respectively. In general, large farms in these
two zones tend to be highly mechanized and so use less manual, especially family, labor.
Of specific interest to this study, farmers in marginal areas use more than 20% on average
more total (both family and hired) labor on maize farming than those in high potential
regions, mainly because maize technologies in the former regions tend to be more labor

intensive.

On the allocation of labor to maize relative to other agricultural activities, small and large
farms used about 58% of their family labor on maize. However, there were differences on
hired labor: small farms allocated 26% of hired labor to maize compared to 47% by large

farms. There was little inter-zonal and intra-zonal variation on these, except in the Moist

33

$5... a” gamma Hts—3.. E... 3% 5.8 3. 5...... mmuo 2:— No.8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7a.... .=»_1e8=:u_Ne..8 I. _. 5.8....»— No.5»
5.3.522 55. 333 DJ. gown» >= Noam
Pet—nan 23:52:? 33:53.? 5.5.552. Hun—.32.... Ema—2::
ZEN.“ 355. C32. ZOE»)?
was: 3...... N3 3o SN :0 Eu 3.. .3
Evan we... .3 an .8 - 5 No mm.
39.8 $3.. have: EE$~
man: 35: N. B Nu Nm wu G Mm
Penna 3...... 3 mm on - .5 No 3
gas was. .82. 76...»):
was: 3...... NAM .8 .9. .3 3a .3 EN
Foam.“ 35... .3 50 N5 - mo .3 co
Gem gumo ~98. 5%
m8»: m3... 3b who am... mwb mo. Sb mum.
henna mun... umb .86 mum - «PM 2.... a...
.80 8.50 ”in. 5%
was: man: 8.”. oo. um 30 Sb .3.» mm...
Emma 35. 3... 3o 3 - 3.. m9“. qmb
.x. 3.55. 5&2 Guam 2. ZEN».
m5»: mad. as MN 3 me u» ac mm
gnu man... mm 3 mm - «A MN mm
.x. wanna have 9.2. o: ZEN.”
m3»: 3...... .o No N; we w. nu ma
Ema mun: NN um um - mm A.» 3

 

 

34

Transitional zone. Likewise, there was little differential between the wage rate paid out
by small and large farmers in 1998 and 1999. In 1998, large farmers paid more than small
farmers, on average, by about Kenya Shillings (Kshs.) 6.40 per manday (MD). In 1999,
the difference was Kshs.9.60 per MD.7 The observed inter-zonal wage differential was
inadequate to compensate for any perceived or real transaction costs of migrating; and (2)

there exists a high information cost to determining the wage differential.

2.5.3 Fertilizer

Soil nutrient depletion is widespread in Kenya (KARI, 1998). In the wake of declining
per capita agricultural productivity, one of the nagging issue is how to manage soil
fertility and improve crop production (Heisey and Mwangi, 1997). Because of land
scarcity, dealing with declining land productivity may involve intensification of
production through use of fertilizers and other land-augmenting technologies. Past
experience indicate inorganic fertilizers hold great potential for nutrient replenishment

and productivity-enhancement (Naseem and Kelly, 1999; Omamo and Mose, 2001).

The adoption and use of fertilizer and hybrid maize was instrumental for past increases of
maize productivity in Kenya (Hassan et al., 1998; Karanja, Jayne and Strasberg, 1999).
The decline in real price of, and concomitant government subsidy on, fertilizer between

1965-80, and expansion of fertilizer marketing channels through recruitment of fertilizer

 

7 One man-day equals the amount of labor from one adult working 8 hours a day.

35

traders in almost every local trading center, contributed greatly to increased access and

adoption of fertilizer.

However, foreign exchange constraints and input market liberalization accompanying the
structural adjustment program of the late 19805 and early 19905 led to substantial
increase in fertilizer prices and subsequent decline in fertilizer use (Hassan and Karanj a,
1997; Argwings-Kodhek, 1997). In fact, whereas increase in land productivity between
1970-90 was due to increased use of hybrid maize and fertilizers, and supportive
government policies, the decline in productivity between 1990—95 is attributed partially to
a reduction in the use of fertilizer and hybrid seed (N yoro and Jayne, 1999). Indeed,
Table 6 suggests that there was a marginal decline in hybrid seed use and the level of
fertilizer application. This, in part, was because input prices more than doubled after
maize market reforms. Other constraints cited include poor infrastructure, lack of credit
and markets, and high information and learning costs associated with specific agronomic

recommendations (Heisey and Mwangi, 1997; Omamo and Mose, 2001).

Although the rate of adoption of fertilizer remains at about 58%, almost equal to the rate
assessed in 1992, the level of application declined. Increases in the relative input-output
price ratio, increased marketing costs and weather variability have left farmers more risk
averse on fertilizer use, even in the high potential regions where maize response to

fertilizer is highest. Further, use of fertilizer has become increasingly risky as incidences
of poor quality seed and variations in onset of rains increase, resulting in a low fertilizer

demand in both marginal and high potential regions.

36

Since most of the fertilizer is applied before or immediately after the rains, a failure or
false onset of rains can actually lead to greater yield losses with than without fertilizer
use. Also, since the outlay for fertilizer is in cash, use of adequate amounts of fertilizer
becomes a hard decision for small, resource-constrained farmers, who have no access to
credit. Unless fertilizer prices are brought down through improvement of the marketing
chain and rural roads infrastructure as well as the establishment of a competitive private
sector transport and trading regime, the fertilizer revenue-cost ratio will not be attractive
enough to stimulate adoption and use to the desired rates for increased maize
productivity. The increase of private traders, ranging from specialized large-scale
importer-distributors to diversified small-scale retailers, is a promising sign of an

emerging competitive market.

2.6 Maize Marketing and Consumption

Because of the strategic importance of maize in Kenya, the government has literally
controlled both domestic and international marketing of the commodity since 1942. The
government set both producer and consumer prices each year at the beginning of the
planting season, maintained domestic maize movement controls, controlled all external
trade and took care of a national strategic reserve. However, this changed dramatically in
the late 1993 when, under pressure from donors and the structural adjustment program,
the government eliminated movement controls on maize trading, deregulated maize and

maize meal prices, and eliminated direct subsidies on maize sold to registered millers

37

(Jayne and Kodhek, 1997). Hence, the sole government maize marketing agent, the
National Cereals and Produce Board, was reduced to “a buyer of last resort and

custodian of a national strategic food reserve”.

Prior to that, the board facilitated maize marketing by opening maize depots in almost all
trading centers and building large silos in surplus maize producing locations. But due to 1
high maintenance costs of the network and bureaucracy, the board posted net losses for
14 consecutive years (World Bank, 1995). Its operations reflected a long-lasting
government policy dilemma of maintaining profitable producer prices and affordable
consumer prices that became fiscally unsustainable and led to its succumbing to maize

market reforms.

The reform process was expected to reduce maize marketing costs by encouraging more
private sector participation and enhancing market competition. Instead, the process has
been slow, frustrating and marked by a series of advances and reversals, especially with
regard to private sector participation (Nyoro, Kiiru and Jayne, 1999). The resistance by
the government to let go of maize market control was partly based on the premise that
after liberalization maize producers and consumers will be exposed to predatory practices
of private traders and discomfort that leaving maize to supply and demand market forces

would be a national food security risk.

Decades of market domination by the NCPB lefi a relatively underdeveloped private

sector grain trading capacity. No wonder, after maize marketing was liberalized in 1994

38

and the government pulled out there was such a vacuum in the market that producer
prices plummeted by 50-70%, which was good for consumers but not producers.
However, over the years, an increasing number of private traders have come into the
market. They still complain of harassment by police, reminiscent of the previous
controlled maize market era, but they are getting more engaged. But because of credit
constraints and poor infrastructure, their ability to purchase and move more maize is

constrained and, as a result, farmers are frustrated about the changes.

Specifically, in the 1998 maize harvest season, there was a serious lack of marketing
outlets. There had been a flooding of the local maize market with imports just before the
harvest season that severely depressed maize prices and farmers were unable to sell their
produce. Local traders could not offer a better price and the government could not, due to
budgetary constraints, step in and purchase surplus maize from farmers. The overall
result was a huge loss for farmers, which caused many to consider reducing their
production in the subsequent years, unless a long-term solution was found to what they

saw as an emerging “perennial problem”.

Karanja, Jayne and Strasberg (1999) found evidence that the process of maize market
reforms may have reduced maize productivity by depressing producer prices and inflating
input prices but Jayne and Argwings—Kodhek (1997) and Mukurnbu and Jayne (1995)
noted that the reforms conferred substantial benefits to consumers through lower
consumer prices and maize processing costs. Therefore, the aggregate effects of maize

market reforms need to be netted out between the gains to consumers and losses to

39

producers. Such an analysis must consider that most of the Kenyan maize producers,

particularly the small farmers, are also maize consumers.

Table 9 shows the per capita maize produced and consumed, the proportion of farmers
who are net sellers and the proportion of maize sold, the average maize selling and
purchasing prices and the market system preference shown by farmers. On average, large
farmers produced 26 times more on per capita basis than small farmers, but this
proportion varied by zone and region. Similarly, large farms consumed about 2.5 times
more maize, on per capita basis, than small farmers. Looking at the difference between
per capita production and consumption of maize, it is evident that only farmers in the
high potential regions produced considerable surplus maize for sale. Most of the

households in the marginal regions were net buyers of maize.

Table 9 also shows that fewer farmers sold their maize to the government marketing
board, the NCPB, in 1998 compared to 1992 with only 3% of the farmers selling in 1998
compared to 19% in 1992. More small farmers (88%) sold to private traders compared to
large farmers (71%). Most farmers sold their maize at harvest, when the selling price was
low, and bought latter at a higher price. In fact, the average selling price in 1998 was
Kshs.11.40 per Kg while the average purchase price was Kshs. 15.05 per Kg, a
differential of Kshs.3.65 per Kg or 32% of the selling price. Expectedly, maize prices
were lower in the high potential regions compared to marginal regions, with the
differential between producer and consumer prices being higher in the latter at Kshs.4.05

per Kg compared to the former at Kshs.3.21 per Kg.

40

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

H.950 w. H2. 0» .3 .522. $22.53.... 2... 02.2.... 32.. 2... 32.8.... a 3...... m2. 2:. N2...
.52. .3. ~22...»— N2.8 I. .. ‘22:...— N25“
5.2.522. OJ. .522 5.... .522 >= N652
Pei—unnu 33.5.2...” 33:52.30 dun—5.22. H3332. mum—Luau
F: 03.8 KEN.“ 509.32.. Wm“
was: men... .8 o. .A. me Now uau .m.
Ema 3.3. .Nm 8 Now 9.3 mom» ~«mu.
..2. 0228 2.22. OoBEdgo...
Km” 3. .8 .NA am .3 .8 .8
m8»: .uwnu .8 mm .2 - #3 “.3 New
beam.” .22.
.x. 22 M253. em Kenn.
was: mad... 3 3 mm MN .3 an A.
Ema 3.3. e um um - 0.. 0A .3
332.30.. em 73qu moE A2...”
was: 3...... o .N .m .N mm u. no
Ema mun: o N. we - mo .3 m...
..\o 955m 8 .0158 $2.2!
MB»: 3.3. o 3 mm 3 mm 3 mm
.tmnmo 3.3. o .8 ac - um mm q.
:22. mm.....m .018. agmu
m8»: .42... Boo 0.3 3.3 3.3 .93 .93 ...mo
Sam... was... 3.5 :um .98 - Pmo 5.8 2.8
7..qu 1525mm vino. .Amrham.
was: $3. 5.3 3.8 3.3 3.3 3.3 Ebb Gum
rwnmo .33. .98 3.3 3.8 - ...mo .98 .~.d
$2.02 0.52: 32.8. $.22...
ax... um mm H... .3 Am 3 mo
man: 55.. 3 mo Nu - u. 3 mo
Emma v.22.

 

 

 

 

 

 

 

41

Generally, large farmers preferred the former marketing system before trade
liberalization whereas small farmers showed no preference. Most farmers were
dissatisfied with the low producer price at harvest time caused by large maize imports.
Also, lack of on-farm storage facilities meant that they could not take advantage of the
seasonal price fluctuations. Small farmers experienced problems under both systems, the
major ones being delayed payments in the former system and lack of reputable markets in

the current one.

42

CHAPTER 3

3.1 Impact of Technological Change

Estimating the relative poverty reduction potential of alternative research activities
requires demographic assessment of who the users are, their locality, economic status, the
types of income-generating activities, ways in which specific technologies affect their
incomes and welfare, and how markets distribute output gains between different regions
and households. Existing evidence on these issues is mixed (Renkow, 1993). Therefore, it
is hard to justify any a priori rationale as a satisfactory basis for research investment

without a thorough evaluation.

However, empirical evidence suggests that research costs are greater and the time lags are
longer in achieving equivalent output gains in marginal areas compared to more favorable
ones. For instance, in wheat and rice, genetic gains due to crop improvement for irrigated
and high rainfall environments have tended to be higher on an annual basis than they
have been in drier environments, and this difference is evident from experimental results
to farmers’ fields (Byerlee, 1996). Moreover, relatively more of the gains from research
in marginal environments are likely to come from crop management research, which is
more location-specific, but tend to be more costly than crop breeding research (Edmeades
et al., 1998; Byerlee, 1994; Pingali and Heisey, 1999). However, if the bulk of past

agricultural research investments have been made in favorable areas, incremental rate of

43

return on investment may decline to the point where it competes with investments in

marginal environments (Byerlee and Morris, 1993).

Empirical evidence is mixed concerning the welfare effects of improved technologies.
Renkow (1993) found that investment in wheat research in favored, irrigated areas of
Pakistan yielded greater overall income growth than investment in rainfed regions. In
contrast, Fan and Hazell (1999) found that investment in improved technologies and rural
infrastructure positively contributed to productivity growth and rural poverty reduction,
with effects in rainfed marginal regions being greater than in high potential irrigated
regions. Thus, in this case, public investments in the marginal regions had better returns,
enhanced productivity growth and reduced poverty. These two examples, and others,
indicate that one cannot generalize on the effects of technological change since they will
tend to vary from country to country and region to region, and will depend on specific

characteristics of the regions, technology and the economy.

Nevertheless, the impact of agricultural technology on poverty alleviation is a source of
controversy. Proponents point to a large body of evidence showing that R&D has been
instrumental in introducing agricultural innovations that have raised agricultural
production, stimulated economic growth and helped poor people through lower food
prices and higher incomes (Lipton and Longhurst, 1989). Evidently, agricultural
technologies have helped food production grow faster than population growth, thus
avoiding the widespread food shortages and catastrophe envisioned by Malthus

(Plucknett, 1991). But there are concerns that research-led technological change in

44

agriculture has favored wealthy farmers in high potential regions at the expense of the

poor farmers and marginal regions (Freebaim, 1995).

Wealthy farmers are considered more likely to become “early adopters” of new
technologies since they have better access to information and credit, and have more
capacity to take risks. Similarly, some technologies may be too “lumpy” and may be
profitable on large and not small farms. Benefits of such technology may also differ from
one region to another. If it raises productivity in high potential regions but not in
marginal regions, the increase in overall production may cause prices to fall, if in a closed
economy, such that farmers in marginal regions are made worse off, thereby exacerbating

poverty.

These competing effects of technological change can be demonstrated by considering a
hypothetical example: supposing that an increase in agricultural productivity has four
impacts: producers have higher output, labor suppliers receive higher wages and better
employment, consumer prices fall and greater economic growth increases overall sales
and employment. Consider also, as could be the case in Kenya, that producers are also

consumers and providers of farm labor.

If technological change causes consumer prices to decrease, as would be the case in a
typical closed economy, this means less income to producers but more disposable income
to consumers. For people who are dual producers and consumers, whether they benefit

from such technological change depends on which of the income effects is greater and

45

whether they are net sellers or net buyers of that commodity. Moreover, the effect of
rising wages on income will depend on whether they are net buyers or net sellers of labor.
For instance, a net seller of labor and net buyer of food would gain unambiguously from
a technology that lowers food price and raise wages. A technological change can raise
incomes of adopting farmers but not of non-adopting farmers, change the demand for
agricultural labor, reduce food prices or dampen food price increases and stimulate
economic growth, thus, generating addition employment and increases in wages. The
extent of these impacts on the overall economy vary greatly and depend on a variety of
conditioning factors, such as the extent of adoption between regions, the indirect impact
on prices of inputs, product and competing or alternative products, and the nature of

markets and government interventions (Renkow, 1993).

Therefore, in evaluating the impact of technologies on farm welfare, the following
variables are important: their direct impacts on output, input demand and input use;
spillover effects mediated through product and labor markets that change relative prices;
the relative numbers of different types of households within a particular region; the share
of farm income in total household income and the nature of the market, whether in an
open or closed economy. It is difficult, however, to state a priori which of these variables
are the most important in determining the appropriateness of targeting one or the other
type of production region. In Pakistan, it was agriculture’s share of household income
that appeared to exercise the dominant influence (Renkow, 1991). In Southeast Asia,

labor mobility and the responsiveness of agricultural workers to perturbations in the labor

46

markets were the key factors mediating the benefits of technological change (David and

Otsuka, 1994).

3.1.1 Partial Equilibrium Analysis of Technological Change

As discussed above, technological change can have variable effects on farm income both
within and across regions, partly because farmers are a diverse population and will
typically adopt the technology gradually and at times partially. Thus, depending on the
rate and spread of adoption, the effects of technological change may be concentrated
within a region or widespread. Further, the distributional implications for different farm
households and regions will depend heavily on policies and institutions that condition the
incentives and constraints that influence the adoption decisions (Mills, 1998). New
technologies impact farm incomes and welfare through the product and input markets.
But the extent and nature of such impacts depend also on whether the economy is closed
or open, and whether the impact is restricted to one particular zone or spills over to other

zones (Renkow, 1 991 ).

3.1.2 Indirect Effects through Commodity Markets

To illustrate the nature of the market effects in a simplified partial equilibrium

framework, consider a maize importing country that has a maize technology whose

impact is limited to a maize net-exporting, high potential region and its marginal region is

a net importer of maize. Figure 3 illustrates the impact of the technology, captured as a

47

Enid u" 5:9w18 geno— 3. finesse—cums: Ora—sun
Em: H.233»— Wommos KERR: 38:3»— Woman

 

 

m.

