AGRICULTURAL DEVELOPMENT IN THE CONTEXT OF FARM STRUCTURE
CHANGE IN ZAMBIA
By
Chewe Nkonde

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Community, Agriculture, Recreation and Resource Studies—Doctor of Philosophy
2017

ABSTRACT
AGRICULTURAL DEVELOPMENT IN THE CONTEXT OF FARM STRUCTURE CHANGE
IN ZAMBIA
By
Chewe Nkonde
The rise of domestic medium-scale investor farms (holdings between 5 and 100 hectares)
is ushering in a change in farm structure that perhaps signals an alternative pathway to
agricultural development in sub-Saharan Africa (SSA). This dissertation examines this change
and what it implies for agricultural development in the region through three standalone but
closely linked essays using the case of Zambia.
The first essay examines the causes and consequences of the rise of domestic mediumscale investor farms in Zambia using a mixed-methods approach. By locating the study within
the broader political economy and new institutional economics literature, the study gleans some
important findings. Results show that a positive change in society’s perception of farming, a
change in enforcement of informal rules of land governance, and unintended consequences of
public spending agricultural support programs have played a pivotal role in farm structure
change. In addition, farm structure change has been associated with agricultural land
concentration, a growing informal land rental and sales market and a skewed level of agricultural
commercialization.
The second essay re-examines the farm size – productivity relationship hypothesis in the
context of farm structure change using a comprehensive set of productivity indicators and a wide
range of farm sizes. Results show that the IR hypothesis is not consistently upheld across
alternative indicators of productivity. While the study finds that small-scale farms are more
productive when the outcome variable is either land productivity or cost of staple food

production per metric ton produced, relatively larger farms have higher labor productivity.
Relaxing the constant returns to scale assumption helps to isolate sources of productivity
differences between less than five-hectare farms and those above five hectares.
The third essay investigates recent trends in agricultural mechanization use, its effects on
agricultural household production and the factors contributing to the rise of mechanization
initiatives in Zambia. Results show that mechanization use for land preparation has remained low
and stagnant. For users, its effects on cropland expansion have been positive but this has not
translated into net gains in crop productivity. The study shows that actors on the supply side have
become active in promoting agricultural mechanization. On the demand side, however, the level
of mechanization use has remained low in part because the changes in the rental rates for
mechanization relative to agricultural labor costs have probably not changed to a level that
induces an unequivocal shift to labor saving technologies such as mechanization.
Three main conclusions are drawn from this dissertation research. First, in order to
achieve agricultural development that is broad-based and inclusive, policy strategies should
address inequalities in land access, enhance a more inclusive formalization of land ownership
and continue to support small-scale agriculture production. Second, findings on the relationship
between productivity and operated farm size, while important, should not be the decisive factor
in guiding agricultural development and land policies in SSA because there are many other
important considerations. Third, this research demonstrates that if identified pitfalls of
mechanization initiatives are not quickly addressed, sustainability of these initiatives may be
compromised.

Copyright by
CHEWE NKONDE
2017

In loving memory of Ena Manda Nkonde.

v

ACKNOWLEDGEMENTS

My doctoral studies journey would not have come to its successful completion without
the unwavering support, encouragement and guidance of a number of people. First and foremost,
I would like to recognize and appreciate each member of my PhD guidance committee—Dr.
Robert Richardson, Dr. John Kerr, Dr. Maria Claudia Lopez and Dr. Thomas Jayne—who
contributed immensely to my scholarly development while I was at Michigan State University
(MSU).
To Robby, thank you for your guidance and effective leadership throughout the last three
years. Your critical review of my work has made me a better researcher and my sincere hope is
that we will continue to collaborate beyond this dissertation research. To John, you played a
pivotal role in strengthening my scholarship especially during the time I took your course:
Foundations of Community Sustainability. I learned a lot from the way you quickly responded
and gave insightful feedback to students’ work. Maria Claudia, thank you for accepting to be on
my committee and for sharing literature and critical insights on institutions and the critical role
they play in development. And Thom, I cannot thank you enough for the important role you have
played in my career development. You faithfully supported my research assistantship position,
which catered for my tuition, living and medical expenses for the entire duration of my stay in
the U.S. It is, however, the research experience that I gained from being part of your team that I
will forever cherish. Zikomo kwambiri!!
I would also like to acknowledge the support of Dr. Scott Loveridge, Dr. Scott Swinton,
Dr. Bob Myers, Dr. Nicole Mason and other members of faculty in the Departments of
Agricultural, Food, and Resource Economics (AFRE) and Community Sustainability (CSUS).

vi

During my time at MSU, I was blessed to meet a number of very good people who I am now
proud to call friends. Because I run the risk of forgetting somebody’s name, I will not list the
names of fellow graduate students, members of staff and other friends in and around the Greater
Lansing community. To all of you (you know yourselves), I say thank you guys. You made my
time in East Lansing and Michigan much more enjoyable than I had initially envisaged.
I am indebted to my employers, the University of Zambia, for granting me study leave for
five (5) years to enable me pursue my PhD studies. Special thanks also go to Mr. Chance
Kabaghe and his entire team of directors, researchers and support staff at the Indaba Agricultural
Policy Research Institute (IAPRI) for allowing me to use their data and office space during my
dissertation research. I would also like to thank the Borlaug LEAP fellowship, and in particular
Ms. Susan Johnson, for funding my qualitative field research and providing me an opportunity to
do my internship at the International Food Policy Research Institute (IFPRI) in Washington DC
under the mentorship of Dr. Frank Place.
Finally, I would like to thank my parents who continued to believe in me. Your first-born
son is now a PhD, congratulations Mr. and Mrs. Nkonde! To my mother in law Ms. Sylvia
Mudenda, aka MIL, thank you for your love and for taking care of our two oldest sons while
both Ntombi and I were busy burning the midnight oil studying. I honestly cannot imagine how
we would have done this without you. To my dear wife Ntombi, you still remain the epitome of
beauty, elegance and intelligence. Thank you for your support, love and constant reminder that I
could do this. To my sons—Penjani, Tabiso and Aiden—who have had to endure my absence, I
hope to make it up to you moving forward. I love you boys. To crown it all, I would like to thank
God for his presence throughout this journey even when I had a lot of doubts.

vii

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................................... xi
LIST OF FIGURES ..................................................................................................................... xiii
KEY TO ABBREVIATIONS ...................................................................................................... xiv
CHAPTER 1: INTRODUCTION ................................................................................................... 1
REFERENCES ............................................................................................................................... 7
CHAPTER 2: THE RISE OF DOMESTIC MEDIUM-SCALE FARMS IN ZAMBIA: CAUSES
AND CONSEQUENCES ............................................................................................................. 10
2.1. Introduction ................................................................................................................... 10
2.2. Literature review ........................................................................................................... 13
2.2.1.
Small-scale agriculture: opportunities and challenges .......................................... 13
2.2.2.
Farm structure change in SSA: empirical insights and gaps ................................. 15
2.2.3.
Agricultural transformation and political economy of institutions ....................... 16
2.2.4.
Farm structure change in the context of land institutions in Zambia .................... 19
2.3. Data and methods .......................................................................................................... 23
2.3.1.
Phase 1: quantitative data collection ..................................................................... 24
2.3.2.
Phase 2: qualitative data collection ....................................................................... 27
2.3.2.1. Farmer in-depth interviews ............................................................................... 27
2.3.2.2. Community leader in-depth interviews ............................................................. 28
2.3.2.3. Interviews with national-level stakeholders...................................................... 28
2.3.3.
Data analysis ......................................................................................................... 28
2.3.3.1. Quantitative data analysis ................................................................................. 28
2.3.3.2. Qualitative data analysis ................................................................................... 29
2.4. Results and discussion .................................................................................................. 30
2.4.1.
Perceived causes of farm structure change ........................................................... 30
2.4.1.1. Change in society’s perception of agriculture .................................................. 30
2.4.1.2. Changes in enforcement of informal rules of land governance ........................ 32
2.4.1.3. Unintended consequences of agricultural subsidy programs ............................ 34
2.4.2.
Characterizing farm structure change in Zambia .................................................. 36
2.4.2.1. Nature of change in farm structure ................................................................... 37
2.4.2.2. Characteristics of medium-scale farms ............................................................. 40
2.4.2.3. Agricultural land ownership by urban-based households ................................. 43
2.4.3.
Agricultural land use in the context of farm structure change .............................. 45
2.4.4.
Land titling and markets in the context of farm structure change ........................ 49
2.4.4.1
Extent of land titling ........................................................................................ 49
2.4.4.2. Extent of participation in land markets ............................................................ 53
2.4.5.
Level of agricultural commercialization ............................................................... 58
2.5. Conclusions and policy implications ............................................................................ 62
APPENDIX ................................................................................................................................... 65

viii

REFERENCES ............................................................................................................................. 79
CHAPTER 3: REVISITING THE INVERSE FARM SIZE-PRODUCTIVITY RELATIONSHIP:
A CASE STUDY OF ZAMBIA ................................................................................................... 86
3.1. Introduction and background ........................................................................................ 86
3.2. Data and methods .......................................................................................................... 95
3.2.1.
Computation of productivity indicators ................................................................ 98
3.2.2.
Empirical framework and estimation strategy .................................................... 102
3.3.
Results and discussion ............................................................................................... 107
3.3.1.
Descriptive results ............................................................................................... 107
3.3.2.
Testing the IR with alternative productivity measures over a wide range of farms .
........................................................................................................................... 118
3.3.3.
Understanding productivity differences across farm sizes ................................. 135
3.4.
Conclusions and implications for policy.................................................................... 142
APPENDIX ................................................................................................................................. 146
REFERENCES ........................................................................................................................... 149
CHAPTER 4: AGRICULTURAL MECHANIZATION IN ZAMBIA: TRENDS AND EFFECTS
ON HOUSEHOLD PRODUCTION........................................................................................... 154
4.1.
Introduction ................................................................................................................ 154
4.2.
Background ................................................................................................................ 157
4.2.1.
Agricultural mechanization in SSA and Zambia: historical perspective ............ 157
4.2.2.
Agricultural mechanization in SSA and Zambia: recent developments ............. 160
4.3.
Data and methods ...................................................................................................... 162
4.3.1.
Phase 1: quantitative data collection ................................................................... 163
4.3.2.
Phase 2: qualitative data collection ..................................................................... 165
4.3.2.1. Farmer in-depth interviews ............................................................................. 166
4.3.2.2. Community leader in-depth interviews ........................................................... 166
4.3.2.3. Interviews with national-level stakeholders.................................................... 167
4.3.2.4. Interviews of mechanization service providers............................................... 167
4.3.3.
Data analysis ...................................................................................................... 168
4.3.3.1. RQ1: Quantitative analysis of trends in mechanization use and tractor hiring
services ......................................................................................................................... 168
4.3.3.2. RQ2: Quantitative analysis of effects of mechanization on household level
production ....................................................................................................................... 168
4.3.3.3. RQ3: Quantitative analysis of the rise of agricultural mechanization services ....
......................................................................................................................... 172
4.3.3.4. Qualitative data analysis ................................................................................. 173
4.4.
Results and discussion ............................................................................................... 174
4.4.1.
Trends in agricultural mechanization use and tractor hiring services ................. 174
4.4.1.1. Mechanization use of small- and medium-scale farmers by province ............ 174
4.4.1.2. Mechanization use of medium-scale farmers by selected districts ................. 175
4.4.1.3. Tractor use per area cultivated in Zambia....................................................... 177
4.4.1.4. Demographic and production characteristics of users versus non-users of
mechanization ................................................................................................................. 178
4.4.1.5. Access to and provision of tractor hiring services in selected districts .......... 180

ix

4.4.2.
Effects of mechanization on area cultivated and net value of crop production .. 183
4.4.2.1. Effects of mechanization on area cultivated ................................................... 184
4.4.2.2. Effect of mechanization on net value of crop production per hectare ............ 189
4.4.3.
Explaining the rise of agricultural mechanization services in Zambia ............... 192
4.4.3.1. Factor price ratios: are they changing enough to induce demand for
mechanization? ............................................................................................................... 193
4.4.3.2. Insights from supply side actors and their role in the rise in mechanization .. 195
4.4.3.2.1. Drivers of current mechanization initiatives........................................... 195
4.4.3.2.2. Performance of these initiatives .............................................................. 196
4.4.3.2.3. Constraints .............................................................................................. 196
4.4.3.2.4. Summary .................................................................................................. 197
4.5. Conclusions and recommendations............................................................................. 198
REFERENCES ........................................................................................................................... 202
CHAPTER 5: CONCLUSIONS ................................................................................................. 206

x

LIST OF TABLES

Table 1: Changes in farm structure among small- and medium-scale farmers in Zambia (2001 –
2015) based on official national survey data ........................................................................ 39
Table 2: Demographic characteristics of medium–scale farmers in selected districts of Zambia by
mode of land acquisition ....................................................................................................... 41
Table 3: Summary statistics of agricultural land ownership of urban based households in Zambia
(2007 – 2013/14) ................................................................................................................... 44
Table 4: Comparison of cultivated and uncultivated agricultural land across landholding size
categories in Zambia (2012 – 2015) ..................................................................................... 45
Table 5: Characteristics of uncultivated land by medium-scale farmers in selected districts of
Zambia .................................................................................................................................. 46
Table 6: Extent of land titling by small- and medium-scale farmers in Zambia by landholding
size categories ....................................................................................................................... 51
Table 7: Maize production in Zambia by landholding size categories ......................................... 60
Table 8: Marketed maize output in Zambia by landholding size categories ................................ 60
Table 9: Gross value of production (crops only) in Zambia by landholding size categories ....... 61
Table 10: Gross value of marketed crops in Zambia by landholding size categories................... 61
Table 11: Proportion of farms in Zambia by landholding size categories .................................... 62
Table 12: Productivity measures and farm sizes in selected past IR studies ................................ 90
Table 13: Variable explanation ................................................................................................... 103
Table 14: Demographic, land, and input use characteristics by landholding size categories ..... 108
Table 15: Crop production characteristics of main crop categories by landholding size categories
............................................................................................................................................. 112
Table 16: Crop production costs and productivity measures by landholding size categories .... 115
Table 17: Parameter estimates (OLS) of the relationship between operated farm size and net
value of crop production per hectare .................................................................................. 120

xi

Table 18: Parameter estimates (OLS) of the relationship between operated farm size and net
value of crop production per family labor/total labor day .................................................. 124
Table 19: Parameter estimates (OLS) of the relationship between operated farm size and cost of
maize production per metric ton ......................................................................................... 128
Table 20: Parameter estimates (OLS) of the relationship between operated farm size and total
factor productivity............................................................................................................... 132
Table 21: Summary of key findings for the relationship between each measure of productivity
and operated farm size ........................................................................................................ 135
Table 22: Production function estimates used for TFP estimation ............................................. 148
Table 23: Tractor use per thousand hectares cultivated in Zambia ............................................ 177
Table 24: Comparison of demographic and agricultural production characteristics of households
using mechanization versus non-users over a three-year period ........................................ 179
Table 25: Access to and provision of tractor hiring services among medium-scale farmers in
selected districts of Zambia ................................................................................................ 181
Table 26: Comparison of means between users and non-users of mechanization among mediumscale farmers in selected districts of Zambia ...................................................................... 184
Table 27: Full information maximum likelihood estimates of the endogenous switching
regression for area cultivated of users of mechanization versus non-users ........................ 186
Table 28: Expected log of hectares cultivated of users of mechanization versus non-users ...... 187
Table 29: Full information maximum likelihood estimates of the endogenous switching
regression for net value of crop production per hectare of users of mechanization versus
non-users ............................................................................................................................. 190
Table 30: Expected log of net value of crop production per hectare cultivated of users of
mechanization versus non-users ......................................................................................... 191

xii

LIST OF FIGURES

Figure 1: Proportion of rural households by landholding size categories stating that market based
land transfers are possible in customary areas ...................................................................... 54
Figure 2: Map of Zambia showing location of ACS households.................................................. 66
Figure 3: Bivariate relationships between factor input ratios and landholding size ................... 116
Figure 4: Box plots of measures of productivity by landholding size categories ....................... 118
Figure 5: Operated farm size versus alternative predicted values of net value of crop production
per hectare (with and without CRS assumption) ................................................................ 138
Figure 6: Operated farm size versus alternative predicted values of net value of crop production
per family labor day (with and without CRS assumption) ................................................. 139
Figure 7: Operated farm size versus alternative predicted values of net value of crop production
per labor day (with and without CRS assumption) ............................................................. 139
Figure 8: Operated farm size versus alternative predicted values of cost of maize production per
metric ton produced (with and without CRS assumption) .................................................. 140
Figure 9: Operated farm size and alternative predicted values of total factor productivity (with
and without CRS assumption) ............................................................................................ 141
Figure 10: Map of Zambia showing location of ACS and RALS households............................ 147
Figure 11: Proportion of small- and medium-scale farming households using mechanization for
agricultural land preparation by province between 2012 and 2015 .................................... 175
Figure 12: Proportion of medium-scale farming households using mechanization for agricultural
land preparation by selected districts .................................................................................. 176
Figure 13: Changes in factor price ratios for mechanization and agricultural labor (adjusted for
inflation) by selected provinces in Zambia ......................................................................... 194

xiii

KEY TO ABBREVIATIONS

ACS

Agricultural Commercialization Survey

AFRE

Agricultural Food and Resource Economics

AMSECS

Agricultural Mechanization Service Enterprise Centers

AU

African Union

CAADP

Comprehensive Africa Agricultural Development Program

CFS

Crop Forecast Surveys

CFU

Conservation Farming Unit

CRS

Constant returns to scale

CSO

Central Statistical Office

CSUS

Community Sustainability

DHS

Demographic and Health Surveys

DRC

Democratic Republic of Congo

FAO

Food and Agriculture Organization

FIML

Full Information Maximum Likelihood

FISP

Farmer Input Support Program

FRA

Food Reserve Agency

GIS

Geographical Information Systems

GoG

Government of Ghana

GPS

Geographical Positioning System

GRZ

Government of the Republic of Zambia

IAPRI

Indaba Agricultural Policy Research Institute

xiv

IFPRI

International Food Policy Research Institute

IR

Inverse Relationship

ISPs

Input Subsidy Programs

LGP

Length of the Growing Period

MAL

Ministry of Agriculture and Livestock

MoA

Ministry of Agriculture

MSU

Michigan State University

NIE

New Institutional Economics

OLS

Ordinary Least Squares

RALS

Rural Agricultural Livelihoods Survey

RD

Research and Development

RNFE

Rural Non-Farm Economy

RTS

Returns to Scale

SRTM

Shuttle Radar Topography Mission

SSA

Sub-Saharan Africa

TFP

Total Factor Productivity

TGCC

Tenure and Global Climate Change

TSPs

Tractor Service Providers

USAID

United States Agency for International Development

ZLA

Zambia Land Alliance

ZMW

Zambian Kwacha

ZNFU

Zambia National Farmers’ Union

xv

CHAPTER 1: INTRODUCTION
Agricultural development has continued to receive unprecedented attention in the
developing world mainly because a significant proportion of its inhabitants depend on agriculture
for livelihoods (Timmer, 1998; Barrett et al., 2010). Yet, agricultural development—defined in
this dissertation as the process that creates the conditions for the fulfillment of agricultural
potential (de Laiglesia, 2006)—in sub-Saharan Africa (SSA) has not progressed at the same pace
as other developing regions of the world. A critical examination of an important outcome of the
agricultural development process such as increased agricultural productivity reveals that SSA has
witnessed a modest upward trend in agricultural total factor productivity since the 1980s (Fugile
& Rada, 2013). However, agricultural productivity in the region still significantly lags behind
that of other developing regions—for example, Latin America or South East Asia—in part
because of a number of constraints that have continued to stifle the pace of growth. The
constraints include low technology use, market failures and inconsistent policies that continue to
distort the allocation of resources to enhance the much needed agricultural development (Diao et
al., 2007). As a consequence, food security concerns and high poverty levels have continued to
persist leaving more than 20 percent of the population in the region undernourished (FAO,
2014).
This dissertation explores an important issue that has re-emerged in the development
discourse—farm structure change—and whether this change represents an opportunity for
countries in SSA to address the myriad of challenges confronting agricultural development
outlined above. In this dissertation, farm structure or the structure of agriculture refers to a
number of attributes such as “the number and size of farms; ownership and control of resources;
and managerial, technological, and capital organization of farming” (Knutson, Penn, and

1

Flinchbaugh, 1998, p296). By extension, change in farm structure describes how the
aforementioned attributes of agriculture, in SSA for example, have evolved over time driven by
factors that are either endogenous (internal) or exogenous (external) to the entire system.
Research on the dynamics of farm structure is not new to development scholars. A review
of the literature reveals that research on farm structure change and its implication for
development has been of interest to development scholars for more than three decades (Berry &
Cline, 1979; Kydd & Christiansen, 1982; Binswanger, Deininger, & Feder, 1993). While there is
quite a diversity in the specific lines of inquiry pursued by a myriad of studies, the overarching
themes that have been investigated in relation to farm structure change include: (1) migration of
labor from the small to the large farm sector (e.g., Bryceson, 1996; Barrett et al., 2001;
Bryceson, 2002); (2) large-scale land acquisitions (Kydd & Christiansen, 1982; Deininger &
Byerlee, 2012; German et al., 2013; Nolte, 2014), and; (3) the expansion of domestic land
investors (Jayne et al., 2014; Sitko & Jayne, 2014; Jayne et al., 2016).
The general focus by scholars studying farm structure change in the current development
discourse has largely been on the development implications of large-scale land acquisitions of
foreign origin. Farm structure change associated with the rise of domestic medium-scale farmers
has received less attention. While foreign driven large-scale land acquisitions have played a role
in altering the agricultural landscape in SSA, the increase in pace at which land is being acquired
by domestic land investors—for example in Ghana, Kenya, Tanzania and Zambia—has
completely changed the farm structure outlook (Jayne et al., 2014; Jayne et al., 2016). In the four
case study countries mentioned above, the total area under the control of domestic medium-scale
investor farms now exceeds that of foreign and previously established domestic large-scale
holdings combined. And in the case of Zambia, it has been estimated that medium-scale farms (5

2

to 100 hectares landholding size) control about 2.5 million hectares while small-scale farms
(those with less than 5 hectares landholding size) control approximately 2.1 million hectares of
land (ibid). The distinction between small- and medium-scale farms employed in this study is
based the following. First, we rely on how African governments in general define these
categories. They use 5 hectares as the distinction between small- and medium-scale holdings.
Second, we want to be consistent with how the emerging scholarly work on the rise of mediumscale investor farms in SSA has defined this scale of farming; they refer to medium-scale farms
as those between 5 and 100 hectares of land
The domestic land investor dynamic has ignited interest in the development discourse to
understand whether or not this new phenomenon signals an alternative pathway for achieving
agricultural development in parts of SSA. Therefore, this dissertation examines farm structure
change associated with domestic medium-scale investor farms and its implications for
agricultural development in sub-Saharan Africa. The study has three specific objectives: (1) to
investigate the causes and consequences of the rise of domestic medium-scale investor farms; (2)
to re-examine the farm size – productivity relationship hypothesis in the context of farm structure
change, and; (3) to explore recent trends in agricultural mechanization use, its effects on
agricultural household production and the factors contributing to recent initiatives aimed at
enhancing mechanization in Zambia. The study was conducted in Zambia because the country
has been experiencing major changes in farm structure, the most salient of which is a major
increase in cultivated area under the control of farms cultivating between 5 and 100 hectares. To
this end, the dissertation consists of three essays summarized here in turn.
The first essay, presented in Chapter 2, examines the underlying political economy and
institutional factors driving the change in farm structure and explores the effects of this change

3

on indicators inextricably linked with a vibrant agricultural sector. It uses a mixed-methods
approach that relies on quantitative—secondary and primary—and qualitative sources of data
collected in Zambia. The secondary quantitative data includes various years of national level
agricultural household data that demonstrate the changing dynamics of agricultural land
ownership and use over the last 10 to 15 years. The primary quantitative data, collected in 2013,
is a cross-sectional dataset that investigates the life history and pathways to agricultural
commercialization of medium-scale farm households in six districts located in central and
southern Zambia. The qualitative data are in-depth interview transcripts of purposively sampled
respondents from the six aforementioned districts and Lusaka, the capital city of Zambia. In
total, 48 in-depth interviews were conducted comprising 24 small- and medium-scale farmers, 12
community leaders and relevant district officials, and 12 national level representatives.
Results indicate that farm structure change has been driven by factors such as a positive
change in perception of farming by society that has ushered in new entrants with a desire to farm
at a level above small-scale agriculture. Further, the change in enforcement of informal rules of
land governance has enhanced the rate of land acquisitions by domestic medium-scale investor
farms. Lastly, the unintended consequences of public spending agricultural support programs
have been favorable to the rise of domestic medium-scale farms. The research also finds that the
change has been associated with agricultural land concentration, perpetuation of a maize
monoculture, accumulation of idle land among medium-scale investor farms and increasing land
constraints among small-scale farmers. Further, the evidence points to low land titling especially
among small-scale farms, a growing informal land rental and sales market and a skewed level of
agricultural commercialization. The study concludes that in order to achieve broad-based
development, policy strategies should address inequalities in land access, the process of

4

formalization of land ownership and low production and productivity among small-scale farm
households.
The second essay, presented in Chapter 3, revisits the inverse farm size-productivity
hypothesis by extending the methodological approach in previous research and provides practical
implications for policy. Past studies have remarkably been limited to data from farms of less than
ten hectares, yet their findings have been extrapolated beyond this farm size range. Moreover,
productivity has mainly been restricted to land productivity when exploring the hypothesis.
Further, previous studies have consistently used the assumption of constant returns to scale when
examining the farm size-productivity relationship, which may not be appropriate when including
large farms. This paper accounts for these issues by relaxing the ceteris paribus assumption and
finds that the inverse farm size-productivity relationship is not consistently observed across
alternative indicators of productivity.
While the study finds that small-scale farms are more productive when the outcome
variable is either land productivity or cost of staple food production per metric ton produced,
relatively larger farms have higher labor productivity. Relaxing the assumption of constant
returns to scale strengthens the hypothesis of an inverse relationship, particularly when
productivity is measured in terms of land productivity and total factor productivity. The paper
concludes that findings on the relationship between productivity and operated farm size, while
important, should not be the decisive factor in guiding agricultural development and land policies
in sub-Saharan Africa because there are many other important considerations.
The third essay (Chapter 4) examines the trends and effects on agricultural household
production of agricultural mechanization and explores the factors contributing to recent
developments aimed at enhancing mechanization in Zambia. The essay is motivated by the

5

observation that there is a dearth of research that examines the contribution of agricultural
mechanization to agricultural development in SSA given the current environment in which
policy makers, the private sector and farmer organizations have teamed up to promote tractor use
for land preparation.
The third essay has three key findings. First, the evidence across a three year period
(2012 – 2015) shows that agricultural mechanization use by small- and medium-scale farming
households has remained low and stagnant with only 1.5 percent of these households reporting
that they used mechanization for land preparation. Second, the study finds that use of
mechanization for land preparation increased farmers’ area cultivated by about 60 percent
relative to those who did not use mechanization. However, the study finds that mechanization
use did not necessarily translate into net gains in productivity for farmers. The study attributes
this to low levels of intensification by farmers in Zambia of productivity enhancing inputs such
as improved seed, fertilizer and others.
Third, although results indicate that the price ratios for rental of mechanization relative to
labor costs had changed over time in favor of a shift to labor saving technologies, demand for
mechanization has not increased correspondingly. Therefore, the study established that
mechanization initiatives in Zambia have mainly been driven by the supply side. The study
concludes by recommending more involvement of government extension in training potential
users of mechanization services, which could in turn stimulate demand for services. Further,
some key areas for remodeling mechanization initiatives are proposed.
The last essay of the dissertation, Chapter 5, summarizes the dissertation by synthesizing
the key findings and policy options crucial to achieving agricultural development.