 

 

 

 

 

 

 

 

 

 

no a.

dag: Woman Zones». 33on

48

 

 

m

 

 

 

 

 

 

 

 

 

 

 

.\\

maize supply shift, on various regional maize markets in an open economy. The price of
maize is determined exogenously and, thus, the expansion of maize supply from s to s' in
the high potential region does not affect the maize price, such that producers in the region
gain unambiguously from increased output at the same price, as depicted by the area
ABCD. Producers and consumers in the marginal region experience no change in welfare
since neither their production nor prices are altered. Urban consumers are likewise
unaffected. At the national level, increased production in the high potential region shifis
the supply curve from S to S', such that part of the national maize demand (Qle) is met,

and imports are reduced from Qon to QIQZ.

In a closed economy, the domestic price is determined by the intersection of aggregate
supply and aggregate demand. A shift in maize supply in the high potential region causes
the aggregate maize supply to shift out from s to s', in turn shifting the national maize
supply from S to S' and accompanied by a decrease in price from P0 to P1 (Figure 4).
Urban consumers gain unambiguously from the fall in price leading to an improvement in
their welfare, represented by area PoMNPl. In the marginal region, the consumers gain
(area PoKLPl) while the producers lose (area Pol 1P1) from the decrease in maize prices,
resulting in a net increase in total surplus equal to the area IJKL. In the high-potential
region, like in the marginal and urban regions, consumers gain from lower maize prices
but the impact on producers is indeterminate because the negative effect of falling prices
is offset by an increase in output. The final results depend on the elasticities of supply

and demand curves and the magnitude of the supply shifi.

49

ENE.» A" Emir—«#58 zan— S. flooraaonmnu— Graeme

P

me
P

Em: 38:3»— ”ammo:

 

/

11>

 

o>\m.
\\

 

 

57/

m m

 

 

can: Woman

 

 

Z
/

 

 

2/

 

 

we
P

we

38m?»— wonosna ”ammo:

 

._ HA

 

_ /

 

./

 

 

 

 

 

232.»— 7358»

 

m.

 

 

 

 

 

50

Such an analysis becomes complicated where, like in Kenya, producers are also
consumers. In such a case, a change in price has both a positive and negative effect. A
supply shift that causes a price drop reduces farm profits and cheapens food price at the
same time, such that the net impact on welfare is determined by whether the household is
a net producer or net consumer of maize. These different effects are depicted in Figure 5
for four types of households differentiated by whether they are adopters or non—adopters

of the technology and whether they are net consumers or net producers.

It is evident that both adopting and non-adopting net consumers unambiguously gain
from a fall in maize prices due to the supply shifi caused by technological change. Net
producers lose from the price decline but for those adopting the technology, the loss in
welfare may or may not be fully compensated by productivity gain. For non-adopting net
producers, there is no offsetting gain in productivity and, therefore, they suffer a net
welfare loss. Net consumers in both urban and non-adopting marginal regions gain
unambiguously from the price decline while non-adopting net producers in the latter

unambiguously lose.

Clearly, the analysis may be further complicated if technology adoption is not uniform
among household groups within a specific region, for example, if the technology is
adopted more by large than small farmers in the high potential region. Differential
adoption may be caused by information asymmetry, differential access to credit,
extension or research, and differences in risk-taking. The effect may also be skewed by

complementary inputs, infrastructure or information needed to adopt the new technology.

51

Ems—d m" Hearse—emmn... 5.32 2. 2.3.2:“ 2:— 2e=->aeu:=n mafia—55m

Zen-

>aova=m Zen 00:25.9.

 

   

 

........................... lr .
.. ___._w________:_:__:_:_::_____z____‘

0.......::.

  

 

29 Out.

 

 

>835 29 Cesar—Boa

 

 

......... b

v: ..............

  

  

 

 

 

 

_____:_____::_____§___._______,

 

 

 

we ...................

w.

Zo:->aouasm 23 $3982

 

      

22 ream

.V///////.///d¢ ...................
‘

 

 

 

558:3 Zn" 309.com

 

 

 

 

 

 

 

52

Skewed product preferences may also contribute to differential adoption patterns between
different households in the same zone. For instance, small farmers focusing on home
consumption may seek certain qualities in a maize variety (e. g. taste, storability) that

commercial large farmers may have little interest on (Kumar, 1994).

3.1.4 Indirect Effects through Labor Market

There is substantial evidence that new, labor-using agricultural technologies tend to

A “(IT‘Ju-l‘

increase the demand for labor (Lipton and Longhurst, 1989; Renkow, 1993). Whenever

 

 

labor supply is less than perfectly elastic, such changes in labor demand put upward
pressure on wage rates in local labor markets, thus affecting incomes in technology-
adopting regions, especially where agricultural labor is also a source of income. The
effects on income may be transferred to the non-adopting region through rural-rural
migration of workers if resultant real wages rise sufficiently to cover migration costs.
Consequently, migration from the non-adopting region puts upward pressure on wages

there as well, benefiting workers who remain behind.

Figure 6 illustrates these effects: initially, equilibrium wage, W0, prevails in both high-
potential and marginal regions. Adopting a new, labor-using technology in the high
potential region shifts the labor demand from d to d’ and raises the prevailing wage from
W0 to W1. The difference in wage rates between the two regions induces migration from
the non-adopting marginal region to the adopting high potential region. The influx of

workers from the marginal regions shifts out labor supply in the high potential region

53

Enid a" 33:8. en flange—on?»— Orsunn e: 5.33 3»..on

Em: H5852»— ”ammo:

 

a a. m

 

’3 /

 

 

.\/
a. As.

 

 

 

 

 

 

<5
cf
<5

33mg”; H3852»: Woman

 

 

./ «w.

 

 

 

 

 

 

 

54

from f to f , putting a downward pressure on wages there. On the other hand, the outflow
of laborers fi'om the marginal region puts upward pressure on the wage rate there due to a
decrease or shift to the left of the labor supply curve from m to m'. This continues until a

new equilibrium wage, W2, is established.

Empirically though, there are instances where such effects may not be observed and
wages may stagnate or experience only small changes. This happens where there is a high
level of unemployment or underemployment in adopting areas and high population
growth or migration into adopting areas. Another reason is that interregional migration is
often costly and non-instantaneous. In the short run, labor tends to be immobile but fairly
mobile in the long run if the wage differential is greater than the reservation wage and
related transaction costs. Usually, the complex and dynamic nature of migration makes it
hard to gauge the extent and direction of its contribution to the distribution of

technological benefits.

3.1.4 The Role of Policies and Institutions

Use of farm inputs does not necessarily skew the gains from technological change if such
inputs are adopted and used by all households. Problem arises when the market does not
work well such that the supply is unreliable or when the inputs are targeted to one group
of farmers and not the other, such that only one category of farmers have access to it and

gain from it. The household’s resource capacity is another factor influencing access. Most

55

large farmers tend to be early adopters of new technologies because they have better

access to extension and credit than small farmers have.

Lipton (1989) found many cases where early adopters gained and late adopters were
made worse off, especially where economic policies related to input supply and land
tenure favored one group of farmers more than another. He suggested that supportive
policy reforms were necessary where technological change could be otherwise harmful.
As agricultural economies evolve, market distortions have become an implicit means of
skewing income distribution, quite often towards the large, wealthy farmers. A good
example is the contentious agricultural subsidies in the United States and European
Union, as well as past implicit tax support to large farmers in Kenya (BFWI, 2003; World

Bank, 1995).

Often, governments in developing countries intervene in the agricultural sector through
pricing policies and through public-sector investment projects. Policies affecting the
prices of agricultural commodities are pervasive and can influence production,
consumption, marketing or international trade. The policies may also be designed to
generate government revenues, subsidize urban consumers, secure food self-sufficiency,
earn foreign exchange, improve rural incomes or a combination of any of these. Often,
the government is the principal provider of infrastructure and agricultural support
services such as research and extension. Like pricing policies, such investments can be
expected to have a strong impact on farm production, incomes and welfare of agricultural

households.

56

Self-sufficiency in maize production has been an explicit policy objective of the
Government of Kenya since the early 19605. In order to ensure this objective is met, the
government continued with the colonial policy, started in 1942, of using government-
controlled cereal marketing boards to determine producer and consumer prices, to
purchase and distribute maize grain, and import and export maize. Under the
government-controlled regime, maize prices often fell within an export and import parity

band regulated through maintaining a buffer stock (Pearson, 1992; Pinckney, 1988).

However, these maize policies changed dramatically in 1994 when the government,
under pressure from donors and the Structural Adjustment Program of the International
Monetary Fund (IMF), liberalized maize grain marketing in Kenya. Subsequently, the
role of the marketing board, the National Cereals and Produce Board (N CPB) changed
from complete domination of maize marketing to that of “custodian of strategic reserves
and buyer of the last resort”. In addition, maize price determination was left to market
demand and supply forces and greater participation of private traders in maize marketing

encouraged (Nyoro, Kiiru and Jayne, 1999.).

The details of specific agricultural policy changes and impacts between 1994-2002 are
beyond the scope of this study. Many studies, including Omamo and Mose (2001),
Karanja, Jayne and Strasberg (1999), Jayne and Mukumbu (1995) have investigated
specific impacts of post-liberalized market policies. As a result, several pertinent issues
related to the role of the government are worth noting. Firstly, the maize market reform in

Kenya was characterized by an unfortunate back-and-forth policy reversal by the

57

government that sent conflicting signals to producers and traders, and contributed to

speculation and market uncertainty.

Secondly, there was little effort made to map out a clear reform process that ensured a
smooth transition from a govemment-controlled to a market-controlled system and create
of an enabling institutional framework and environment for such changes. Failure to
establish such a process led to an institutional failure in maize marketing, which may
have resulted in significant maize productivity and welfare losses (Karanja, Jayne and
Strasberg, 1999). Lastly, there are serious institutional and infrastructure impediments to
the benefits of market reforms, and these include public-funded rural infrastructure.
Omamo (1998) and Renkow, Hallstrom and Karanja (2003) demonstrate how high
transaction costs from poor roads and transport systems could hinder rural market

participation and reduce gains from agricultural production.

In summary, government action or inaction affects the level and distribution of gains or
losses to producers, consumers and the economy. Government interventions alter the
range of distribution outcomes and serve to drive a wedge between consumer and
producer prices (Renkow, 1991). The common tendency is to hold consumer prices
below import parity and either subsidize or tax producers. In cases where both consumers
and producers are subsidized, a shifi in supply due to technological change represents a
greater drain on public fiscal resources. But where price policies subsidize consumers and
tax producers, as is often common, a supply shift increases the effective tax on producers

and there is less drain on the govemment’s fiscal resources.

58

When governments have substantial control over producer and consumer prices, as in
Kenya before 1994, the prices may be regarded as fixed in the short run such that the
supply shift is not accompanied by price changes. This is similar to the open economy
scenario where the prices are set exogenously. In the long run, however, the prices
change as a result of various things, which may include political pressure or government
constraints. The government also has other intervention options that can influence the
distribution of gains or losses from technological change that are beyond the discussion
in this study. However, the next section outlines the analytical framework chosen for this
study, which also includes assessment of the effects of potential government

interventions.

3.2 The Multi-Market Model for Technological Change

3.2.1 The Analytical Framework

A multi-market model is used to analyze the impacts of maize technological change on
farm incomes and their distribution in Kenya. The choice of the model is influenced by a
need to evaluate the differential effects of the technologies among different types of
households, some of which are net producers and others are net consumers, and in
different regions. The model is implemented in six major maize growing regions and
considers also an urban sector. In each maize growing region, the households are split

into two groups, based on farm size: small and large. Production of two commodities is

59

assumed, for simplicity, in each region, r; Q is the output of maize, and Q2 is the output

of an alternative commodity.

Technology is modeled as an exogenous shift variable, T, labor (L) and fertilizer (F) are
assumed the only variable inputs and land (2) is considered a fixed input, at least in the

short and medium term. Output supplies and input demands depend on the prices of the
two commodities (P1, P2), the prices of the variable inputs, W and f for labor and

fertilizer, respectively, and, except for Q2, the technology shifter, E.

It is assumed that variable inputs are not differentiated by crop and that regional wage
rates are endogenously determined in competitive labor markets, an assumption that is
later dropped. The regional labor markets are assumed to clear in isolation due to labor
immobility. The respective output supply and input demand equations, in rate-of-change

notation, are represented by:

erh=gll’hPI+812th2+81LrhW+£lprhf+Elrh

Q2»: = 32’” PI + 522”: P2 + 82L». W + 62m f

.d _ A .. “ “ .
L r]._flL/thl+flLZth2+flLLrhW+flLFrhf+ELrh

fir]: : 16F”). 131 + flFZrh P2 + flFLrhW + flFFrh f + EF’”

60

where the subscripts r and h denote region and household type, i=1, 2; j =F,L,° k=1,2,L,F;
aim. and Bjkrh are region and household group specific elasticities of output i or input j

with respect to price k, and 12*,” is the exogenous proportional shift in output supply or

input demand over time due to the new technology holding fixed inputs constant, and ""
implies “rate of change’. The changes in the total regional output are given by the sum of

group-specific output changes weighted by each group’s share of total regional output

(Kirk):
Q" = g lqumJQz, = 2:, lzrhézrn
Likewise, changes in input demands are given by:
L“: = 2":5rI-ijhi and fird = § ‘1’». Fri

where 5,}, and VP”. are the shares of total regional demands for labor and fertilizer

accounted for by household group h.

Individual rural labor supply, 1:, , is considered a function of real wage rate, w = W/ 1):, ,
where P::. is an endogenous, group-specific consumer price index. Denoting the

population of household group h in region r as N,;,, group-specific labor supply is simply:

61

3 _ 3
Lrh—lrthrh

Letting gm, be the wage rate elasticity of household labor supply, proportional changes

in regional labor supply are given by:
13:}. = 8m. W + A7,].

This can be aggregated to derive changes in regional labor supply as:
L”: = 2.: w. 13..

where um, is the share of labor supply in region r accounted by household group h.
It is assumed that all residual farm profits not attributed to variable inputs accrue as
returns to land. Therefore, letting II”. be farm profits of household group h in region r,

then:

Uri! = Per + P2 QZrh ‘ Wth ’ fF,;,

Changes in farm profits are, therefore, given by:

A _ A A A A A Rd A “
1'1”, - arm. (P, + Q,,,,) + 7:2”. (P; + QM) + In». (W + L,,.) + mm. (f + F,,.)

62

where am. is group-specific profit shares accounted for by outputs and variable inputs
(positive for outputs and negative for inputs). The household output demand, Dim , is a

function of the prices of these commodities and group-specific nominal income (Y,,,):
Dirh=Dirh(PI:P2rth) i=122

On the other hand, group—specific consumption, Cirh, is the product of household output

demand and the group population grth in region r:
Cm. = Drrh X th
Group changes in consumption of both commodities are given by:
(in... = 77,... 13, + 77.2.. 132 + my... in. + All.
C}... = 772,... P, + 7722,. 132 + 772m. in. + 19..
where mm, is the group-specific cross-price elasticity of demand for commodity i with

respect to the price of commodity j. Letting am, be the share of regional consumption of

good i accounted for by group h, changes in regional consumption are:

éir=2ai’héirh i=1'2

63

Group-specific nominal income within a region, Y, , is the sum of the net returns to
factors rented out by that group. These factors included labor income, farm profits, and

other exogenous sources (X) such as non-farm labor:
th = WLih+nrh+th

Letting um, denote the group-specific share of income attributed to income source k,
where k = L, 1'1, X, and assuming no change in the distribution of land holdings, changes

in real income (y = Y/Ptrh) are given by:
JAG}. = I” Lrh( W +£ih)+1unrh fir}: + ”XX‘II ' 16:}.

where PL. is a region/group-specific consumer price index. Changes in this price index are
given by

Prhzzwi’hpirh 1:1’2
i

where mm, is the group-specific expenditure share for commodity k.

64

3.2.2 Closing the Model and Simulation Scenarios

Closing the model requires specification of the conditions under which maize and labor
markets clear. Table 10 summarizes the basic characteristics of two sets of analyses done
in this study. The first is a fixed-price model, in which maize prices are exogenous and
labor is assumed mobile so that it clears nationally. In addition, imports are endogenous
and are used to bridge the gap between national demand and domestic production. This
closely represents the pre-1994 period in which the government set both producer and
consumer prices of maize according to its desired price policy. Alternatively, this may

also happen when the world market determines the maize prices.

The second represents a flexible-price model in which domestic maize prices are
determined endogenously by demand and supply conditions, similar to a closed economy.
Labor is, as in the previous model, considered mobile so that it clears at the national
level. However, this model is set up to allow some potential influence by the government
through exogenous imports, presumably as part of its trade policy. Therefore, this closely
matches the current system in which the maize market is liberalized with respect to
internal trade and the government covertly influences maize trade by indirectly

sanctioning large maize imports that often depresses maize price at harvest time.

65

Table 10: Multi-Market Model Characteristics

 

 

 

 

 

 

 

 

Parameter Fixed-Price Model Flexible-Price Economy
Maize Market Open Closed

Maize Price Exogenous Endogenous

Regional Labor Market Integrated Integrated

Maize Imports Endogenous Exogenous

 

66

 

These two models assess long-run impacts of technological change and, thus, assume that
labor markets are sufficiently integrated to allow wages to clear and be determined at the

national level.8 The labor market clearing condition is represented by:

 

 

The maize market clearing condition for the two models is based on the identity

C1 = Q1 +G, where C1 is the total national consumption, Q; is the total national
production, and G is the quantity of govermnent-influenced maize imports. In the fixed-
price model, G is endogenous and makes up the difference between production and
consumption, either through net imports or alteration of government buffer stocks
whereas in the flexible-price model, G is exogenous. Therefore, the rate-of-change

notation, the identity becomes:
%é,.=(1-r22%é,,+ m

where F=G/C1 is the share of imports in the national consumption. Second-round effects

are not estimated for any of the models

 

8 However, this ignores transaction costs of moving from one region to another.

67

3.3 Parameter Selection and Share Computations
3.3.1 Production and Consumption Shares

Inter-regional production shares are computed using the 1996-98 data on area and maize
production from the Department of Remote Sensing and Resource Surveys while intra-
regional production shares are computed using the following formula and the1998 farm
survey:

Qi aiWryi

Q 2 army.-

 

where Q=maize output for household 1', Q=maize output in maize zone, a,=share of total
cropped area farmed by the household 1', wi=share of maize in total cropped area for
household i and yi=ratio of average yield for household 1' to the average yield in the maize
zone. The data was transformed from district to zonal data using a GIS classification

template developed by KARI (Mills, 1998).

Table 11 presents the production and consumption shares of maize and the alternate
commodity by zone and household type. The alternative crops were selected based on
farmers’ preferences and corroborated by regional crop production data and assessment
from Kenya’s agricultural extension offices. Farmers in the Lowlands zone chose cassava
as their most prominent alternative crop to maize. The zone accounted for nearly 30% of
Kenya’s cassava or about 140,000 Mt between 1996-98. The crop was also the alternative
crop chosen by farmers in the non-urban Rest of Kenya (ROK) zone, although it

accounted for a small fraction of the national total production.

68

He...» .7 Someone... Doug—=33... use... 62.5.... »s: wow—.53.... 9.3.8 duo.— .5 :5 2558.28

 

3....«3... 5.8....»— N38

an: .58....»— N38

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5.2.588. DJ. 30.8. DJ. 3...:

Pei—2...“ 3.9.5.53 Zia—:3...“ 5.2.8.3.... €358.33 Eur—2:...
Zine 5698...... Ame—5:51.35 mas—.8.
ax. 3. Heme—.18.»:
o m3»: 538 PM 3.8 Pm 8.3 0.. A. .5 wb Ga. .mb 2...... a... Sub.
o momma 35.8 o. 8S a... 3.3 m... 3.3 PM 8.6 N... 3.3 ...~ 0.8
22735.8 5.2—5...... Adena—.533...
9.2.8.1}. a. 5.33.... 8...:
o 3.83:5 03.. 0.5.8 0833. DJ. woe; 3...... <55». $2.8. <55...
c was: 3.3... 3.15.3 .9... CS3 3.15.8 ab 8.8 A... Q... .md 3....
o momma $5.8 9a 3.3 PM 3.3 ob 3.3 .mb 3.8 .9. A. .3 am... 3.8
Hrs—.2. 52.5.... Ame—5.5 2.5.8..
fix. A... 8...:
. m3... 3.3.8 P. 3.3 .9. Cu... .Nb Cu... 9. 0.3 3... BF... Bu Dub.
o .tmnmn 3.38 .b 2... Po 0.8 gm 2... ob A. .8 Mo 3.3 NM 33
3.1.52. 525...... m......8.a
ax. on 8.... 3.4.52. 325......
o m3»: $5.8 Peg hoax. Pox. 9.x. 3.x. 56$
0 henna madam 99$ 9mg We... Pix. 363 Pmfi
woe—.33.... @338.
fix. 9.. 8.9..
o ma»... 3.3.8 PAR. 2.0.x. 2.9x. Aux 3.9%: 29$
0 Fame 3.3.8 Pix. Nix. ....x. 99%.. who; mix.

 

 

mocaoau a W035 288. a .93....» :83. a >c...o...m 0033.85... 8382.8 matinee: .5 33min. 2... Em: .5853... 8min? 2:. .328 9..
503-388 WOW No.5 3.85 E... 5...... .6825 ANNXV.

69

The alternative crop in the Dry Midaltitude zone was dry beans. This zone accounted for
22% of the country’s total dry beans production. Farmers in the Moist Midaltitude chose
millet as their alternate crop and the zone accounted for about one-third of the total millet
production. Wheat was the choice in the Dry Transition, Moist Transition and Highland
zones, with data indicating that large farmers in these zones produced nearly 70% of all
of Kenya’s wheat. The choice was maintained for small farmers as well since their crop
choices, potatoes and beans, which are intercropped with maize on small areas rather than

substitute for maize.

Maize consumption shares were computed from zonal population estimates and the
KMIS consumption data. First, the population was transformed from district to zonal data
and the average per capita consumption values from the farm survey data used to estimate
the share of maize consumption by zone and household. For the urban households, the
average national per capita consumption level was used. A similar procedure was used to

compute consumption shares for alternate crops.

The results indicated that most of the maize was consumed on-farm by small farmers
while large farmers, mostly in high potential regions, accounted for large marketable
surpluses. Comparison of production and consumption shares reveal that in the marginal
region only large farmers in the Moist Midaltitude zone were net producers of maize; the
rest were net consumers. In the high potential region, only small farmers in Moist
Transition zone were net consumers; the rest were net producers. At aggregated zonal

level, only the Moist Transition and the Highland zones were net exporters of maize, the

70

rest had varying degrees of maize insufficiency. For instance, the Lowland zone met only
32% of their maize consumption needs; Dry Midaltitude (52%), Moist Midaltitude
(84%), Dry Transition (76%) and ROK (54%). In general, Kenya had a maize deficit of
22%, a wheat deficit of 39% but was self-sufficient in sorghum, millets and dry beans

over the period 1996-98.

3.3.2 Input Demand, Input Supply and Population Shares

Table 11 also tabulates shares of labor and fertilizer demand, labor supply and
population. Labor demands were computed from farm level data and aggregated to
household and zone levels. Small farms demanded about 70% of total labor and 48% of
total fertilizer used on maize. Farmers in marginal regions demanded approximately 60%
of the total labor and only 30% of fertilizers. This may be due to a higher labor-intensity
of maize technology for marginal regions. In contrast, the high potential regions
demanded 70% of fertilizer and 38% of total labor for maize production. The two high
potential zones and the Lowland zone had surplus labor while the rest experienced

deficits.

Labor supply shares were computed from population census and distribution of
agricultural households reported in Kenya (1994). Since supply of farm labor for maize
production from the landless and large farms is rare in ma] areas, it was assumed that all
hired labor comes from small farms. The assumption was supported by lack of significant

labor flows between villages. In general, family labor accounted for 82% of all labor

71

supply for maize production, with the proportion being higher among small farms (87%)
than among large farms (73%). Population shares showed that 78% of the population
lived in rural areas in 1998, about half of it the high potential regions and the other half in

the vast marginal regions.

3.3.3 Income, Profit and Expenditure Shares

Table 12 presents profit, income and expenditure shares by zones and household type.
Income shares were computed from farm production, sale of agricultural labor and non-
agricultural activities, reflecting the relative contribution to household income. Farm
profits for the major agricultural production activities, namely the production of maize,
alternate crops and other crops/livestock, were derived by subtracting total inputs cost
from revenues accruing to these activities. Profit shares were then computed as the
proportion of revenue accruing to these activities, with profit shares of farm inputs

recorded as negative profit.9

Maize production accounts for over one third of the total household profits in all farms in
the Moist Midaltitude and Moist Transitional zones, among large farms in the Lowlands
zone and small farms in the Highlands zone. Overall, profits from maize production are
particularly important in small and large farm households in the high potential regions,
and are a relatively insignificant fraction of total household income for small farms in the

Lowlands zone (7%) and large farms in the Dry Transition zone (9%).

 

9 Family labor’s reservation wage was assumed to be half the hired labor rate.

72

H9...» 5" 563.. :8...» a...— muu»...=::.» 2.2.8 :8».— ... 9.553....“

 

Swan...»— w...»..:.._ N38

mm": 5.8....»— N38

 

 

 

 

 

o Cam» 3.3.».

 

 

 

 

 

 

 

5.3.582 UJ. 3e... 63. 3......
rot—258 3595:...» 3335:...» H3558 €238.22. Eur—v.38

ZEN» 5.3.: mama»...
A..\e S. 3...... 1.538 3...... SEN».
o m3»: 3.3.8 99X. .mbX. uo.a.X. NNbX. uA.N.X. uu.m.X.
o ream» 3.3.8 ”EQX. 3.9X. “SEX. PuX. wm.m.X. NPNX.
5.3....“ 5.8...» 9.38.
AX. 3. 5. £8...» 3...... 3...... 2.35
o m3»: $5.8 kBEX. am.m.X. Ao.,\.X. mq.m.X. muLX. .EhX.
. Eqm» mad... 33X. flue} m~.w.X. 33$ 89X. ew.m.X.
r32... :8...» 2.2.8.
AX. e». E. 7.8...» 3...... an .35...
o m3»: 1..an SEX. SEX. N_.~.X. BQX. PNX. QéX.
o rmnm» 3.an w.§X. GNX. AbX. :wuXc obX. odX.
ZEN» aux—.2255.» 9.2.8..
1X. 3. E. 22.2.32...» 2. Sums».
o m3»: 3.3.8 NPGX. GHQ. NA...X. SEX. SEX. SQX.

BSEX. NbX. GQX. APmX. SEX. $.mX.

 

 

mesa»? .. 5.3....» :88“ . W9...» G33

73

The importance of farm income to total household income, as is the share of maize profits
in farm profits, varies widely across zones and farm types. But still, farm profits are the
major source of income for all households except in the Lowlands zone and among small
farms in the Moist Transition zone (about 40%). Among large farms in the high potential
zones, farm profits accounted for over 90% of total household income. In addition to
farm profits, small farm households, especially in marginal regions, obtain a sizable
fi'action of household income from returns to farm labor, with zonal levels ranging from
6-9% in the high potential regions and 10-21% in the marginal regions. Thus, factors that
affect agricultural wage rates, such as shifts in labor demand attributable to diffusion of

new technologies, may have important impacts on the welfare of these households.

Overall, 62% of the rural households’ income comes from farm production, 28% from
non-farm activities and 10% from sale of agricultural labor. These proportions vary
between high potential and marginal regions. In the marginal regions, farm profits
account for 56% compared to 79% in high potential regions. Similarly, off-farm activities
and sale of agricultural labor account for 32% and 12% of household income in marginal
areas while the accounting for 17% and 4%, respectively, in high potential regions. Thus,
ceteris paribus, an increase in the profitability of maize production from technological
diffusion will invariably affect more strongly those households in which maize profits

represent a large share of household income.

Expenditure shares were estimated from the 1994 Kenya Welfare Monitoring Survey

data, which had detailed food and non-food consumption expenditure data and used a

74

similar same sampling frame as this study. For simplicity, household food and non-food
preferences were assumed not to have changed significantly between 1994-98, and if they
did, the changes were assumed to have occurred proportionately across zones and
households. On average, maize accounted for 19% of the total household expenditures,
including for urban households. There was significant inter- and intra-zonal variation,
with the largest expenditure shares being in large farms in the Lowland zone (47%) and
in the Dry Transition zone (50%). Households with the largest maize expenditure shares

would be affected the most if there is a huge price change caused by technological shift.

3.3.4 Maize Technology Shifter

Mills (1998) estimated the measure of technology shift used in this study as two distinct
processes: (1) technology development; and (2) technology adoption. The technology
development process is viewed as a triangular distribution of possible yield gains or
losses, adjusted for incremental input use or savings, to generate the expected net yield
gain, which was computed from: (1) the probability of exceeding the net yield gain
dissemination threshold; and (2) the expected net increase conditional on the

dissemination threshold being exceeded.

Figure 7 depicts the distribution of expected net yield gains. Letting K represent the net
yield gain of an innovation, the minimum net yield gain is K], the most probable net yield
gain is Km and the maximum net yield gain is Kh. The minimum net yield gain necessary

for an innovation to be released for dissemination is K“, (a 3% gain in this example). For

75

E E.» q. 32.2...“ a.» 56.33:... 5.8.1.252. on ”82.3.. 658.38

 

256:5.
.u..0....m.U=..< 3A.

3056 N” 4.30 Uinzoczo: om mxumonma Zm. <6... 023m

 

 

 

O:33c.m:<m

IX.

 

 

 

.mo\o

Vacuum—0:5. mOo\o

 

.OOo\o

 

0. <5... Om...

20. <55 Om...

 

mesa». 2...; :83

76

 

every K there is a corresponding probability f(K) which is assumed to follow a triangular
probability distribution. The probability of achieving Kal , Pr(K 2 K“), is given by the
cumulative density function. In the example, the probability of a net yield gain from

research above 3% is quite low, approximately 15%. The expected net yield gain is

simply the expected value of K, conditional on Ka being achieved: E[KIK 2 K3 ].

For a triangular probability density fimction the cumulative probability of producing an

innovation with a net yield gain above K“ is:

 

 

(k*-k()2 .
Pr(K 2 K*)=1— :fk, s k* < km and
(kl. ’19ka —k1)
._ * 2
Pr(K2K*)= 0‘” k ) ifkm sk*<k,,.

(kl. - k1)(k;. — km)

The expected net yield gain, E[K], given the threshold value for dissemination is reached

can be calculated as:

 

 

 

 

 

 

 

 

 

K... K.
{ [ 2 )6/3K’ — 1/2K2K,)+[ [ 2 )Q/zKZK. -l/3K’)

K. (K. - K.)(K.. - K.) K... (K. - K.)(Kh - Km)

E[K|K2K*]= K... K.
2 2 2 2
(1/2K — K.) + (K. —l/2K )
[K* (K. - Kn)(K.. - K.) [Km (K. - K.)(K. - K...)

For KI S K“ < Kmand

K.r

2 )6/2K2K. —1/3K’)

K.((K. —K.)(Kh-Km)

E[K|K 2 K‘] =- "K.
2 (K. — 1/2K2)
.K“ (Kb - Kl)(Kh — Km)

For Km 3 K“ < Kb.

77

This net yield gain is contingent upon the rate and extent of technology adoption.
Therefore, the adoption profile is estimated using a simple, linear approximation,
represented by Figure 8. The profile is envisaged having a trapezoidal functional form
with a: research development phase (A); increasing adoption phase (B); adoption plateau

phase (C); and a declining adoption rate phase (D).

The final adjusted net yield gain is the product of the estimated probability of
dissemination and the conditional expected net yield gain. Table 13 presents results of
this analysis, with net yield gains presented in percentage form, and generated by
research themes - breeding, crop management and technology transfer - and by zones
used in this study. The net yield gains are summed up across the themes to generate an
estimate of a maize supply shift used in the model simulations. This shift ranges from

about 7% in the Lowlands zone to about 30% in the Moist Transition zone.

The projected yield improvement in marginal regions is largely attributed to change in
crop management practices, whereas breeding research has greatest relative impact the
high potential regions. The implications of this is that labor demand due to technological
change is higher in the marginal regions (11%) and lower in the high potential region
(6%), since agronomic innovations are more labor intensive. Estimated fertilizer demand
associated with technology adoption in both marginal and high potential regions is not
significantly different because the former hardly use any fertilizer while the latter will

probably use just as much as their former levels.

78

Ems.» m. Zia» H»»....e€..e..ae= 5......»

2.2.88

 

.OOX.

X. o. 38:22

moX.

9X.

20....» u. don-50.03 2.2.20: 5.03.0

 

 

M h m m

<8. mac. «8: o. 8329

 

.o .N .5 .»

 

 

moan». 7...; 38m.

79

as...» .u. 2». 5»... 0...... 2... .....=. 5252.... >8..».2».. 3.... Nona-mug...» ...»».........m.8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m...:.»» e.. 1»... m2... AX... .53.... 52:2... .u»......~»..

N2.» 0...... H»».......em< 2». 5»... .5268». 52:2...

.w..»».....n 7......4..h»...»... 5.2.8.2. O»... .3268».
52.2.... EX. 3X. AQX. SAX. . . 6X. mX.
DJ. 352......» 9X. 3X. $.X. .wa. ...w.X. NX.
2.0.8. 7.32.3...» 2X. 3X. 3X. EhX. ...m_X. MX.
DJ. 5.53.2. on. 8X. 3X. Nm.~.X. .o.m.X. NX.
7.0.8.. 528.30.. 92X. 9X. 3X. MPHX. mbX. 9X.
Enigma 3X. uX. MNX. m.m.X. muX. QX.

 

 

moan». 3...? 3882. 2... 3225. Geo... . >=...o..m 8.......2.o...

80

3.3.5 Elasticities and Model Structure

The elasticities used in the modeling in this study (Table 14) indicate behavioral response
of households to changes in product and input prices that they face as consumers,
producers or suppliers of labor. Because elasticities for this study could not be estimated
from primary data, the necessary elasticities were borrowed from other studies. Munyi
(2000) used time series data (partially used in this study) to estimate output supply and
input demand elasticities for different maize production regions in Kenya. Labor and
fertilizer supply elasticities were borrowed from Pitt and Sumodiningrat (1999), and

consumption demand elasticities adapted from Renkow (1991).

Once all the parameters are assembled, each of the multi-market models is constructed in
the form HU = K, where H is a matrix of elasticities and shares, U is a vector of
“unknown” endogenous variables (including proportionate changes in production and
consumption of maize and non-maize commodities, input demand and supply, consumer
price index, wage rate, real income, farm profits and maize price (in flexible-price model)
or maize import (in fixed-price model) for each region and group), and K is a vector of
exogenous variables (including proportionate changes in the technology shifter,
population, fertilizer price, price of non-maize crop, exogenous income, price of maize
(in fixed-price model) and maize import (in flexible-price model). Pre-multiplying both
sides of the equation by the inverse of matrix H yields a solution for U for specified
values of exogenous variables in K. Figure 9 illustrates the structure of the multi-market

models, using arrows to show links between various components in the model.

81

due—a 39%—88.. 593.23.... Cm...— ... 3......— £55.52;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

game was... Zmfin 2.3.... ZEN... 9.3.7. 7.9.8 82.7. 02.5.... 9.3.5..
N9... £95.05. @3305. 0.9305. $89.90. 29305. c.- 0.330....
«as... 32.8» «Qua-hwfim. 9...... .mcoq $5.... mo... Boo Siam 2... o.- Bu§o ta...
9...... Hana .508 8 52......
Sign...
9.8 -obmu .93.. -o. . mo -925 ohms
37.5%....“ 9.8 -98.. -98.. -93.. -98.. 98..
3......
2.595950 ouom 66mm 5.8.. -93.. -93.. obao
ca. .3358 9... -9 . 3 -98.. -98.. 9.... 9.8
3......
$2.92.... chum ob... 5me .93.. -ohoo oboe
maze... 9.8 -98. -98.. 9.... -98.. 98..
C3»: - - - - .93.. 9.3..

 

 

mos-on. 7.93.. 383. .u... 9.... m5o........m3. C8...

82

..-9..... .9 kneung

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.96.... 9.59.... .99.... .0892. .96.... 9039.... .96... 9939.... 53:92 9.5.9.... 33:9...- .-9....~ .83....
No.5 9.93.90. $.99 c.9383. 8...... 0.9.0.305. 2...... 29303. 2.9.. 0.930.... 9...... 339.... «.9305.
.99... .58 89.90 .58 mo... .58 9.83.. .58 39.90 .58 9.939... 2....
9.83.. .58
998.99....
.93 9o~o .993 98c 98c 98.. 9.5..
cc. Sagas... 9.. 98.. 99... 98.. 9o... 99.... 9......
7.0....
3.99:.qu -93 9on0 ..9owo 980 980 98o 9.5..
.5 9.2.2.5.. 9.. 98.. -98.. 98.. 9o... 98.. 9......
2.05. -
9.9.5.3.... -9..m 98o -9omo 98o 9 . oo 9 . mo 9a....
$99.8... -9... 98.. 9...... 98.. 9.8 9.... 99...
C39: - - - - - - -

 

 

mos-no. w... 9.... mgo...=...m~9. C8... .3332 Con...