6

REFERENCES

7

REFERENCES

Barrett, Christopher B., Michael R. Carter, and C. Peter Timmer. 2010. “A Century-Long
Perspective on Agricultural Development.” American Journal of Agricultural Economics
92 (2): 447–68. doi:10.1093/ajae/aaq005.
Barrett, Christopher B., T. Reardon, and P Webb. 2001. “Nonfarm Income Diversification and
Household Livelihood Strategies in Rural Africa: Concepts, Dynamics, and Policy
Implications.” Food Policy 26: 315–31.
Berry, Robert Albert, and William Richard Cline. 1979. Agrarian Structure and Productivity in
Developing Countries: A Study Prepared for the International Labour Office Within the
Framework of the World Employment Programme. John Hopkins University Press.
Binswanger, Hans, Klaus Deininger, and Gershon Feder. 1993. “Power, Distortions, Revolt, and
Reform in Agricultural Relations.” In Handbook of Development Economics, edited by
Jere Behrman and T.N. Srinivasan. Vol. IIIB. Amsterdam: North-Holland.
http://core.ac.uk/download/pdf/6645289.pdf.
Bryceson, Deborah Fahy. 1996. “Deagrarianization and Rural Employment in Sub-Saharan
Africa: A Sectoral Perspective.” World Development 24 (1): 97–111.
———. 2002. “The Scramble in Africa: Reorienting Rural Livelihoods.” World Development 30
(5): 725–739.
de Laiglesia, Juan Ramón. 2006. “Institutional Bottlenecks for Agricultural Development.”
http://www.oecd-ilibrary.org/development/institutional-bottlenecks-for-agriculturaldevelopment_671788081061.
Deininger, Klaus, and Derek Byerlee. 2012. “The Rise of Large Farms in Land Abundant
Countries: Do They Have a Future?” World Development 40 (4): 701–14.
Diao, Xinshen, Peter B. R. Hazell, Danielle Resnick, and James Thurlow. 2007. The Role of
Agriculture in Development: Implications for Sub-Saharan Africa. Intl Food Policy Res
Inst.
FAO. 2014. “Food Security Data, FAOSTAT.” http://faostat3.fao.org/home/E.
Feder, Gershon, Richard E. Just, and David Zilberman. 1985. “Adoption of Agricultural
Innovations in Developing Countries: A Survey.” Economic Development and Cultural
Change 33 (2): 255–98. doi:10.1086/451461.
Fugile, Keith, and Nicholas Rada. 2013. “USDA Economic Research Service - Research Raises
Agricultural Productivity in Sub-Saharan Africa.” USDA Economic Research Service.
May 6. http://www.ers.usda.gov/amber-waves/2013-may/research-raises-agriculturalproductivity-in-sub-saharan-africa.aspx#.UzRXIF660oY.

8

German, Laura, George Schoneveld, and Esther Mwangi. 2013. “Contemporary Processes of
Large-Scale Land Acquisition in Sub-Saharan Africa: Legal Deficiency or Elite Capture
of the Rule of Law?” World Development 48 (August): 1–18.
doi:10.1016/j.worlddev.2013.03.006.
Jayne, T. S., Chamberlin, J., Traub, L., Sitko, N., Muyanga, M., Yeboah, F. K., … Kachule, R.
(2016). Africa’s changing farm size distribution patterns: the rise of medium-scale farms.
Agricultural Economics, 47(S1), 197–214. https://doi.org/10.1111/agec.12308
Jayne, Thomas S., Antony Chapoto, Nicholas Sitko, Chewe Nkonde, Milu Muyanga, and Jordan
Chamberlin. 2014. “Is the Scramble for Land in Africa Foreclosing a Smallholder
Agricultural Expansion Strategy?” Journal of International Affairs 67 (2): 35–53.
Johnston, Bruce F., and John W. Mellor. 1961. “The Role of Agriculture in Economic
Development.” The American Economic Review 51 (4): 566–93.
Knutson, Ronald D, J.B. Penn, and B.L. Flinchbaugh. 1998. Agricultural and Food Policy. 4th
ed. Upper Saddle River, N.J.: Pearson/Prentice Hall.
Kydd, Jonathan, and Robert Christiansen. 1982. “Structural Change in Malawi since
Independence: Consequences of a Development Strategy Based on Large-Scale
Agriculture” 10 (5): 355–75.
Nolte, Kerstin. 2014. “Large-Scale Agricultural Investments under Poor Land Governance in
Zambia.” Land Use Policy 38 (May): 698–706. doi:10.1016/j.landusepol.2014.01.014.
Sitko, Nicholas J., and T. S. Jayne. 2014. “Structural Transformation or Elite Land Capture? The
Growth of ‘emergent’ Farmers in Zambia.” Food Policy, Boserup and Beyond: Mounting
Land Pressures and Development Strategies in Africa, 48 (October): 194–202.
doi:10.1016/j.foodpol.2014.05.006.
Timmer, C. Peter. 1998. “The Agricultural Transformation.” In International Agricultural
Development, edited by Carl K. Eicher and John M. Staatz, 113–35. Johns Hopkins
University Press.

9

CHAPTER 2: THE RISE OF DOMESTIC MEDIUM-SCALE FARMS IN ZAMBIA:
CAUSES AND CONSEQUENCES
2.1. Introduction
Most rural households in sub-Saharan Africa (SSA) directly or indirectly depend on
agriculture for their livelihoods (Diao et al., 2010). With more than two-thirds of these
households residing on small farms of less than two hectares (Hazell et al., 2010), development
practitioners and policymakers in the region have in the last 50 years or so championed
agricultural development strategies that support small-scale agriculture. Yet, agriculture in SSA
has continued to underperform relative to other developing parts of the world. This is partly
because of low technology use, market failures and inconsistent policies that distort allocation of
resources in agriculture (Diao et al., 2007). As a result, food security concerns and high rural
poverty levels persist in SSA—almost 23.8 percent of the population is undernourished—and the
contribution of agriculture to overall economic development in the region has been dismal at
best. In essence, prospects of achieving agricultural development and in the process structural
transformation through a strategy that is focused on small-scale agriculture have continued to
suffer major setbacks.
However, recent evidence has shown that there is a rapid expansion in farmland
acquisitions in parts of SSA by domestic land investors or medium-scale farms—farms with
landholdings between 5 and 100 hectares (Jayne et al., 2014)—and foreign owned large-scale
farms (Deininger & Byerlee, 2012; German et al., 2013; Nolte, 2014; Schoneveld, 2014; Land
Matrix, 2015). This evidence suggests that farm structure in affected countries of SSA has been
undergoing changes symbolized by the relative increase in the number of larger farms. Does this
change signal an emerging alternative pathway to address the myriad of challenges confronting

10

agriculture in SSA? Is this change likely to help achieve the main goal of rural and overall
economic development in SSA?
In view of the aforementioned questions, the current study examines the change in farm
structure characterized by the rapid expansion of medium-scale investor farms and addresses its
implications for rural development in SSA using the case of Zambia. In Zambia, recent estimates
have shown that medium-scale farms now account for more land than the holdings in the entire
small-scale sector (Jayne et al., 2014). Understanding the causes and consequences of this farm
structure change is vital as this seems to be a crucially important topic and one on which more
applied research evidence is required.
In this paper, we define medium-scale investor farms as farm holdings between 5 and 100
hectares. As observed by Jayne et al (2016), definition of what each scale of farming implies is
rather arbitrary and could mean different things in different regions of the world. Therefore, our
decision to use cutoff points of 5 and 100 ha has been informed by two things. First, the choice is
based on how a number of African governments define medium-scale farm households. They use
5 hectares as the distinction between small- and medium-scale holdings (ibid). Second, we
wanted to be consistent with how the emerging scholarly work on the rise of medium-scale
investor farms in SSA has defined this scale of farming; they refer to medium-scale farms as
those between 5 and 100 hectares of land (see for example, Jayne et al., 2014; Jayne et al., 2016).
Farm structure or the structure of agriculture, in this essay, refers to a number of
attributes such as “the number and size of farms; ownership and control of resources; and
managerial, technological, and capital organization of farming” (Knutson, Penn, and
Flinchbaugh 1998, p296). By extension, farm structure change implies the evolution of farm

11

structure over time driven by factors that are either endogenous (internal) or exogenous
(external) to the entire system.
While there are studies that have examined recent changes in farm structure (e.g., Jayne
et al., 2014; Sitko & Jayne, 2014), some gaps in the literature still exist. First, past studies have
not employed a rigorous analytical framework to comprehend the political economy and
institutional dimensions contributing to changes in farm structure. Second, given the complexity
of this topic, it is remarkable that past studies have largely employed quantitative methods of
inquiry at the expense of using a mixed-methods approach that elicits deeper understanding of
observed changes in farm structure.
This study addresses these gaps by developing an analytical framework that combines
quantitative and qualitative methods to understand the direct and indirect effects of farm
structure change symbolized by rapid expansion of medium-scale investor farms. The current
study addresses five research questions that extend beyond what has been examined previously:
1. What do farmers, community leaders and stakeholders involved in land governance

issues perceive to be the key drivers of farm structure change in Zambia?
2. How has farm structure changed and what are the characteristics of domestic land

investors that have emerged in the last decade or so in Zambia?
3. How has land use evolved in the context of farm structure change?
4. How have attributes such as agricultural land titling and participation in land markets

evolved in the context of farm structure change?
5. What relationship exists between farm structure change and the level of agricultural

commercialization in the rural economy of Zambia?

12

The rest of the paper is organized into four main sections. The next section presents a
review of the literature by outlining the following: (1) a discussion of the opportunities and
challenges affecting small-scale agriculture; (2) insights from empirical studies on farm structure
change; (3) the conceptual foundations used to develop the analytical framework for this study,
and; (4) a historical perspective of land institutions in Zambia. The “Data and Methods” section
explains the primary and secondary sources of data and the qualitative and quantitative methods
employed in the study. The results are presented and discussed in the penultimate section and the
last section concludes and provides implications for policy.
2.2. Literature review
2.2.1. Small-scale agriculture: opportunities and challenges
Before examining recent developments pertaining to farm structure change, it is first
important to outline reasons why development practitioners and policymakers in SSA champion
agricultural development strategies that support small-scale agriculture. A popular explanation
relates to the often-observed inverse farm size-productivity relationship—a hypothesis that
postulates that small farms are more productive per unit of land than large farms (Berry & Cline,
1979).
Although the existence of an inverse farm size-productivity relationship is widely
contested, one cannot ignore the preponderance of evidence from studies conducted in Asia that
have empirically found this relationship (e.g., Sen, 1962; Ghose, 1979; Berry & Cline, 1979;
Carter, 1984; Chen et al., 2011). Likewise, there has been an increase in the number of empirical
studies in SSA that have also shown the existence of an inverse farm size-productivity
relationship (e.g., Kimhi, 2006; Carletto et al., 2013; Ali & Deininger, 2015; Larson et al., 2014).

13

As a consequence, findings of an inverse relationship have been used as the basis for supporting
agricultural strategies that promote small-scale agriculture.
Another explanation relates to the argument that a smallholder-led or ‘unimodal’ farm
structure (featuring relatively egalitarian land distribution) is more effectively advantageous for
achieving poverty reduction and broader economic growth than a large-scale farm structure
(Mellor & Johnston, 1984; Timmer, 1998; Barrett et al., 2010). Asia still provides a good
example of a developing country region where a smallholder-led strategy has led to pro-poor
growth. Hazell et al. (2010, p. 1351) have argued that, “Asia’s green revolution demonstrated
how agricultural growth that reaches large numbers of small farms could transform rural
economies and raise enormous numbers of people out of poverty.” In such an environment,
universal intensification of a great number of small farms not only maximizes output and
employment in agriculture, but also provides a strong impetus to the development of the Rural
Non-Farm Economy (RNFE) culminating into long-run structural transformation (Hayami, 1977;
Haggblade et al., 1989; Delgado, 1998; Lanjouw & Lanjouw, 2001). This is presumably because
increased income is widely distributed and strengthens local demand for what the RNFE
produces.
Meanwhile, agricultural growth continues to elude most of SSA as evidenced by poor
agricultural performance (Larsson et al., 2002). In Zambia, for instance, Hichaambwa & Jayne
(2014) showed that there has been very little production growth among the bottom 60 percent of
smallholders. They also showed that the top five percent of farmers with over 10 hectares are the
category of farmers who have achieved almost all of the growth in agriculture over the past
decade. In general, smallholder production has been hampered by many challenges that have
undermined its potential (Jayne et al., 2010).

14

2.2.2. Farm structure change in SSA: empirical insights and gaps
The literature is replete with empirical studies of agrarian transformation or farm
structure change in Africa covering the colonial period through to the post-independence era.
While the line of inquiry pursued regarding farm structure change is quite diverse, three main
topical areas that have been investigated include studies that aim to understand how farm
structure has been affected by: (1) migration of labor from the small to the large farm sector
(e.g., Bryceson, 1996; Barrett et al., 2001; Bryceson, 2002); (2) large-scale land acquisitions
(Kydd & Christiansen, 1982; Deininger & Byerlee, 2012; German et al., 2013; Nolte, 2014), and;
(3) the expansion of domestic land investors (Jayne et al., 2014; Sitko & Jayne, 2014). Since this
paper is focused on the direct and indirect effects of domestic land investors, the rest of this
subsection is only dedicated to the discussion of empirical studies that have investigated the
expansion of domestic land investors.
The general focus by scholars studying farm structure change in the current development
discourse has largely been on the development implications of large-scale land acquisitions of
foreign origin. While the increase in foreign-driven-large-scale land acquisitions cannot be
overemphasized, the increase in pace at which land is being acquired by land investors from
within the boundaries of SSA countries (e.g. Ghana, Kenya and Zambia) has completely changed
the farm structure outlook (Jayne et al., 2014). In the three aforementioned countries, the area
under these domestic land investors (mainly medium-scale farms) now exceeds that of foreign
and previously established domestic large-scale holdings combined. And in the case of Zambia,
it has been estimated that medium-scale farms (5 to 100 hectares landholding size) control about
2.5 million hectares while small-scale farms (less than 5 hectares landholding size) control
approximately 2.1 million hectares of land (ibid). As a result, the domestic land investor dynamic

15

has ignited interest in the development discourse to understand whether or not this new
phenomenon signals an alternative pathway for achieving agricultural development in parts of
SSA.
Studies examining the rise of domestic land investors have investigated how the rate of
land acquisitions by these new investors relates to the amount of economically attractive arable
land and changing conditions of land access, inequality, and alienation of small-scale farmers
(Jayne et al., 2014; Sitko & Chamberlin, 2015). The other issues explored include pathways by
which the emergent farming sector has grown (Sitko & Jayne, 2014) and the causes and likely
effects of this structural change (Jayne et al., forthcoming). Based on the aforementioned issues,
these studies have provided important insights that demonstrate that the rise of domestic
investors: (1) threatens to exacerbate localized land scarcity; (2) may be largely comprised of
urban based or elite farmers and very few farmers who have organically developed through
farming, and; (3) is probably not addressing national goals of poverty reduction and agricultural
development. The current study builds on this emerging literature by focusing on existing gaps;
the need to employ a rigorous analytical framework and mixed-methods of inquiry to understand
the direct and indirect effects of farm structure change symbolized by rapid expansion of
domestic land investors.
2.2.3. Agricultural transformation and political economy of institutions
This study is situated within the broader context of two main strands of literature that
discuss: (1) the agricultural transformation process (e.g. Johnston & Mellor, 1961; Timmer,
1998), and; (2) the political economy of institutions in agricultural development (extensively
reviewed by Hoeffler (2011)). The agricultural transformation process can be described as one
by which individual farms shift from highly diversified, subsistence-oriented production towards

16

more specialized production oriented towards the market or other systems of exchange (Staatz,
1998). When examining the political economy of institutions, the focus is on the analysis of
formal and informal rules underlying political powers, bureaucratic agencies or social and
private organizations in the agricultural development process (The World Bank, 2008). Each of
these conceptual foundations is outlined in turn.
Timmer (1998) demonstrates that overall economic development of most nations is
preceded by rapid growth in the agricultural sector. In the process, however, the importance of
agriculture begins to diminish as other sectors of the economy take center stage. The reason for
this apparent paradox is that when agriculture undergoes this transformation process, there is a
greater reliance on input and output delivery systems and increased integration of agriculture
with other sectors of the domestic and international economies (Staatz, 1998). This leads to
overall economic structural transformation whereby agriculture plays a lesser role in economic
output and employment generation.
The literature further shows that the evolutionary process of agricultural transformation
goes through four distinct phases (Timmer, 1998). First, there is a noticeable increase in the
returns to agricultural labor, which helps to enhance agricultural surplus production. Second, the
increase in surplus production invariably leads to the development of the nonagricultural sector
of both the rural and urban economies. Third, there is a progressive integration of the agricultural
sector into the macro economy through improved infrastructure and market equilibrium linkages.
And as already explained above, the role of agriculture gradually becomes less noticeable due to
the economy-wide structural transformation.
Given that developing countries have largely remained committed to the desire to
stimulate the much-needed economic growth, is the current change in farm structure (explained

17

above) likely to engender the theoretical predictions embodied in the agricultural transformation
process? Interestingly, proponents of a broad based agricultural development process question
whether this apparent concentration of land ownership and subsequent concentration of upstream
agricultural production is the ideal path for SSA agriculture. Therefore, locating this paper within
the context of the agricultural transformation literature is important for understanding of the
effects of farm structure change on rural development in SSA and Zambia in particular.
The agricultural transformation process happens within the context of political decisionmaking and institutions that exist in a given society. Hoeffler (2011, p. 31) explains that, “the
agricultural sector shows so many specific historic, agro-ecological, economic, social and
political features that a careful analysis of the agents, institutions, markets and networks involved
as well as their respective interests and powers within the sector is mandatory for agricultural
policy analysis and reform implementation.” Therefore, it is vital that a study that seeks to
understand the causes and consequences of farm structure change should clearly understand
existing institutions.
This essay mainly draws from the New Institutional Economics (NIE) framework to
develop an analytical framework for unpacking the study’s overarching research questions. The
NIE framework is a useful tool for assessing institutions and how institutions interact with
organizational arrangements (Ménard & Shirley, 2008). The range of applications where the
framework has been extensively applied include studies exploring transaction costs economics
(North, 1990, 2008; Williamson, 1985) and property rights (Ostrom, 1990; Schlager & Ostrom,
1992; Kim & Mahoney, 2005; Libecap, 2008).
Scholars, for example North (1994), have also generally been interested in understanding
institutional change and have therefore developed and implemented an analytical approach with

18

attributes that help the researcher to: (1) identify individuals—actors—who are at the center of
institutional change; (2) comprehend the sources of institutional change, whether external or
internal to the actor’s environment; (3) understand whether institutional change takes the form of
changes to formal rules (legislative, judicial, regulatory or constitutional changes) or changes in
informal constraints that constitute norms, conventions, or personal standards of honesty or both;
(4) understand the role of path dependence—the idea that history matters, and; (5) determine
whether any revolutionary change has occurred that has led to institutional change. This study
adapted this analytical framework to get insights on farm structure change in Zambia and the
implications this has for overall rural and economic development.
2.2.4. Farm structure change in the context of land institutions in Zambia
The institutional environment for land administration in Zambia has undergone changes
over time alongside the country’s evolution of political governance chronologically categorized
as follows: (1) the pre-colonial period (before 1924); (2) the period of colonial rule (1924-1964);
(3) the first post-independence era (1964-1972); (4) the second post-independence era (19731991), and; (5) the third post-independence era (beyond 1991).
The pre-colonial period land institutions mainly operated within the context of customary
law. In this context, people were linked to land through ethnic ties to a particular region of the
territory where the present Zambia exists (van Loenen, 1999). Although the laws differed from
one ethnic group to the other, there were a number of similarities that cut across most ethnic
groups. Land was never viewed as a saleable commodity, landownership was communal and
systems of regulating communal rights existed (Chinene et al., 1998).
The coming in of colonial rule probably represents the first wave of noticeable changes in
farm structure in Zambia. The British Colonial Authority took over land administration in 1924

19

and initially created a dual land tenure system: Crown land for the minority European settlers
and Native Reserves for Africans (natives). The location of Crown land was mainly in parts of
the country considered to have the best potential for agriculture and along the line of rail
constructed in the early parts of the 20th Century (Adams, 2003).
Three aspects or developments emerged during the colonial land tenure system. First,
European settlers occupying Crown land were granted freeholds or leaseholds in this land with
title while natives occupying Native Reserves were not allowed to obtain title (van Loenen,
1999). Second, European settlers or non-natives were also allowed to obtain land in Native
Reserves but for only five years or less (ibid). However, Africans did not have access to land in
Crown territory Third, colonialists introduced a new law that allowed reclassification of suitable
land not previously alienated, which therefore introduced a third category of land called Trust
land. This new law was only introduced a couple of decades after colonialists assumed land
administration.
The introduction of Trust land came about because of the pressure colonialists received
from both natives and European settlers. Natives had perpetually resisted the concept of Native
Reserves as reserves were associated with overcrowding and impoverishment while European
settlers had also continued to demand for larger tracts of land (Chinene et al., 1998). Therefore,
Trust land was accessible to both the natives and the non-natives. Unfortunately, the introduction
of Trust land did little to circumvent the non-egalitarian land distribution of the colonial era. This
is because the good fertile land was still reserved for European settlers.
When Zambia gained independence in 1964, the new political leadership led by Kenneth
Kaunda largely maintained the land tenure system left by colonialists (Chileshe, 2005). Although
a large proportion of European settlers left when Zambia became a sovereign state in 1964, most

20

of the best farm land remained in the hands of the remaining settlers (van Loenen, 1999). During
the first post-independence era, the few changes made were that Crown land was renamed State
land and was now vested in the President rather than the British colonial government and that the
new Zambian government was allowed by law to acquire undeveloped and unutilized lands
particularly those lands held by absentee landowners (Brown, 2005).
The second post-independence era when multiparty politics were abolished probably
represents the period when the Kaunda-led administration made significant changes to land
policies inherited from colonialists. Although socialism and nationalism were economic models
that the country embraced at independence, the second republic was the period when these ideals
became truly entrenched and thus had a significant impact on a number of policies including
those related to land. Through an act of parliament, the Land (Conversion of Titles) Act of 1975,
freehold tenure was abolished and converted to statutory leasehold while the market for land was
suppressed and the state directly administered all land transfers (van Loenen, 1999; Brown,
2005). Further, when selling or buying land, no intrinsic value was attached to such property
unless the property in question had improvements such as buildings and other infrastructure.
To curtail foreigners from owning large tracts of land, parliament passed legislation in
1985 that restricted the alienation of land to foreigners (transferring of land acquired from
customary landowners by government to foreigners), with the exception of presidentially
certified investors and charitable organizations (ibid). Therefore, farm structure during the
second post-independence era was characterized by highly subsidized state farms, a few largescale commercial farms owned by European settlers and small-scale farmers who remained in
the majority and still occupied Reserve and Trust lands.