83

3mm; 9. 7.9...-39979. 2.999. 9.99.9.9

 

 

 

 

 

 

 

 

 

.999. .999. 9.9999. ....9..99.
99...... 09899.. .9898. 99......
39.... 9999.99 9 ..
......... . .3... Alva... . .2... Alllnv 9.9.95... 1.99.19... 16...... .9“ .L
_ 9.999999%... . .
.999. 7.9....9. 39999. 7.9....9.
4
% ...9..8 3.9.... A... s.
.9889 $98 $999999... .mx9m99999
>m...99.8..9. ¢ .9889 T .9889
.98.. m9.9 . ..
9... . C...
A mm. + a. s. ..9..9.9..99 09999.99. .....99
A ., . ; f .998.
2.. :9...
W99. .u9. 09....9
3999999...
.9889

C...

84

CHAPTER 4

4.1 Results and Discussions from Simulations

The multi-market model framework was used to simulate the impact of an exogenous
shock caused by maize technological change on farm profits and income, with the latter
being used as a measure of farm welfare. Two models are estimated: (1) a fixed-price

1.l0 For each model, three scenarios are simulated:

model and (2) a flexible-price mode
1. When technology diffusion is limited to the marginal regions;
2. When technology diffilsion is limited to the high potential regions; and

3. When technology diffusion occurs in both the marginal and high potential regions.

The results of each model are compared to a baseline scenario, which assumes no
technological change in any of the regions but an annual growth in exogenous income by
2.3% and population by 3.1% (Kenya, 1994).11 A further comparison is made in the
fixed-price model by assuming maize prices increase exogenously by 20%, which is
consistent to fluctuation in domestic and world market prices as shown in Figure 10 and
Figure 11, respectively. In the flexible-price model, comparison is made to when maize
imports are doubled exogenously. In general, all the simulations assume no change in
market institutional structure, so that marketing margins remain constant and there are

equal changes in producer and consumer maize prices.

 

'0 Both models estimate long-run scenarios in which labor is mobile and labor markets
clear nationally.
H A 15-year period, starting 1998, is considered for this analysis.

85

 

;
_

A.

Ems—6 3" W258 9.58:0 ZEN.” $18 «$133... a»: Sea - Una Eww.

 

Aobo

No.00 .

 

 

% Price Variation

0

o

o

_1

$
'=:_—
:=

 

 

 

-Nobo .: . \ I 11 r \ Ill \ ]\¢ it - - . r‘ ‘%i\;\\---i.

 

 

 

 

2.033 to: 8-08 03 g

 

86

man—=6 :m €33 gun—8» gamma $18 5.1.52: a»: Sea .. 58 Sam.

 

 

 

 

 

 

8.8
8.8
.m
v 0.8 -
a
m
%
-38
L68

 

 

 

2.03:6 Cm: 8.08 we

 

 

. «mob. ancamv. £38 83 318

87

4.1.1 Fixed-Price Model Simulations

The results of simulations under the fixed-price model are presented in Tables lSa-f.
Table 15a presents results of the baseline scenario, which indicates a general decline in
real income for all households in the absence of technological change and change in
maize prices. An increase in population grth stimulates growth in labor supply, which
puts a downward pressure on wages in turn causing growth in aggregate farm profits. But
as population growth dominates exogenous household income growth, real per capita
income of all households, including urban households, decline. Real agricultural wages
decline by 60% and maize import demand increases by 165% over current levels (1998).
The national maize production level (weighted by zonal output shares) increases by only
4.3% while the aggregate real income decline by 36% over the current level. Evidently,

this is not a sustainable option for Kenya.

Table 15b presents results of the simulation when technology diffusion occurs in the
marginal regions. Maize production increases in the marginal regions, relative to the
baseline, as a result of technology adoption there. Overall, national maize production
increases by 4.1%, resulting in a decline in maize import demand by 15% compared to
the baseline scenario. In addition, increasing maize production in marginal regions raises
labor demand and puts upward pressure on the real agricultural wage, which increases by

6.8% relative to the baseline case.

88

duo—o Hm”: Ewan $18 32.2 wnmiau wean—=8 man—Slo— ax. Ore—£8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3a.. :5. $22.2»— N¢=nm m: a 5:23»— Noam
5.2.522. UQ 32% OJ. 32m,
rot—ago gas—22...... 23:53.? H3339: 56552. ammo—2:.» 61:5 238:».
gang 3.2—=33: who who who mum 9.3 Pd Pow
3...... 533%
. was: 335 who .93 who mun ohm 4.3.
Ah—

. W35; 8 ESQ 9: 3.3 2.8 3:94. 2.8 8.8 _o oo
. henna madam Pow uh» aha 93 9.3 o.oo .

. ”2.55 8 En Srmo 3:3 Nmuo :.m~ Sou o.oo
”2: 3.8:.» £2. 33.»
- m3»: «35.4. .533 .3?“ LP; -uoum L93 -moba 3 so
u wanna 9.3.... -wobo -m~._~ $3.3 -uo.mm -muho -ow.~o .
. C3»: roamnroam 33.3
‘18 :33
a was: 3:3 o.oo o.oo o.oo o.oo o.oo o.oo
. ream.“ 3:3 o.oo o.oo o.oo o.oo o.oo o.oo
. Eng: socmoraam o.oo
wan— >m12==n£ game #6.;
02.2.». $18 :53 o.oo
552.3 abhm

 

 

 

 

 

>mm=30m so 033%.. E Rama H.108. man 95 so guano 5 rocmoroE $850 @100 5&8? >= $.58 mango oQonnSmo 033mg 5 £38: so
«nogoammom. ognmo Bwom Ewen Ba «33: 5 vows—no.0: Ea oxomoaocm 50an 3:92 883 33039

N on 39:8 8382: 839:3 883m 8 Foo go 935. Econ

89

H.928 .m... 3.8.. 5.8 7.8....— flam=.8. H8.........n.8. Ora—.8 ... 785...... ”8.2.“. fix. 9.2.8.

 

.55. .5.— wo.o=..n_ No.88

 

 

 

 

 

 

 

 

 

 

l. .. 5.8....»— 82.8

  

   

 

 

 

 

 

 

 

 

 

 

 

5.12.8

 

 

.88....”an DJ. 78:. UJ. 78$.
Poi—n58 3.8.28.8 3...»...88 H.833: 56.85.... I. ....:...u :18: 2.....5...

7.8.8 3.2.8.8.. How 58. .900 3...... .9: on.“ a. .
5...... 3.58.
. was: 9...... 5.3 who who «.3 on: no.3

. .888 8 5:... -..~o -w..m ohm who L... -...m .38
. ream.“ 9.35 Fm”. Poo who ~84 -o...o o.oo -o...

- .8255 8 .12... -..m.. -..m.\ -Nom -93 o.oo -ob.
was. 88:... 8.. 8......
. was: 8:8 ohm -o.o. Nbu who ohm -o.~. o.oo
. .86.. 9:8 92. Pub -..~. o.oo .o.m~ o.oo
. C8»: 39.8.8.8 o.oo
5.8 :53
. m3»: 935 o.oo o.oo o.oo o.oo o.oo o.oo
. beam." 958 o.oo o.oo o.oo o.oo o.oo o.oo
. 9...... 30:88.8 o.oo
”a... >n...a=_.....n_ .298 0.3
028....— .v...8 3...: o.oo

.93

 

 

 

.8858 .8 0.5.5.. ... 3&8 3.8m

2.8 88...... 82.2.... 3.8.8 3....
~ me... .858 8688:. 0.5.6.8.. 8835 8 .3... E... 83.... .88.

. E... 88 .8 2.88 ... 8:88... 88...... .58 .888.

>.. $.88 8.88 88888 0......me 8.3.8 8

90

Changes in farm profits for the different household groups are influenced by whether
they are technology adopters and the labor-intensity of their maize farming.12 Therefore,
for adopters, farm profits increase or decrease depending on which is dominant between
the positive effects of productivity gains from adopting new technology or the negative
effects of increasing cost of production from higher wages caused by adoption of new,
labor-inducing technology. Table 15b shows that, apart from small farm households in
the Lowlands zone whose productivity gain is insufficient to compensate for increased
labor costs, all households in marginal regions experience gains in farm profits. For non-
adopting households in high potential regions, increasing cost of labor without a
corresponding productivity gain leads to a decline in farm profits, except for smallholder

households in the Moist Transition zone.

Changes in real household income are influenced by changes in labor wages and farm
profits, and the relative importance of these components in the household income (Figure
10). Apart from small households in the Moist Transition zone, for whom labor earnings
are an important source of household labor, households in high potential regions suffer
declines in real per capita incomes, again primarily because of higher labor costs without
productivity gains from technological change. In the marginal regions, all households
except small farmers in Dry Midaltitude and large farmers in Moist Midaltitude zones,
experience gains in real per capita incomes. Real per capita income for urban households
remains unchanged since maize prices do not change. Overall, aggregate farm profits,

which are computed as the sum of farm profits for all households weighted by zonal

 

‘2 Farm profits refer to combined returns to land and family labor.

91

population share, increase by 1.5% while aggregate real income, computed as the sum of

real income for all households weighted by zonal population share, increase by 0.3%.

Table 15c shows the impact of maize technology when it is limited to high potential
regions. Weighted national maize production increased by about 15%, compared to 4.1
for marginal regions. All adopting households in high potential regions enjoyed higher
farm profits accruing from large output effects, which overcome induced labor costs. All
non-adopting households in marginal regions suffer declining profits since their labor
costs increase as wages rise by 2%, without any productivity gain. This wage effect is
smaller than when the technology diffilsion takes place in the marginal regions (6.8%)

because marginal regions’ technologies are more labor demanding (Table 13).

Demand for maize imports declines by 50%, relative to the baseline, compared to a
decline of 15% for the marginal region scenario (Table 15b). The aggregate farm profits
for the high potential region scenario (2.9%) is almost double that of the marginal region
scenario (1.5%). The aggregate income increased by 1.8% compared to the baseline
scenario, compared to 0.3% when the technology diffusion occurs only in the marginal

region.

Table 15d shows simulation results when technology diffusion occurs in both marginal
and high potential regions. The overall impact on the weighted maize production is
greater, raising national production by about 19%, than when the technology is limited to

either of the regions. This puts upward pressure on real agricultural wages, which rise by

92

Hus—a Go. 3x2. $18 :2.»— womiau Hearse—emmai Ora—fin m: Ema—E:— wonmeufi fix. 0.5.53

 

 

 

 

 

 

 

 

 

 

 

 

 

562.8

 

 

 

Sunni». 5:25». Nona En: wan—3». NEEm
5.3532. 51% 39mm» 53. 332
rein—.3 gay—=33 gas—23% H3359: 5.2552. Emu—95am can: 233:».
also 3.2—:32. -93 -93 -95 -95 59% 9% 3.3
5.2: 36:8»
. was: 3:3 .9: 9: .95 .95 Eh... 5.3
5.8
. Won—Sm 8 55¢ -95 .93 9.3 -93 Sam Now
Pmm

. Ema 335 -9: 5.5 .95 9: Sam :3

u flag 8 5:5 -93 -93 -93 -93 :50 #3
”an. 38:8 12. SE?
. 93: mi; 98 98 9.: 98 92 rd
. r» m :3

«mo ”new 98 95 -93 93 2.3 #3
I can: rocmawoam o.oo
$18 :53
. m8»: 935 o.oo o.oo o.oo o.oo o.oo o.oo
. Penna mama o.oo o.oo o.oo o.oo o.oo o.oo
. CE»: 3833:; 98
”2: >n1n===~£ gun» 5.05
025.». $18 :53 98

L95

 

 

>mm=Bnm so 053% 5 33% 2.38. BE 95 so 035% :— rocmoroE $5050 v18 590%. 3»: 3:58 @958 608339». 035mg 8:520 8
9o Gawain mangle Admin Gav.

~ on mmfiom 35833 83:53 3:55 8 Fun 25 mun—5, Eco...

93

as...» a... 3x2— vloa 32.n— wauiau accrue—ennui Osman» E >= ”ammo—a. ax. Ora—£3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

53.2.8

 

 

 

gaunt—u. 5:25-— Nesam En: $22.2». Ne—Sm
wens—=32. DJ. Zea" UJ. gown
rot—2:: Saw—33% gains—an 5.25:3: #3532. Emu—25m :3»: Zeno—z.—

Zumna 2.2—=23: 93 5.3 5.3 NS: 598 m. S 5.8
~32: Edna“
. MB»: «ms-Em .9: 5.: Que PNA 5.5m Pam

. wage... a 53 -Co a8 -93 EN 93 0.8 93

fum

- beam.“ 925.. 5.3 raw aha Pm: 5.5m r5

. W253 8 F25 5...: 5.8 -93 -93 5.3 95
was— FSSa can 2.3:.
. ma»: sea 93 -3“. New Mom 93 3..
. 5% was 9: 93 -_ .3 2o 8.: 93 N:
- C3»: rocmaroam o.oo
118 :3:
. m3»: man—mm o.oo o.oo o.oo o.oo o.oo o.oo
- ruin 355m o.oo o.oo o.oo o.oo o.oo o.oo
. CG»: socmnuoam o.oo
”an. >«1o:_.=..»_ Sena 9.3
025:: $18 =5: 98

-333

 

 

_ .5238 so 035% 5 33% Beam. Ba 9% :o 255% 3 795285 «magma v18 5&08

So can—So 825:0 Ada—En 33.

N WoE :28 3389: 83658 3255.. 8 .85 85 335‘ .33.

. >= A‘s—com @958 wonogSMn crummom 35:5 8

94

8.8%, relative to the baseline scenario. The increase in maize production from both
regions depresses maize import demand by about 64%, compared to 15% for marginal

regions and 50% for high potential regions.

All households in the high potential regions and marginal regions, except small
households in the lowlands zone, post gains in farm profits, again indicating that their
productivity gains from technology adoption are greater than the increases in labor costs
associated with the technologies. Households in the Moist Transition zone have the
greatest profit gains because of their large output effects. The aggregate profit level is
also higher (4.7%) than when the technology diffusion is limited to either one of the
regions. As in Table 15a to 15c, welfare in urban households remains unchanged relative

to the baseline because maize prices remain unchanged.

The aggregate real income changes by about 2.2%, compared to 1.8% for the highland
region scenario and 0.3% for the marginal region scenario. All households, except small
farm households in Dry Midaltitude zone and large farm households in the Moist
Midaltitude, obtain increases in real income, with farm households in the Moist
Transition zone getting the highest gains driven by huge farm profits that are, in turn,

influenced by large output effects.

Table 15e assumes no technology change takes place in either the marginal or high

potential region or both, but instead maize prices are exogenously increased by 20%,

either by the government or effects of the world market. The price increase considerably

95

as!» Hun. Ema $18 32.0— 5825" 2: Hague—ounce. 9.2.“? :25 $18 3.2.38 3. N52... fix. 9.253

 

:2. :5. $22.22 N255

 

 

 

m: _. $22.2». N255

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

552.5

 

 

 

$22222. OJ. :22 53. :2:
Poi—2:5 3322...? 23222:.” 5.2532. 5.2552. Eur—255 9.92. 225:».

:25 2.2—:25: Pam flmm Mom 9% who 93 33
$2.2 $535”
_. man: 335 9: P3 3.3 5.3 5.?“ wan

. woe—.25 8 r25 -93 wk. 5.2 93 mm; Nd 98
_. fiwnmo @525 3.: Pd ~93 P: 0.3 wk» 93

n 50:55 8 ~25 who Poo 3.8 cu— ch... +5
”2: $825 $2. 2.2:.
u was: 935 NAN 5.8 who 5.: 3.3 5.8 3.3
. roam.“ 335 who 3.3 ~98 on; ~93 no.3
. Gaga 3252555 -._o
$18 55:
u was: 335 95 PE 9% Nu; Nb; who
.. 52mm @525 93 Pm» Pf o.oo who foo
. can: 3250—555 NE
”on. >m1n:_:=.2 ¢<una .98
02.2.»— $18 :53 NE

-323

 

 

>mmc3nm 5 253mm 5 BEN." $185. 25 3:5 5 22in 5 325255 mun-.050 $3.8 5&8?

30 25150 582:5 Quin 33.

N .55 $558 8282: 8:558 8835 8 25 25 $25. 525.

>= 558 @258 $o~oo=8mo 05258 5:58 8

96

boosts farm profits in all households, except among small farm households in the
Lowlands zone for which maize profits is a small share of overall farm profits, resulting
in an aggregate profits level that is higher than any of the preceding simulations.
However, the weighted national maize production is much lower (6.8%), since there are
no gains in maize productivity, than when there is technological change in the high

potential regions (15%) or in both the marginal and high potential regions (19%).

Real incomes improve for all rural households, especially in the high potential regions,
with the aggregate income increasing substantially by about 10% relative to the baseline
scenario. However, urban household suffer loss in welfare from the price increases,
which is different from previous simulations (Tables 15a-15d), which had no change in
maize prices. The real agricultural wages are depressed by about 1%, partly due to lack of
technological-induced labor demand increases, and overall maize import demand declines

by about 55%, relative to the baseline.

When the maize price increase occurs in conjunction with technological diffusion in both
marginal and high potential regions (Table 150, the magnitude of increase in the
weighted national maize production is greatest (25.8%) among all the previous
simulations, relative to the baseline scenario. This huge increase in national maizr
production causes maize import demand to decrease by 119% compared to the baseline
case, indicating strong synergistic production effects between maize price increase and
technological change. As in Table 156, farm profits and income increase substantially,

apart from a decline of the former in small households in the Lowlands zone and urban

97

5...... .m... $.83 $18 .52.... ”8...... Hon—.5238... 9.2.8 ... >= $78.25. .528 $18 1.2.258 3. Nae... ax. 9.2.“...

 

 

 

 

 

 

 

 

 

 

 

 

98

 

 

 

 

 

 

 

 

3.... .5». $22...»— Nbaou m... .. $225.... N285
5.3.222. OJ. 7.8.5. .2... .525.
.8325. 3.8.22.8 3.8.22.8 5.2552. $25.22. mum—.82.. 61.2. 22.2.».

.528 $35.28.. .93 No.3 5.3 3.3 ”3.3 3.3 Nu.mN
$2.... 2.2.5.
. m3... .225 5.: 5...... .93 BAN No.3 Pa. 9 d

. .8835 8 r2... -..om -o... .N... m... .m.uN «.3 «.8
. 5.... an... 9% a... 5... N... .9... u... .