21

The land policy environment in the third post-independence era represents the current
system, which began when the country reverted to multiparty politics in 1991. The new
government led by Fredrick Chiluba was committed to making substantive land reforms and by
1995, they had managed to enact the 1995 Lands Act (Government of the Republic of Zambia
(GRZ), 1995) which paved way for investment in land including by foreigners who were
restricted by the previous regime (Nolte, 2014).
According to Brown (2005), the new land law came with four fundamental changes.
First, the law significantly strengthened the property rights of titleholders on State land, which
made it possible for owners of land to sell land even without any improvements to it. A second
aspect of the Act was that it eased restrictions on land ownership by foreigners. Third, the Lands
Act created a Lands Tribunal to protect leaseholders and customary rights holders from abuse
and to ease congestion within the High Court. Fourth, the Act made both cosmetic and
substantive changes to the administration of customary land tenure. The cosmetic change made
was the amalgamation of Reserve and Trust lands into a new category called customary lands.
The substantive change made it easier for both outside investors (foreign or domestic living
outside customary lands) and indigenous Zambians living on customary lands to acquire private
title to customary land.
Despite the good intentions of the Lands Act, its implementation has probably led to
adverse effects on the allocation of unutilized land in customary land areas. First, while the law
intends to benefit all potential land owners with respect to acquisition of private title to
customary land, the clear beneficiaries have not been the majority of small-scale farmers but
elites both at the local level and those coming from urban areas (Sitko & Jayne, 2012; Malambo,
2014). Second, although the original intent of allowing purchases of land by foreigners was to

22

encourage investments important for national development, the nature of land transactions that
have been processed with foreign companies have been questionable at best and have largely
excluded local land users from the process (Economist, 2011; McMichael, 2012; Nolte, 2014).
Therefore, the third post-independence era has been characterized by land acquisition by
individuals and firms who have the social and financial capital to benefit from a market-based
environment at the expense of poor small-scale farmers. It is for this reason that civil society
efforts spearheaded by the Zambia Land Alliance (ZLA) have been aimed at lobbying and
advocating for a just land policy and law that takes into account the interests of the poor.
2.3. Data and methods
In this section, the data and methods used are discussed in detail. An explanatory
sequential mixed-methods approach was used to address the main research questions. First, what
are the perceived causes of farm structure change? Second, what are the characteristics of farm
structure change? Third, how is land use evolving in the context of farm structure change?
Fourth, what is the extent and pattern of titling of agricultural land and farmer participation in
agricultural land markets? Fifth, what is the relationship between agricultural commercialization
and farm structure change?
The sequential mixed-methods approach involved the following steps. The first step
involved collecting and analyzing secondary and primary quantitative household survey data.
The surveys provided data and analytical results related to a number of themes such as
proportional changes in the number of small- versus medium-scale farms, demographic attributes
of medium-scale investor farmers, extent of participation in land markets and other measurable
attributes. The second step involved qualitative research methods. The qualitative methods
helped with exploring the views of farmers, community leaders and land governance actors with

23

knowledge about farm structure change and agricultural development. The mixed-methods
approach helped the researcher to triangulate findings generated from the two unique methods,
an approach that enhanced the validity and generalizability of the study.
2.3.1. Phase 1: quantitative data collection
The quantitative data were collected using secondary and primary data sources (survey
instruments are available upon request). The secondary data sources were drawn from three
nationally representative surveys conducted in Zambia in the last decade or so. The first
nationally representative source of secondary data that the study relied on was various years of
the Crop Forecast Surveys (CFS) of small-, medium- and large-scale farms in Zambia beginning
from 2001.1 The Central Statistical Office (CSO) of Zambia conducts the CFSs on an annual
basis in collaboration with the Ministry of Agriculture and Livestock (MAL). The CFSs are
designed to collect data from an average of 12,000 households on key agricultural production
parameters during each agricultural season that runs from October 1 to September 30 the
following year. Because it is not practical to collect all policy relevant information in this annual
survey, this study relied on another secondary source of nationally representative data—the
Rural Agricultural Livelihoods Survey (RALS) of small and medium-scale farming households
in Zambia collected in 2012 and 2015.
The RALS is a longitudinal survey conducted by the Indaba Agricultural Policy Research
Institute (IAPRI) in conjunction with the two aforementioned public offices. The RALS used the
new sampling frame derived from the 2010 census. A total of 8,839 and 7,934 households were
interviewed during the 2012 and 2015 surveys respectively. The survey collected data on a
1

Similar to (Jayne et al., 2016), the three farm size categories in this study are defined based on
landholdings as follows: small-scale (0-5 hectares); medium-scale (5-100 hectares); large-scale
(more than 100 hectares).
24

number questions related to the following main themes: demographic characteristics of
household members; farmland and use; crop sales from own production; input and credit
acquisition; livestock ownership and marketing; off-farm income sources; food security
indicators, and; other themes such as kinship ties of the household head. In addition, soil samples
were collected from largest maize fields of sub-sampled households during the 2012 survey.
The third source of secondary data was the Demographic and Health Surveys (DHS)
collected in Zambia in 2007 and 2014. These datasets were used primarily because they capture
information about land acquisitions of not only rural households but urban-based households as
well.
To get a better understanding of the changes in farm structure observed in Zambia
especially with respect to medium-scale farms, the study also used primary data collected
through the Agricultural Commercialization Survey (ACS). The ACS was conducted because
both the CFS and RALS had two main limitations. First, it was established that the sampling
frame for both surveys did not adequately capture medium-scale farms especially those between
20 and 100 hectares of land owned. Second, both the CFS and the RALS were not designed to
capture information about the timing of land acquisitions, sources of income for these
acquisitions and the employment history of the rising group of relatively large domestic
landowners. The ACS was therefore designed to fill this gap but for only selected regions of
Zambia.
The ACS of 2013 was a follow-up to a survey of medium-scale farms conducted in
Zambia in 2011 by Sitko & Jayne (2012) with assistance from the author of this paper. The main
objective of the ACS was to understand the history, characteristics, land use decisions and
patterns, and level of commercialization of medium-scale farmers observed to be a growing class

25

of farmers in Zambia. Similar to the first survey, the 2013 ACS survey purposively selected
research sites (administrative districts) based on the concentration and number of medium-scale
farms as reported in the 2010/2011 CFS.
Two main criteria were used to select the districts included in the survey. First, at least 3
percent of farming households had to be classified as medium-scale farms in order to ensure a
reasonable population from which to sample. Second, the research sites were selected along a
continuum of concentrations to ensure geographical diversity. But as noted by Sitko &
Chamberlin (2015), because medium-scale farms are concentrated along what is referred to as
the ‘line of rail’ in Zambia, the selected districts were in close proximity to this region. The line
of rail is a region of Zambia that is served by the railway linking the Copperbelt province with
Lusaka, and with the border town Livingstone in the south of the country.
Therefore, the six selected districts were Chibombo, Choma, Chongwe, Kalomo,
Mpongwe and Mumbwa (see Figure 2 in Appendices for the map). A total of 482 households
were randomly selected from a list of medium-scale farms generated in consultation with the
Zambia National Farmers’ Union (ZNFU) and the Ministry of Agriculture and Livestock (MAL)
district offices. The Ministry of Agriculture and Livestock (MAL) block areas were identified as
the sampling units with help from local district offices. Whereas the sampling procedure was
supposed to ensure reasonable representativeness of farmers in the 5-100 hectare category within
the selected districts, the sample may have not been statistically representative of all mediumscale farmers in Zambia. This is because medium-scale farmers were not sampled from the other
65 districts. But we can say that the chosen districts are understood to contain the highest
proportions of farms in this size class.

26

2.3.2. Phase 2: qualitative data collection
Qualitative data were collected through in-depth interviews of 48 purposively sampled
respondents. For each interview, detailed notes were taken; and where the interviewee consented,
an audio recorder was used to record the full interview. At the end of each day, the field notes
and audio recordings were fully transcribed in readiness for analysis.
The respondents included: (1) farmers (small- and medium-scale farmers); (2)
community leaders and relevant district officials, and; (3) national level stakeholders. We
describe each of these sources of qualitative data in turn.
2.3.2.1.

Farmer in-depth interviews
The in-depth interviews with farmers allowed the researcher to get further insights on

issues not adequately addressed during the survey such as the following: (1) why they became
farmers and what motivated them to acquire land; (2) agricultural land use and farmers’
perception of critical drivers; (3) agricultural technology choices, and; (4) relationship between
farm structure change and prevailing institutions and policies.
A total of 24 farmer in-depth interviews were conducted, with12 small-scale farmers and
12 medium-scale farmers. The in-depth interviews were conducted in the six districts were
household survey data was collected using the ACS. The study first identified two medium-scale
farmer in-depth interview participants per district using the ACS list. Selection of the participants
was based on the trajectory the household followed to achieve medium-scale farmer status:
organic growth through small-scale farming or lateral entry through non-farm income. Using
snowball sampling, the study then selected small-scale farmers who were neighbors to the
medium-scale farmers who participated in the in-depth interviews.

27

2.3.2.2.

Community leader in-depth interviews
Community leaders were included in this study because they have vast knowledge

pertaining to land issues and the observed phenomenon of the change in farm structure in their
respective areas. The community leader interviews targeted two community leaders in each of
the six districts. In each district, the study undertook interviews with one village headman from
the agricultural blocks visited during the survey and one high level official working for one of
the key government ministries at district level with knowledge about the evolution of farm
ownership in the district. In total, the study interviewed 12 community leaders.
2.3.2.3.

Interviews with national-level stakeholders
The national level stakeholder perspectives helped to broaden the study’s understanding

of the causes and the consequences of the rise of medium-scale farms in Zambia. A total of
twelve national level stakeholder interviews were conducted with senior officials from relevant
government ministries, civil society organizations dealing with land issues in Zambia such as the
Zambia Land Alliance (ZLA), farmer organizations such as the Zambia National Farmers’ Union
(ZNFU), private sector firms and experts on land issues in Zambia.
2.3.3. Data analysis
2.3.3.1.

Quantitative data analysis
In order to quantitatively address research questions two to five, the study conducted

descriptive analysis of secondary and primary sources of survey data assembled using Stata 14
software. Analysis of all secondary sources of data (CFS, RALS and DHS) involved comparison
of means across small- and medium-scale farming households disaggregated by landholding size
categories. Further, analysis of the secondary data accounted for the survey design by using the
svy command available in Stata. For the primary source of data (ACS), the study compared
28

sample means among the survey respondents disaggregated by their mode of entry to this level
of operation through: (1) farming from family acquired farm; (2) farming from non-family
acquired farm, and; (3) through non-farm income sources.
The first mode of entry described medium-scale farmers that began farming by acquiring
a small piece of land (less than five hectares) from another family member and expanded their
farms with revenue generated from small-scale farming. The second group described mediumscale farmers that began by acquiring a small piece of land (less than five hectares) from
someone other than a family member, and expanded their farms with revenue generated
primarily from small-scale farming. The third mode of entry described medium-scale farmers
that were primarily involved in non-farm employment or business to begin with but afterwards
acquired land (more than five hectares) using their non-farm income sources. Based on the
dichotomy developed by Sitko & Jayne (2014), the first and second mode of entry captured
medium-scale farmers that followed an agricultural-led growth strategy while the third group
captured farmers that followed a lateral entry into medium-scale farming.
2.3.3.2.

Qualitative data analysis
The data collected using key informant interviews were used to gain insights for research

questions one to four. Transcripts were developed for each of the 48 key informant interviews
captured through audio recordings and hand written notes taken during the interviews. Based on
grounded theory approach (Strauss & Corbin, 1994), the study used an iterative process to
develop a coding scheme in NVivo 10 software. The coding scheme specified the concepts and
themes, their definition and the rules for applying the codes. The codes developed enabled the
researcher to easily retrieve similar information across individual transcripts. Summary

29

statements were written to represent the diversity of responses and where possible, quotes from
the respondents were used to clearly represent their perspectives.
2.4. Results and discussion
This section presents and discusses the key findings in line with the study’s main research
questions. To remain consistent with the explanatory mixed-methods research design employed,
the quantitative results are first discussed followed by the qualitative results except in the case of
research question 1 that exclusively relied on qualitative data and analysis.
2.4.1. Perceived causes of farm structure change
Jayne et al. (2016) conjecture several key drivers of the rise of medium-scale investor
farms in SSA. These include the spike in food prices experienced in the latter parts of the last
decade, improved access to inputs and technology, relative profitability of larger-scale farming,
and vociferous farmer lobby groups whose agenda is pro large farms. Using a framework based
on New Institutional Economics, the current study elicited views from key informants about
what they perceived to be the main causes of farm structure change unique to the Zambian case.
We found that change in society’s perception of agriculture, change in informal rules and
customary land governance, change in the land policy environment and unintended consequences
of public spending in agriculture have played a key role in changing the farm structure.
2.4.1.1.

Change in society’s perception of agriculture

All categories of key informants who participated in the in-depth interviews alluded to
the fact that the change in society’s perception of agriculture has attracted new entrants. Some
respondents stated that prior to the liberalization of the Zambian economy in the early 1990s,
agriculture was viewed as a domain for either resource constrained small-scale farmers or highly

30

capitalized large-scale farms owned by European settlers or government state farms. While
relatively medium-scale farms that were better capitalized than small-scale farms existed, they
were few and far between. The general perception that prevailed at the time was that you had to
be poor to be a farmer by default or extremely rich to reap the benefits of farming. Because of
this perception, the majority of Zambians with levels of education high enough to escape the trap
of farming by default, looked to work in industries other than primary agriculture. But with time,
there has been a general realization that agriculture can also be a profitable venture and society’s
perception of primary agriculture in the last two decades has changed.
The change in society’s perception of agriculture has influenced ongoing farm structure
changes in two ways. First, individuals not primarily involved in agriculture but have salaried
jobs or sources of income in other sectors, now view agriculture as a sustainable option for
securing their current and future livelihoods. One respondent put it this way: “People have
realized that there is money in agriculture now. Previously, people wanted to work in mining or
other sectors. But now, they have realized that actually they can make money out of agriculture.
So, everyone wants to have land.” One of the national-level key informants also explained it as
follows: “In the past we [Zambians] were working as workers with no interest in farming. We
now have realized it’s important to have land. The villagers have sold the land along the road.
They’ve just moved further and a lot of people are buying, not just the Chinese, but also
Zambians because they want to be growing things. I have colleagues at the University [of
Zambia] who have small-holdings.”
Second, the change in society’s perception of agriculture has also influenced land
acquisition decisions of people involved in primary agriculture for most of their lives. One
farmer explained it this way: “I think people have woken up. They are learning from people

31

coming from urban areas. A lot of people from the urban areas are coming to get land, so people
have realized that land is finishing and are being compelled to acquire agricultural land to
increase production.”
2.4.1.2.

Changes in enforcement of informal rules of land governance

The current study also focused on understanding the role that recent developments in
formal and informal land governance rules have had on farm structure. Of particular interest to
this study were changes in policy provided for in the 1995 Lands Act and changes in customary
land governance.
All categories of respondents were asked whether they knew anything about changes in
land policy in Zambia before eliciting their views on the link between land policy and farm
structure change. The majority of farmers, mostly small-scale farmers, indicated that they had
not been exposed to any information related to changes in land policy. Knowledge of changes to
land policies for other categories of respondents—community leaders and national-level
stakeholders—varied among the respondents with some showing limited knowledge while others
were extensively familiar with relevant land policies obtaining in Zambia. However, the majority
of these respondents did not attribute the changes in farm structure to changes in land policy or
enactment of the 1995 Lands Act. They argued that the Act is not very clear and that there is no
proper mechanism for implementation of existing provisions. Therefore, they contended that the
changes in land policy have not directly induced the rapid expansion of domestic medium-scale
investor farms. The following statements highlight this view:
“I don’t think policy has played any role. It is only because people have realized that they
need to go back to the land” (community leader, August 2014).
“The problem is that the current land policy or Lands Act is not very clear and therefore
difficult to implement. It is just that people in this era are waking up to the idea that they

32

need some land; they need some land to do some farming” (land policy consultant,
August 2014).
Nonetheless, the general consensus among the respondents was that changes in
enforcement of customary rules of land governance have played a key role in changing the farm
structure landscape. Through in-depth interviews, the study asked the respondents to highlight
some of the key features of customary land governance, whether these features have changed
over the last two decades and the effects of these changes, if any, on farm structure in Zambia.
Despite a few variations in customary norms and values in the different sites visited for this
study, the following features of customary land governance cut across all the communities
visited: (1) headmen administer land at village level on behalf of a chief; (2) land is freely
allocated if available; and (3) community members are not allowed to rent out or sell land.
Across all categories of respondents, there was widespread consensus that these salient
features of customary land administration have not, in principle, changed. However, a number of
them indicated that with increased demand for land, land allocation has not been free in reality
and a land rental and sales market has emerged in some communities; the changes in
enforcement of informal rules of land governance have been the genesis of this. The result of this
emerging land rental and sales market is that land acquisition has disproportionately been by
those who can afford to pay for it. As one community leader stated, “The rules have not changed,
but if one has money, they are likely to get land than someone without money.” One mediumscale farmer interviewee also described the prevailing situation in this manner: “I do not think
the rules have changed, but these days you have to buy to acquire land [in customary areas].”
The majority of the in-depth interview respondents perceived that the subtle changes in
the governance of customary land have had an effect on farm structure in Zambia. They said that

33

the development of the informal market for land has significantly contributed to the acquisition
of individual land parcels especially between 10 and 50 hectares of land in rural areas closer to
urban centers of Zambia. This has effectively ushered in new participants in agricultural
production who are perceived to be relatively more passionate about farming than an average
small-scale farmer.
2.4.1.3.

Unintended consequences of agricultural subsidy programs

The Zambian government signed onto the Comprehensive Africa Agricultural
Development Program (CAADP) during the African Union (AU) Heads of State summit held in
2003 in Maputo, Mozambique. The objective of CAADP is to “help African countries reach a
higher path of economic growth through agriculture-led development, which eliminates hunger,
reduces poverty and food insecurity, and enables expansion of exports” (African Union, n.d.). A
key commitment made by each member state at this summit was to allocate at least 10 percent of
their national budgets to agriculture.
Since the establishment of CAADP, Zambia has steadily increased its public spending
and has in some years (since 2007) surpassed the program’s spending objectives (Kuteya &
Kabwe, 2015). When critically analyzed, however, there are two related issues of concern about
this increase in spending which threaten attainment of the program objective in Zambia. First, the
increase in spending has been primarily achieved because of increased spending in two subsidy
programs that account for more than 50 percent of the agriculture budget—Farmer Input Support
Program (FISP) and Food Reserve Agency (FRA) (Sitko & Chamberlin, 2015). Second, the
subsidy programs disproportionately favor relatively larger farms that have the capacity to cover
the cost-sharing requirements of the FISP and to produce surplus maize to sell to the FRA at
prices above equilibrium market prices (ibid).

34

The in-depth interviews inquired further on the latter concern. Although all the
interviewees indicated that the two subsidy programs were targeted towards both small- and
medium-scale farms, they shared a number of implementation challenges that have implicitly
distorted the intended benefit structure. With respect to the FISP, the respondents shared three
main implementation challenges. First, the respondents overwhelmingly indicated lack of
transparency in the way some cooperatives—farmer organizations through which farmers access
FISP inputs at local level—operate and that such have only benefited a few. They further
indicated that those with financial capacity are the ones who have greater control of existing
cooperatives. The following statements summarize the respondents’ concerns:
“There are loopholes you know. FISP is politics. What usually happens is that even
political cadres who are not engaged in farming form cooperatives and access program
inputs. So you find that a local political cadre, his spouse, nephews and people that are
dead become members of that cooperative.”
“When you join the cooperative you just continue benefiting even if the rules say one is
supposed to graduate after a few years. You find that it is the same people who are
members of the cooperative for years.”
“It’s not the small farmers [who are benefitting] from FISP. Cooperatives, through which
FISP inputs are sold, need financial resources and logistics to be able to travel between
the community and districts to try and lobby for input packs. The people that have these
financial resources and connections tend to be those who are on a higher scale—a very
small-scale famer will be part of the [cooperative] executive but not necessarily at the
forefront of making decisions to access this fertilizer.”
Second, some respondents complained about the late delivery of FISP inputs as a major
implementation challenge. As a result, farmers, especially small-scale farmers who rely on
subsidized inputs, plant late, have poor harvest and therefore little to no surplus production for
the market. Third, most of the farmer respondents mentioned that the input packs delivered to
some cooperatives are fewer than the expected allocation. Therefore, members of a cooperative

35

are either compelled to share or some are completely left out from accessing the few available
input packs.
The respondents highlighted two main implementation challenges of the maize output
market provided by the FRA. The first challenge mentioned was that the FRA rarely pays the
farmers on time after they have delivered their maize. This delay has a negative impact on the
farmers’ ability to meet contemporaneous household needs and future plans for the next
agricultural season. The second challenge indicated was that some of the procedures put in place
by FRA of repackaging grain at their buying depots have effectively removed some resourceconstrained farmers from extracting the benefits of selling their maize at prices above market
equilibrium. As one respondent put it: “the problem is that you have to repackage the maize
again when you had already done that at home. So that process is costly because you pay for
labor when doing that at the depot again.”
Put together, the unintended consequence brought about by the implementation
challenges of the two subsidy programs is that the incentive structure has been distorted. The
distorted incentive structure has worked in favor of relatively resource-rich farmers who are
more likely to be medium-scale farmers.
2.4.2. Characterizing farm structure change in Zambia
In order to have a clear characterization of farm structure change in Zambia, the current
study conducted the following series of analyses using both quantitative and qualitative data.
First, the study assessed the change in the number of farms and total landholdings of households,
disaggregated by classifications based on landholding size. Second, we examined the
demographic characteristics of domestic investor farms and, third, we explored available data

36

sources to understand the extent to which urban households are entering into agriculture. We
discuss each of these in turn.
2.4.2.1.

Nature of change in farm structure

This study sought to explain the nature of change in farm structure in Zambia using two
Crop Forecast Survey (CFS) datasets separated by a period of 15 years. Table 1 summarizes a
number of important statistics that demonstrate the nature of these changes between 2001 and
2015. The table uses landholding size to characterize farm structure in Zambia classified as
follows: (1) 0-2 hectares; (2) 2-5 hectares; (3) 5-10 hectares; (4) 10-20 hectares, and; (5) 20-100
hectares.
Three important changes can be identified from this table. First, although the number of
farms for all categories has increased during the reference period, the change in the absolute
number of farms with five hectares and above has been more rapid. When analyzed at national
level, the change in the number of farms in the 0-2 hectares and 2-5 hectares categories is less
than 40 percent while farms above 5 hectares have increased by more than 60 percent. When we
isolate six districts that were the site for further investigation of medium-scale farm growth, the
increase in the number of farms above 5 hectares is over 100 percent.
Second, the results show that the proportion of small- and medium-scale farmland owned
by farmers with five hectares and above has increased by 11 percent from 58 percent in 2001 to
69 percent in 2015 when analyzed at national level. If the focus is on the six districts, we find a
similar positive trend (15 percent increase) in the proportion of total small- and medium-scale
farmland owned by farms with five hectares and above during the two time periods.
Third, the results indicate that average landholding has decreased the most for the
smallest category of farms with less than 2 hectares at both national level and the six selected

37

districts. On the other hand, the largest category (20-100 hectares) of medium-scale farmers has
seen an increase in average landholding of about 11 and 34 percent at national level and the six
districts respectively.
Findings from the qualitative analysis also reveal the general perception that there has
been an expansion of medium-scale farms in Zambia in recent years. As one medium-scale
farmer observed: “[the] number [of medium-scale farmers] is increasing. It is like a veil has just
been removed from people, everyone has arisen and looking for land.” Other key informants,
especially national level stakeholders, gave more insights on how the expansion of medium-scale
farms is happening across the whole country including regions that previously had minimal
agricultural activity. One national-level stakeholder explained it this way: “I have had an
opportunity to travel around the country and in my tour of duty I think that [medium-scale farm
expansion] is happening countrywide. People have just realized that land has value and as such
they are strategically positioning themselves to use it [land] now to earn a living or leave it for
future investment or to use it as an inheritance for their children.”
The insights from both the quantitative and qualitative analysis suggest that the changes
in farm structure have been characterized by expansion of medium-scale farms. The quantitative
analysis, however, demonstrates that these changes have further led to land concentration in
fewer households both nationally and in the six districts analyzed.

38

Table 1: Changes in farm structure among small- and medium-scale farmers in Zambia (2001 – 2015) based on official national
survey data
Landholding
Number of farms
Change in
% of total small- and
Total landholdings (ha)
% change in
size categories
number of
medium-scale farms
total
farms (%)
landholdings
2001
2015
2001-2015
2001
2015
2001
2015
2001-2015
National level
0-2 ha
597,310
704,776
17
15
8
1.1
1.0
-9
2-5 ha
343,372
466,051
36
27
23
3.2
3.3
3
5-10 ha
122,318
204,793
67
24
28
6.8
6.8
0
10-20 ha
35,535
73,154
106
16
22
13.4
13.2
-1
20-100 ha
12,897
22,226
72
18
19
36.8
41.0
11
Total
1,111,019
1,469,136
32
100
100
3.1
3.7
19
Six selected
districts*
0-2 ha
63,811
81,050
27
11
5
2-5 ha
44,037
58,992
34
26
17
5-10 ha
16,339
39,953
145
28
34
10-20 ha
5,444
12,067
122
21
25
20-100 ha
1,174
2,920
149
14
20
Total
130,805
194,982
49
100
100
Source: CSO/MAL Crop Forecast Surveys (CFS) various years
Notes
* Six selected districts = Chibombo, Choma, Chongwe, Kalomo, Mpongwe and Mumbwa.
ha = hectares

39

1.1
3.2
6.9
13.3
33.9
3.3

1.0
3.4
7.0
13.5
45.5
4.4

-9
6
1
2
34
33

2.4.2.2.

Characteristics of medium-scale farms

The ACS of medium-scale farms conducted in 2013 captured valuable information on the
demographic characteristics of domestic land investors, as this is an important step to knowing
who these farmers really are. Table 2 presents summary statistics on demographic characteristics
of medium-scale farmers by comparing them based on three modes of entry to this scale of
operation. About 41 percent of the respondents indicated that they achieved medium-scale
farming status through farming from a family acquired farm while those who achieved this scale
of operation either through non-family acquired farm or non-farm income accounted for 43 and
16 percent of the respondents respectively.
A number of key lessons emerge from the comparison presented in Table 2. First,
medium-scale farm owners who enter through non-farm income sources are on average older (53
years) than those entering through farming from family acquired farms (49 years) or those
entering through farming from non-family acquired farms (50 years). Second, medium-scale
farm owners in the non-farm income source category are more likely to have held a salaried job
other than farming than those entering through alternative means. Third, more than two thirds of
respondents who achieved medium-scale farming either through non-family acquired farms or
non-farm income sources are more likely to have been born outside the areas where they own
agricultural land. On the other hand, less than 45 percent of medium-scale farm owners who
entered farming through family acquired farms were born outside the areas where they own
agricultural land. Fourth, while we are cognizant of the fact that Zambia has undergone shifts in
the generational composition of its population, the results still suggest that more than 75 percent
of land acquisitions by domestic land investors for all three modes of land acquisition have
happened during the last two periods (1990 – 99 and 2000 – later).

40

Table 2: Demographic characteristics of medium–scale farmers in selected districts of Zambia
by mode of land acquisition
Mode of land acquisition
1
2
3
Full
sample
% Cases
41
43
16
100
% Men
90
91
89
90
Respondent’s marital status
Single (%)
2
2
0
2
Monogamous marriage (%)
59
64
81
65
Polygamous marriage (%)
27
27
8
24
Divorced (%)
4
2
3
3
Widowed (%)
7
5
8
6
Age (years)
49
50
53
50
Years of education of head
9
9
10
9
Have held a salaried job other than farming (%)
31
44
66
42
Parents to the head owns/owned land (%)
89
90
80
88
Where was the head born?
Born in the area (%)
56
34
31
42
Not born in the area but settled and live in the
42
57
58
51
area (%)
Not born in the area, acquired farm in the area
2
9
11
7
but spend most of time elsewhere (%)
When did HH head acquire land?
1969 or earlier (%)
2
2
0
2
1970 – 79 (%)
5
3
7
5
1980 – 89 (%)
15
15
12
14
1990 – 99 (%)
33
31
33
32
2000 or later (%)
45
49
48
47
Notes
Source: Agricultural Commercialization Survey of medium-scale farms in Zambia (2013)
Mode of entry: (1) Achieved through farming from family acquired farm; (2) achieved through
farming from non-family acquired farm; (3) achieved through non-farm income
The qualitative part of this study investigated the origins of medium-scale farm owners.
Although the general consensus by all 48 key informants interviewed was that a sizeable
proportion of medium-scale farmers are individuals who are considered local, they
acknowledged that there has also been an increase in the number of medium-scale farmers

41

coming from other parts of Zambia.2 From the in-depth interviews, three main categories of
“non-local” medium-scale farmers were identified.
The first group consists of medium-scale farm owners who have retired from salaried
jobs based either in the nearest or outside district town. These farmers have been able to acquire
land through established social networks. The second group of medium-scale farm owners
consists of those migrating from areas with a number of limitations such as land availability
constraints, land degradation and climate change, and lack of important infrastructure to support
agricultural expansion. The third group of medium-scale farms is individuals working in urban
areas that have acquired land and started farming. One of the national level stakeholder
respondents ably described the third group in this manner: “It is not only indigenous people
coming from a given community that own medium-scale farms; it is also people like us
[professionals/working class]. When one goes for work [field visits in some of these
communities], community members usually make offers to sell land.”
In sum, findings from both the quantitative and qualitative analysis suggest that the
expansion of medium-scale farms in Zambia is not only characterized by farm owners who are
local to the community but individuals migrating from other parts of the country. Moreover,
individuals who have had or still have access to wage income have also driven the expansion in
medium-scale farms and that this surge in land acquisitions has gathered momentum in the last
20-25 years.

2

A farmer is considered local if he or she has family ties in an area and is most likely to have
ethnic ties to people in that settlement. A household head is local if he/she is CONSIDERED
local. In many cases people will consider themselves as non-local if they are denied certain
benefits as a result of having migrated into the area. In other cases migrants will still consider
themselves as local if they feel that they have been accepted into the community and have the
same basic rights as other residents (CSO, MAL, and IAPRI 2012, p.77).
42

2.4.2.3.

Agricultural land ownership by urban-based households

Anecdotal evidence in Zambia suggests that urban-based households have also been a
major factor in farm structure changes unfolding in the country. With limited data, however,
studies looking at farm structure in Zambia have generally just hypothesized that the rapid
expansion in domestic land investors and hence farm structure change could also be associated
with urban based households. Only recently did Jayne et al. (forthcoming) empirically assess the
extent to which urban-based households control agricultural land in six sub-Saharan countries
including Zambia using the Demographic Health Surveys. Using the Zambian DHS data
collected in 2007 and 2013/14, the current study adapts the analysis conducted in the
aforementioned study but further extends it to show the following: (1) changes in the proportion
of agricultural landholdings held by urban households by landholding size categories; (2)
changes in average landholding, in hectares, held by urban based households, and; (3) how these
changes link to the rapid expansion of domestic land investors.
Results presented in Table 3 suggest that although the number of households who stated
that they own agricultural land between 2007 and 2013/14 seemed to have reduced, the
proportion of agricultural land that is controlled by urban households has increased during the
same period. When disaggregated by landholding size categories, the results show that the
change in the proportion of agricultural land controlled by urban-based households has decreased
for farms below ten hectares, marginally increased for farms between 10 and 20 hectares but
significantly increased for farms above 20 hectares. In terms of average landholding by urban
households analyzed by landholding size categories, the results indicate a marginal reduction,
marginal increase, modest increase and significant increase for farms in the 0-2 hectares, 2-10
hectares, 10-20 hectares and 20-100 hectares range respectively. The findings in Table 3 lends

43

credence to the view that urban-based households have been playing a key role in the control of
agricultural farmland in Zambia. It is, however, interesting that the level of control in terms of
landholding size is more likely to lead to an increase in medium-scale farms and a probable
reduction in small-scale farms primarily owned by rural based households.
Table 3: Summary statistics of agricultural land ownership of urban based households in Zambia
(2007 – 2013/14)
2007
2013/14
Change
(A)
(B)
(B – A)
Number of households interviewed
7,164
15,920
% of urban households owning agricultural land
27.4
24.8
-2.6
% of rural households owning agricultural land
88.3
88.1
-0.2
% of all households owning agricultural land
67.2
61.7
-5.5
% of total agricultural landholdings held by
19.5
26.6
7.1
urban households
% of total agricultural landholdings held by
urban households by landholding categories
0-2 ha
2.9
0
-2.9
2-5 ha
2.2
0.7
-1.5
5-10 ha
2.5
1.7
-0.8
10-20 ha
2.6
2.8
0.2
20-100 ha
9.4
21.3
11.9
Average agricultural landholding size (ha) held
by urban households by landholding categories
0-2 ha
0.2
0
-0.2
2-5 ha
3.9
4.0
0.1
5-10 ha
8.6
8.9
0.3
10-20 ha
15.4
16.9
1.5
20-100 ha
52.7
63.5
10.8
Source: Demographic Health Surveys (2007, 2013/14).
Notes: Households with landholding above 95 hectares were reclassified as owning 95 hectares
because the DHS did not collect actual landholding data from such households.
ha = hectares

44

2.4.3. Agricultural land use in the context of farm structure change
To understand how land use has evolved in the context of farm structure change, the
current study examined available quantitative data by generating the following statistics: (1)
average size of area cultivated by landholding size; (2) average proportion of uncultivated area
by landholding size, and; (3) average proportion of uncultivated land used for grazing by mode
of entry into medium-scale farming. Results for (1) and (2) are reported in Table 4 and those for
(3) are reported in Table 5.
Table 4: Comparison of cultivated and uncultivated agricultural land across landholding size
categories in Zambia (2012 – 2015)
Landholding size categories
Year 0-2 ha 2-5 ha 5-10 ha 10-20 ha 20-100
Full
ha
sample
Number of respondents
2012
4020
3002
1244
440
133
8839
2015
3388
2589
1242
492
223
7934
Landholding (ha)
2012
0.9
3.1
6.9
13.4
42.6
3.8
2015
1.0
3.2
7.0
13.5
44.6
4.7
Area cultivated (ha)
2012
1.1
2.4
4.5
6.3
6.5
2.3
2015
1.1
2.4
4.0
5.6
6.3
2.4
Area uncultivated (ha)
2012
0.1
0.7
2.3
7.1
36.0
1.5
2015
0.2
1.0
3.2
8.1
37.3
2.4
Maize area planted (%)
2012
64
55
57
58
59
59
2015
63
58
58
59
59
60
% Uncultivated land
2012
13
22
34
53
85
40
2015
16
30
45
60
84
52
Source: Rural Agricultural Livelihoods Surveys (2012, 2015)
Note: ha = hectares

45

Table 5: Characteristics of uncultivated land by medium-scale farmers in selected districts of
Zambia
Full
Mode of entry into medium-scale
sample
farming
1
2
3
% Cases
100
41
43
16
Uncultivated land:
Fallow (%)
38
44
32
38
Forested (%)
42
32
54
35
Rented out (%)
6
1
4
20
Grazing/pastures (%)
14
23
10
7
Source: Agricultural Commercialization Survey of Medium-Scale farms in Zambia (2013)
Mode of entry: (1) Achieved through farming from family acquired farm; (2) achieved through
farming from non-family acquired farm; (3) achieved through non-farm income
Results show that the average number of hectares cultivated increases with landholding
size and this is true across the two rounds of RALS data. The cropping system is dominated by
maize with an average of 60 percent of area cultivated planted to maize. This dominance of
maize production is strikingly similar across all landholding size categories and has also
remained consistent over the two survey years.
The proportion of area that is uncultivated is increasing with landholding size and this is
evident across both survey years. In 2012, for example, households in the 0-2 hectare category
only had an average of 13 percent uncultivated land while about 85 percent of total landholding
for households in the 20-100 hectare category was uncultivated. This similar pattern emerged in
2015 where households in the 0-2 hectare category had 16 percent uncultivated land but those in
the 20-100 hectare category had 84 percent uncultivated land. The fact that farmers with
relatively large landholdings cultivate less than 20 percent of the land they own suggests that a
lot of land is being left idle by this category of farmers. However, one has to be cautious with
interpreting these results since there is a possibility that some of the uncultivated land
particularly by medium-scale farms is dedicated to animal grazing and not necessarily left idle.

46

To establish whether farmers apportion most of their uncultivated land for animal grazing
purposes, the current study further analyzed the medium-scale farmer survey conducted in
selected districts of Zambia. The survey asked the respondents to give estimates of how they
apportion uncultivated land based on the following categories: (1) fallow land; (2) forested land;
(3) land rented out, and; (4) grazing/pasture land. Results reported in Table 5 show that more
than 70 percent of uncultivated land is either fallow or is forested across the three modes of entry
into medium-scale farming. The proportion of uncultivated land that is dedicated to grazing
animals is only about 23 percent, 10 percent and 7 percent for farms categorized based on the
three respective modes of entry (farming from family acquired farm, farming from non-family
acquired farm, and non-farm income sources). These results affirm the conclusion that most
medium-scale farms leave idle a greater proportion of their land.
The study also explored stakeholder perceptions of changes in land use in the context of
farm structure change. The line of inquiry specifically focused on getting the views of all key
informants in relation to changes in crop and livestock production in the last five to ten years.
The majority of all respondent groups agreed that crop production patterns had changed over the
last decade. The consensus by most respondents, who perceived that crop production patterns
had changed, indicated that crop production had now become maize centric. As one respondent
put it, “… in the 80’s and early 90’s, farmers were growing a diversity of crops. People grew
different types of crops including those mainly grown for the fertility of the soil but now it is just
maize. I have seen the mono-cropping culture resurfacing.” Some respondents, however,
highlighted changes in cropping patterns, which have witnessed a growth particularly in oilseeds
(e.g. soya beans) as a consequence of crop diversification efforts promoted by either the
government or the private sector.

47

The qualitative research revealed a number of reasons for the changes in cropping
patterns, but the following were mentioned the most by the interview participants. First, the
respondents argued that favorable market conditions for maize supported by government policies
have been key to the resurfacing of the maize monoculture. One of the favorable market
conditions cited included near proximity of maize buying points (Food Reserve Agency (FRA)
or private traders) to maize producing households. In addition, a number of respondents stated
that the low gross margins in some of the traditional cash crops such as cotton has played a role
in farmers moving back to maize production. Second, the respondents also mentioned labor
constraint as a factor that has contributed to producers moving away from more labor-intensive
crops (e.g. cotton) to less labor-intensive crops. Third, some respondents attributed the shift to
other crops such as oilseeds to changes in the environment and climate.
Respondent views about livestock production also suggested that production patterns
have evolved in the last decade. Four notable developments pertaining to livestock production
emerged during the key informant interviews. First, while cattle production is still common
particularly in regions where it has been part and parcel of tradition, several communities have
embraced small ruminant production such as goats. One respondent gave an interesting
illustration of how things have changed in his community in terms of goat production. He said it
as follows: “when I was growing up, when you introduced a goat in the area, dogs would think
it’s a wild animal and they would even catch it. But now, almost everyone has a goat.”
Second, a number of communities have become more proactive in terms of disease
control compared to two decades or so ago when they were more reactive, waiting for
government to do something about animal diseases. Third, livestock production in some
communities has also seen a shift to more intensive methods of livestock production such as use

48

of feedlots. Fourth, the growing urban population and change in consumption patterns have also
led to the growth of poultry production in Zambia.
2.4.4. Land titling and markets in the context of farm structure change
To gain further insights into the effects of changes in farm structure in rural Zambia, the
current study used the various survey data and stakeholder perceptions to analyze the extent and
pattern of titling of agricultural land and farmer participation in land markets. We discuss the
quantitative and qualitative findings for each of these attributes in turn.
2.4.4.1

Extent of land titling
To assess the extent and pattern of land titling in the context of farm structure changes,

the current study examined both the survey and qualitative data assembled. Using the 2015 round
of the Rural Agricultural Livelihoods Survey (RALS), Table 6 reports information on tenure
status disaggregated by landholding size. Seven tenure status categories were reported in the
2015 RALS: (1) state land with title; (2) state land with title in the process; (3) state land with no
title; (4) former customary land with title; (5) former customary land with title in the process; (6)
customary land with no title, and; (7) customary land with chief certificate.
The first three categories refer to land not under the jurisdiction of traditional leaders
differing only on account of the land owner having either processed title deeds or has
commenced the process of obtaining title deeds or has no title deeds at all. The fourth, fifth and
sixth categories refer to land under the jurisdiction of traditional leaders differing only on
account of the stage of processing title deeds similar to the case of state land. For the chief's
certificate (category seven), it refers to land under the jurisdiction of traditional leaders except in
this case; a particular chief gives a certificate of occupancy to any individual requesting for it to

49

abate conflicts in future. While this is a practice that existed prior to the 1995 Lands Act, it has
increased of late.
The following can be summarized from Table 6 about the extent and pattern of land
titling. First, the proportion of either state land or former customary land that is with title or is in
the process of being titled is clearly an increasing function of landholding size. For instance, only
four percent of land (both state and customary land) controlled by households in the 0-2 hectares
category is either under title or is in the process of being titled. On the other hand, more than 20
percent of land (both state and customary land) controlled by households in the 20-100 hectares
category is either titled or undergoing the process of titling.
Second, households with large landholding sizes are more likely to process the traditional
landholding certificate granted by a chief than households with small landholdings. About eight
percent of households in the 0-2 hectares category have acquired traditional landholding
certificates while 17 percent of households in the 20-100 hectares category have managed to
process these certificates. The traditional landholding certificate initiative has been popularized
by civil society organizations like the Zambia Land Alliance (ZLA) in collaboration with
bilateral partners like the USAID’s Tenure and Global Climate Change (TGCC) project. The aim
of this initiative has been to provide landowners on customary land acquire formal
documentation in order to minimize land disputes. The current study suggests that households
with larger landholdings are the ones taking advantage of this initiative aimed at helping
vulnerable households who are most likely land constrained.

50

Table 6: Extent of land titling by small- and medium-scale farmers in Zambia by landholding
size categories
Landholding size categories
0-2
2-5
5-10 10-20 20 -100
Full
ha
ha
ha
ha
ha
sample
Number of respondents
3388 2589 1242
492
223
7934
% Land owned by tenure status
State land with title
2
State land with title in the process
1
State land with no title
12
Former customary land with title
1
Customary land with title in the
0
process
Customary land with no title
76
Chief certificate
8
Source: Rural Agricultural Livelihoods Surveys (2015)
Note: ha = hectares

3
1
7
1
1

2
1
8
2
2

4
2
11
3
1

10
3
19
6
3

5
2
12
3
1

82
5

79
6

69
10

42
17

67
10

The qualitative research phase of the study elicited respondents’ views about reasons why
the extent of land titling has remained low especially among smallholder farmers. Three main
explanations emerged. First, the farmer in-depth interviews generally revealed a ubiquitous lack
of awareness of existing provisions that allow individuals such as smallholders living in
customary land areas, to process title for agricultural land under their control. Interestingly, the
interviews also revealed that smallholder farmers who are aware about these provisions have not
proceeded with getting title because such farmers have not understood the process involved.
Second, respondents from all categories of key informants suggested that traditional
leaders have continued to express reservations about approving conversion of customary land to
titled land—a key policy introduced in the 1995 Lands Act. According to the respondents,
traditional leaders have continued to contend that the aforementioned provision threatens to shift
the balance of power away from chiefs/headmen and render them redundant; customary land
administration by and large embodies authority of traditional leaders. The following statements

51

summarize the views of key respondents interviewed about whether traditional leaders allow
community members to freely apply for title deeds:
“We discourage them [community members] because when you get title for the land the
headman loses authority over the land. We don’t like the issue of the title deeds”
(headman, August 2014).
“I have not [applied for title deeds]. I found it difficult because headmen fear to give title
deeds because they know they won’t have power [since] everyone will be a headman on
his own” (medium-scale farmer, August 2014).
“No [we have not applied for title deeds]. This is because we are afraid that the village
headman may feel as though his powers are being threatened” (small-scale farmer,
August 2014).
The third explanation that was also widely mentioned by key informants as the reason for
the low extent of land titling among smallholders had to do with the lengthy and bureaucratic
process of obtaining title. Other than technical delays such as getting survey diagrams of land in
question promptly processed, there are other administrative hurdles that exacerbate bureaucracy.
This is because a number of aspects pertaining to processing of titles are not decentralized and
require approval by government officials based at the Ministry of Lands headquarters in Lusaka.
Unfortunately, this issue of the Government of the Republic of Zambia perpetuating a highly
centralized system of carrying out functions pertaining to land has been an issue of concern for
decades (Bruce et al., 1995). The undesirable outcome of this lengthy and bureaucratic process is
that documents sometimes go missing somewhere between local government authorities at
district level and the Ministry of Lands. In such cases, applicants have been forced to restart the
process of applying for title. Ultimately, the lengthy process has proven to be quite expensive
and has effectively eliminated a number of smallholder farmers from applying for title deeds.

52

2.4.4.2.

Extent of participation in land markets

The study also examined (both quantitatively and qualitatively) the extent of participation
in land markets particularly in customary land areas. Using data from the two waves of RALS,
the study quantitatively estimated the proportion of rural households that thought that market
based land transfers were possible in customary land areas. The qualitative data analysis, on the
other hand, explored in detail the evolution of land markets in the context of farm structure
change.
The current laws of Zambia allow for conversion of land under customary land tenure to
leasehold tenure (Government of the Republic of Zambia (GRZ), 1995). This law in turn allows
owners of “former customary land with title” to participate in market-based land transfers. This
study assessed whether or not it is possible to buy or sell “customary land with no title” as this
has implications for changes in farm structure. Figure 1 presents results of the proportions of
rural households that stated that market based land transfers were possible for parcels of land
categorized as “customary land with no title”. A Chi-square test was used to see the relationship
between the aforementioned proportions and our five-landholding size categories. The test
results are reported at the bottom of Figure 1 for each year of the RALS survey data.

53

50%
45%
40%
35%
30%
25%

31%
28%

26%
22%

22%

21%

25%

25%

5 - 10 ha

10 - 20 ha

29%

22%

20%
15%
10%
5%
0%
0 - 2 ha

2 - 5 ha

2012

> 20 ha

2015

Figure 1: Proportion of rural households by landholding size categories stating that market
based land transfers are possible in customary areas
Source: Rural Agricultural Livelihoods Surveys (2012, 2015)
Note: ha = hectares
Pearson Chi-square test within years:
2012—Pearson chi-square, F-stat = 22.3, p-value = 0.000
2015—Pearson chi-square, F-stat = 14.9, p-value = 0.005
The following findings emerged. First, more than 20 percent of households across
different landholding size categories thought that it was possible to buy or sell land in customary
land areas. Second, landowners with large landholding sizes were more likely to state that land
under customary tenure could be transferred using markets. In 2012, for instance, more than 30
percent of landowners with landholding size greater than 20 hectares stated that it was possible
to buy or sell land in customary land areas in comparison to 25 percent of landowners in the 0-2
hectare category. Based on the chi-square test results, these proportions are statistically different
in both 2012 and 2015. Third, the proportion of households thinking that it was possible to buy
or sell land in customary land areas was similar across the two survey years indicating that
respondents’ views remained consistent between the two time periods. Put together, these three
findings suggest that market based land transfers in customary land areas are happening despite

54

legal requirements that do not allow such transfers to take place. The outcome of these
‘vernacular’ land markets—a term borrowed from Chimhowu & Woodhouse (2006) of the
categorization of land markets in customary land areas—is that they are likely to contribute to
the increase in pace of land consolidation by elite domestic land investors.
The information gathered through key informant interviews on land markets in customary
land areas could be described as follows. First, the interviews provided more insights on whether
land markets in customary land areas in Zambia exist and why. Second, the interviews asked
respondents to explain their local understanding of the mode of land acquisition described as
‘allocated/given’ in most survey data collected from rural households. A follow up hypothetical
question was asked to most of the respondents to consolidate the understanding of the
aforementioned mode of land acquisition. The question was phrased as follows: “When an
individual negotiates with the headman to have him/her allocate 50 hectares to that individual,
and some money changes hands informally, would you still describe this mode of land
acquisition as ‘allocated/given’?” The second set of questions was included to find out whether
the mode of land acquisition categorized as ‘allocated/given’ inadvertently masks the full extent
of informal land markets.
The general view of the key informants interviewed for the study was that informal
markets exist and are widespread in customary land areas in Zambia. Almost all of the
informants interviewed seemed to suggest that renting of land has been the common feature of
these emerging markets while selling or buying of land has been done in a subtler manner.
The respondents gave the following reasons when asked to share their views of why land
markets have emerged in customary land areas. On the supply side (renting out or selling land),
most respondents indicated that some households living in customary land areas are too poor to

55

afford inputs to sustain agricultural production. Therefore, renting or selling part of their land
allows them to meet input requirements as well as immediate consumption needs. On the
demand side (renting in or buying land), some respondents argued that the improved economic
fortunes associated with farming have attracted individuals from outside affected communities
including urban-based households and hence the emergence of land markets. One respondent
explained it this way: “people right now want to go into farming so we see a lot of people come
to the village asking a person with a big piece of land to rent them land.”
Farmer and community leader key informants were asked to explain their local
understanding of the mode of land acquisition categorized as ‘allocated/given’ that is mainly
associated with customary land. The consensus among the respondents was that when land is
allocated or given, it implies that the recipient of the land does not pay a market value price.
However, the tradition has been that after a person has been shown and given the land, they are
expected to give a token of appreciation to the headman who showed them the location of the
land in question.
The respondents’ views to the follow up hypothetical question—whether the mode of
land acquisition with an informal payment for a relatively large piece of land after negotiations
with a headman, can still be considered as ‘allocated/given’—were evenly split. On the one
hand, some respondents (both farmers and community leaders) still maintained that they would
consider this mode of land acquisition as ‘allocated/given’ and that whatever amount of money is
paid, is just a token of appreciation. The following statements summarize the views of the key
respondents who had this view:
“When someone comes to ask for land from the [village] headman and land is found to be
available, the [village] headman shows the person the land who will then give a token of
appreciation. So, the land is still given to him. The only thing he has done is to say thank
you.”

56

“That is just to appreciate what the parent [headman] has done for you; the land is still
given to you.”
“Yes, I would still say it’s given. This is because it’s not selling as such. This is just part
of the procedure that one has to follow; it’s just a token of appreciation.”
The second group of respondents (also composed of both farmers and community
leaders) used the following explanations to argue against the fact that the aforementioned mode
of land acquisition could be considered as ‘allocated/given’. First, the key informants with this
counterview explained that village headmen have devised a system of asking potential land
seekers to pay for a unit of land similar to a market based system. Usually, when someone gives
the headman a token of appreciation, the payment is in kind and not cash. Alternatively, if the
token of appreciation is in cash form, it is generally small, not based on a pricing system similar
to market based transactions and is determined by the person looking for land. However, if a
headman uses a payment system reminiscent of a market-based system, the mode of land
acquisition cannot be categorized as ‘allocated/given’ but ‘purchased’. One of the community
leader informants described it this way: “there are those [headmen] who literally sell their land.
They even have rates on a per hectare basis. For example, they would put the rate at ZMW5,000
[Zambian Kwacha] per hectare. That is selling and not “allocated/given.”
Second, some key informants argued that because land has increasingly become scarce in
a number of communities, the mode of land acquisition where land is allocated or given freely no
longer exists. To a great extent, whether or not land is given freely is likely to vary by scarcity of
land of a given location or region of the country. Key informants explained that cases where
people were allocated land freely happened years back but this has not been the case in the last
one to two decades. In this light, if an individual negotiates with a headman for a piece of land as
large as 50 hectares, the exchange of land with money is part of the negotiation process. The
57

individual looking for land knows that they need to pay something close to market value to
expedite the land allocation process. The key informant interviews further revealed that an
individual seeking a relatively large piece of land diminished their chances of getting it if no
payment was made to the headman.
In sum, the qualitative research component of this study found that the extent of land
markets in customary land areas has been expanding over the years and has been partly a
consequence of those with money. One of the small-scale farmer respondents put it this way:
“These days they just go through the village and approach people and ask if they can buy some
land from them. And you know these days, if you have money it is easy to buy land.”
2.4.5. Level of agricultural commercialization
The final set of analysis conducted in this study focused on understanding how farm
structure in Zambia (using quantitative data) is associated with the level of agricultural
commercialization especially in agricultural output markets as this has implications for rural
development and economic growth. The analysis assessed production and sales data of farming
households in Zambia using Crop Forecast Survey (CFS) data of small and medium-scale
farming households for 2009, 2012 and 2014. To capture production and sales across the whole
farm size continuum in Zambia, the analysis also included data from Large-Scale CFS data: an
additional survey that is conducted by the CSO of farming households cultivating more than 20
hectares. Disaggregating the data by two landholding size categories— 5 hectares or less (smallscale farming households) and > 5 hectares (medium- and large-scale farming households)—the
study specifically conducted the following analyses in line with the fifth study objective (to
examine the relationship between farm structure change and level of agricultural
commercialization).