. .8255 8 .2... o... ..oo .33 -9... No.8 Pam
.8... .82.... $8.. 8......
_. was: 3.35 u..N 5.3 2.3 .9... .9... 3...». .N mm
. .t2mo m5... N.AN 3.... NN... .mA nib. NT... .
. C32. 72.88.... N...
$18 5...:
. m3... 3.25 9.5 w... :3 N... N.N.. Nu...
_. ~28 m2... of. PMN Po» o.oo 93 ..oo
- :82. 52.88.... N...
.8... >n...o.....=.u. $38 8....
02.2.... $18 1...: Nb.
mafia-4m n— _©.w0

 

 

 

 

 

>5mcaom .8 1.2.8 .3 228 2.85. 2... 9.5 .8 2.2.8 .5 5258.8... 2.8.»... $18 .5985. >.. 8.85 8.88 $209.88. 1.2.85 8.2.8 8
2.0 6.58:8 583103.210 3....

N me... $558 8382.. 83.....2. 88...... 8 .2... 2... 83.... 5.85

households, resulting in a 9.7% increase in aggregate farm profits and a 12.6% increase
in aggregate incomes, relative to the baseline scenario. Households in high potential

regions gain but urban households are hurt by the maize price increases.

In summary, the simulations in the fixed-price model suggest a few things. First, large
output effects and subsequent maize import declines accompany diffusion of maize
technologies in high potential regions, with or without price changes. In contrast, wages
are affected most when technological change occurs in marginal regions rather than in
high potential regions, because of greater labor intensity and labor demand associated
with adoption of maize technologies targeted to marginal regions. This is closely related
to the fact that these technologies derive theiryields gains mostly from labor-intensive
agronomic recommendations compared to technologies for high potential regions (Table

13).

Second, the magnitude of the farm profits depend on whether a household is an adopter
or non-adopter and on how labor intensive their maize farming and how large their wage
effect is. Therefore, farm profits for technology adopters vary depending on the relative
size of their output effects and the increase in production costs associated with adopting
the new technologies. For non-adopters, farm profits decline as production costs increase

without corresponding gains in maize productivity.

Third, impacts on real income vary depending on changes in labor wages and farm profits

and the relative importance of these in the total household income. Labor from sale of

99

agricultural labor is a significant source of income among small farmers in all regions,
and more so in marginal regions (Table 12). The largest income effects accompanying
technological change occur in the Moist Transition zone because of large output effects
(Table 13). In general, positive income effects occur for almost all households when
technology diffusion occurs in the high potential region but the effects are more mixed

when technology diffusion occurs only in marginal regions.

Lastly, raising the price of maize exogenously without technological change increases
farm profits and real incomes much more than when only technology change takes place
in either one or both of the regions. However, this has less impact on maize production
and maize import demand compared to when there is a technological change, and urban
households suffer from the maize price increase. Seemingly, raising maize prices and, at
the same time, encouraging diffusion of maize technologies complement each other:
increases in national maize production, increases in real incomes, as well as decline in
maize importation, are greatest in this scenario than all the other simulations under the
fixed-price model. However, such increases in maize prices have to be generated from
somewhere, presumably through increases in tax, so that the income gains attributed to

increasing prices amount to an income transfer.

100

4.1.2 Flexible-Price Model Simulations

Tables 16a-f present simulation results of impacts of technological change in a flexible-
price model in which maize prices are determined endogenously by maize demand and
supply, similar to a closed-market economy, and maize imports or government’s buffer
stocks are used exogenously to bridge the gap between domestic production and
consumption demand. One big difference between this model and the previous one
(fixed-price model) is that here maize prices will respond to changes in quantity
produced, thereby declining when diffusion of maize technologies results in increases in
production. The baseline scenario (Table 163), like in the fixed-price model, assumes no
technological change in any of the regions but there is growth in exogenous income by
2.3% and growth in population by 3.1% per year for the lS-year period under

consideration.

The baseline results predict that without technological change, and with population
grth outstripping growth in non-exogenous income, there will be declines in real per
capita income for all households (including urban households) and large (58%) increases
in maize prices as excess labor supply depresses real agricultural wages by 63%.
Although individual and aggregate farm profits are positive, especially due to cheap
agricultural labor, household and aggregate real incomes decline significantly, the latter

by as much as 41%.

101

Haw—o a». Emir—n $18 :28. Womiau wane—:8 moons—.8. fix. Owummov

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32M?»— wongzu. Ne=om an: 5.83:: Nbaan
3332: ed :2: H5 2...:
rot—Ea“ 33:52:? 33:52:? 5.2552. Hun—.222. Ewan—Em :3»: 2:22.».
3&3 USA—:28: 3.3 wade 8.?“ No.8 No.3 ~93 who.»
3:5. 303%
n was: 835 ram 3.; Aqua 3.: umbc uuba
Nun:

. W253 8 5:8 9.3 ”5.3 3.3 uobc 3.? 3.3 meo
. anmn 935 3.5 Sum APem Lbe 3.3 2.:

. ”$55 8 ESQ APB Nagm 3.x» 9% 3.3 8.3
ia». 585... can 25:»
. was: was -28 -38 .83 ago -3; .38
. Emma 8:3 -mmbm L38 Lba $5.3 -waO -mw.- Lfibo
- can: rocmnroam -wabq
~18 Hanan
. m3»: madam 3.3 5.3 Sum Sea who mam
. Fawn 953 3.8 :3 5.: wam 2.5 9.3
. CG»: remove—am 4.5
”on. >m1o==:..n_ «5&0 -o~.u_
O32.»— wlaa =5: 0.3
”an. :18 118 3.3

 

 

80:30.38». 0:850 85% Ewan 28 «329 E nova—wag Ea nxomosocm 50an 8:02 88:. @283?

~ on mmfig 3380:" 839:3 383m 8 Sam Ea 852v. Econ

>mmcaom no 035mm 3 BEN.” unnom. Ba 9cm :0 035mm 5 3950305 £8850 v18 59.02... >= <mEow @258 “58388 035mg 5 $38: no

102

Table 16b represents the scenario when technology diffusion occurs only in the marginal
regions. Relative to the baseline, real agricultural wages rise by 7%, due to technological-
induced labor demand in marginal regions, and maize prices fall by 3.6%. The change in
the weighted national maize production is lower (2.7%) than the counterpart scenario in
the fixed-price model (4.1%). Farm profits increase for adopters in marginal regions,
except in the Lowlands zone where, output effects fail to compensate for increasing labor
costs, and there is decline in farm profits for all non-adopters in the high potential
regions. Real per capita income gains are experienced only by 5 out of the 8 households
in marginal regions while the rest of the rural households suffer decline in real incomes,
mostly as a result of increasing labor costs and the maize price decline associated with
increased maize production. Urban households gain unambiguously fiom the price
decline. Overall, the aggregate farm profits increase by only 0.3% relative to the baseline,

while the aggregate real income increase by only 0.2%.

When technology diffusion occurs only in the high potential regions (Table 16c), the
weighted national maize production increases by about 10%, depressing maize prices by
11.6%, both relative to the baseline scenario. The real agricultural wage increases by
2.7%, which is (like in the fixed-price model) smaller than the wage effect when
technology diffusion occurs only in the marginal regions. All non-adopting households in
marginal regions experience loss in farm profits, save for small farmers in the Lowlands
zone and large farmers in the Dry Transition zone — whose proportion of farm profits
from maize are least. What is different from the fixed-price model is that only adopters in

the high potential regions post gains in farm profits, specifically large and small

103

.—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.u—._o a? 32.58 $18 32.2 #858. Humane—ommoa 9.950 m: gonna.»— wmnmouu. fix. Cream».
33“....»— wo.2.:»_ 82.8 En: 132...». No.5...
33:38.. dd. 78.2 US 78.2
Pot—2.3 33:53am 339.28% 5.25:8: €33.32. Enigau :13... 2:22.»—

ZMEn v2.9.2.2. 9N“. 3.0m 3.3 Naxfi -N.: -_ .me who
5...... 3.2-.8.
. m3»: 8an Ar». uhu Pan flow Lbu bow

. $83... 8 $2. -. .8 LE -2: we. -3... -NS Wm.
. rams as; -5. 5.. “.3 9.... -N...» -2... - .

. W253 8 P25 Qua 39...: -m.oo -93 -u.oo Lbo
”as. 58...» c2. 26:»
. was: one .3 5.8 2... w... -9... -8.
. 5.2 one 5. -9... -23 .2 a: -...: 9:.
. C32. roomaroam 9.3
~18 =33
. m3»: 935 -_.~w -93 -..ON .93 -95.. -93
I wanna 3an -.oo -9: -93 -.: -9qo -958
. C32. rocmnroam -93
”2.. >m...e==:.§_ gun» o.oo
02.2.». $18 :33 -93
5o». 32% V18 - wo.

 

 

>mm=Bnm .8 2.250 5 328 E88. 25 5% no 255.. 5 725283 woman-8 “.18 598.... >= $.88 @288 382.8%. 2.253 8.276 8

9o gun—So 82.38 3.2....

33.

o WoE mmfig 5282: 832:3 38...... 8 .25 25 85...... Eco...

104

.55... .9. Son...» 5.8 .52.... ”8...? Hoor:o.on.nw. 0.5.50 ... =.m......... ”£83. ax. 05...“...

 

.S»..m...... 5:2...»— Noam

En: 532...». N28.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5.2.522. 53. 78.2 OJ. 78.3
F2383 :3»...an 2.32:8: H3528: €33.22. amiss... 9...... 228......

2....8 32828.. -9... -93 L20 #8“. Eu. Au“. .9NN
3...... @338.
- m3»: 9...... 9.8 b...» hbm -.mu m..u -mb.

. .383... 8 CE. -9... a... -98 a... A... -2: Arum
. San 9:3 A... -9... -3. .8 u... -93 - .

. .325... 8 5.... Arum -93 .8. 9Q. .93 Lb.
”2.. 82...; ..2. 8......
. ma»: 9:... So 9 a -93 -98 98 -98
. ruin madam Poo Lac .9: .8. 90m -93 C.
. C32. 8828.3 .bw
H18 :8:
. m3»: 9an -99. -..w -93 ...mo -..u~ -..-E
. 5E0 935 -9... -98 .93 -9-3 .9? -..~o
- C32. 852.2% -93
”2.. >m...n===..... fiano N3
02.2.... ~18 =52 -~.oo
”on. .538 5.8 - Sue

 

 

$525.8 .8 ornamo 8 .528 9.03. 3... 9.5 .8 0.55% ... 30.5283 €850 .58 8908. >.. <m.:om 9288 “582.550 2.258 8.3.5 8
:8 630...... 82.2.8 3.8.0 .9...

N we... mmcam 8982.. 889.52. 88...; 8 .2... 2... $3.... .83.

105

households in the Moist Transition zone with sufficiently large output effects that
compensate for both the increases in labor costs and decreases in maize prices. All
households in the Highlands zone suffer loss of farm profits as well, since their output

effects are not large enough to counter the negative labor costs and price decline.

Similarly, real per capita income gains are highest for households in the Moist Transition
zone whereas those in the Highlands zone suffer loss of welfare for reasons discussed
above. In addition, half of the households in marginal areas gain in real per capita
incomes. Urban consumers gain because of the larger price decline compared to when the
technology diffusion occurs only in marginal regions. Households with large maize
expenditure shares, such as large farmers in the Lowlands and Dry Transition zones, also
gain from the price decline as well. Overall, the aggregate farm profits decline by about

0.7% while real aggregate incomes increase by 1.3%.

Table 16d displays results from the simulation when technology diffusion occurs in both
marginal and high potential regions. The impacts on farm profits are almost similar to
those in Table 16c, except that now small households in the Dry Transition zone join
those who experience profit gains and also the gains are relatively smaller as a result of
higher wages (9.7%) compared to 2.7% when technological change takes place only in
the highland regions. Similarly, only 5 of the 8 households in marginal regions and the
households in the Moist Transition zone experience gains in real per capita income. The
rest, including households in the Highlands zone, suffer welfare loss. Urban households

gain more here than when the technology diffusion occurs in either on the regions.

106

Haw—o an. Eons—a $18 32:: ”8:5" Hague—emmna. Cream» ”5 >= wanna—5. ax. Ornamov

 

gunnin— weaazn. Nesou

En: $22.3». Noam

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5.3522. DJ. 3%. DJ. 32%
rot—aim gas—23% zap—:83 #3359: 5.2552. Ewan—Eu :3»: Zeno—B.

ZEN” 33:28.. 93 9% was. ~53 3.3 P3. 5.3
3.2: 365%
a was: 33.5 PE -_.: Luz. mhm uh. .93 .c A»

. $23... 8 53 -er -33 -52 -3: 98 -98 Mme
. Ffimo 935 .93 L be .9: 5.3 P3 .98 .

. flag 8 5:5 .53 -93 Luna 9% Eco -93
”2.. 38:5 .5. 25:»
. was: and... 98 .93 :3 we 93 $8 _ am
. rmnmn «~36 mum Poo -m.mw Sac “£3 98 .
I C3»: gauge—am :3
$18 :3:
._ m3»: 335 -93 -~.mo Lbo .25 .25 -~.Nm
- Sumo $58 -92 -93 -wu— -muq -Nba -_ be
. C3»: cognac—um ._.m.\
”an. >n1n===2z fiumo 0.3
02.2.». $18 :33 -wa
we: 3:: ~18 - 3.8

 

 

>mmcaom :o guano 5 Human @103. Ba 9%. no 035% 5 roping—Q $850 “58 5&8? >= <qum @0308 383mm.” 03258 83:20 8
90 98:2” 82:50 Quin 33.

N on mnemom 3380:" 85338 8:55 8 5.5 23 935‘ Econ

107

Overall, maize production increases by 12.9%, which cause a 15% decrease in maize

prices, while aggregate farm profits decline by 0.4% and real income increases by 1.5%.

Table l6e represents the scenario when there is no technological change but the baseline
level of imports is doubled from the current national maize import demand of 22% to
44%. Relative to the baseline scenario, the national maize production declines by 12%
and domestic maize prices decline by 30%, indicating a crowding out of domestic
production by imports, a situation commonly observed whenever relatively large maize
imports flood the domestic market. All households, except small farm households in the
Lowlands zone and large farm households in the Dry Transition zone, experience losses
in farm profits, mainly because of the huge maize price declines unaccompanied by any

productivity gains.

The decline in farm profits does not spare even households in the Moist Transition zone,
which show resilience when technology diffusion occurs in either or both the regions
without change in maize imports. In fact, the negative trends are observed for real per
capita incomes for all the households in the high potential regions and 4 out of the 8
households in the marginal regions. In general, the effects are good for net consumers and
bad for net producers. Overall, real agricultural wages increase by 1.8% while aggregate

farm profits and aggregate incomes decline by 8% and 2.3%, respectively.

When doubling of maize imports occurs concurrently with technology diffusion in both

marginal and high potential regions (Table 161), the decline in maize production is only

108

Haw—o an. 32:38 5.8 7.8%. ”8...? 28 Hague—anal 6.5:“? 625.2. 758.8. fix. 03:52

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32.“.5. 132:...— Nbsom En: $22...»— 82.8
5.3332 c: 2%. a: 2...:
F312...» 3.8.28.8 3.3.2.8..» #3352. ans—5:8: Ewan—Em :32. 228:».

.528 512.539: -abo -Ebo Lohu -. :3 LPG -. .bo LNG
3.2: 133%
. man: «mam ran .92 -53.. -98 -28» Lake m em

. ”2.55 8 .28 9.3 .82. -wao -HS -39. .92 .3 MN
. 58¢ mafia .93 .9... -mbm 9m“. .3... .3: - .

. Won—Bu 8 C58 .500 .83 .Nqu Now -38. .98
”on. 8253 :2. 2.8.8
. was: 93.... 8.3 9:. -_ um -93 -98 8.: N 2
. 5:2 9:3 was 3.3 -33 3.3 -58 -9: .
. C8»: 8:883.» PH.
M18 :5:
. m3»: 953 -38.. -m.mm -mua L28 -ubm Lam
I wanna madam -u...m .93 -300 -380 -u.mo -93
. C8»: regurgmm -u..$

. .8

”2.. >n.._o==:_.u. gnno
02.2.». #18 :33 -m.-
”2.. 35.8 $18 -25.

 

 

.5288 so 255%.. 5 5280 @108. 28 9cm .8 255mm 8 .8828... $59.30 .58 .5908.

En Gawain 82510 Quin .9...

n we... mmfiom 8326:. 83:82. 88:5 8 .25 2:. 88.7. .33.

>.. <m.=om @288 “882.8%" 035mg 8.2.5 8

109

H020 .8... 30......0 5.00 2.2.0. ”00...? €098.88... 0.850 .: .9: ”88:0. 50:20.. .5810. ax. 0.850.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.0.8:». $8.05.... 88:00 in: 530:2... N300
5.2.80.2. UJ. 30.0. 55. 28.0.
rot—.5... 3.9.58.8 van—=28? 5.80.08: €35.20: Enigma 9...»: 238:...
3:80 1880.8: .9... -3... .26 0.3 chm -m.mm -9...
5.0:. 13......
.. man... 83... ....u usu— -Na.am 5.8. LN...» -3...
-25..
. .882... 8 CE: -28 -32. bob. 9?. -3... --.~. o
-3. N
. Em... qua... Lab“. -32. Luau 4.3 .93 -.~.um
. W083. 8 .58 -Nmbw -.~.oo L..oo PM. .98 .38.
”0... 502:0 10.. 0......»
. was: :35 a... -0... c... N... -2: -0...
u .5 m _ -..~o
80 03:0 .93 -99. .3... we... -93 -83
. C8»: 8:00.850 Moo
5.00 830:
. wan: 3.80 .93 -m.wm -.~.ma -99. -93 -9...
- SE0 33.0 -8... -73 Lou. -8... -m.mo -93
. C3»: 8508:? .980
. N
”0... £12.83. (<30 .. a
00:02.. 5.00 .80: -23
N0... .5580 3.00 LG m.

 

 

3.0.5.00 8 0.8..«0 .: 8:80 .5000. 0:: 9.5 :o 0.5:«0 .: 8508.: $0050 U100 .8800.

50 €000.80 30:38 9.8.0 .9...

N we... :28 33000:. 83:80... 380:... 8 .08 :8 83.... .08:

>= 583 80:90 $0803.60 0.8:«00 00.3.80 8

110

0.7% compared to 12% (without accompanying technological change). However, maize
prices decline by 46% and real agricultural wages increase by 11%, compared to 30%
and 2%, respectively, when there is no technological change. These changes in maize
price and wages lead to losses in farm profits for most houses, except small farms in the
Lowlands zone and large farms in the Dry Transition zone. Also, all households in high
potential regions and 3 out of 8 of households in the marginal regions suffer losses in real
income. Households that experience gains in real income are those with larger shares of
their income from labor earnings and also those with a high maize expenditure share.
Overall, the aggregate real income declines by 1.3% while aggregate farm profits decline

by 10.3%.

In summary, a few points can be deduced from these simulations. First, because output
effects from technology adoption are accompanied by a maize price decline in this model,
impacts on farm profits and real income tend to be lower in these flexible-price models
than in the fixed-price models. Since these output effects are larger when the technology
is adopted in high potential regions that when adoption is confined to marginal regions,

the price declines are greater in the former than in the latter model scenarios.

Second, maize price declines accompanying output effects in the flexible-price model
represent a positive effect especially for net consumers with large maize expenditure
shares such as large farmers in the Lowlands and Dry Transition zones. Whether these
positive price effects and earnings from sale of agricultural labor are sufficient to

outweigh declines in farm productivity and increasing labor costs as agricultural wage

111

increases is what determines whether households experience gains or loses in real
income, relative to the baseline. Lastly, urban consumers gain unambiguously because of
the price declines, a significant difference compared to unchanged welfare under the

fixed-price models.

4.2 Aggregate Income and Farm Profit

Table 17 presents a synthesis of aggregate income and farm profit effects of the various
technology scenarios discussed in the previous two sections.13 The aggregate income
effects are computed as sums of real per capita income changes for all households in a
specific simulated scenario, weighted by population shares, and represent the average
percentage change in real income per capita for that scenario. Similarly, the aggregate
farm profits effects are computed as sums of farm profit changes for all households in a
specific simulated scenario, weighted by population shares, and represent the average

percentage change in farm profits for that scenario.

The results presented in Table 17 indicate that, irrespective of the technology change
scenario, aggregate income effects are greater in the fixed-price model (when maize
prices are exogenously determined) than in the flexible-price model (when maize prices
are endogenously determined). In addition, within each of the models, the aggregate
income effects are greater when technological change occurs in high potential regions

than when it occurs in marginal regions. However, the greatest income effects are

 

'3 The individual results are assembled from Tables 15a-f and Tables 16a-f.

112

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Haw—o Hug—dunno 5:4: $35 5:— Foeaa madam. sun 0:: 083033 aw Hague—bum Drs.—ma wan—5:8
32?. 0.8:...“ Hague—om! Drs.—mo ~556qu 5...:— >nn3nu8 Eco—5m QETOoanmaan
02552. man—51¢ 3.3.: Ore—fin axe. Drs.—no Agvn
«<5. Unex:
3x8 mica wan—Ea. 20 H8: magma +3 $3.3 obs
Exam 348 38%: ”ammo: 03v~ T3 9.3 98.»
3on Human Em: magnu— Wmmmoa 0:? Nmo :3 9N8
3an ~58 >= Wommosm Paw N. H q chug
«<5. annex."

Exam Wino memo—Bo. 20 H8: Osgmo 98 5.: 93¢
3an $18 >= Wommonm 0.3 Cum cbwm
€5— Quack."

28:20 mica wmmo==9 Zo Hon: Oswamo 8.“: L: :8 98¢
Sofia—o vino 73:.me— Wommoa 0:; 98 9; PM:
39320 $18 5%. wanna": ”ammo: Os? bub. 7): ONE
383% $18 >= Wommosm -93 #3 93¢
«<3. OH— 3.x:

29120 $18 mama—Bo. Zo door 033% -mbm Eh: 9&3
39:20 vino >= Wommosm Lobm -_ .Na ohma

 

 

 

 

 

 

 

_ Goa—2:3 3 m8: cm a: roan—55% 3:: Bow: arm—swam $0553 3 nous—mac: .928.

~ 02$.an mm ME: om u: gauge—am. :2: 58:5 2552.. Summing g nous—mac: $53.?

113

achieved when the technology diffusion occurs in both marginal and high potential

regions.

Exogenously raising maize prices by 20% in the fixed-price model increases,
substantially, the aggregate income effects, relative to the baseline, compared to without-
price-change scenarios. The changes are larger when technology diffusion occurs in both
regions compared to when it occurs in either of the regions. In contrast, the flexible-price
model exhibits lower aggregate income effects, which become worsened by doubling of

maize imports.

As in the case of aggregate incomes, the aggregate farm profit effects are higher for the
fixed-price model than for the flexible-price model for every corresponding scenario. The
effects are also greater when technology diffusion occurs in the high potential regions
than in marginal regions in the fixed-price model, because of larger output effects
associated with technology adoption in the former, and the effects are larger when the
technology diffusion occurs in both marginal and high potential regions than when it is

confined to either one of the regions.

But this is not true for the flexible-price models, where aggregate farm profits are greater
when the technology is confined to marginal regions than when it is confined to the high
potential regions. Whereas raising maize prices exogenously by 20% in the fixed-price
model results in higher aggregate farm profits, increasing maize imports in the flexible-

price model results in much lower aggregate farm profits, relative to the baseline.

114

4.3 Income Distribution Effects

Equitable distribution of income in Kenya has been expressed as a desirable policy goal
for Kenya’s agricultural and economic development (Kenya, 1998). Gini coefficients
were computed to explore the impact of maize technology diffusion on income
distribution as simulated by the fixed-price and flexible-price models. A Gini coefficient
is based on estimation of a cumulative frequency curve, the Lorenz curve, which
compares the distribution of a specific variable with uniform distribution that represents
equality (Figure 12). The greater the deviation of the Lorenz curve from the diagonal, the
greater is the inequality. The coefficient can also be calculated as the ratio of the area
between the Lorenz curve and the diagonal line, and the entire area below the diagonal

line.

The results of the Gini computations are shown in the last column of Table 17. The
baseline Gini coefficient is 0.262 for the fixed-price model and 0.259 for the flexible-
price model. In general, although diffusion of maize technologies in high potential
regions have greater aggregate impacts on farm profits and real income, the Gini
computations reveal that their income distribution effects are inferior to diffusion of
technologies in marginal potential regions. In the fixed-price model simulations, the
computed Gini coefficients indicate that income distribution improves under both the
marginal region and the high potential region scenarios, compared to the baseline, but
that income distribution is more even under the former. In the flexible-price models, the

computed Gini coefficient for the marginal region simulation is lower than for the

115

Figure 12: Calculation of the Gini Coefficient.

 

 

 

 

 

 

Area A
4
Area B
’ The Gini Coefficient is calculated as: Area A/(Area A+B)

116

Equality
Diagonal

Lorenz
curve

baseline, signifying better income distribution in the former, whereas the Gini coefficient
for the high potential region is greater than that of the baseline, indicating poorer

distribution effects in the highlands compared to the baseline.
4.4 Sensitivity Analysis

In order to test how sensitive the simulation results were to changes in various elasticities
and shares, a sensitivity analysis was done. First, elasticities of maize output and input
supply and demand were systematically doubled, one by one, and simulated changes in
endogenous variables noted. Likewise, shares of input demand and supply, profit, income
and expenditure were also systematically doubled, except for income shares, which were
increased by 30%, and changes in endogenous variables noted.l4 Tables 18a and 18b

present the results of these analyses for the fixed-model and flexible model, respectively.

In general, the two models are fairly insensitive to most of the changes in elasticities and
shares. However, in the fixed-price model, a doubling of the labor demand and supply
elasticities yield changes in the wage rates as large as 37% for the labor demand and 31%
for the labor supply. Apart from a 19% decline in aggregate farm profits with a doubling
of labor demand elasticity, none of the other endogenous variables changed by more than
3%. When the labor supply elasticity was doubled, none of the other endogenous

variables changed by more than 6%. Similarly, the maize import demand was sensitive to

 

‘4 Doubling income shares on large farms in the highlands zones would exceed 100%.

117

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

H.920 Han" moumHnHi >3.— mHu on $0.020: 5820300 23 mus—.00 3.. 2.0 Esau—.100 3.50—
3.352: HSwHNm >mn..0nan0 >mmn0nun0 $30 38 fl H.352
fix. CHE—50H b39890: menabnoma 5850 again
6

5x0: H.100 -HwHeuaoHnHa" \o QE—io
Kass 2%: 38$. -39.. 3.x. 5.x. o: ._ 1x.
H.069 Q0885 A+ H 8.5 - H .x. - H 0.x. -unx. 3.x. 60.x.
H063 263% $595 -o..H.x. -Qx. -H .ufi. 3.x. 5.0.x.
7330 00885 $595 9x. 0.x. 9x. 9x. -uNR.
Hid—.3504 Q0825 A+H 8.5 9x. 9x. 5.8.x. oo\o 0.x.
H2830 00803 $595 ox. 9x. -988 9x. Hoo\o
Hanna H.100 fire—.00" .
HUB? A+ H 8.5 ox. - H 9x. -u H 0.x. 9x. oix.
mxH00=aHH=~0 A+Hoo$v chug wdfi. NHX. 3.9x. -w.m_.\°
5850 $8.5” 9x. ox -wa ox. 3.x.

 

 

 

 

 

 

 

W033 8 256100 0539?.
N 8.x. Hm :08 000050 Hfim0 H.380 3&8 5830 0303 £25 9808 89x. Hm 3:200.

118

 

H.020 Ha? $529.3. >350? o_.w0H020.H Manama E:— m—Bgm 3.. E0 3355.330 30:0.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

130508.. gin.” >nm~0mfi0 f Hymn—dung $50 00.0 Hag— 1:50
ax. Ora—50V Uzi—.020: v H.025 36:3 5850 .500

6
E0530 H.100 -HwHummangu k. Ora—i.“
2:50 26un $59§V~ -89.} .9.\.. -H wax. Hoax. max.
W053 Q0385 $H 8.x; 6.5.x. mix. H 1x. 5.9x. 5.1x.
r009 26un $595 box. 99x. Max. Kg 5.0.x.
H338 @0385 $595 -H 9x. ix. 99x. 9x. 5.9x.
$352 50303 $59§ 9x. 9x. -9899 9x. 9x.
H3830 Q0385 $59K; bx. -mfi. -Hmfi. 9.9x. 3.x.
E0120 H.100 Imam—.00"
HUSH: $595 .33 52.x. -odfi. Pix. 5.0.x.
9:002:83 $H 595 ix. P99: 3.x. - H 9x. 95.x.
38:5 395“ 0.x. -H 0.x. 5.0.x. Mex. 8.x.

 

 

 

 

 

W308 8 0256100 053050

u 5.x. Hm smog c0850 Hfim0 H.350 5530 5830 038.0 55:5 0x003 59x. HH. 50:205.

119

the output demand elasticity. Finally, aggregate income changes were somewhat sensitive

to profit and income shares.

In the flexible-price model, real maize price changes are sensitive to both maize supply
elasticity and the expenditure share, exhibiting an increase of over 20% with a doubling
of the former and a decrease of 26% with a doubling of the latter. The maize supply
elasticity used in the simulations was one of the elasticities that were drawn from a study
that is specific to Kenya and was adapted from Munyi (2000), which is base on similar
time-series data. The expenditure shares were estimated from Kenya (1994), whose
sampling frame is the foundation of this study. Whereas the labor demand and supply
elasticities were not computed from the Kenyan maize case, I am fairly confident of the
parameters computed in this study and those used from Munyi (2000). Since getting
appropriate elasticities proved to be a great challenge for this study, it is comforting that

the choices made were not overly distorting.

120

CHAPTER 5

5.1 Conclusions and Implications for Research and Policy

The objectives of this study were threefold. First, to determine the status of production,
consumption and technology adoption for maize, the most important food crop in Kenya.
Second, to evaluate the impact of “on the shelf’ technologies on farm profits and income
distribution for different households and regions in Kenya. The third objective was to
enable maize researchers and managers to use this information to make R&D investment
decisions. In this regard, Chapter 2 presented detailed findings on the first objective while
Chapter 4 provided detailed simulations on the second objective. This chapter synthesizes
these findings and discusses the implications for future research and policy. The study
used a multi-market model, rather than other models such as the economic surplus model,
in order to explore the effects of regionally differentiated technological change at a
household level and accommodate differences between households, some of which are

joint consumers and producers of maize, and sometimes also providers of farm labor.
5.1.1 From Descriptive Analysis

Chapter 2 presented an update of the status of maize production, consumption and
technology use in Kenya, and differentiated these by small and large farms in different

production regions, classified as marginal and high potential regions. Traditionally,

agricultural policies in Kenya have been made using these farm size and regional

121

classifications. This study used agro-climatic regions based on geo-referenced data and
information that allow tailored recommendations for research managers and policy

makers.

The findings from the descriptive analysis in Chapter 2 confirmed maize is the most
important food staple in Kenya, as it accounts for large production and consumption
shares for the bulk of the population. Therefore, any technological, institutional and
policy changes affecting maize production and prices are bound to have significant
effects on the incomes and welfare of many people, both producers and consumers. The
high potential regions are the dominant maize production zones and together accounted
for more than two-thirds of the total output and nearly half of the total area under maize
in 1996-98. These regions have about 1.9 million agricultural households and almost half

of Kenya’s population.

Maize technologies used by farmers differed by regions, farm households and seasons.
Therefore, it is important that research recommendations be differentiated accordingly to
maximize the technologies’ potential productivity. The different conditions, locations and
seasons under which maize is grown in the country present a great challenge to research
resource allocation, because they require tailored research, recommendations and,
sometimes, technology transfer. However, the effort to do this has been greatly enhanced

by the new regional classification, as used in this study.

122

Considerable human, physical and fiscal resources have been spent on maize research
and extension in the past decade or so, but there seem to be little change in the level and
type of technologies used, and the level of productivity achieved by maize farmers. For
instance, between 1992 and 1998, there was a 5-6% decrease in the use of hybrid seed by
small and large farmers, and a 9% and 16% increase in the use of basal fertilizer in small
and large farms, respectively. These changes, though slight, took place afier restructuring

of the Kenya’s maize marketing.

Most farmers prefer maize variety (H614, first developed in 1976 and renewed in 1986)
and unimproved “local” maize varieties. This issue raises three questions: (1) where are
the new hybrids and open-pollinated varieties that were released after H614; (2) why are
these technologies “on-the-shelf’; (3) why were fewer varieties released from the
research institution in the 19905 compared to the 19603, 1970s and 19805? These
questions are beyond the scope of this study but should be considered by research

managers and policy makers.

It is worth noting that the increase in the number of farmers using fertilizer on maize in
1998 was accompanied by a decline in the amount of fertilizer used at planting, with most
farmers attributing the change to higher input costs and lower post-liberalization market
incentives. Combining this with the fact that there seems to be declining soil fertility and
little change in varietal choices for farmers, it is not surprising that maize yields have
stagnated over the past decade. In fact, the average yield for small farmers was 1.38 t/Ha

in 1992 and 1.59 t/Ha in 1998, and for large farmers these were 2.62 t/Ha and 2.73 t/Ha,

123

respectively, signifying almost no change in productivity over the 1992-98 period. In
addition, the gap between farmers’ and research yields at 50—70% is large, and points to
an unexploited potential. Meanwhile, the outcome of stagnating maize productivity in the
face of a rapid population growth is increases the reliance on imports to meet firture
maize demands, which is hard for an economy that is increasingly lacking foreign

exchange earnings.

Two important trends from the data in Chapter 2 are worth highlighting because they
raise serious policy concerns for the future of maize production in Kenya. First, there was
a large decline (45%) in the number of farmers that had access to extension services in
1998 compared to 1992. According to Karanja (1996), the increase in maize production
in the late 19605 and 19705 is attributed to adoption of improved maize seeds and
complementary technologies made possible by a dynamic synergy between the research,
extension and the seed multiplication and distribution programs.15 However, a recent
expiration of the Training and Visit (T&V) extension program of the World Bank and
lack of alternative funding sources are blamed for the decline in extension access, and

underscore the need for sustainable funding of such essential services.

The second trend of concern is the continued lack of access to farm credit, especially
among small farmers, which has been identified as a significant constraint to agricultural

development in Kenya (KARI, 1993; Karanj a, 1996). In the past, only about 10% of

 

15 Other studies, such as Karanja (1990) and Hassan, Karanja and Mulamula (1998), have
indicated the importance of agricultural extension in the expansion and growth of
Kenya’s maize production.

124

maize farmers have been able to access low-interest credit offered by the now collapsed
govemment-controlled Agricultural Finance Corporation, mainly as seasonal crop loans
(Hassan, Karanja and Mulamula, 1998). So far, none of the commercial banks lend to
smallholder maize farmers, and this is bound to limit the level of agricultural

intensification needed to raise agricultural productivity.

The effects of maize market liberalization in Kenya are not clear; the government
backtracked several times on their commitment to fully liberalize maize markets, such
that intended policy and structural changes took long to occur (Jayne and Kodhek, 1997;
World Bank, 2000). However, the data also reveals that, with maize liberalization, there
was an increase in the number of farmers who are net sellers of maize. This is may be
partly because there was an increase in the number of maize traders, who operated
bicycles and small pick-up trucks, and managed to penetrate remote areas, compared to
the monopoly marketing board that opened maize-buying depots in major maize-growing
regions during the previous, govemment-controlled maize marketing era. In general,
farmers had no preference for either the current or the former marketing system; the
current marketing system offers prompt payment for produce purchases but only controls
small volumes of sales at any one time, and tend to bid for low prices. In contrast, the
former marketing boards offered higher producer prices, bought large volumes but

delayed payments (N yoro, 1992).

Two characteristics related to the farmers raise concern - age and gender. The average

age of maize farmers is high (44 years) and almost equivalent to Kenya’s average life

125

expectancy of 47 years, at a time that there is a high level of unemployment for young
people in rural areas. This is a tough policy issue to resolve since it involves traditional
land bequeathing systems and lack of land ownership rights for women and young men.
Controlling for experience, age is inversely related to technology adoption, and has
implications on farm productivity (Feder and Umali, 1993; Hassan and Karanja, 1997).
The second issue is gender. In 1998, women farmers comprised one-half of the small
farmers and less than one-quarter of the large farmers that were interviewed, reflecting a

gender bias on land allocation.

In summary, there are numerous technological, institutional and policy issues that require
urgent attention if maize production is to be increased to meet growing consumption
demand. The past decade has shown little progress in form of new technologies,
increased adoption of these technologies and productivity increases. Therefore, there has
been stagnation in maize yields, which when coupled with limited arable land for maize
area expansion, is cause for diminished productivity and increased reliance on maize
imports to meet the growing maize consumption demand. However, importing maize is
often a difficult experience for Kenya for three reasons. First, the timing of importation
has been a problem in the past conflicting with domestic production by depressing local

prices at harvest.
Second, Kenya’s economy is hard-pressed and will find it hard to finance large imports,

Finally, the roads infrastructure has deteriorated significantly in recent years such that the

cost of distribution of such maize will make it expensive to the detriment of consumer

126

welfare. Therefore, a viable long-term strategy must include developing or adapting
maize technologies (mostly agronomic and yield-loss reduction rather than breeding new
high-yielding varieties) and developing supportive institutional capacity for extension,

credit and infrastructure that lower the cost of production and increase productivity.