58

First, given the important role of maize in Zambia’s agricultural sector, the study
estimated maize production (Table 7) and marketed output (Table 8) separately across
landholding sizes. Second, the study also estimated the gross value of production (Table 9) and
marketed output (Table 10) of non-maize crops across landholding size categories. It was
inevitable to use the gross value of production and marketed output because the study was
aggregating production volumes of more than one non-maize crop. Third, the proportions of
farms across landholding sizes categories (Table 11) were also estimated.
According to results reported in Table 7 and 8 for maize production and marketed output,
the following can be summarized. First, maize production in Zambia has increased across the
years for both landholding size categories. Small-scale farming households produce more than
50 percent of total maize production. However, the proportion of total maize produced of
farming households with five hectares and above has steadily increased during the five year
period considered. For example, the proportion of national maize production (Table 7) of the
aforementioned households was about 41 percent in 2009. In 2014, households in this category
produced about 46 percent of total national maize production. Second, marketed maize output
(Table 8) has also increased across the years for both landholding size categories. But, the
proportion of national marketed output is greater for medium- and large-scale farms than that of
small-scale farms.

59

Table 7: Maize production in Zambia by landholding size categories
Year
0 – 5 ha
> 5 ha
Maize production (Thousand metric tons)
2009
1132
772
2012
1619
1334
2014
1809
1541
% of national maize production
2009
59.47
40.53
2012
54.84
45.16
2014
54
46
Source: CFS data (2009, 2012, 2014)
Note: ha = hectares
Table 8: Marketed maize output in Zambia by landholding size categories
Year
0 – 5 ha
> 5 ha
Marketed maize output (Thousand metric tons)
2009
345
491
2012
785
850
2014
824
1088
% of total marketed maize output
2009
41.23
58.77
2012
48.01
51.99
2014
43.09
56.91
Source: CFS data (2009, 2012, 2014)
Note: ha = hectares
Table 9 and 10 report the gross value—in Zambia’s currency, Kwacha (ZMW)—of
production and marketed output of non-maize crops produced by both landholding size
categories. The results demonstrate that production and marketed output have both increased
over the five-year period. While small-scale farms had a higher level of production relative to
medium- and large-scale farms of non-maize crops in 2009, the gross value of production of the
latter group exceeded that of former group in 2012 and 2014. This finding illustrates the increase
in the role that relatively large farms in Zambia play in the production of crops outside the
politically influenced maize production. Similar to maize marketing results reported above, gross
value of marketed outputs of non-maize crops is higher for medium- and large-scale farms than
that of small-scale farms.

60

In sum, although agriculture is still dominated by small-scale farms in Zambia (more than
80 percent as reported in Table 11), medium- and large-scale farms have steadily increased their
share of national crop production especially non-maize crop production. In terms of crop
marketing (maize and non-maize crop sales), small-scale farms continue to contribute to national
marketed output. However, the level of participation in marketing of all crop types by mediumand large-scale farming households suggests that relatively large farms have an important role in
Zambia’s agricultural growth and development.
Table 9: Gross value of production (crops only) in Zambia by landholding size categories
Year
0 – 5 ha
> 5 ha
Gross value of crop production (million ZMW)
2009
2810
2560
2012
3750
3920
2014
3880
4260
% of total gross value of crop production
2009
52.33
47.67
2012
48.89
51.11
2014
47.67
52.33
Source: CFS data (2009, 2012, 2014)
Note: ha = hectares
Table 10: Gross value of marketed crops in Zambia by landholding size categories
Year
0 – 5 ha
> 5 ha
Gross value of marketed crops (million ZMW)
2009
1050
2200
2012
1880
3310
2014
1730
3120
% of total gross value of marketed crops
2009
32.11
67.89
2012
36.22
63.78
2014
35.67
64.33
Source: CFS data (2009, 2012, 2014)
Note: ha = hectares

61

Table 11: Proportion of farms in Zambia by landholding size categories
Year
0 – 5 ha (%)
> 5 ha (%)
2009
89
11
2012
84
16
2014
82
18
Source: CFS data (2009, 2012, 2014)
Note: ha = hectares
2.5. Conclusions and policy implications
Although the rise in foreign owned large-scale farms has somewhat altered the
agricultural landscape in SSA, recent evidence has also shown that the rise in farmland
acquisitions by domestic medium-scale investor farms—between 5 and 100 hectares—represents
a more significant change in farm structure. In Zambia, recent estimates have shown that
medium-scale farms now account for more land than the holdings in the entire small-scale sector.
In view of the relative underperformance of small-scale agriculture in SSA that has been
documented by a number of scholars, one important question that has emerged is this: is the
change in farm structure characterized by the rapid expansion of medium-scale investor farms,
signaling an alternative pathway for agricultural and rural development?
This study investigated the causes and consequences of the rise of domestic mediumscale investor farms in Zambia using a mixed-methods approach. By locating the study within
the broader political economy and new institutional economics literature, the study gleaned some
important findings.
Three main causes of farm structure change associated with domestic medium-scale
investor farms emerged from the key informant interviews. First, society’s perception of
agriculture has changed from being viewed as a domain for those who are poor to a domain that
is seen as a profitable sector. Second, the governance of customary land has undergone some
changes that have led to the development of informal land markets contrary to established norms.

62

Third, public spending in agriculture has had the unintended consequence of shifting support to
relatively better capitalized medium-scale investor farms at the expense of small-scale farm
households. Put together, all the three drivers identified by this study had contributed to the rise
of domestic medium-scale investor farms and change in farm structure in Zambia.
The consequences of the change in farm structure are closely associated with agricultural
land concentration, perpetuation of a maize monoculture, accumulation of idle land among
medium-scale investor farms and increasing land constraints among small-scale farmers. Further,
the evidence pointed to low land titling especially among small-scale farms, a growing informal
land rental and sales market and a skewed level of agricultural commercialization.
These findings point to three main conclusions. First, while there are some positives that
can be drawn from the identified causes of farm structure change, growth in Zambian agriculture
is not being driven by organic growth of existing small-scale farmers. Rather, a considerable
number of entrants in the medium-scale farming sector are either local elites with connections to
traditional leaders responsible for land administration in customary land or urban-based
Zambians with the financial capacity to acquire land. If small-scale farmers are not growing
organically, stated public policy goals of addressing household food security and reducing rural
poverty (currently standing at 78 percent headcount) will be difficult to achieve. As a result, this
threatens prospects of achieving overall economic development through agricultural
development.
Second, the system of land acquisition is unintentionally being rigged in favor of the
wealthy even though the provisions of the 1995 Lands Act, which allow for conversion of
customary land to titled land, are not designed to distort things. The prevailing system has
implications for prospects of small-scale agricultural expansion especially when recent evidence

63

clearly shows that small-scale farmers face serious land constraints (Chisanga & Chapoto, 2015).
Third, the limited participation of small-scale farmers in agricultural output markets presents
another set of challenges. Our analysis demonstrates that although more than 80 percent of
farmers in Zambia fall within the small-scale farming category, approximately two-thirds of
gross value of marketed agricultural crops is by medium- and large-scale farms. This suggests
that small-scale farmers remain predominantly subsistent farmers. While this is not a problem in
terms of addressing food security concerns if the level of subsistence enhances self-sufficiency,
limited participation of the majority of farmers in output markets may weaken prospects of a
vibrant commercialized agricultural sector, exacerbate high levels of poverty and rural
unemployment and further frustrate rural development efforts.
Policy strategies should therefore address not only the inequalities that exist in terms of
land access and ownership but also address the bottlenecks associated with obtaining title or
indeed traditional landholding certificates. Although there is considerable debate on the link
between agricultural productivity and tenure status (Place, 2009), we demonstrate that the
process of obtaining title is mainly being exploited by farmers who are categorized as mediumscale farmers. If small-scale farmers are not part of this process, there is a danger that, with time,
some households may become landless. To enhance small-scale farmers’ role in agricultural
markets, policy should also endeavor to promote research and development initiatives that would
uplift their production levels and generate surplus for markets. Obviously, this strategy should
not be maize centric but should extend to other crops or livestock suitable for small-scale
agriculture.

64

APPENDIX

65

Figure 2: Map of Zambia showing location of ACS households

66

Appendix A1: Key informant interview guides
Participant’s Code #_________________________________________
Place: ____________________________________________
Date: ________________________
Time______________________
Hello [Participants name]
My name is Chewe Nkonde currently pursuing my doctoral studies at Michigan State University
(MSU) in the United States of America (USA). Thank you for taking time to meet with me and
participate in this interview.
This interview is part of a study that seeks to understand stakeholder views on the constraints
affecting and opportunities influencing agricultural development in the context of rising large
landholders (emergent farmers) in Zambia. You have been identified as a key informant among
different stakeholders, which includes farmers, traditional authorities, district officials, and
stakeholders and thought leaders at national level. In sum, I anticipate to interview about 48 key
informants.
This component of the broad study is designed to address the following main objectives:
1. Assess the direct and indirect impacts of emergent farms on income growth and
distribution
2. Examine how land accumulation in agriculture is changing over time and the implications
for small-scale agriculture
3. Characterize land use of small-scale and emergent farmers and explore the factors that
influence land utilization
4. Explore the institutional changes (formal or informal) that are contributing to the rise of
emergent farmers
The key informant interviews are designed to supplement the data collected from a survey that I
coordinated called the Agricultural Commercialization Survey (ACS). The ACS for emergent
farming households in Zambia was conducted in 2011 and 2013 in six districts namely
Chibombo, Choma, Chongwe, Kalomo, Mpongwe and Mumbwa. A total of 482 households
participated in the survey.
Overall, this study is part of the requirement for my doctoral program. Ultimately, I expect to
present the results from this study in Zambia and East and Southern Africa to promote
discussions among African policy makers and local analysts regarding how land policy in Africa
can be harmonized more compatibly with other national policy objectives. Therefore, your
voluntary participation in this interview will be very much appreciated.
Today’s interview should take about one hour to complete. All questions have been designed to
help me draw on your knowledge to help enrich the quality of my research. A follow up
interview may be requested if more information is desired. It is important to know that there are
no right or wrong answers. I just want to know what your views about the topic are with respect
to your farm, community or the whole country. Your participation is voluntary and you may
choose not to participate at all, refuse to answer certain questions, or stop your participation at
67

any time without any consequences. If you are interested, you will be provided with a final
document of the whole research at the end of the study.
To ensure that I do not miss anything we talk about, I would like to audio-record our
conversation. All your responses will be kept confidential and your privacy will be protected to
the maximum extent allowable by law. All reports and publications resulting from this interview
will be written and shared using pseudonyms and code numbers.
If you have any questions or concerns regarding your participation in this study, you may contact
my dissertation advisors:
1. (Major advisor) Robert B. Richardson, PhD, Associate Professor, Department of
Community Sustainability, Michigan State University, 480 Wilson Road, Rm 305, East
Lansing, MI 48824-1222, Tel: +1(517)-355-9533, e-mail: rbr@msu.edu
2. (Advisor) Thomas S. Jayne, PhD, Professor, Department of Agricultural, Food and
Resource and Economics, Michigan State University, 446 W. Circle Dr., Rm 317c, Justin
S. Morrill Hall of Agriculture East Lansing, mi 48824-1039 Tel: +1(517)-604-1572 Fax:
+1(517)-432-1800 email: jayne@anr.msu.edu

At this point, do you have any questions?

Yes

No [if yes, answer questions and proceed]

Is it okay for you if I audiotape our interview?
May I begin?

Yes

Yes

No

No

[If no, thank and end]

Interview guide 1: Questions for medium-scale farmers (Emergent farmers)
Theme 1: Farmer categorization, land access and distribution
1. In the survey we conducted in 2013, we classified you as an emergent farmer based on the
size of land you owned at the time of the survey (we defined emergent farmers as those
owning between 5 and 100 hectares of land). Do you agree with our assessment of your
status? Please elaborate on why you agree or disagree.
2. Why did you become a farmer?
3. Based on our definition of an emergent farmer, what is your assessment of the number of
emergent farmers in your community? Has the number increased or reduced in the last 20
years? 10 years? What do you think has led to this increase (decrease) in the number of
emergent farmers in your community?
4. If someone wants land in this area, is there land available?
5. About 62% of emergent farming households interviewed in 2013 indicated “allocated/given”
as the mode of land acquisition for the land they owned. Could you please clarify to me what
“allocated/given” really means? Is the land given freely without any cash or in-kind
transaction?
6. When an individual negotiates with the headman to have him/her allocate 50 hectares to
him/her, and some money changes hands informally. Would you still describe this mode of
land acquisition as “allocated/given”?
68

7. In the 2013 survey, the largest proportion (49%) of emergent farmers, by our definition, said
that they achieved that status through farming from non-family acquired farm and using nonfarm income sources. Is it difficult to achieve emergent farmer status by farming from a
family acquired farm (growing from small-scale to emergent farmer status)? Please share
your thoughts with us on this.
8. What are the primary barriers, from your experience, for achieving a larger scale of
operation?
9. Are there any other reasons that hinder some farmers from graduating to emergent farmer
status?
Theme 2: Land use and drivers of utilization
1. How many crops do you cultivate in a serious way? (Ignore small garden plots, etc.)
2. Please describe at most three main crops that you have consistently planted in the last 3-4
farming seasons.
3. Has the choice of three main crops that you have planted in the last 3-4 seasons changed
from what you planted in the first 3-4 seasons when you began farming? If there has been a
change, please explain what has led to this change.
4. Do you anticipate changing your current three main crop portfolio in the near future?
5. What crops or alternative land use do you envisage to change to? Please explain why.
6. What are some of the main factors that determine how you utilize the land that you own?
Explain how each factor influences land utilization.
7. Results from the 2013 Survey indicated that only 6.2% of emergent farming households
cultivated the entire land owned. Are you in the category of farmers who usually do not
cultivate the entire land they own?
8. In what state is the land that you leave uncultivated? Is it cleared? Forested? Fallow rotation?
Theme 3: Direct and indirect effects of emergent farms on household income growth and
distribution
1. What is your current main source of income?
2. Has this been your main source of income since you became an adult?
3. If farming is your current main source of income, what is your assessment of your annual
gross farm income from the time you started farming? Has your income been increasing or
decreasing from one year to the next since you started farming? Please explain.
4. What are some of the favorable factors that have contributed to your ability to sustain an
increase in annual gross farm income?
5. Do you hire labor from surrounding small farms to work on your farm?
6. Is this primarily piece work/ganyu, salaried-casual or salaried-formal?
7. How do you hire and how often?
8. Why do you hire labor?
9. Why don’t you hire people permanently?
10. What do you think of the quality of hired labor?
11. Are people readily available from small farms to work on your farm?
12. Please explain whether your level of agricultural production (crop and livestock) has
increased, decreased or remained unchanged in the last 5-10 years? Explain why it has
increased, decreased or remained unchanged.
69

13. How do you interact with small-scale farmers in your vicinity? Do you provide some form of
training or knowledge transfer to them? Explain how and the type of training.
14. Do you buy agricultural output from small-scale farmers in your locality? Please explain
how and specify the types of output.
15. Do you sell agricultural inputs and services to small-scale farmers surrounding you? Please
explain the terms of such transactions.
Theme 4: Technology choices
1. More than 96% of emergent farming households stated that they used either animal or
mechanized draft power for land preparation during the 2012/2013 farming season in the
communities we visited including yours. Which of the two methods do you prefer using and
why?
2. Are there hiring services for animal draft power in your community?
3. Do you provide these [animal draft power] services?
4. Do you hire out when you have completed all your tasks on your farm or do you do it
whenever somebody asks for your services?
5. Why do you hire out your oxen draft power?
6. In general, do you think that the use of animal draft power for land preparation has increased
in your community in the last 5-10 farming seasons?
7. Are there tractor-hiring services for land preparation and other farming activities in your
community?
8. Do you provide these [tractor hiring] services?
9. Do you use these [tractor hiring] services?
10. Why do you use the tractor hiring service?
11. In general, do you think that the use of mechanized draft power (tractors) for land preparation
has increased in your community in the last 5-10 farming seasons? And among what type of
farmers?
12. Can farmers access tractor-hiring services on credit?
13. Are you aware of any initiatives (government, farmer organizations, donors, NGO’s, private
sector) that have been recently promoted in your community to enhance tractor use as a
source of power for land preparation?
14. If you are aware of such initiatives, what categories of farmers do they target? Please explain
why they target those categories.
15. Have you been consistent with using improved seed varieties, inorganic fertilizers and other
agricultural intensification technologies over the last 5-10 farming seasons? Could you say
your use of such inputs has increased or reduced over the stated period?
Theme 5: Institutions and policies
1. Are you aware of any government policies designed to empower landowners with respect to
tenure security? Have you heard of the lands act?
2. Have formal land rental and sales markets become common in your community?
3. Have informal land rental and sales markets become common in your community?
4. Are you aware of any government policies designed to empower farmers with respect to
enhancing farm productivity, food security and marketing of surplus production?
5. What are these policies and could you describe some of their key provisions?
70

6. Have you been a consistent beneficiary of these policies designed to improve farm
productivity, food security and marketing of surplus production?
7. Does the presence of emergent farms (including yours) influence the types of services that
are available locally (e.g. vets, mechanics, transportation services, extension, input
suppliers)? Please distinguish this effect in the village versus impacts on services available at
the nearest district.
8. Do larger farms (including yours) attract better/different types of buyers of agricultural
output? If so, why? If not, why
9. Do you belong to a farmer group (association/cooperative)?
10. If yes, how long have you been a member?
11. Is the membership restricted to farmers at your scale of operation and larger? Or, do you
have small-scale farmers who are members?
12. What are some of the benefits that you get from being a member?
13. What are the rules and norms governing land administration (allocation and provision of
tenure security) in your community?
14. If you were involved in off-farm employment prior to settling in this area, to what extent did
the economic changes in the 1990’s when the Zambian economy was liberalized compel you
to acquire land and become a farmer? Please explain.
Interview guide 2: Questions for small-scale farmers
Theme 1: Farmer categorization, land access and distribution
1. In your community, has the number of farmers owning between 5 and 100 hectares of land
(aka emergent farmers by our definition) increased or reduced in the last 10 years?
2. What do you think has led to this increase (decrease) in the number of emergent farmers?
3. Why did you become a farmer?
4. Why did you acquire land?
5. About 73% of small-scale farming households interviewed in 2012 indicated
“allocated/given” as the mode of land acquisition for the land they owned. Could you please
clarify to me what “allocated/given” really means? Is the land given freely without any cash
or in-kind transaction?
6. What are your plans with respect to growing from small-scale farming to emergent farming?
7. Is it difficult to achieve emergent farmer status by growing from small-scale farming? Please
share your thoughts with us on this.
8. Does the presence/growth of large farmers in this area make it harder for small farmers to
access/acquire additional land here? (In other words, is the growth of larger holdings
contributing to scarcity?)
9. Do large farmers bring benefits to small farmers in this area? If so, what are these benefits?
(Could be: new farming technologies/knowledge, wage-income opportunities, traction
services, transport services, aggregation/bulking opportunities, lower cost of inputs, etc….)
Theme 2: Land use and drivers of utilization
1. How many crops do you cultivate in a serious way? (Ignore small garden plots, etc.)
2. Please describe at most three main crops that you have consistently planted in the last 3-4
farming seasons.

71

3. Has the choice of three main crops that you have planted in the last 3-4 seasons changed
from what you planted in the first 3-4 seasons when you began farming? If there has been a
change, please explain what has led to this change.
4. What are some of the main factors that determine how you utilize the land that you own?
Explain how each factor influences land utilization.
5. Do you anticipate changing your current three main crop portfolio in the near future?
6. What crops or alternative land use do you envisage to change to? Please explain why.
Theme 3: Direct and indirect effects of emergent farms on income growth and distribution
1. What is your current main source of income?
2. Has this been your main source of income since you became an adult?
3. If farming is your current main source of income, what is your assessment of your annual
gross farm income from the time you started farming?
4. Has your income been increasing or decreasing from one year to the next since you started
farming? Please explain.
5. What are some of the favorable factors that have contributed to your ability to sustain an
increase in annual gross farm income?
6. Are there wage-earning opportunities on emergent farms in this area?
7. Do you or any of your family members earn income from this source?
8. Do they pay well/fairly?
9. Is this primarily piece work/ganyu, salaried-casual or salaried-formal?
10. Please explain whether your level of agricultural production (crop and livestock) has
increased, decreased or remained unchanged in the last 5-10 years? Explain why it has
increased, decreased or remained unchanged.
11. Have you gained any knowledge about farming and other related activities from them?
Explain how?
12. Do you sell any of your agricultural output to them? Please explain how and specify the types
of output.
13. Do emergent farmers sell agricultural inputs and services to you? Please explain the terms of
such transactions.
Theme 4: Technology choices
1.
2.
3.
4.
5.

What is your preferred source of draft power for land preparation? Explain why.
Are there hiring services for animal draft power in your community?
Do you use these services?
Who provides these services? (i.e. are they primarily provided by emergent farmers?)
In general, do you think that the use of animal draft power for land preparation has increased
in your community in the last 5-10 farming seasons?
6. In general, do you think that the use of mechanized draft power for land preparation has
increased in your community in the last 5-10 farming seasons?
7. Are you aware of any initiatives (government, farmer organizations, donors, NGO’s, private
sector) that have been recently promoted in your community to enhance tractor use as a
source of power for land preparation?
8. If you are aware of such initiatives, what categories of farmers do they target? Please explain
why you think they target those categories.
72

9. Have you been consistent with using improved seed varieties, inorganic fertilizers and other
agricultural intensification technologies over the last 5-10 farming seasons?
10. Could you say your use of such inputs has increased or reduced over the stated period?
Theme 5: Institutions and policies
1. Are you aware of any government policies designed to empower landowners with respect to
tenure security?
2. Have you acted upon the provisions of the policies to improve your land tenure and security?
3. Have informal land rental and sales markets become common in your community?
4. Have you ever used these informal land markets either as a renter (landlord) or buyer
(seller)?
5. Are you aware of any government policies designed to empower farmers with respect to
enhancing farm productivity, food security and marketing of surplus production?
6. When did you learn about these policies?
7. Do you sell your produce to FRA?
8. Do you belong to a farmer group (association/cooperative)?
9. For how long have you been a member?
10. If you are a member of a farmer group, what are some of the benefits that you get from being
a member?
11. Are these benefits available to all farmers in your community or specific category of
farmers? Please explain in detail.
12. What are the rules and norms governing land administration (allocation and provision of
tenure security) in your community?
a. Is there any communal land were farmers can have there cattle to graze from?
b. Is it allowed for my cattle to graze from my friend’s farm?
13. Have these rules and norms been the same since you started farming in this community?
Explain if there are any changes that have occurred.
14. If the rules and norms have changed in your community, has the change improved the way
land is administered? Please explain.
Interview guide 3: Local community leaders (includes traditional authorities and district
officials)
Theme 1: Role in community/district
1. How long have you lived in this community/district?
2. How long have you been operating in your capacity as a traditional authority/district official
in this community/district?
3. Please explain the role that your position plays in the administration of agricultural land in
this community/district.
4. What kind of plans are you dealing with especially with agriculture land? Are these farmers
who have title to land or do you also provide services to farmers living in customary land?
5. How do you interact with farmers in this community/district in relation to land issues?
6. What challenges do farmers in this community/district share with you about land related
issues?
73

7. What challenges do you see with respect to land related issues in your community/district?
8. What do you think should be done to address both land related challenges discussed above?
Theme 2: Emergent farmers, land access and distribution
1. Results from national surveys of small- and medium- scale farming households in Zambia
reveal that the area controlled by medium-scale holdings in the 5 to 100 hectare category
now exceeds that of small-scale farms of less than 5 hectares in Zambia. Is this large land
acquisition by so-called emergent farmers something that you have observed in your
community? Please explain.
2. How long has this phenomenon of large land acquisition been going on in this community?
3. Our retrospective survey for emergent farming households in selected districts of Zambia
showed that about 42% of the household heads were born in the communities visited while
the other 58% came from neighboring districts or other provinces of Zambia. From your
assessment, are farmers acquiring large tracts of land community members from within,
Zambians based in urban areas or foreign investors? Explain.
4. What do you think has led to the increase in large land acquisitions in your community, if at
all it is happening?
5. A high proportion of small-scale and emergent farming households that we have interviewed
have indicated “allocated/given” as the mode of land acquisition for the land they own.
Could you please clarify what “allocated/given” really means? Is the land given freely
without any cash or in-kind transaction?
6. When an individual negotiates with the headman to have him/her allocate 50 hectares to
him/her, and some money changes hands informally. Would you still describe this mode of
land acquisition as “allocated/given”?
7. From survey results, less than 45% of farmers perceive that traditional authorities still have
unallocated arable land to give in the area they live. To what extent would you attribute land
availability constraints in customary land areas to the increase in large land acquisitions?
8. Does the presence/growth of large farmers in this area make it harder for small farmers to
access/acquire additional land here? (In other words, is the growth of larger holdings
contributing to scarcity?)
Theme 3: Land use and drivers of utilization in community/district
1. Please explain whether land use in terms of crops grown and livestock raised has evolved in
the last 10-20 years by farmers in your community/district.
2. Has the change in land use happened across all categories of farmers in your
community/district? Explain.
3. What factors have contributed to the change in land use by farmers in your
community/district?
4. Has the use of improved seed varieties, inorganic fertilizers and other agricultural
intensification technologies in your community/district increased or reduced over the last 510 farming seasons? What are the main factors that have contributed to the increase
(decrease) of use of such inputs?
5. What are some of the main factors that determine how farmers (small-scale and emergent
farmers) utilize the land that they own? Explain how each factor influences land utilization.

74

Theme 4: Direct and indirect effects of emergent farms
1. Please explain whether the level of agricultural production (crop and livestock) has
increased, decreased or remained unchanged in the last 5-10 years in your
community/district? Explain why it has increased, decreased or remained unchanged.
2. What effect has the presence of emergent farms had on income growth for small-scale and
emergent farmers in your community/district?
3. Has use of animal draft power (ADP) been on the rise?
4. Has use of mechanical power (tractors) been on the rise?
5. What could you say about hiring services for both ADP and tractors in your community?
6. Are there wage-earning opportunities on emergent farms in this area?
7. Is this primarily piece work/ganyu, salaried-casual or salaried-formal?
8. Does the presence of emergent farms influence the types of services that are available locally
(e.g. vets, mechanics, transportation services, extension, input suppliers)? Please distinguish
this effect in the village versus impacts on services available at the nearest district.
9. Do larger farms attract better/different types of buyers of agricultural output? If so, why? If
not, why?
10. Do emergent farmers provide some form of training or knowledge transfer to small-scale
farmers? Explain how and the type of training.
11. Do emergent farmers offer training to small-scale farmers?
12. Do emergent farmers buy agricultural output from small-scale farmers? Please explain how
and specify the types of output.
13. Do emergent farmers sell agricultural inputs and services to small-scale farmers? Please
explain the terms of such transactions.
Theme 5: Institutions and policies
1. Have formal land rental and sales markets become common in your community?
2. What role has the development of formal land markets contributed to the increase in number
of emergent farmers in your community?
3. Have informal land rental and sales markets become common in your community?
4. What role has the development of informal land markets contributed to the increase in
number of emergent farmers in your community?
5. Which type of farmers in your community/district benefit from government policies designed
to empower farmers with respect to enhanced farm productivity, food security and marketing
of surplus production?
6. Are there farmer groups (associations/cooperatives) offering member services to farmers in
your community?
7. What are some of the benefits that one gets from being a member of the groups that you have
given as examples?
8. Are these benefits available to all farmers in your community or specific category of
farmers? Please explain in detail.
9. What are the rules and norms governing land administration (allocation and provision of
tenure security) in your community/district?
10. Have these rules and norms been the same since you started living in this community?
Explain if there are any changes that have occurred.

75

11. If the rules and norms have changed in your community, has the change improved the way
land is administered? Please explain.
12. To what extent did the economic changes in the 1990’s when the Zambian economy was
liberalized, compel those laid off from wage earning jobs to acquire large tracts of land and
become farmers? Please explain.
13. Should government policy be focused on supporting emergent farmers more than small-scale
farmers?
Interview guide 4: National level stakeholders (government officials, land experts and
thought leaders)
Theme 1: Roles and land related challenges
1. What roles do you or does your organization play in land administration (especially
agricultural land) in Zambia?
2. What land related challenges have farmers shared with you during your interaction?
3. How do you think you are going to address the problem of customary land when most
farmers are still living in the customary lands?
Theme 2: Emergent farmers, land access and distribution
1. Results from national surveys of small- and medium- scale farming households in Zambia
reveal that the area controlled by medium-scale holdings in the 5 to 100 hectare category
now exceeds that of small-scale farms of less than 5 hectares in Zambia. Is this large land
acquisition by so-called emergent farmers something that you have observed countrywide or
is it something that is localized? Please explain.
2. Our retrospective survey for emergent farming households in selected districts of Zambia
showed that about 42% of the household heads were born in the communities visited while
the other 58% came from neighboring districts or other provinces of Zambia. From your
assessment, are farmers acquiring large tracts of land community members from within,
Zambians based in urban areas or foreign investors? Explain.
3. What do you think is leading to the increase in large land acquisitions in parts of Zambia, if it
all it is happening?
4. From survey results, less than 45% of farmers perceive that traditional authorities still have
unallocated arable land to give in the area they live. To what extent would you attribute land
availability constraints in customary land areas to the increase in large land acquisitions?
5. Does the presence/growth of large farmers in parts of Zambia make it harder for small-scale
farmers to access/acquire additional land? (In other words, is the growth of larger holdings
contributing to scarcity?)
Theme 3: Land use and drivers of utilization in Zambia
1. Please explain whether land use by farmers in terms of crops grown and livestock raised has
changed in the last 10-20 years in Zambia.
2. If there has been a change, has the change in land use happened across all categories of
farmers in Zambia? Explain.

76

3. Has the use of improved seed varieties, inorganic fertilizers and other agricultural
intensification technologies in Zambia increased or reduced over the last 5-10 farming
seasons?
4. What are some of the main factors that determine how farmers (small-scale and emergent
farmers) utilize the land that they own? Explain how each factor influences land utilization.
Theme 4: Direct and indirect effects of emergent farms
1. What effect has the presence of emergent farms had on income growth for small-scale and
emergent farmers in Zambia?
2. Results from nationally representative surveys have shown that approximately 35% of smalland medium-scale farmers used animal draft power as their source of power for land
preparation during the 2010/2011 farming season. The ACS survey for emergent farmers’
results indicated that 90% of these farmers used ADP as their main source of power for land
preparation during the 2012/13 farming season. Would you consider these percentages as an
increase in ADP use in comparison to a decade ago?
3. If the use of ADP has increased in the last ten years, why do you think this has occurred? If
not, why?
4. Results from the nationally representative data have shown that only 1% of smallholders
used tractors as their source of power for land preparation during the 2010/2011 farming
season. The ACS survey for emergent farmers’ results indicated that 10% of medium-scale
farmers used tractors as their main source of power for land preparation during the 2012/13
farming season. Would you consider these percentages as an increase in tractor use in
comparison to a decade ago?
5. If the use of tractors has increased in the last ten years, why do you think this has occurred?
If not, why not?
6. Are there wage-earning opportunities on emergent farms in some parts of Zambia?
7. Is this ganyu/piecework of salaried employment?
8. Does the presence of emergent farms influence the types of services that are available locally
(e.g. vets, mechanics, transportation services, extension, input suppliers)? Please distinguish
this effect in the village versus impacts on services available at the nearest district.
9. Do larger farms attract better/different types of buyers of agricultural output? If so, why? If
not, why?
10. Do emergent farmers provide some form of training or knowledge transfer to small-scale
farmers? Explain how and the type of training.
11. Do emergent farmers buy agricultural output from small-scale farmers? Please explain how
and specify the types of output.
12. Do emergent farmers sell agricultural inputs and services to small-scale farmers? Please
explain the terms of such transactions.
Theme 5: Institutions and policies
1. What changes in national government policies are contributing to the increase in large land
acquisitions? The 1995 Lands Act?
2. Have formal land rental and sales markets become common in some communities in
Zambia? If so, since when?

77

3. What role has the development of formal land markets contributed to the increase in number
of emergent farmers in Zambia?
4. Have informal land rental and sales markets become common in some communities in
Zambia?
5. Which type of farmers in Zambia benefit from government policies designed to empower
farmers with respect to enhanced farm productivity, food security and marketing of surplus
production? Please explain.
6. What actions have you taken in the last decade to ensure/support/advocate a more inclusive
pattern of agricultural development?

Conclusion
Is there anything else that you think I have not asked about that might be relevant to the study I
am undertaking?
Thank you very much. All the information you provided is very helpful. In the event that I have a
few further questions or need you to clarify something we discussed today, can I please contact
you again? [If yes, confirm contacts and asked for preferred contact mode]
Thank you.

78

REFERENCES

79

REFERENCES

Adams, M. (2003). Land tenure policy and practice in Zambia: issues relating to the
development of the agricultural sector. Draft Document for DFID. Lusaka, Zambia.
Retrieved from
http://www.mokoro.co.uk/files/13/file/lria/land_tenure_policy_and_practice_zambia.pdf
African Union. (n.d.). About CAADP | African Union. Retrieved April 23, 2016, from
http://pages.au.int/caadp/about
Ali, D. A., & Deininger, K. (2015). Is There a Farm Size–Productivity Relationship in African
Agriculture? Evidence from Rwanda. Land Economics, 91(2), 317–343.
Barrett, C. B., Carter, M. R., & Timmer, C. P. (2010). A Century-Long Perspective on
Agricultural Development. American Journal of Agricultural Economics, 92(2), 447–
468. http://doi.org/10.1093/ajae/aaq005
Barrett, C. B., Reardon, T., & Webb, P. (2001). Nonfarm income diversification and household
livelihood strategies in rural Africa: concepts, dynamics, and policy implications. Food
Policy, 26, 315–331.
Berry, R. A., & Cline, W. R. (1979). Agrarian Structure and Productivity in Developing
Countries: A Study Prepared for the International Labour Office Within the Framework
of the World Employment Programme. John Hopkins University Press.
Brown, T. (2005). Contestation, confusion and corruption: Market-based land reform in Zambia.
In S. Evers, M. Spierenbug, & H. Weis (Eds.), Competing Jurisdiction: Settling Land
Claims in Africa (Vol. 6, pp. 79–102). Leiden: Brill.
Bruce, J., Kachamba, F., & Hansungule, M. (1995). Land administration: Processes and
constraints. Land Tenure, Land Markets and Institutional Transformation in Zambia,
Research Paper, Land Tenure Centre, University of Wisconsin-Madison, 47–77.
Bryceson, D. F. (1996). Deagrarianization and rural employment in sub-Saharan Africa: A
sectoral perspective. World Development, 24(1), 97–111.
Bryceson, D. F. (2002). The scramble in Africa: reorienting rural livelihoods. World
Development, 30(5), 725–739.
Carletto, C., Savastano, S., & Zezza, A. (2013). Fact or artifact: The impact of measurement
errors on the farm size–productivity relationship. Journal of Development Economics,
103, 254–261. https://doi.org/10.1016/j.jdeveco.2013.03.004
Carter, M. R. (1984). Identification of the Inverse Relationship between Farm Size and
Productivity: An Empirical Analysis of Peasant Agricultural Production. Oxford
Economic Papers, 36(1), 131–145.
80

Chen, Z., Huffman, W. E., & Rozelle, S. (2011). Inverse Relationship Between Productivity and
Farm Size: The Case of China. Contemporary Economic Policy, 29(4), 580–592.
http://doi.org/10.1111/j.1465-7287.2010.00236.x
Chileshe, R. A. (2005). Land tenure and rural livelihoods in Zambia: case studies of Kamena
and St. Joseph. University of the Western Cape. Retrieved from
http://www.researchgate.net/profile/Roy_Chileshe/publication/30758061_Land_tenure_a
nd_rural_livelihoods_in_Zambia_case_studies_of_Kamena_and_St._Joseph/links/02e7e5
27e118c67533000000.pdf
Chimhowu, A., & Woodhouse, P. (2006). Customary vs Private Property Rights? Dynamics and
Trajectories of Vernacular Land Markets in Sub-Saharan Africa. Journal of Agrarian
Change, 6(3), 346–371. http://doi.org/10.1111/j.1471-0366.2006.00125.x
Chinene, V. R. N., Maimbo, F., Banda, D. J., & Msune, S. C. (1998). A Comparison of
Customary and Leasehold Tenure: Agriculture and Development in Zambia. In P.
Groppo (Ed.), Land Reform (Vol. 2). Rome, Italy: Food and Agricultural Organization.
Retrieved from http://www.fao.org/docrep/x1372t/x1372t07.htm
Chisanga, B., & Chapoto, A. (2015). Under-appreciated Facts about Zambia’s Agriculture. In
Agriculture in Zambia: Past, Present and Future (pp. 25–36). Lusaka, Zambia: Indaba
Agricultural Policy Research Institute (IAPRI).
CSO, MAL, & IAPRI. (2012). The 2012 Rural and Agricultural Livelihoods Survey for small
and medium-scale holdings: Interviewer’s Instruction Manual.
Deininger, K., & Byerlee, D. (2012). The Rise of Large Farms in Land Abundant Countries: Do
They Have a Future? World Development, 40(4), 701–714.
Delgado, C. L. (1998). Agricultural Growth Linkages in Sub-Saharan Africa. Intl Food Policy
Res Inst.
Diao, X., Hazell, P. B. R., Resnick, D., & Thurlow, J. (2007). The Role of Agriculture in
Development: Implications for Sub-Saharan Africa. Intl Food Policy Res Inst.
Diao, X., Hazell, P., & Thurlow, J. (2010). The Role of Agriculture in African Development.
World Development, 38(10), 1375–1383. http://doi.org/10.1016/j.worlddev.2009.06.011
Economist. (2011, May 5). The surge in land deals: When others are grabbing their land. The
Economist. Retrieved from http://www.economist.com/node/18648855
German, L., Schoneveld, G., & Mwangi, E. (2013). Contemporary Processes of Large-Scale
Land Acquisition in Sub-Saharan Africa: Legal Deficiency or Elite Capture of the Rule
of Law? World Development, 48, 1–18. http://doi.org/10.1016/j.worlddev.2013.03.006
Ghose, A. K. (1979). Farm Size and Land Productivity in Indian Agriculture: A Reappraisal.
Journal of Development Studies, 16(1), 27.

81

Government of the Republic of Zambia (GRZ). The Lands Act, Pub. L. No. Chapter 184 of the
Laws of Zambia (1995).
Haggblade, S., Hazell, P., & Brown, J. (1989). Farm-nonfarm linkages in rural sub-Saharan
Africa. World Development, 17(8), 1173–1201. http://doi.org/10.1016/0305750X(89)90232-5
Hayami, Y. (1977). Review of Agriculture and Structural Transformation: Economic Strategies
in Late-Developing Countries by Bruce F. Johnston; Peter Kilby. Economic Development
and Cultural Change, 26(1), 176–181.
Hazell, P., Poulton, C., Wiggins, S., & Dorward, A. (2010). The Future of Small Farms:
Trajectories and Policy Priorities. World Development, 38(10), 1349–1361.
http://doi.org/10.1016/j.worlddev.2009.06.012
Hichaambwa, M., & Jayne, T. S. (2014). Poverty Reduction Potential of Increasing Smallholder
Access to Land. Michigan State University, Department of Agricultural, Food, and
Resource Economics. Retrieved from
http://www.researchgate.net/profile/Munguzwe_Hichaambwa/publication/254388054_S
mallholder_Commercialization_Trends_as_Affected_by_Land_Constraints_in_Zambia_
What_are_the_Policy_Implications/links/541829cf0cf2218008bf2a5f.pdf
Hoeffler, H. (2011). The Political Economy of Agricultural Policies in Africa: History,
Analytical Concepts and Implications for Development Cooperation. Quarterly Journal
of International Agriculture, 50(1), 29–53.
Jayne, T. S., Chamberlin, J., Traub, L., Sitko, N., Muyanga, M., Yeboah, F. K., … Kachule, R.
(2016). Africa’s changing farm size distribution patterns: the rise of medium-scale farms.
Agricultural Economics, 47(S1), 197–214. https://doi.org/10.1111/agec.12308
Jayne, T. S., Chapoto, A., Sitko, N., Nkonde, C., Muyanga, M., & Chamberlin, J. (2014). Is the
Scramble for Land in Africa Foreclosing a Smallholder Agricultural Expansion Strategy?
Journal of International Affairs, 67(2), 35–53.
Jayne, T. S., Mather, D., & Mghenyi, E. (2010). Principal Challenges Confronting Smallholder
Agriculture in Sub-Saharan Africa. World Development, 38(10), 1384–1398.
http://doi.org/10.1016/j.worlddev.2010.06.002
Johnston, B. F., & Mellor, J. W. (1961). The Role of Agriculture in Economic Development. The
American Economic Review, 51(4), 566–593.
Kim, J., & Mahoney, J. T. (2005). Property Rights Theory, Transaction Costs Theory, and
Agency Theory: An Organizational Economics Approach to Strategic Management.
Managerial and Decision Economics, 26(4), 223–242.
Kimhi, A. (2006). Plot size and maize productivity in Zambia: is there an inverse relationship?
Agricultural Economics, 35(1), 1–9. http://doi.org/10.1111/j.1574-0862.2006.00133.x

82

Kuteya, A., & Kabwe, S. (2015). Zambia Agricultural Sector Performance, 2000 - 2015. In A.
Chapoto & N. J. Sitko (Eds.), Agriculture in Zambia: Past, Present, and Future (pp. 37–
54). Lusaka, Zambia: Indaba Agricultural Policy Research Institute (IAPRI).
Kydd, J., & Christiansen, R. (1982). Structural change in Malawi since Independence:
Consequences of a development strategy based on large-scale agriculture, 10(5), 355–
375.
Land Matrix. (2015). Zambia - LAND MATRIX. Retrieved June 19, 2015, from
http://www.landmatrix.org/en/get-the-detail/by-targetcountry/zambia/?order_by=&starts_with=Z
Lanjouw, J. O., & Lanjouw, P. (2001). The rural non-farm sector: issues and evidence from
developing countries. Agricultural Economics, 26(1), 1–23.
http://doi.org/10.1111/j.1574-0862.2001.tb00051.x
Larson, D. F., Otsuka, K., Matsumoto, T., & Kilic, T. (2014). Should African rural development
strategies depend on smallholder farms? An exploration of the inverse-productivity
hypothesis. Agricultural Economics, 45(3), 355–367. http://doi.org/10.1111/agec.12070
Larsson, R., HolmĂŠn, H., & HammarskjĂśld, M. (2002). Agricultural Development in SubSaharan Africa. Lund: Lund University. Retrieved from
http://lup.lub.lu.se/record/761418
Libecap, G. D. (2008). State Regulation of Open-Access, Common-Pool Resources. In C.
Ménard & M. M. Shirley (Eds.), Handbook of New Institutional Economics (pp. 545–
572). Springer Berlin Heidelberg. Retrieved from
http://link.springer.com/chapter/10.1007/978-3-540-69305-5_22
Malambo, A. H. (2014). Land Administration in Zambia After 1991: History, Opportunities and
Challenges From the 1995 Lands Act. Journal of Geography and Geology, 6(1).
http://doi.org/10.5539/jgg.v6n1p139
McMichael, P. (2012). The land grab and corporate food regime restructuring. Journal of
Peasant Studies, 39(3–4), 681–701. http://doi.org/10.1080/03066150.2012.661369
Mellor, J. W., & Johnston, B. F. (1984). The World Food Equation: Interrelations Among
Development, Employment, and Food Consumption. Journal of Economic Literature,
22(2), 531.
Ménard, C., & Shirley, M. M. (2008). Handbook of new institutional economics. Berlin:
Springer. Retrieved from
http://public.eblib.com/EBLPublic/PublicView.do?ptiID=367382
Nolte, K. (2014). Large-scale agricultural investments under poor land governance in Zambia.
Land Use Policy, 38, 698–706. http://doi.org/10.1016/j.landusepol.2014.01.014

83

North, D. C. (1990). Institutions, institutional change, and economic performance. Cambridge;
New York: Cambridge University Press.
North, D. C. (1994). Institutional Change: A Framework Of Analysis (Economic History No.
9412001). EconWPA. Retrieved from
http://ideas.repec.org/p/wpa/wuwpeh/9412001.html
North, D. C. (2008). Institutions and the Performance of Economies over Time. In C. Ménard &
M. M. Shirley (Eds.), Handbook of new institutional economics (pp. 21–30). Berlin:
Springer.
Ostrom, E. (1990). Governing the commons: the evolution of institutions for collective action.
Cambridge; New York: Cambridge University Press.
Place, F. (2009). Land Tenure and Agricultural Productivity in Africa: A Comparative Analysis
of the Economics Literature and Recent Policy Strategies and Reforms. World
Development, 37(8), 1326–1336. http://doi.org/10.1016/j.worlddev.2008.08.020
Schlager, E., & Ostrom, E. (1992). Property-Rights Regimes and Natural Resources: A
Conceptual Analysis. Land Economics, 68(3), 249. http://doi.org/10.2307/3146375
Schoneveld, G. C. (2014). The geographic and sectoral patterns of large-scale farmland
investments in sub-Saharan Africa. Food Policy, 48, 34–50.
http://doi.org/10.1016/j.foodpol.2014.03.007
Sen, A. (1962). An Aspect of Indian Agriculture. Economic Weekly, Annual Number 14.
Sitko, N., & Chamberlin, J. (2015). The Anatomy of Medium-Scale Farm Growth in Zambia:
What Are the Implications for the Future of Smallholder Agriculture? Land, 4(3), 869–
887. http://doi.org/10.3390/land4030869
Sitko, N. J., & Jayne, T. S. (2012). The Rising Class of Emergent Farmers: An Effective Model
for Achieving Agricultural Growth and Poverty Reduction in Africa? (Working Paper No.
69). Lusaka, Zambia: Indaba Agricultural Policy Research Institute (IAPRI). Retrieved
from http://www.aec.msu.edu/fs2/zambia/wp69.pdf
Sitko, N. J., & Jayne, T. S. (2014). Structural transformation or elite land capture? The growth of
“emergent” farmers in Zambia. Food Policy, 48, 194–202.
http://doi.org/10.1016/j.foodpol.2014.05.006
Staatz, J. M. (1998). Agricultural Transformation Definition. Retrieved March 26, 2016, from
http://fsg.afre.msu.edu/ag_transformation/Def_Trans.htm
Stanton, B. F. (1991). Farm Structure: Concept and Definition. Department of Agricultural
Economics, Cornell University Agricultural Experiment Station, New York State College
of Agriculture and Life Sciences, Cornell University. Retrieved from
http://ageconsearch.umn.edu/bitstream/121378/2/Cornell%20SP%2091-06.pdf

84

Strauss, A., & Corbin, J. (1994). Grounded Theory Methodology: An Overview. In N. K. Denzin
& Y. S. Lincoln (Eds.), Handbook of Qualitative Research (pp. 273–285). SAGE
Publications.
The World Bank. (2008). The Political Economy of Policy Reform: Issues and Implications for
Policy Dialogue and Development Operations (No. 44288–GLB). Washington, D.C:
Social Development Department, The World Bank.
Timmer, C. P. (1998). The Agricultural Transformation. In C. K. Eicher & J. M. Staatz (Eds.),
International Agricultural Development (pp. 113–135). Johns Hopkins University Press.
van Loenen, B. (1999). Land tenure in Zambia. University of Maine. Retrieved from
https://www.researchgate.net/profile/B_Loenen/publication/242672704_LAND_TENUR
E_in_ZAMBIA/links/546e04f90cf29806ec2e6c15.pdf
Williamson, O. E. (1985). Transaction Cost Economics. In The Economic Institutions of
Capitalism (pp. 15–42). New York: The Free Press.