5.1.2 From Multi-Market Model Simulations

The decision by research managers and policy makers about where to invest scarce
research resources becomes extremely difficult without information on how research
products or technologies affect peoples’ livelihoods. This study chose maize because it is
the most popular food crop in Kenya, and also because it is the most researched, thus, it
has numerous data useful for such elaborate simulations. This section summarizes the
findings of the impacts of “on the shelf” maize technologies on income and profits for
different farms household in different regions, and also for urban consumers, under
different maize market policy scenarios in Kenya. Basing the analysis on farm size and
regional categories fits the way Kenya’s agricultural development debate is currently

framed.

Simulations in Chapter 4 show that in 15 years, in the absence of injection and adoption
of new streams of maize technologies, maize import demand will increase by 165% over
the current levels of imports if a fixed-price market regime is pursued and maize prices
will increases by nearly 60% if a flexible-price maize policy regime is followed. In both

cases, real agricultural wages will decrease by 60% as population grth leads to

127

increases in labor supply, which in the absence of labor-demanding technological change,
or emigration, puts a downward pressure on wages. Technological change, especially in
the high potential regions, is associated with increased maize productivity and subsequent
reduction in maize import burden, as well as increased farm profits and real income.
More specifically, the net impact of new technologies on real household incomes depends
on change in wages and profits engendered by adoption of those technologies and the

relative importance of these two components in the household income.

Simulation results indicate that the positive effects of wage increases on farm labor
income generally outweighed the negative profit effects in the overall distribution of
benefits and losses. The impacts on labor earnings were particularly beneficial for small
farmers, especially in marginal regions where the maize technology is more labor
intensive. Therefore, wages are more profoundly affected in scenarios in which
technology adoption occurs in the marginal regions. In contrast, diffusion of improved
technologies in the high potential regions has a much stronger positive impact on national
maize output than scenarios in which diffusion is confined to marginal regions, because

of larger output effects in the former.

Typically, changes in the farm profits of different farm households in different locations
depend on two factors: whether or not the household is a technology adopter and how
labor intensive maize farming is for the particular households. Farm profits inevitably fall
for non-adopters since maize productivity stays the same but the cost of production rises

due to higher wage rates. For adopters, profits rise or fall depending on the relative

128

balance between yield increases brought about by the new technology and higher cost of
production caused by wage increases. In terms of the magnitude of simulated income
effects of technology adoption, it is clear that the largest increases in profits are achieved

in the high potential regions.

The results of the simulations suggest some general conclusions regarding the potential
welfare effects of diffusion of maize technologies currently on the shelf in Kenya. First,
maize technologies that have been developed for high potential regions will continue to
have greater aggregate impact on maize production and will, thus, lead to greater
reductions in import demand if prices are controlled or reduction in maize prices if maize
prices are flexible. Second, diffusion of technologies in these regions is likely to have
substantially greater positive impacts on aggregate real income and farm profit, with
effects being greater for simulations in which technology diffusion occurs in the high
potential regions, with or without diffusion in marginal regions. This finding is consistent
with other analyses comparing welfare impacts of technical change across production

environments (e. g., Renkow, 1991; Coxhead and Warr, 1991).

Third, the way in which the maize market clears has important ramifications for both the
magnitude and distribution of gains and losses from various scenarios of technology
adoption. When maize prices are exogenously determined, aggregate income increases
are generally somewhat greater and the number of household types that suffer real

income losses is smaller than when prices are endogenously determined. A notable

129

exception to this latter point is urban households, for whom welfare increases when

prices are endogenous and is unchanged when prices are exogenous.

Simulations assuming exogenous maize prices increases of 20 percent in the case of
fixed-price model or doubling of current levels of maize imports in the case of the
flexible-price model, with or without technological change taking place, have larger
effects on aggregate income and farm profits. The former generated large positive
impacts on farm profits and real income, greater than that generated by technological
change in either or both the production environments. This is because technological
change are limited by relatively low net yield potential in most zones and the fact that
technological change is accompanied by negative wage effects, especially in marginal

regions where the technology is more labor-intensive.

However, increasing maize prices alone has minimal impact on reducing import demand
and uncertain output effect. Indeed, the aggregate effects on income are maximized when
the exogenous price rise is accompanied by technological change, indicating there is a
synergy between both policy options. But it is important to note that there is a cost to
achieving any of these options, including funds to boost maize prices and facilitate
adoption of these technologies, and the money must come from somewhere, a discussion

that is currently beyond this study.

On the other hand, an increase in maize imports in the flexible-price model diminishes

aggregate farm profits and real income. The accompanying decline in maize prices, when

130

output increases from adoption of technology in one or both the production regions or
from more importation of maize, increases the welfare of urban households and other net-
consuming households that benefit fiom food price declines. In contrast, net producers
suffer loss from price declines if their output effects from technological change do not
compensate them from price decline. This demonstrates that policy choices matter and

may have great ramifications on who gains and who loses.

The analysis in this study also show that the diffusion of maize technologies in marginal
regions has better income distribution outcomes although the aggregate output impacts on
incomes are smaller than technologies targeted for high potential regions. As to which of
these goals - increasing income levels or increasing equity - is more desirable policy
choice is a matter that policy makers will have to decide, especially in lieu of the wider
market liberalization policy debate, and important decisions on maize research and

development resource allocations.

In conclusion, this study set out to simulate the impacts of technological change on farm
profits and income both within and across different production environments. The overall
results indicated that the effects are indeed different, both for different household types
(including urban consumers) and different production regions. Prior knowledge of these
effects before selection and implementation of any of the policy strategies is extremely
important. It is wrong to assume that, as has been the case, raising producer prices will
solve all the problems and make everyone better off, since it is evident that some

households will gain and others lose. Also, it is important to note that there are other

131

critical avenues of transmission of technological benefits and losses, rather than just
commodity markets. The very design of appropriate technologies targeted for different
regions must take that into consideration. This study shows that, beyond the commodity
market effects, there were significant labor market effects, especially for small farmers.
Such information and ex-ante knowledge of potential income distribution effects of
technologies at the design phase can be a great asset to researchers and policy makers to

achieve desirable equitable development objectives.

5.2 Study Limitations

Several limitations of the multi-market model used in this study are worth noting. First,
the model is simplified to enhance the ability to interpret the results, which means that its
capacity to capture “real world” effects is deliberately compromised. A two-input, two-
output production system and a two—product consumption system were assumed to allow
following through of the impacts of changes in production, consumption, price, etc. due
to simulated external technological changes. However, failure to do these would yield

results that are too complex and meaningless.

Second, data is always a major limitation in studies like this. Lack of data and time to get
actual estimates of various elasticities made it necessary to “borrow” values from other
studies. Although sensitivity analysis was done on these parameters to test the level of
distortion introduced on the base output by using these values, it is always more certain

when using actual data applicable to a specific region. Related to this, making meaningful

132

income distribution assessments requires assembling information on the initial income
distribution across the various households. This facilitates comparison with simulated
posterior distributions. Meanwhile, the ability to geo-reference demographic, agro-
climatic and socio-economic data allows for an integrated spatial analysis that is more
realistic. Therefore, the use of GIS to collate needed data and information may be useful

in enhancing the accuracy of such models.

Third, the model results underestimate the impacts of technological change since these
are limited to effects of changes in farm profits, changes in labor demand hence real
wages, and changes in maize prices. For instance, the model does not account for changes
in rural incomes from multiplier effects of investing surplus agricultural income on rural
non-agricultural activities, it does it include effects of non-agricultural income, which is
considered exogenous, and also ignores changes in nominal urban wages from changes in

food prices and rural-urban migration.

Fourth, this model represents snapshots of two static time periods, one before
technological change and the other after the effects have occurred, rather similar to
assuming instantaneous changes. Such a model, therefore, cannot capture interesting
dynamic effects of market equilibration. However, the snapshots of “with” and “without”
technological change provide important information on technology effects and policy

implications.

133

Lastly, this modeling does not account for the cost of implementing the said
technological and price changes, and therefore, cannot be used to estimate potential
benefit-cost ratios of different technology and policy options that may be needed for a
more comprehensive analysis of these scenarios to inform priority setting and definite
resource allocation. But still, the current analysis makes important contribution to
estimating the direction and scale of impact of different technologies on different
households in different regions, and these can be readily updated as more information

becomes available.

134

APPENDICES

135

APPENDIX A

1999 Farm-Level Survey Questionnaire

136

ACCEPTANCE TO PARTICIPATE/ RUHUSA YA KUHOJIWA:

Tunafanya utafiti kuhusu umuhimu wa teknolojia ya mahindi kwa kuongeza mazao na
mapato kutokana na ukuzaj i wa mahindi katika sehemu mbali mbali za Kenya.
Tutashukuru kwa usaidizi wako wa habari kuhusu ukuzaji wa mahindi na jinsi wewe
unavyoj aribu kuongeza au kudumisha ukuzaji.

MAJIBU YAKO NI YA HIARI NAITAWEKWA SIRI KABISA. Baadaye
yatajumulishwa na mengine mia tatu ili kutoa ripoti kuhusu teknolojia ya mahindi.

1. Location Identification (LOC): Complete this section last.

 

 

 

 

District: Village:

Division: Cluster Code: Household No:
Location: Date of Interview:

Sub-Location: Name of interviewer:

 

 

 

 

2. Farmer Identification:
a. Name of farmer: b. Sex: ( ) Male; ( ) Female
0. Age (Years): (1. Education: ( ) None; ( ) Primary; ( ) Secondary; ( ) Degree/Diploma

3. Farm Production System:

 

 

 

 

 

 

a. Total farm size (acres): Own__; Rented_; Shared—

b. Total area under maize:

i. Variety acres_____ ( )pure; ( )intercropped with
ii. Variety acres___ ( )pure; ( )intercropped with
iii. Variety acres___ ( )pure; ( )intercropped with
iv. Variety acres__ ( )pure; ( )intercropped with

 

 

c. Maj or enterprises besides maize (acres):

 

 

 

 

 

 

 

 

i. acres ( ) pure; ( )intercropped with
ii acres ( ) pure; ()intercropped with
iii acres ( ) pure; ( )intercropped with
iv acres ( ) pure; ( )intercropped with
v acres ( ) pure; ( )intercropped with

 

 

d. Major maize production season: ( ) 1st rains (Mar-Aug); ( ) 2nd rains (Oct-Feb); ( ) Both

137

e. List major constraints to maize production on your farm in the order of importance:

i.

 

ii.

 

iii

 

iv

 

V

 

4. Importance of Maize Production:
a. List the reasons why you grow maize (in order of importance):

i

 

ii

 

iii

 

iv

 

b. Among the crops you grow, which one(s) is (are) the major source of food (in the order

 

 

 

 

of importance)?

i iv
ii v
iii vi

 

 

Among the crops you grow, which one(s) is (are) the major source of cash (in the order of

 

 

 

 

importance)?

i iv
ii v
iii vi

 

 

138

 

5.a. Maize Production Information by Major Plot:

 

 

 

 

 

 

 

Attribute Long Rains 1998 Short Rains 1998

Area of major maize (Acres) (Acres)

plot

Location ( )Home; ( )Away; Ifaway, ( )Home; ()Away; If away,
distance from home (km) distance from home _(krn)

Slope of field ( )0-25%; ()25-50%; ( )0-25%; ()25-50%;
()50-75%; ()>75% ( )50-75%; ( ) >75%

Soil erosion problem? ( )yes; ( )no ( )yes; ( )no

 

Erosion control

( )none; Method

 

( )none; Method

 

 

 

 

 

method?
Land Tenure ( )own; ( )use right; ( )rent; ( )own; ( )use right; ( )rent
( )share; ( )other ( )share; ()other
Cropping system ( ) pure; () intercrop; ( ) other ( ) pure; ( ) intercrop;
‘ ( ) other
If intercrop, with what? main _; minor main ; minor

 

139

 

5.b. Maize Technology Characteristics & Availability

 

Attribute

() Long Rains 1998; () Short Rains 1998

 

Name of Major Variety

; Acres

 

 

Seed rate on major plot

Kg/Plot or Kg/Acre

 

Source of Seed

( )buy; ( )own;

other

 

 

Reason for choosing major variety

( )drought tolerant; ( ) stores better;
( )sweeter; ()yields higher; ( )stable yield;
( )better fodder; other

 

If using non-Hybrid:

Why not use hybrid

( )expensive; ( )not available;

( ) don’t know how to use; ( ) need
additional labor; ( )not aware of seed;
( )other

 

Seed available at right time:

( )always; ( )sometimes; ( )never

 

Seed available at local store:

( )always; ( )sometimes; ()never

 

Appropriate seed available

( )always; ( )sometimes; ()never

 

Use Basal fertilizer
Basal Amount
Topdress Fertilizer
Topdress Amount
Manure

Manure Amount

 

 

( )none; ( )type
Kgs or (_Kg-Bags/plot)
( )none; ( )type
Kgs or L_Kg-Bags/plot)
( )yeS; ( )no
Kgs or units:

 

 

Ifnot using fertilizer,
why not?

( )expensive; ( )not available;
( )don=t know how to use;
( )other

 

Is fertilizer available at the right time?

()always; ( )sometimes; ( )never

 

Is fertilizer available at the local store

( )always; ( )sometimes; ()never

 

Is the appropriate fertilizer available?

( )always; ( )sometimes; ()never

 

 

Is manure available?

 

( )always; ( )sometimes; ()never

 

140

 

5.c. Other Maize Farm Operations

 

 

 

 

 

 

Attribute () Long Rains 1998; () Short Rains 1998
Tillage method ()tractor; ( )oxen; ( )hoe; ( )oth
Tillage date Month Week: 1 2 3 4
Planting method ( )tractor; ( )oxen; ()hoe; ( )oth
Planting date Month Week: 1 2 3 4
Relation to Onset of Rain Before At After

Number of weeding(s)

Method of lst weeding ( )tractor; ( )oxen; ( )hoe; ( )chem
Date of lst Weeding Month week: 1 2 3 4
Method of 2nd weeding ( )tractor; ( )oxen; ( )hoe; ( )chem
Date of 2nd Weeding Month week: 1 2 3 4
Control field pest ( )no; ( )chem; ( )other

 

Other chemical(s) used

( )none; name:

 

 

Amount of chemical used

units

 

 

Chemical availability

( )always; ( )sometimes; ()never

 

Harvest green?

How much?

( )no; ( )yes;
00-25%; ( )25-50%; ()50-75%;
( )75-100%

 

( )yes; ( )no

Is there land to rent?

 

Ifnot renting land, why? ( )expensive; ( )not available;

 

 

 

 

5.d. Alternative Crop:

i. After maize, what is the next most important crop you grew in 1998?

 

ii. Would you vary the amount of land between maize and this crop? ( )yes; ( )no
iii. Would you vary your labor allocation between maize and this crop? ( )yes; ( )no

iv. Would you vary your fertilizer between maize and this crop? ( )yes; ( )no

 

v. Planting date for the alternate crop in 1998: Month—week: l 2 3 4
vi. Harvesting date for the alternate crop in 1998: Month—week: l 2 3 4
vii. Yield of alternate crop in 1998: (units__/acre)

viii. Yield of alternate crop in a normal year: (units__/acre)

 

141

6. Total Farm Production in 1998 Long and Short Rains

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Crop or Season Total Total Sale Total Purchase
Livestock Production Sales Price Bought Price
Product (Kgs, Ltr) (Kgs, Ltr) (Ksh/unit) (Kgs,_Ltr) (Ksh/unit)
1. Long
Short
2. Long
Short
3. Long
Short
4. Long
Short
5. Long
Short
6. Long
Short
7. Long
Short
8. Long
Short
9. Long
Short

 

142

 

7.a Production Costs on An Acre of Maize and Alternative Crop

 

Attribute Maize Alternate Crop

 

Seed Quantity

 

Fertilizer Quantity

 

Manure Quantity

 

Pesticide/Herbicide Quantity

 

Tillage costs: Tractor

 

 

(Hours)

 

 

Oxen (Hours)
Hoe (Days)

 

Planting costs: Tractor

 

 

(Hours)

 

 

Oxen (Hours)
Hoe (Days)

 

lst Weeding (Days)

 

 

2nd Weeding (Days)

 

Fertilizer application (Days)

 

Pesticide/Herbicide
application (Days)

 

Harvesting (Days)

 

Bagging (DaYS)

 

Total Transport cost (Ksh)

 

Total Marketing cost (Ksh)

 

Other costs (Ksh)

 

 

 

 

 

 

 

143

7.b. Maize Labor Source, Cost and Availability in 1998 (indicate units of measure)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Activity Tillage Planting Weeding Harvesting Shelling Selling
ays of Family Labor:
dults:
ids (<12yrs):
ours of Hired Labor
n maize
* 1=Family; 2=Hired; 3=Both; ** l=Always; 2=Sometimes; 3=Never
8.a. Input Use (Compute out of 100%):
Maize Alternate Other Livestock Total
Crop Crops
Family Labor 100%
Hired labor 100%
Fertilizer 100%
Other Inputs 100%
8.b. Family Labor:
Name Sex Age Yrs spent in Occupation Days devoted Days devoted Days devoted to
School to maize to other farm off-farm
production production activities
activities on- activities on-
farm farm
1
2
3
4
5

 

144

 

 

 

8.c. Total Output for the Year (Quantity or Value):

 

Attribute

Maize

Alternative

Crop

Other
Crops

Livestock

Off-Farm
Labor

Other

Income

 

Total Output
_(Kg or Bags)

 

Gross
Revenue

(Kshs)

 

 

 

 

 

 

 

 

9.1!. Marketing Channel and Prices

 

Attribute

Maize

Alternate Crop

 

Month of main crop harvest

 

Month of most crop sale

 

How much sold in that month (Kgs or Bags)

 

Price Received (Ksh/ )

 

Main buyer“

 

 

Current price in 1999 (Ksh/ ):
Lowest price in 1998 (Ksh/ ):
Highest price in 1998 (Ksh/ ):

 

 

 

"' 1=Government/NCPB; 2=Trader; 3=Miller; 4=Direct consumers; 5=others

145

 

 

9.b. Access to Support Services & Infrastructure

3. Received extension advise? ( )never; ( )regularly; ( )once in the past 5 yrs;

( ) once in past 10 yrs

a. Major source of your maize advise? ( )MoA agent; ( )other farmers; ( )input seller; ( )NGO;

( )local church; ()others

b.Have you received credit for maize production? ( )never; ( )regularly; ( )once in the past 5 yrs;
( )once in past 10 yrs

0. Source of maize credit: ( )AFC; ( )KFA; ( )NGO; ( )Church; ( )Bank; ( )neighbour;

( )others

d.Have you received credit for other farm activities? ( )yes; ( )no. If yes, for what

e. Source of other credit: ( )AF C; ( )KFA; ( )NGO; ( )Church; ( )Bank; ( )neighbour;

( )others

g. How far is your farm from a tarmac road? krns; from a murram road? krns
b. How far is your farm from a seed/fertilizer dealer? krns

i. How far is your farm from a T&V demostration site? krns

10. Final Question

i. Would you ever reduce the amount of your land under maize? ( )yes; ( )no

ii. If yes, why?