85

CHAPTER 3: REVISITING THE INVERSE FARM SIZE-PRODUCTIVITY
RELATIONSHIP: A CASE STUDY OF ZAMBIA

3.1. Introduction and background
Agriculture in sub-Saharan Africa (SSA) has been experiencing gains in productivity
since the early 1980s (Block, 2010). However, there remain lingering doubts whether an African
“green revolution” can be achieved given the patchy progress in African smallholder agriculture
(Mosley, 2002). The pessimism about African smallholder agriculture could be attributed to a
number of challenges such as: (1) slow productivity growth and uptake of modern inputs (Diao
et al., 2007); (2) shrinking size of most smallholder farms due to rural population growth and
rising land scarcity (Hazell 2005; Jayne et al., 2012), and; (3) widespread land degradation (Lal,
1995). In addition, recent evidence showing a rapid expansion in farmland acquisitions by
domestic land investors (Jayne et al., 2014) and foreign owned large-scale farms (Deininger &
Byerlee, 2012; Land Matrix, 2015) in SSA is depleting available arable land, casting doubts on
the viability of a smallholder-led agricultural growth.
This article revisits the farm size-productivity relationship hypothesized to have an
inverse relationship (IR) in development economics literature. The IR hypothesis postulates that
small farms are more productive per unit of land than large farms (Berry & Cline, 1979). While a
few studies that have found no evidence of a farm size-productivity inverse relationship (Kevane,
1996; Dorward, 1999), a large number of studies have found evidence upholding the relationship
(Sen, 1962; Carter, 1984; Barrett, 1996; Barrett, Bellemare, & Hou, 2010; Chen, Huffman, &
Rozelle, 2011; Carletto, Savastano, & Zezza, 2013; Larson, Otsuka, Matsumoto, & Kilic, 2014;
Ali & Deininger, 2015). Binswanger, Deininger, & Feder (1995) demonstrated that without
distortions that seem to favor large farms, small farms can be as efficient if not more efficient

86

than large farms. As a consequence, findings affirming the existence of the IR have been used as
the basis for supporting agricultural strategies that promote small-scale agriculture especially in
land-scarce countries. In SSA, smallholder agriculture has remained an integral part of the
region’s agricultural development strategies and as such, development scholars have also taken an
interest to examine the farm size-productivity relationship in this context.
Research examining the inverse farm-size productivity hypothesis has explored the
following questions. Is the observed relationship fundamentally because small farms are in and
of themselves more productive than large farms? Or, are there other factors that explain the
evidence that farm size and productivity are inversely related? A number of studies have
empirically examined three contested explanations for why an inverse relationship is
predominantly observed in developing countries. We discuss these in turn.
The first explanation relates to market imperfections in factor markets such as labor,
insurance and credit. Most studies on the IR have mainly focused on imperfections in labor
markets. In an imperfect labor market, the cost of searching and supervising hired labor
(transactions costs) is relatively high. Therefore, this creates an advantage for small farms who
mainly rely on family labor for most farming activities. Large farms, on the other hand, depend
substantially on hired labor to help with farming operations. According to Eswaran & Kotwal
(1986), hired labor have a propensity to “shirk” if not supervised, and therefore, the labor that
can be hired on the market is an imperfect substitute for one’s own time.
Using Pakistan data, Heltberg (1998) found an inverse relationship between farm size and
productivity explained by the presence of a supervision constraint with regard to labor, where
outside workers were imperfect substitutes for family labor. Toufique (2005) compared two
regions in Bangladesh with differences in transactions costs in rural labor markets. The study

87

found that in a region with higher transaction costs, output per acre on smaller farms was found
to be higher than on larger farms. The opposite was true for the region with lower transactions
costs.
However, other studies have questioned imperfection in labor markets as the rationale for
the existence of an IR. Assunção & Braido (2007) argued that the imperfect factor markets
explanation does not explain the inverse farm size-productivity relationship based on evidence
from India. To control for imperfect factor markets, their study analyzed household data with
multiple plots in each season. They found that the inverse relationship remained virtually
unchanged implying that the IR could not be explained by imperfection in labor markets. An
empirical study of rice producers in Madagascar also concluded that only a small portion of the
inverse farm size-productivity relationship could be explained by market imperfections (Barrett
et al., 2010).
The second explanation for the often-observed inverse relationship is the omission of
important variables such as land quality in estimation equations. Since the estimation of the
relationship between farm size and productivity is operationalized using econometric procedures,
omission of relevant variables might lead to a biased coefficient on the variable of interest. If
small farms have better (poorer) quality land, omitting important control variables for land
quality might amplify (diminish) the inverse relationship between farm size and productivity.
Bhalla & Roy (1988) found that when exogenous land quality variables are accounted for, the
inverse relationship observed weakens, and in many cases, disappears. In other cases, failure to
account for unobserved plot level characteristics such as soil and land quality overstated the
inverse relationship (Lamb, 2003; Assunção & Braido, 2007). Conversely, Barrett et al. (2010)

88

found that none of the inverse relationship could be explained by the omission of soil quality
measurements.
The third explanation examined in the literature relates to the role of measurement error
(Lamb, 2003; Holden & Fisher, 2013; Carletto et al., 2013). Measurement error, in this case, is
considered as the inaccuracy in the variable measuring farm size (Lamb, 2003). Using
econometric techniques that control for measurement error inherent in farmer self-reporting of
their farm size, it has been observed that a statistically significant IR, reported prior to estimation
adjustments, completely disappears (ibid). Recent studies (e.g., Carletto et al., 2013; Holden &
Fisher, 2013) have addressed the issue of measurement error by using rich data that include selfreported land size information complemented by plot measurements collected using Global
Position System (GPS) devices. They have found that using an improved measure of land size
(GPS measure) strengthens the evidence in support of the existence of the inverse relationship.
Despite these contestations, however, there are some methodological issues that have not
been adequately addressed in previous empirical studies, especially those conducted in SSA.
First, most research has been limited to data from farms of less than 10 hectares, yet their
findings have been extrapolated beyond this farm size range. Moreover, productivity has mainly
been restricted to one indicator of productivity—land productivity—when exploring this highly
contested hypothesis. Table 12 (column 2) shows that most IR studies have predominantly used
land productivity as the measure of productivity and that there are few cases when more than one
measure of productivity has been used in a single study.

89

Table 12: Productivity measures and farm sizes in selected past IR studies
Author
Study
Measure of
How did they compute the measure
location
productivity
of productivity?
Carter (1984)

India

Land productivity

Log of total annual farm output per
hectare

Heltberg (1998)

Pakistan

Land productivity

Net value of total farm output per
operated holding size

Dorward (1999)af

Malawi

Land productivity

Net value of output per area
planted

Lamb (2003)

India

Assunção & Braido
(2007)

India

Land productivity
Labor productivity
Land productivity

Log of household profits
Log total hours by gender
Log output per acre

Kimhi (2006)af

Zambia

Land productivity

Yield of maize

Barrett et al. (2010)af

Madagascar Land productivity

Yield of rice

Carletto et al. (2013)af

Uganda

Land productivity

Net agricultural revenues per area
operated

Larson et al. (2014)af

Kenya

Land productivity

Yield of maize (Kenya)

90

Salient farm size
characteristics of sampled
households
~90% of sampled
households have farm sizes
of 12 ha or below
Bottom quartile of sampled
households own 2.5% of all
land, while the top quartile
hold 71.5%
~90% of sampled
households have farm sizes
of 2 ha or below
ND
95% of sampled households
have farm sizes of 16.8 ha or
below
87% of sampled households
have farm sizes of 5 ha or
below
95% of sampled households
have farm sizes of 1 ha or
below
90% of sampled households
have farm sizes of 4 ha or
below
75% of sampled households
have farm sizes of 1.67 ha or
below

Table 12 (Cont’d)
Author

Study
location

Measure of
productivity

How did they compute the measure
of productivity?

Malawi

Yield of maize (Malawi)

Tanzania

Yield of maize (Tanzania)

Uganda

Yield of maize (Uganda)

Holden & Fisher (2013)af

Malawi

Land productivity

Net agricultural return per unit area

Verschelde et al. (2013)af

Burundi

Land productivity

Net value of agricultural output per
unit of land

Li et al. (2013)

China

Land productivity

Value per unit area

Labor productivity

Value/working days of the labor or
value/number of farm’s labor force
Ratio of total output to total input

af

Ali & Deininger (2015)

Rwanda

Total factor
productivity
Technical efficiency
Land productivity

Stochastic frontier analysis
Logarithm of the value of crop
output per hectare

Source: Authors’ own compilation from previous IR studies
Notes: af = Studies conducted in an African country ND = distribution of farm sizes not provided

91

Salient farm size
characteristics of sampled
households
75% of sampled households
have farm sizes of 0.74 ha or
below
75% of sampled households
have farm sizes of 2.10 ha or
below
75% of sampled households
have farm sizes of 1.88 ha or
below
90% of sampled households
have farm sizes of 2.23 ha or
below
95% of sampled households
have farm sizes of 4.5 ha or
below
95% of sampled households
have farm sizes of 1.4 ha or
below

98% of sampled households
have farms of 1 ha or below

In our review of the literature, we are only aware of one study (in China, by Li et al.,
2013) that has used a comprehensive set of productivity measures to reexamine the IR hypothesis
such as land productivity, labor productivity, profit ratio, total factor productivity (TFP), and
technical efficiency. Nonetheless, the aforementioned study also has another issue; the data used
covers a narrow range of farm sizes—zero to seven hectares—when exploring the farm sizeproductivity relationship.
The use of a comprehensive set of productivity indicators provides a number of insights,
which cannot otherwise be gleaned from one indicator. For example, land productivity, the most
commonly used indicator, is important if one needs to assess food security especially when land
is a highly binding constraint. Labor productivity is an important measure for capturing farm
profitability while total factor productivity comprehensively captures farm production by
accounting for labor, land and capital inputs simultaneously.
Second, a number of studies conclude that small farms are more productive than large
farms but by extrapolating their findings outside the range of the data available. The last column
in Table 12 presents the operated farm size ranges reported in selected IR studies. While some
studies do not report detailed descriptive statistics, the studies where such statistics are available
reveal that very few include farms outside the 0-10 hectare range. Interestingly, these studies
have mostly been done in countries where land is scarce and population is very dense, which
probably justifies why farm size ranges analyzed have been within the 0-10 hectare range. What
about countries like Zambia that is relatively less dense? Can findings from past IR studies be
generalized? In the Zambian context, tests of the IR hypothesis covering a wide range of farm
scales, take on even greater policy importance in light of recent papers questioning the viability
and even the objectives of promoting small-scale agriculture in Africa (Collier & Dercon, 2014).

92

Third, previous studies have been consistently based on the assumption of constant
returns-to-scale when examining the ceteris paribus farm size-productivity relationship. While
the assumption may have been appropriate for studying the IR across small farms, it is less
relevant when extending beyond this range to include medium- and large-scale farms. This study
argues that findings from such studies cannot address the salient policy questions related to
desired land allocation patterns given the rapid rise of farms between 5 and 100 hectares
observed in some SSA countries such as Ghana, Kenya and Zambia (Jayne et al., 2014).
Against this background, this essay extends the methodological approach used in the
literature to examine the IR hypothesis in the context of farm structure change. The article has
four specific objectives. First, it investigates the IR hypothesis over a much wider range of farm
sizes using a statistically representative sample of small- and medium-scale farming households
(both types of households mainly grow basic grains) with agricultural landholdings between 0
and 100 hectares from central and southern Zambia. Including this wide range of farms in our
analysis can help clarify the IR hypothesis beyond the 0-10 hectare range and inform current
policy discussions about how governments should allocate unutilized/underutilized land in order
to achieve national equity and productivity goals.
In this study, small-scale farmers are defined as agricultural households owning land less
than or equal to five hectares and medium-scale farmers are defined as those whose landholding
size ranges between 5 and 100 hectares.3 These definitions are based on how a number of
African governments distinguish small- from medium-scale farm households. They use 5
hectares as the distinction between small- and medium-scale holdings (Jayne et al., 2016). Also,
3

In Zambia, medium-scale farmers are alternatively referred to as emergent farmers. This term
became popularized in the 1970s when it was used to describe farmers that leave subsistence
farming sufficiently far behind to sell at least half their produce to the market in an average year
(Lombard, 1972).
93

we want to be consistent with how the emerging scholarly work on the rise of medium-scale
investor farms in SSA has defined this scale of farming; they refer to medium-scale farms as
those between 5 and 100 hectares of land (see for example, Jayne et al., 2014; Jayne et al., 2016).
Second, the paper examines the relationship between farm size and productivity using five
alternative measures of productivity: net value of crop production per hectare planted; net value
of crop production per family labor day; net value of crop production per farm labor day; cost of
maize production per metric ton produced, and; total factor productivity. A more detailed
explanation of each productivity measure is provided in the “Data and Methods” section of this
paper.
A number of past studies have exclusively used yield or value of output per unit area—
both measures of land productivity—to investigate the farm size–productivity relationship.
While land productivity is an important measure that can be used to assess differences in farm
productivity, we argue that a more comprehensive look at agricultural productivity is vital, more
so because findings from such studies are used to draw implications for land policies and
agricultural development strategies. For each measure of productivity, all production costs are
accounted for including less commonly considered costs such as family labor and additional
costs associated with fixed assets.
Third, the paper explores reasons for potential differences in productivity within and
between farm size categories. The study further reassesses the validity of the standard practice in
the extant literature of holding everything else constant to examine the relationship between farm
size and productivity. Since important sources of productivity differences could be input or
management decision variables that tend to vary with farm size, we predict the relationship
between farm size and productivity by relaxing the Constant Returns to Scale (CRS) assumption

94

and assess the extent to which this approach affects our interpretation. Fourth, the paper assesses
whether findings of the relationship between productivity and operated farm size should be a
decisive factor in guiding agricultural development and land policies under different situations
characterized by differences in relative factor abundance.
Zambia is an interesting case for the present paper because the country has been
experiencing major changes in farm structure, the most salient of which is a major increase in
cultivated area under the control of farms cultivating between 5 and 100 hectares (Sitko, Jayne,
& Hichaambwa, 2013; Jayne et al., 2014). Kimhi (2006) is the only notable IR study we are
aware of on Zambia. There are two things that are important to note about the aforementioned
study. First, it uses data from the crop year of 1993-1994. Second, the study addresses the IR issue
using data with 86 percent of the sampled households having farm sizes under 3 hectares. The
changes in farm structure since that study was conducted have warranted an empirical review of
the IR hypothesis that covers the relevant range of farm sizes with the goal of informing policy
makers on how land policies can be harmonized more compatibly with other national policy
objectives.
The rest of the article is organized as follows. The following section presents the data and
the estimation strategy used in this study. The study findings are outlined and discussed in the
third section while the final section offers conclusions and draws out policy implications.
3.2. Data and methods
The study uses two main sources of household survey data collected in Zambia. The first
source of data is the Agricultural Commercialization Survey (ACS) of emergent farming
households in Zambia conducted in 2013 by the Indaba Agricultural Policy Research Institute
(IAPRI) of Zambia. The survey was conducted in six administrative districts of Zambia out of 72

95

districts: Chibombo, Choma, Chongwe, Kalomo, Mpongwe and Mumbwa (see Figure 2 in
Appendices B1 for the map). The aforementioned districts were purposively selected based on
the concentration and number of farmers owning over five hectares of land. Using the nationally
representative Rural Agricultural Livelihoods Survey (RALS) of 2012, districts with at least
three percent of all farmers owning over five hectares, were purposively selected. Medium-scale
farming households in this study were defined as farmers with landholding between 5 and 100
hectares. The proportion of medium-scale farming households was 27 percent, 22 percent, 20
percent, 17 percent, 12 percent and 5 percent for Mumbwa, Kalomo, Choma, Chibombo,
Mpongwe, and, Chongwe respectively.
The main objective of the ACS was to understand the characteristics, behavior, land use
patterns and agricultural output production of medium-scale farms. A total of 482 households
were randomly selected from a list of emergent farming households compiled in consultation
with the Zambia National Farmers’ Union (ZNFU) and the Ministry of Agriculture and
Livestock (MAL) district offices. The Ministry of Agriculture and Livestock (MAL) block areas
were identified as the sampling units with help from local district offices. Whereas the sampling
procedure was supposed to ensure reasonable representativeness of farmers in the 5-100 hectare
category within the selected districts, the sample may have not been statistically representative of
all medium-scale farmers in Zambia. This is because medium-scale farmers were not sampled
from the other 65 districts. But we can say that the chosen districts are understood to contain the
highest proportions of farms in this size class.
We augment the ACS data with a second source of data called the Rural Agricultural
Livelihoods Survey (RALS) of small- and medium-scale farming households in Zambia
conducted in 2012. The RALS was implemented by IAPRI in collaboration with the Central

96

Statistical Office (CSO) of Zambia and the Ministry of Agriculture and Livestock (MAL). The
purpose of this survey was to provide policy relevant information that is not practical to collect
annually from the government agricultural surveys. While the RALS is a nationwide survey, we
restrict our analysis in this paper to the six districts where the ACS was conducted (see Figure 10
in Appendices for the location of households interviewed during the two surveys). Therefore, the
number of households in the sub-sample from the RALS is 1000 households bringing the total
number of households in our pooled sample to 1482. For the analysis in this study, we focus on
1429 households because 53 households either reported that they did not cultivate/plant any part
of their land or had missing values on some of the key variables necessary for computing the
measures of productivity. RALS includes both small- and medium-scale farms from 0 to 20
hectares of land while the ACS only has medium-scale farms between 5 and 100 hectares. In
sum, the sample of 1429 households analyzed in this study includes farm households between 0
and 100 hectares.
Both surveys captured the Geographical Positioning System (GPS) coordinates of
surveyed households. This made it possible to generate Geographical Information Systems (GIS)
data for soil and land quality variables included in our analysis.4 While these variables are
deserving of caution as they are coarse estimates and may not correspond to local conditions,
they are generally good proxies for dealing with the omitted variable problem prevalent in such
studies. The data on soil were collected from the Harmonized World Soil Database v 1.2
(Nachtergaele & Batjes, 2012). In this study, we extracted data on soil nutrient availability that
combines soil characteristics such as soil texture, soil organic carbon, soil pH and total
exchangeable bases. We were able to also extract data for the length of the growing period

4

We would like to thank Jordan Chamberlin for extracting all the GIS data used in this study.
97

(LGP) from the GIS data – LGP combines information on temperature and available moisture to
determine the length of time for adequate crop growth (Fischer, Velthuizen, Nachtergaele, &
JernelĂśv, 2000). Elevation and slope variables were extracted from the Shuttle Radar
Topography Mission (SRTM) data.5 Finally, the extracted rainfall data is a time-series of decadal
estimates collected from 1983 to 2013.6 Using these data, we defined rainfall as the total amount
of estimated rainfall, in millimeters, received during the crop production season for each survey.
3.2.1. Computation of productivity indicators
In the current paper, five distinct indicators are constructed as measures of agricultural
productivity. The indicators are: (1) net value of crop production per hectare planted (land
productivity); (2) net value of crop production per family labor day (labor productivity); (3) net
value of crop production per farm labor day (labor productivity); (4) cost of production of maize
per metric ton produced, and; (5) total factor productivity. Valuation of input costs and output
values is in Zambian Kwacha based on 2011/2012 agricultural season prices.7
The first measure of productivity, net value of crop production per hectare planted (𝑌1 ),
is computed as follows:
(1)

𝑌1𝑖 =

∑ 𝐺𝑉1𝑖𝑗 −∑ 𝑉𝐶1𝑖𝑗 −∑ 𝐹𝐶1𝑖𝑗
∑ 𝑎1𝑖𝑗

where 𝐺𝑉1𝑖𝑗 is the gross value of production, 𝑉𝐶1𝑖𝑗 is the measure of variables costs, 𝐹𝐶1𝑖𝑗 is
the measure of additional costs for using fixed assets , 𝑎1𝑖𝑗 is the area planted to each crop and
all these components are measured across each household i and crop j. Variable costs include the
5

The SRTM data are available here: http://srtm.csi.cgiar.org/

6

Precipitation data are available from: http://www.cgiar-csi.org/data/climate/item/104-cru-ts-31climate-database
7

The average exchange rate prevailing at the time was US$ 1 = ZMW 5.3.
98

opportunity cost of family labor, imputed hired labor costs, seeds, fertilizers, chemicals, and
rental costs for power. The opportunity cost of family labor is computed by multiplying the
number of prime age (15-59 years old) household members working on each farm by the median
days allocated to farming by the median daily agricultural wage rate prevailing in each district.
Imputed hired labor costs are computed by multiplying the district level median daily
agricultural wage rate by the number of hired labor days. Additional costs for using fixed assets
are computed for owned agricultural-related assets used during the production season captured
by the survey data. To compute additional costs for using fixed assets, we use the rental rate for
assets with an active rental market. For example, if a household owns a tractor and uses it for
land preparation, the tractor cost is the land preparation rental rate per hectare multiplied by the
number of hectares cultivated. If the household used the tractor multiple times in a year, for
example, for land preparation, weeding, and harvesting, etc., we valued all the uses of the tractor
to get the cost attributable to the tractor for each household.
While the cost measure is computed mainly using the rental rate approach, an alternative
approach is also used for fixed assets with no rental market (e.g., irrigation equipment, pumps).
For the ‘no rental market’ approach, the cost of an asset is computed as follows:
(2)

𝐹𝐶 = 𝑐⁄𝑛 + (𝑐 ∗ 𝑟)

where FC is the cost of an asset; c is the price or cost of an asset when it is new; n is the
estimated asset’s usual life, and; r is the annual straight line depreciation rate. Assets considered
in the computation of additional costs for using fixed assets (rental and non-rental market
approach) included ploughs (ox drawn and tractor), sprayers, weighing machines, tractors, carts,
trailers, water pumps, irrigation equipment, planters, trucks, water source, shelling and weeding
machines.

99

The second measure of productivity, net value of crop production per family labor day
(𝑌2 ), is computed as:
𝑌2𝑖 =

(3)

∑ 𝐺𝑉2𝑖𝑗 −∑ 𝑉𝐶2𝑖𝑗 −∑ 𝐹𝐶2𝑖𝑗
∑ 𝐿2𝑖𝑗

where 𝐿2𝑖𝑗 is the number of days taken for crop production activities by adult family labor during
the crop production season for household i and crop j. All other components are defined in a
similar way as explained for the net value of crop production per hectare planted. However,
please note that 𝑉𝐶2𝑖𝑗 excludes the opportunity cost of family labor.
The third measure of productivity, net value of crop production per farm labor (family
and hired) day (𝑌3 ), is computed as:
𝑌3𝑖 =

(4)

∑ 𝐺𝑉3𝑖𝑗 −∑ 𝑉𝐶3𝑖𝑗 −∑ 𝐹𝐶3𝑖𝑗
∑ 𝐿3𝑖𝑗

where 𝐿3𝑖𝑗 is the number of days taken for crop production activities by adult family labor and
hired labor during the crop production season for household i and crop j. All other components
are defined in a similar way as explained for net value of crop production per hectare family
labor. However, 𝑉𝐶3𝑖𝑗 excludes the opportunity cost of family labor and the cost of hired labor.
This is because this measure of productivity calculates the returns to both family and hired labor.
The fourth measure of productivity, cost of production of maize per metric ton produced
(𝑌4 ), is computed as:
(5)

𝑌4𝑖 =

∑ 𝑉𝐶4𝑖 +∑ 𝐹𝐶4𝑖
∑ 𝑀𝑇4𝑖

where 𝑉𝐶4𝑖 measures the variables costs of maize production, 𝐹𝐶4𝑖 measures the additional costs
for using fixed assets of maize production, 𝑀𝑇4𝑖 denotes the metric tons of maize produced and
all these components are measured across each household i.

100

Our fifth measure of productivity, total factor productivity (TFP), comprehensively
reflects the productivity of the whole crop production process. Based on the approach used by Li
et al. (2013), which they adapt from Fan (1991) and Zhang and Carter (1997), we use the CobbDouglas production function to calculate TFP using the following functional form:
𝛼𝑀
𝐺𝑉𝑖 = 𝐴0 𝑒 𝜂𝑡 𝐾𝑖𝛼𝐾 𝐿𝛼𝐿
𝑖 𝑀𝑖 exp⁡(𝜀𝑖 )

(6)

where 𝐺𝑉𝑖 is the gross value of crop production of farm i, 𝐾𝑖 , 𝐿𝑖 and 𝑀𝑖 represent the value of
capital (all costs of production except imputed family labor costs), total number of labor days
(hired and family labor), and land inputs (operated farm size) of farm i, respectively, 𝛼𝐾 , 𝛼𝐿 , 𝛼𝑀
are the output elasticities for capital, labor and land, correspondingly, t is the time trend term,
and 𝜂 is the rate of technological progress. Using natural logarithm, equation (6) is estimated as
follows:
(7)

𝑙𝑛𝐺𝑉𝑖 = 𝑙𝑛𝐴0 + 𝜂𝑡⁡ + 𝛼𝐾 𝑙𝑛𝐾𝑖 + 𝛼𝐿 𝑙𝑛𝐿𝑖 + 𝛼𝑀 𝑙𝑛𝑀𝑖 + 𝜀𝑖 ⁡

Given that this production function is estimated with cross sectional data, the time trend
variable is t=1 and thus the 𝑙𝑛𝐴0 + 𝜂𝑡⁡term becomes the constant term. To get the TFP indicator,
we first compute the returns to scale (RTS) coefficient, which is the sum of factor output
elasticities (𝑅𝑇𝑆 = ⁡ 𝛼𝐾 + 𝛼𝐿 + 𝛼𝑀 ). We then normalize each factor’s output elasticity and obtain
∗
𝛼𝐾∗ = 𝛼𝐾 /𝑅𝑇𝑆, 𝛼𝐿∗ = 𝛼𝐿 /𝑅𝑇𝑆, and 𝛼𝑀
= 𝛼𝑀 /𝑅𝑇𝑆 and define TFP (𝑌5 ) as:

(8)

𝑌5𝑖 =

𝐺𝑉𝑖
∗
𝛼∗𝐾 𝛼∗𝐿 𝛼𝑀
𝐾𝑖 𝐿𝑖 𝑀𝑖

101

3.2.2. Empirical framework and estimation strategy
The standard empirical framework used in this study is the production function approach
implemented by most studies testing the existence of the IR in farm production. In this article,
the production function is estimated using Ordinary Least Squares (OLS) regression as shown
below:
𝒀𝒊 = 𝛽1 + 𝛽2 𝐴𝑖 + 𝒁𝒊 𝜷𝟑 + 𝑿𝒊 𝜷𝟒 + 𝜀𝑖

(9)

where 𝒀𝒊 represents a vector of the five measures of productivity for each household i, 𝐴𝑖 is the
operated farm size at household level, 𝒁𝒊 is a vector of exogenous variables such as household
head characteristics, land and soil conditions and district fixed effects, 𝑿𝒊 denotes a vector of
crop management practices that influence crop production, the 𝛽′𝑠 are the parameter estimates,
and; 𝜀𝑖 is the error term. Table 13 lists and defines all the variables used in the estimation of the
models specified in this study.
However, in order to critically address the study’s underlying objectives, the study
implements an empirical strategy that is different from previous studies on the topic. The
empirical strategy for establishing the relationship between farm size and alternative measures of
productivity in this paper is first econometrically specified in two ways. For each measure of
productivity, we first specify a restricted model where each productivity measure is regressed on
farm size, a quadratic term for farm size where applicable, household head characteristics, land
and soil conditions, and district fixed effects. The quadratic term is not included in two
production functions: net value of crop production per hectare and TFP regressions based on
joint F-tests of the linear and quadratic terms.