 

iii. What would you plant instead?

 

iv. Would you ever stop growing maize? ( )yes; ( )no

146

APPENDIX B

1999 Village-Level Survey Questionnaire

147

ACCEPTANCE TO PARTICIPATE/ RUHUSA YA KUHOJIWA:

Tunafanya utafiti kuhusu umuhimu wa teknolojia ya mahindi kwa kuongeza mazao
na mapato kutokana na ukuzaji wa mahindi katika sehemu mbali mbali za Kenya.
Tutashukuru kwa usaidizi wako wa habari kuhusu ukuzaji wa mahindi na jinsi
sahemu hii inavyoj aribu kuongeza na kudumisha ukuzaji wa mahindi.

MAJIBU YAKO NI YA HIARI NAITAWEKWA SIRI KABISA. Baadaye
itajumulishwa na habari kutoka sehemu nyingine za Kenya ili kutoa ripoti kuhusu
ukuzaji na teknolojia ya mahindi.

1. Location Identification: Complete this section last.

 

 

 

 

District: Village:

Division: Cluster Code:
Location: Date of Interview:
Sub-Location: Name of interviewer:

 

 

 

 

b. Respondents Name:

 

c. Position in village: ( ) area chief; ( ) extensionist; ( ) politician; ( ) business
person; ( ) church leader; ( ) other (specify)

2. Village Infrastructure:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Infrastructure Is it available in If not, how far is it from
village? village?
Tarmac road Yes _; No_ krns
Murram road Yes_; No_ krns
Tap water Yes__; No kms
Electricity Yes ; No_ krns
Hospital Yes __; No_ krns
Grain Silos Yes _; No_ lans
Demonstration site Yes ___; No_ krns

 

148

3. Village Maize Production Status:

 

Period/Y ear

How much maize produced in the village?

 

Before market liberalization (1994)

( )Less than enough; ( )enough; ( )surplus

 

1994 to 1997

( )Less than enough; ( )enough; ()surplus

 

1998

 

 

( )Less than enough; ( )enough; ( )surplus

 

4: Village Institutions:

 

 

 

 

 

 

 

 

 

 

 

Institution Available in If not, how far % of villagers

village? (krns) using this
service
Agricultural credit ( )yes; ( )no
Govt maize buyer ( )yes; ( )no
Private maize buyer ( )yes; ( )no
Maize seed stockist ( )yes; ( )no
Fertilizer stockist ( )yes; ( )no
Agricultural ( )yes; ( )no
extension

 

5. Alternate Crap:

a. Besides maize, list 3 major crops grown in this village in order of importance:
'. " iii.
b. Will increased production of the alternate crop identified in (3) require:
iv. More labor ( )yes ( )no;

1

11.

i. New markets ( )yes ( )no;

ii. New credit sources ( )yes ( )no;

iii. Better/New roads ( )yes ( )no; vi. Other

 

v. New extension effort ( )yes ( )no

149

9

 

 

6: Prices of Maize and Alternate Crops (Ksh/ ):

 

Crops

Maize

Crop 2

Crop 3

Crop 3

 

1998:

Sale Price (highest):
Sale Price (lowest):
Purchase Price (highest):

Purchase Price (lowest):

 

1999:

Sale Price (highest):
Sale Price (lowest):
Purchase Price (highest):

Purchase Price (lowest):

 

Where sold in 1998?*

 

Who bought in 1998?“

 

 

 

 

 

 

‘1=Within village; 2=Neighbouring village; 3= Urban Market; 4=Export/Import;

5=other

"'* 1=Govt; 2=next door farmer; 3=local trader; 4=trader from outside village;

5=other

150

 

7. Average Prices of Selected Farm Inputs (indicate units)

 

Inputs 1998 1999

 

Maize seed: Variety l (Ksh/kg)
Variety 2 (Ksh/kg)
Variety 3 (Ksh/kg)

 

DAP Fertilizer

 

CAN fertilizer

 

Urea Fertilizer

 

TSP Fertilizer

 

MAP Fertilizer

 

Pesticide

 

Herbicide

 

Manual Labor: Ploughing:
Weeding:

Harvesting:

 

Tractor Hire: Ploughing:

 

Oxen Hire: Ploughing

 

 

Others:

 

 

 

151

 

8. Village Transport and Proximity to other Markets.
a. Do traders come to the village? ( )yes; ()no

b. If not, what do farmers commonly use to transport maize to the market: ( )lorries;

( ) tractors; ( )pick/Ups; ( )bicycles; ( )donkeys/animal; ( )on backs/heads; ( )others_
b. What is the distance and road type from this village to the nearest:

Permanent market: krns; Type of road: ( )tarmac; ( )murram; ()dirt; ( )other
Urban Centre/T own kms; Type of road: ( )tarmac; ( )murram; ( )dirt; ( )other

9. Village Labor Supply and Demand for Maize Production

a. What is the major source of labor in this village? ()family; ( )hired; ( )comrnunal;
( ) others

b. If hired labor, from where? ( )nearby village; ( )within the village; ( )from a far off;
( ) others

c. Is there migrant labor into the village? ( )yes; ( )no;

(1. Is there migrant labor out of the village? ()yes; ( )no;

e. Which month does labor demand peak?
f. Is it hard to get hired labor at peak season? ( )no; ( )hard; ( )very hard; ( )impossible

 

10. Importance of Maize and Maize Technology

a. What proportion of village depend on maize as a major source of food?

( ) 0-25%; ()25-50%; ( )50-75%; ()75-100%

b. What proportion of village depend on maize as a major source of cash income?
( ) 0-25%; ()25-50%; ()50-75%; ()75-100%

c. List major constraints to increasing maize production in this village (in order of importance):
1st

2nd

3rd

4th

(1. How was the maize produce marketing in the village in 1998 compared to the past?
( ) bad; ( ) fair; ()good; ( )excellent

c. How was the maize seed marketing in the village in 1998 compared to the past?

( ) bad; () fair; ( )good; ( )excellent

 

 

 

 

152

BIBLIOGRAPHY

153

Bibliography

Argwings-Kodhek, G. 1997. “Factors Constraining Fertilizer Use in Kenya.” Working
Paper. Tegemeo Institute of Agricultural Policy and Development. Nairobi,
Kenya.

BF WI. 2003. Agriculture in the Global Economy. Hunger Report 2003. Bread for the
World Institute. Washington, DC.

Blackie, M. 1990. “Maize, Self Sufficiency and Policy in East and Southern Afiica.”
Food Policy. 15(5): 383-94.

Byerlee, D. 1994. “Technology Transfer Systems for Improved Crop Management:
Lessons for the Future.” In J.R. Anderson (ed), Agricultural Technology Policy
Issues for the International Community. CAB International, Wallingford, UK. pp.
208-230

Byerlee, D. 1996. “Modern Varieties, Productivity and Sustainability: Recent
Experiences and Emerging Challenges.” World Development. 24(4): 697-718

Byerlee, D. and M. Morris. 1993. “Research for Marginal Environments: Are we
Underinvested?” Food Policy. 18(5): 381-393.

Byerlee, D. and CK. Eicher, 1997. “Accelerating Maize Production: Synthesis.” In
D. Byerlee and CK. Eicher (eds), Afiica’s Emerging Maize Revolution. Boulder,
CO: Lynne Rienner Publishers.

Cobbert, J. 1998. “Classifying Maize Production Zones in Kenya through Multivariate
Cluster Analysis.” In R.M. Hassan (ed), mize Technology Development Ed
_"Ilansfer: A GIS Application for Research Planning in Kenya. CAB International,
Wallingford, UK. pp. 15-25.

Coxhead, I. and P. Warr. 1991. “Technical Change, Land Quality, and Income
Distribution: A General Equilibrium Analysis.” American Journal of Agricultural
Economics 73: 345-360.

David, CC. and K. Otsuka (eds). 1994. Modern Rice Technology and Income
Distribution in Asia. Boulder, CO: Lynne Rienner Publishers.

Dowswell, C. R., R. L. Paliwal and R. P. Cantrell. 1996. Maize in the Third World.
Boulder, CO: Westview Press.

154

Edmeades, G.O., J. Bolanos, M. Banziger, J .M. Ribaut, J .W. White, M.P. Reynolds, and
HR. Lafitte. 1998. “Improving Crop Yields Under Water Deficits in the Tropics.”
p. 437-451. In V.L. Chopra, R.B. Singh and A. Varma (eds), Crop Productivity
and Sustainability - Shaping the Future. Proceedings of Second International Crop
Science Congress. Oxford and IBH, New Delhi.

Fan, S. and P. Hazell. 2001. “Returns to Public Investment in the Less-Favored Areas of
India and China.” American Jouml of Agricultural Economics. 83(5): 1217-1222

FAOSTAT, 2001. Agriculture Data. http://apps.fao.org/page/collections

FAQ-UNESCO. 1974. Soil Map of the World, 1: 5,000,000. Vol. I Legend. UNESCO.
Paris, France.

Hassan, RM. and DD. Karanja, 1997. “Increasing Maize Production in Kenya:
Technology, Institutions and Policy.” In D. Byerlee and CK. Eicher (eds),
Africa’s Emerging Maize Revolution. Boulder, CO: Lynne Rienner Publishers.

Hassan, R., J. Lynam, P. Okoth., 1998. “The Spatial Sampling Frame and Design for
Farmer and Village Surveys.” In R.M. Hassan (ed), Maize TechnM
Development and Tranafer: A GIS Application for Research Planning in Kenya.
CAB International, Wallingford, UK. pp. 27-43.

Hassan, R., K. Njoroge, M. Ngure, R. Otsyula and A. Laboso. 1998. “Adoption Patterns
and Performance of Improved Maize in Kenya.” In R.M. Hassan (ed), Maize
Technology Deve10pment and Tra_Lr_sfer: A GIS Applfition for Resfiearch Planning
in Kenya. CAB International, Wallingford, UK. pp.107-136.

Hazell, P. and J .L. Garrett. 1996. “Reducing Poverty and Protecting the Environment:
The Overlooked Potential of Less-favored Lands.” 2020 Brief No. 39,
International Food Policy Research Institute. Washington, DC.

Heisey P. and W. Mwangi, 1997. “Fertilizer Use and Maize Production.” In D. Byerlee
and CK. Eicher (eds), Afiica’s Emerging Maize Revolution. Boulder, CO: Lynne
Rienner Publishers.

Jaetzold, R. and H. Schmidt. 1983. Farm Management Handboolgof Kenya —— Volume I,
II and III. Ministry of Agriculture. Nairobi, Kenya.

Jayne, TS, and Gem Argwings-Kodhek. 1997. “Consumer Response to Maize Market
Liberalization in Urban Kenya.” Food Policy. 22(5): 447-458.

Karanja, DD. 1990. “The Rate of Return to Maize Research in Kenya: 1955-88.”

Masters Thesis. Department of Agricultural Economics, Michigan State
University. East Lansing, MI.

155

Karanj a, DD. 1996. “An Economic and Institutional Analysis of Maize in Kenya.”
International Development Working Paper No. 57. Department of Agricultural
Economics, Michigan State University. East Lansing, MI.

Karanja, DD. 1999. Unpublished Data. Kenya Maize Impact Survey (KMIS).

Karanja, D.D., T.S. Jayne and P. Strasberg, 1999. “Determinants of Input Use and Maize
Productivity in Kenya: Implications of Cereal Market Reform.” In Conference
Proceedings on Strategies for Raising Productivity in Kenyan Agriculture.
Tegemeo Institute of Agricultural Policy and Development. Nairobi, Kenya.

Kenya, Government of -. 1993. The 1994-98 Kenya Development Plan. Government
Printers. Nairobi, Kenya.

Kenya, Government of -. 1994. _B_asjc Report 11: Kenya Welfafi Monitoring Survey.
Central Bureau of Statistics. Office of the Vice President and Ministry of
Planning and National Development. Government Printers. Nairobi, Kenya.

Kenya, Government of-. 1998. First Report on Poverty in Kenya. Volame II: Poverty and
Scml Indicators. Central Bureau of Statistics. Ministry of Planning and National
Development. Government Printers. Nairobi, Kenya.

Kenya, Government of-. 1999. Unpublished Data. Department of Resource
Surveys and Remote Sensing. Ministry of Planning and National Development.
Nairobi, Kenya.

KMDB (Kenya Maize Database Project). 1992. Unpublished Data. Kenya Agricultural
Research Institute. Nairobi, Kenya.

Lipton, M. and R. Longhurst. 1989. New Seed_s_ and Poor People. Baltimore: John
Hopkins University Press.

Lyne, M., M.J. Roth and B. Troutt, 1997. “Land Rental Markets in sub-Saharan Afiica:
Institutional Change in Customary Tenure.” In R. Rose, C. Tanner and MA.
Bellamy (eds), Issues in Agricultural Competitiveness: Mflets and Policiqa.
I.A.A.E. Occasional Paper No.7. International Conference of Agricultural
Economists, held in Harare, Zimbabwe on 23-29 August, 1994. pp. 58-67.

Matlon, P. and D. Spencer. 1985. “Increasing Food Production in Sub-Saharan Afiica:
Environmental Problems and Inadequate Technological Solutions.” American
Journal of Agriculturaflconomics. 66: 671-676.

Migot-Adholla, S., P. Hazell, B. Blarel and F. Place. 1991. "Indigenous Land Rights

Systems in Sub-Saharan Afiica: A Constraint on Productivity. " Economic
Review. 5(1): 155-175. Washington, DC: The World Bank

156

Mills, B.F. (Ed.). 1998. Agricultural Research Priority Setting: Information Investments
for the Improved Use of Research Resources. The Hague: International Services
for National Agricultural Research.

Mills, B.F., R.M. Hassan, and P. Mwangi. 1998. “Estimating Potential Benefits from
Reearch and Setting Research Priorities for Maize in Kenya.” In R.M. Hassan
(Ed.), Maize Technology DevelopmerL and Tran_sfer: A GIS Application for
Research Planning in Kenya. CAB International, Wallingford, UK. pp. 89-104.

Mukumbu, M. and TS. Jayne. 1995. “Urban Maize Meal consumption Patterns:
Strategies for Improving Food Access for Vulnerable Urban Households in
Kenya.” SD Publication Series Technical Paper No. 8. Office of Sustainable
Development, Bureau of Afiica. Washington, DC: USAID.

Munyi, V. 2000. “Maize Supply Response in Kenya.” Unpublished Master's Thesis.
University of Connecticut, Storrs, CT.

Naseem, A. and V. Kelly. 1999. “Macrotrends and Determinants of Fertilizer Use in Sub—
Saharan Africa.” International Development Working Paper No. 73. Department
of Agricultural Economics, Michigan State University, East Lansing, MI.

Nyoro, J .K. and T.S. Jayne. 1999. “Trends in Regional Agricultural Productivity in
Kenya.” Kenya Agricultural Marketing Policy and Analysis Project. Michigan
State University and Tegemeo Institute of Agricultural Policy and Development.

Nyoro, J.K., M. Kiiru and TS Jayne. 1999. “Evolution of Kenya’s Maize Marketing
Systems in the Post-Liberalization Era.” Working Paper No. 2A. Tegemeo
Instititute of Agricultural Policy and Development. Paper Presented at the 4‘11
Agricultural Transformation Workshop. Nairobi, Kenya.

Ochieng, J. 1999. Personal Communication.

Omamo, SW. 1998. “Transport Costs and Smallholder Cropping Choices: An
Application to Siaya District, Kenya.” American Journal of Agricultural
Economics. 80: 116-123.

Omamo, SW. and L. Mose. 2001. “Fertilizer Trade Under Market Liberalization:
Preliminary Evidence from Kenya.” Food Policy. 26(1): 1-10

Pearson, S. 1992. “Issues and Options in Food Price Stabilization.” In the Conference
Proceedings on Maize Supply and marketing Under Market Liberalization held by
the Egerton University’s Policy Analysis Matrix Project at the Kenya Commercial
Bank, Karen on 18-19 June 1992. Nairobi, Kenya.

Pinckney, T.C. 1988. “Storage, Trade and Price Policy Under Production Instability:

Maize in Kenya.” Research Report No. 71. Washington, DC: International Food
Policy Research Institute.

157

Pingali, PL and PW. Heisey. 1999. “Cereal Crop Productivity in Developing Countries:
Past Trends and Future Prospects.” CIMMYT Economics Working Paper No. 99-
03. Mexico, DF: CIMMYT.

Pitt, M.M. and G. Sumodiningrat. 1991. “Risk, Schooling and the Choice of Seed
Technology in Developing Countries: A Meta-Profit Function Approach.”
International Economic Review, 32: 457-474.

Reardon T., C. Delgado and P. Matlon. 1992. “Determinants and Effects of Income
Diversification amongst Farm Households in Burkina Faso.” Journal of
Development Studifiea. 28(2): 264-96.

Renkow, M.1991. “Modeling the Aggregate Effects of Technological Change on Income
Distribution in Pakistan’s Favored and Marginal Production Environments.”
Economics Paper No. 4. Mexico, DF: CIMMYT.

Renkow, M.1993. “Differential Technology Adoption and Income Distribution in
Pakistan: Implications for Research Resource Allocation.” Americap Journal of

Agg'cultural Economics. 75: 33-43.

Renkow, M. 2000. “Poverty, Productivity, and Production Environment: A Review of
the Evidence.” Food Policy. 25(4): 463-478.

Renkow, M., J. Hallstrom and DD. Karanja. 2003. “Rural Infrastructure, Transaction
Costs and Markets in Kenya.” J owl of Develgrment Economics. Forthcoming.

World Bank. 1995. Kenya Poverty Assesament. Report No. 13152-KE. Population and
Human Resource Division, East African Department. Afiica Region. Washington,
DC: The World Bank.

World Bank, 2001. Can Afiica Claim the 21St Century? Washington, DC: The World
Bank.

158