102

Table 13: Variable explanation
Name
Variable input costs:
Land rental
Hired animal and machine
use

Fertilizer
Seed
Opportunity cost of family
labor

Imputed hired labor

Additional costs for using fixed
assets

Net value of crop production per
hectare
Net value of crop production per
family labor day

Cost of maize production per
metric ton

Description
Computed by multiplying the number of hectares rented by
the district median land rental rate (2010/11 prices)
Computed for main crop production activities where hired
animals or machines were used. The main crop production
activities include land preparation, planting, fertilizer
application, weeding and harvesting. Multiplied each field
area by 2010/11-district median cost per hectare for each
activity and finally aggregated to household level.
Computed amount of fertilizer used in each field and
multiplied by total price to the farm gate
Computed amount of seed used for each crop in each field
and multiplied that by price to the farm gate
Opportunity cost of family labor computed by multiplying
three variables: (i) the number of prime age (15 – 59 years)
household members working on each farm; (ii) family
labor days allocated to crop production activities farming
activities, and; (iii) district median daily agricultural wage
rate.
Hired labor costs computed by multiplying the district
median daily agricultural wage rate by the number of hired
labor days.
The rental rate for assets with an active rental market was
used to compute costs for fixed assets used in the
production process. For example, if a household owned a
tractor and used it for land preparation, the cost was the
land preparation rental rate per hectare multiplied by the
number of hectares cultivated. If the household used the
tractor multiple times in a year (e.g., for land preparation,
weeding, harvesting, etc.), we valued all the uses of the
tractor to get the additional cost attributable to the tractor.
Examples of other assets accounted for include: animal
traction plows, tractor plows, oxen, planters, sprayers, carts
and shellers.
Value of crop production net of variable input and
additional costs for using fixed assets divided by the
operated farm size (Kwacha/hectare)
Value of crop production net of variable input (excluding
opportunity cost of family labor) and additional costs for
using fixed assets divided by the number of days taken for
crop production activities by family labor (Kwacha/day)
Sum of variable input costs and additional costs for using
fixed assets divided by the number of tons of maize
produced (Kwacha/metric ton)

103

Table 13 (Cont’d)
Name
Total factor productivity indicator

Operated farm size (ha)
Landholding size (ha)
Fallow land (ha)
Basal fertilizer per hectare
Top dressing per hectare
Maize
Beans and oilseeds
Other cereals
Traditional cash crops
Roots and tubers
Family labor days per hectare
Hired labor days per hectare
Manual
Animal
Mechanical
Herbicide
Insecticide
Length of growing period
Elevation
Slope
Rainfall
Soil nutrient availability

Male
Age
Employment

Description
Value of crop production (𝑌𝑖 ) divided by the product of
three factors of production: value of capital (𝐾𝑖 ), total
number of labor days (𝐿𝑖 ) - family and hired labor, and the
total amount of land used for crop production (𝑀𝑖 ). Each
factor was raised to the power of the normalized factor
∗
output elasticities (𝛼𝐾∗ , 𝛼𝐿∗ , 𝛼𝑀
):
𝑌𝑖
= 𝛼 ∗ 𝛼∗ 𝛼∗
𝐾𝑖 𝐾 𝐿𝑖 𝐿 𝑀𝑖 𝑀
Total area planted to crops during the reference year.
Sum of uncultivated land (cleared and not cleared) and
operated farm size during the reference year
Cleared land not planted during the reference year
Amount of compound fertilizer (NPK) applied in kilograms
per hectare planted (kg/ha)
Amount of Nitrogen fertilizer applied in kilograms per
hectare planted (kg/ha)
Share of area planted with maize (%)
Share of area planted with beans and oilseeds (%)
Share of area planted with other cereals (%)
Share of area planted with traditional cash crops (%)
Share of area planted with roots and tubers (%)
Number of family labor days by prime age family members
per hectare (days/ha)
Number of hired labor days per hectare (days/ha)
1 = used manual power
1 = used animal draft power
1 = used mechanical power
1 = used herbicide for weed control
1 = used insecticide for pest control
Number of months when moisture conditions are adequate
for plant growth
Meters above sea level
Measure of steepness (degrees)
Rainfall in millimeters received during the reference year
Soil texture, soil organic carbon, soil pH, total
exchangeable bases classified as: 1=No or slight
constraints; 2=Moderate constraints; 3=severe constraints
1 = male head of household
Age of household head (years)
1 = Household head currently of previously involved in
salaried employment

104

Table 13 (Cont’d)
Name
Household head education

Local
Business
Settlement
Survey
District fixed effects

Description
Level of education completed by the household head
classified as: 1 = No formal education; 2 = basic education
(Grade 1-9); 3 = high school education (Grade 10-12); 4 =
tertiary education (higher education qualification)
1 = Household head considered a local
1 = Household head involved in off-farm business
activities
Number of years the head has lived in current settlement
1=Agricultural Commercialization Survey
Unobserved effects captured by the six district dummy
variables for our survey data

After running each regression for all the measures of productivity that include both the
linear and quadratic terms for farm size, a post estimation F-test (not reported in the tables
section) is also conducted to see if the coefficients on farm size and farm size squared are jointly
equal to zero. In the case of net value of crop production per hectare and TFP, we fail to reject
the hypothesis that including the two terms is significantly different from zero. This implies that
including the quadratic term does not add explanatory power and hence the reason why the term
is dropped in the two aforementioned production functions. For the labor productivity
regressions and cost of maize production per ton, the joint F-test of including both the linear and
quadratic term is statistically significant.
The second specification is a full model that includes variables included in the restricted
model as well as crop management practices. In addition, a dummy variable for farms between
0-5 hectares is included in the full model as well as interaction terms between this dummy and
the various crop management practices and input use variables. This approach is implemented to
relax the assumption of CRS that is common to past empirical studies on the IR.
Because we estimate a restricted and full model for each of the five measures of
productivity, a total of 10 production functions are estimated in this paper. All parameter

105

estimates generated from these production functions are stored for later use during the prediction
exercise described below. Overall sample means are then computed for all exogenous variables
included in the restricted and full model specifications. Moreover, means for crop management
practices are computed grouped by two farm size categories: 0-5 ha and >5 ha. For example, for
total inorganic fertilizer application per hectare, two sample means are computed; one for
observations with farm sizes between zero and five hectares and the other for observations with
farm sizes above five hectares.
We compute predicted values of each measure of productivity for the two separate
models using two approaches. For the restricted specification, each parameter estimate obtained
in the restricted model is multiplied by the overall sample mean of the relevant exogenous
variable. We then sum up all the products together with the constant parameter estimate to obtain
the predicted values. The only variable(s) that vary across observations are farm size and farm
size squared. Please note that when computing predicted values, the farm size squared variable is
only included in the labor productivity and cost of maize production per metric ton prediction
equations. This is because it is only in those particular cases where a joint F-test for including
both the linear and quadratic term during estimation, is significantly different from zero. For the
full specification, the inclusion of the operated farm size dummy and the interaction terms
between this dummy and various management practice and input use variables allows us to have
coefficients that are different between the small- and medium-scale farms. Thereafter, we use
different sub-sample means for crop management variables and multiply them with the different
coefficient estimates obtained for 0-5 hectare versus 5 hectares and above farms. For both the
restricted and full model specifications, only statistically significant right hand side variables are
included in the computation of predicted values of each measure of productivity.

106

Finally, we graphically illustrate the relationship between operated farm size and
alternative predicted values of each measure of productivity described above, and we relax the
ceteris paribus assumption by allowing variation in crop management technologies. We present
two separate line graphs in each figure appropriately labeled as: (i) ceteris paribus graphs
(restricted), and (ii) with variations in crop management (full). In addition, a scatter plot showing
individual observations are generated and superimposed on each of the five figures. The scatter
plots depict the relationship between operated farm size and each measure of productivity.
3.3. Results and discussion
The section begins with an extensive discussion of descriptive results and the
econometric results testing the existence of the IR when different productivity indicators and
data with a wide range of farms are used. The section also discusses potential reasons for
productivity differences across farm sizes when the CRS assumption imposed by previous
studies is relaxed.
3.3.1. Descriptive results
The descriptive results are presented by four landholding size categories: (1) five hectares
(ha) and below (0-5 ha); (2) five hectares to twenty hectares (5-20 ha); (3) twenty hectares to
fifty hectares (20-50 ha), and; (4) fifty hectares to one hundred hectares (50-100 ha). This
categorization is intended to mirror the way the Ministry of Agriculture and Livestock (MAL) in
Zambia categorizes different scales of farming households, although in their case the
categorization is based on area cultivated.
Table 14 presents demographic, land and input use descriptive characteristics
disaggregated by landholding size categories. Results indicate that the average age of household
heads increases with landholding size. While there is no clear pattern with respect to the number
107

of years that the household has lived in the current settlement and how that relates to landholding
size, household heads with landholding size between five and twenty hectares have on average
lived the longest in the current settlement (32.9 years). This finding suggests that household
heads with relatively longer years in current settlement do not necessarily have the largest
landholding size. As expected, the majority of households are male-headed and the pattern
suggests that the proportion of male-headed households increases with landholding size.
Table 14: Demographic, land, and input use characteristics by landholding size categories
5 ha and Above 5
Above
Above
Full
below
to 20 ha 20 to 50
50 to
sample
Variables
ha
100 ha
Means and percentages
Age of household head
45.1
48.5
51.2
53.1
47.3
Household head years in current
24.3
32.9
29.2
31.1
28.2
settlement
Household settled in the current
26.5
13.2
14.7
6.0
19.6
settlement within the past 10 years (%
yes)
Male headed households (% yes)
77.8
91.8
91.4
90.0
84.9
Household head’s education attainment
(% yes)
No formal education
6.3
4.6
1.2
0
4.9
Basic (Grade 1 – 9)
77.4
73.3
64.4
50.0
73.5
Secondary (Grade 10-12)
11.8
18.8
23.3
32.0
16.4
Tertiary
4.5
3.3
11.0
18.0
5.2
Household head previously or currently
23.7
23.6
39.9
34.0
25.9
employed (% yes)
Household head involved in off-farm
36.5
51.6
54.6
60.0
44.9
business activities (% yes)
Household head considered local (% yes)
78.3
73.1
31.9
40.0
69.8
Operated farm size (hectares)
2.0
6.4
10.7
18.7
5.2
Operated farm size (% of total
85.7
60.9
31.2
25.1
45.2
landholding)
Cleared land not cultivated (hectares)
0.2
2.4
10.0
24.4
3.0
Uncleared land (hectares)
0.1
1.6
12.8
29.0
3.1
Total landholding size (hectares)
2.3
10.6
34.2
74.3
11.5
Rented land (% yes)
4.8
3.4
5.5
4.0
4.4
Used basal fertilizer (% yes)
66.3
84.5
95.1
92.0
78.6

108

Table 14 (Cont’d)
Variables

Quantity of basal fertilizer (kg/ha)
Used top dressing fertilizer (% yes)
Quantity of top dressing fertilizer (kg/ha)
Number of crops grown per holding
Number of fields per holding
Used family labor (% yes)
Used hired labor (% yes)
Used manual power (% yes)
Used animal draft power (% yes)
Used mechanical power (% yes)
Used hired animal and machines (% yes)
Used own productive fixed asset (% yes)
Used herbicide (% yes)
Used insecticide (% yes)
Number of observations by landholding
size

5 ha and
below

67.1
68.6
66.7
2.2
2.3
93.8
18.6
32.7
73.4
3.2
46.9
29.9
8.9
17.4
695

Above 5
to 20 ha

Above
Above
20 to 50
50 to
ha
100 ha
Means and percentages
72.2
87.8
90.5
89.4
96.9
90.0
70.3
84.0
84.9
2.7
2.7
3.1
2.5
1.3
1.4
95.8
95.7
97.5
36.9
65.6
74.0
10.9
1.2
0
92.7
88.3
78.0
4.6
18.4
18.0
22.8
28.8
22.0
74.1
71.2
72.0
31.3
63.2
62.0
39.5
38.7
26.0
521
163
50

Full
sample

72.2
80.2
70.6
2.5
2.3
95.1
32.5
20.0
82.3
6.0
35.2
52.2
25.1
28.2
1,429

Source: Authors’ own computation from ACS (2013) and RALS (2012)

Household heads with secondary or tertiary education have comparably larger
landholdings. An important aspect about the changes in farm structure currently being
experienced in parts of SSA is that a significant proportion of medium-scale farmers comprise
landowners who have or have had a salaried job besides farming. The results indeed confirm
this, with more than 30 percent of household heads with landholding size of 20 hectares or
greater indicating that they previously held or currently hold a salaried job. On the other hand,
the results show that less than 25 percent of household heads in the bottom two landholding size
categories (0-5 ha and 5-20 ha) were previously or currently employed in a formal job.
One of the survey questions asked the respondents to state whether they were local or
non-local in their current settlement. A household head is considered local if he or she has family
ties in the area and is most likely to have ethnic ties to people in that settlement. Interestingly, the
average proportion of households who consider themselves local in their current settlement

109

reduces with landholding size. In other words, households with larger landholdings are more
likely to be individuals who are not considered local in the area they have settled. This finding
supports the hypothesis that most medium-scale farms in Zambia are owned by people mostly
based in urban areas or retirees who worked in the civil service (Sitko & Jayne, 2014).
Although operated farm size is increasing with landholding size, results in Table 14
clearly show that the proportion of landholding that is operated reduces with landholding size.
For example, farms with landholding size of 0-5 ha cultivate more than 85 percent of total
landholdings while farms with 50-100 ha cultivate only 25 percent of total land. This finding
suggests that farms with relatively large landholdings are underutilizing the land they own for
crop production while farms with small landholdings are over utilizing the land they own and run
the risk of exhausting soil fertility due to reduced fallows and low use of productivity-enhancing
inputs.
Results for input use characteristics by landholding size categories illustrate some
interesting patterns. First, the proportion of households applying inorganic fertilizers (basal and
top dressing) is increasing with landholding size. Second, households with landholding size of
50-100 ha are likely to apply more inorganic fertilizer (basal and top dressing) than households
with smaller landholding sizes. In general, however, the average application rates for inorganic
fertilizers for households surveyed are lower than the nationally recommended fertilizer
application rates for maize production pegged at 300-400 kg/ha.
Third, although the proportion of households using mechanized power is increasing with
landholding size, results show that the proportion of households using mechanization is low
across all landholding size categories. On the other hand, animal draft power seems to be a
popular power method used by more than 70 percent of farms across all landholding size

110

categories. Fourth, both herbicide and insecticide use are positively correlated with landholding
size. Fifth, use of family labor to conduct activities such as land preparation, planting, fertilizer
application, weeding and harvesting is consistent across all landholding size categories. The
proportion of households employing hired labor is positively related to landholding size, perhaps
not surprisingly. In sum, this study shows that use of productivity-enhancing inputs, improved
technologies, and allocation of hired labor are all positively associated with landholding size.
Agriculture in Zambia is dominated by maize production primarily because of maize
centric policies that have been pursued by successive governments since the country gained
independence in 1964 (Chapoto, Kabaghe, & Zulu-Mbata, 2015). In this study, we analyze the
crop production characteristics of maize and four other crop categories: (1) beans and oilseeds
(groundnuts, soya beans, mixed beans, bambara nuts, cowpeas, velvet beans); (2) other cereals
(sorghum, rice, millet, popcorn); (3) traditional cash crops (seed cotton, tobacco, coffee,
sugarcane), and; (4) roots and tubers (sweet potatoes, potatoes, cassava). As shown in Table 15,
more than 98 percent of households surveyed produce maize and this is true across all
landholding size categories. The results also seem to suggest that larger farms are more likely to
have greater diversity of crops compared to smaller farmers. Despite the crop diversification
response by larger farmers, they still allocate a relatively bigger proportion of their land to maize
production probably because of the policy incentives associated with maize production in
Zambia.

111

Table 15: Crop production characteristics of main crop categories by landholding size categories
5 ha and Above 5
Above
Above
Full
below
to 20 ha 20 to 50
50 to
sample
ha
100 ha
Means and percentages
Households planting each crop
category c (% of sampled households)
Maize
96.1
100.0
100.0
100.0
98.1
Beans and oilseeds
56.7
75.2
73.0
90.0
66.5
Other cereals
2.6
2.9
4.9
12.0
3.3
Traditional cash crops
18.0
39.7
31.3
20.0
27.5
Roots and tubers
26.0
17.7
8.6
10.0
20.4
Households planting maize only (% of
sampled households)
Maize yield (MT/ha)
Land allocation to main crop
categories c (% of area planted)
Maize
Beans and oilseeds
Other cereals
Traditional cash crops
Roots and tubers
Total
Gross value per hectare for households
planting main crop categories c
(ZMW/ha)
Maize
Beans and oilseeds
Other cereals
Traditional cash crops
Roots and tubers
Gross value of crop production per
hectare (ZMW/ha)

26.3

11.9

15.3

6.0

19.1

2.0

2.3

2.7

3.1

2.2

70.9
13.3
0.9
10.9
4.0
100.0

71.2
13.4
0.7
12.4
2.5
100.0

73.5
12.7
0.6
11.5
1.7
100.0

75.6
12.2
0.6
10.4
1.2
100.0

71.5
13.3
0.8
11.5
3.0
100.0

2,184.0
1,594.0
1,067.0
2,458.0
2,785.0

2,551.0
1,145.0
964.0
2,505.0
1,619.0

3,367.0
1,521.0
563.0
983.0
879.0

4,123.0
1,273.0
1,142.0
1,698.0
1,564.0

2,527.0
1,384.0
958.0
2,402.0
2,306.0

2,074.0

2,167.0

2,495.0

2,596.0

2,174.0

112

Table 15 (Cont’d)

Contribution of main crop categories c
(% of gross crop value)
Maize
Beans and oilseeds
Other cereals
Traditional cash crops
Roots and tubers
Total

5 ha and
below

Above 5
to 20 ha

Above
Above
20 to 50
50 to
ha
100 ha
Means and percentages

74.6
8.8
0.3
11.6
4.8
100.0

79.0
7.1
0.2
11.6
2.1
100.0

82.6
6.4
0.2
9.4
1.4
100.0

Full
sample

85.2
6.0
0.2
7.7
0.9
100.0

77.5
7.8
0.2
11.2
3.3
100.0

Number of observations by landholding
695
521
163
50
size
Source: Authors’ own computation from ACS (2013) and RALS (2012) survey data

1,429

Notes: c Crop category/categories: 1 = Maize; 2=beans and oilseeds (groundnuts, soya beans,
mixed beans, Bambara nuts, cowpeas and velvet beans); 3=other cereal grains (sorghum, rice,
millet, popcorn); 4= traditional cash crops (seed cotton, tobacco, coffee and sugarcane); 5=roots
and tubers (sweet potatoes, potatoes and cassava).
Table 15 also reports the gross value of crop production per hectare. Two main patterns
emerge with respect to gross value per hectare of main crop categories when analyzed across
landholding size categories. First, there is a positive relationship between gross value of maize
production per hectare and landholding size such that farms in the 50-100 ha category nearly
generate double the value than farms in the 0-5 ha category. Second, the gross value per hectare
for other crop categories does not follow a similar pattern as maize. In general, farms categorized
in the 0-5 ha group have a greater gross value per hectare for beans and oilseeds, traditional cash
crops and roots and tubers than those in the 50-100 ha group. However, when one critically
analyzes the changes in gross value per hectare across the landholding size continuum, the
results suggest a U-shaped or inverse U-shaped relationship between gross value per hectare and
landholding size.

113

Table 16 reports crop production costs and the five indicators of productivity by
landholding size categories. In our study, we account for seven individual cost items that build
up the cost of crop production across farms in Zambia. The cost items include land rental, hired
animal and machine use, hired labor, seed, fertilizer, family labor and additional costs for using
fixed assets. The cost associated with family labor must be considered as an upper bound
estimate and therefore need to be interpreted with caution. As noted by Burke et al. (2011), the
wage assigned to family labor is likely to be well above the economic cost of that labor. Also,
households tend to overestimate the amount of family labor spent on a given farming activity.
Total production cost per hectare planted increases across landholding size categories, which
provides the first indication that small farms are, on average, cost efficient relative to large farms
in Zambia.
The summary results of the five indicators of productivity reveal some interesting
patterns. First, net value of crop production per hectare and the total factor productivity average
values suggest a negative relationship between each measure of productivity and landholding
size. Second, the labor productivity measures (net value of production per family labor day and
net value of production per labor) results show a monotonic positive relationship across
landholding size categories. This implies that large farms are more productive than small farms
when the measures of productivity under consideration are family labor or total farm labor
productivity. Third, cost of maize production per metric ton increases across landholding size
categories. Put another way, smaller farmers are more cost efficient compared to larger farmers
when it comes to maize production.

114

Table 16: Crop production costs and productivity measures by landholding size categories
5 ha and Above 5 Above
Above
Full
below
to 20 ha 20 to 50
50 to
sample
ha
100 ha
Means
Cost of production per hectare – all
crops (ZMW/ha)
Land rental
Only HH with non zero values
221.0
154.0
223.0
151.0
200.0
All HH
10.5
5.3
12.3
6.1
8.7
Hired animal and machine use
Only HH with non zero values
97.0
91.0
90.0
60.0
94.0
All HH
45.7
20.8
25.9
13.2
33.2
Hired labor
Only HH with non zero values
473.0
573.0
658.0
609.0
572.0
All HH
103.0
332.0
610.0
585.0
261.0
Seed
Only HH with non zero values
206.0
168.0
173.0
180.0
187.0
All HH
203.0
168.0
173.0
180.0
186.0
Fertilizer
Only HH with non zero values
542.0
515.0
513.0
527.0
526.0
All HH
365.0
459.0
497.0
497.0
419.0
Family labor
Only HH with non zero values
287.0
356.0
344.0
355.0
323.0
All HH
257.0
347.0
342.0
340.0
302.0
Additional costs for using fixed assets
Only HH with non zero values
219.0
223.0
220.0
217.0
221.0
All HH
65.7
166.0
157.0
156.0
116.0
Total production costs (All HH)
1,050.0 1,497.0 1,817.0
1,778.0
1,326.0
Productivity measures
Net value of production per hectare
1,025.0
670.0
678.0
818.0
(ZMW/ha)
Net value of production per family labor
37.8
100.0
239.0
191.0
day (ZMW/day)
Net value of production per labor day
34.6
89.3
212.0
212.0
(ZMW/day)*
Cost of production per metric ton of
1,188.0 2,069.0 3,027.0
3,296.0
maize (ZMW/MT)
Total factor productivity indicator
270.0
249.0
254.0
255.0
Number of observations by landholding
695
521
163
50
size
Source: Author’s own computation from ACS (2013) and RALS (2012) survey data
Notes: * This includes both family and hired labor days

115

848.0
91.3
83.2
1,804.0
260.0
1,429

Figure 3 illustrates the relationship between factor input ratios and landholding size using
four factor input ratios, specifically the capital-labor, capital-land, labor-land, and labor-capital
ratios. Capital is the sum of all costs of production excluding imputed family labor costs, and is
presented in terms of Zambian Kwacha (ZMW); labor is the number of prime age members,
between 15 and 59 years, living and participating in farming activities in a given household, and;
land is the operated farm size. Both capital-labor and capital-land ratios exhibit a positive
relationship while the labor-land and labor-capital ratios have a negative relationship when
graphed against landholding size. Two main observations are made from these graphs. First,
farms become more capital-intensive as landholding size increases. Second, farms become more
land-using and labor-saving as landholding size increases, but particularly in the range of 1-10
hectares—there is a lot of substitution of factor inputs within this range, but not so much after 10
hectares.

Figure 3: Bivariate relationships between factor input ratios and landholding size

116

The box plots in Figure 4 display boxes bordered at the 25th and 75th percentiles of the yvariable with a median line at the 50th percentile. Whiskers extend from the box to the upper and
lower adjacent values and are capped with an adjacent line. The box plots clearly show that for
each measure of productivity, there are variations in the level of productivity even within a given
landholding size category. Consider the box plots for the net value of crop production per hectare
depicted in the top left hand corner of Figure 4. For farms with landholding size less than five
hectares, the range of the level of productivity between the 25th and 75th percentile is
approximately ZMW 200 to ZMW 1600 with a median value of about ZMW 700. Similarly, the
level of productivity for farms with landholding size above 50 hectares has high variation
ranging between ZMW -600 and ZMW 1400 for the 25th and 75th percentiles respectively and a
median of ZMW 450.
This trend for within-category differences is true across all farm size categories and
measures of productivity. In other words, farms within a defined category do not necessarily
have similar levels of productivity. This suggests that differences in productivity are not
necessarily driven by farm size but that there are other factors that play an important role other
than the scale of farm.

117

Figure 4: Box plots of measures of productivity by landholding size categories
Note: Box plots bordered at the 25th and 75th percentiles of the y-variable with a median line at
the 50th percentile. Whiskers extend from the box to the upper and lower adjacent values and are
capped with an adjacent line
3.3.2. Testing the IR with alternative productivity measures over a wide range of farms
Ultimately, the relationships between operated farm size (𝐴𝑖 ) and the five outlined
measures of productivity, conditional on control variables that potentially explain productivity
differences, are of interest to this study. The sign on coefficient estimates will be interpreted
differently for the various measures of productivity. For four measures of productivity—net
value of crop production per hectare planted, net value of crop production per family labor day,
net value of crop production per farm labor day, and TFP—a statistically significant negative
coefficient on the operated farm size variable upholds the IR hypothesis, while a positive
coefficient suggests otherwise. However, the interpretation of the coefficient on operated farm
size when the measure of productivity is cost of maize production per metric ton is the opposite

118

of the other four measures. With this measure of productivity, the IR is upheld when the
coefficient on the operated farm size variable is positive, while a negative coefficient would
imply otherwise.
Tables 17 to 20 present estimates of the OLS regression analysis exploring the operated
farm size-productivity relationship using five measures of productivity. We first discuss the
estimates for net value of crop production per hectare planted presented in Table 17. Results for
the net value of crop production per hectare for both the restricted and full models uphold the IR
hypothesis. However, the coefficient on the farm size variable is only statistically significant in
the full model specification. For a one-hectare increase in farm size, the value of crop production
per hectare reduces by ZMW 6. This result is significant at 10% confidence level.
According to the restricted specification, factors associated with differences in net value
of crop production per hectare could be summarized as follows. Male-headed households are
more likely to have net value of crop production per hectare that is ZMW 190 higher than
female-headed households. An increase by one year of living in a given settlement is negatively
associated with productivity. Household heads with higher education (tertiary education) have
significantly higher levels of productivity relative to those with only 7 years or less of primary
education. Surprisingly, household heads’ currently or previously employed have a lower net
value of crop production per hectare, relative to those with no formal job experience.

119

Table 17: Parameter estimates (OLS) of the relationship between operated farm size and net
value of crop production per hectare
Explanatory variables
Dependent variable
Net value of crop
production per hectare
(ZMW/ha)
Restricted
Full
Operated farm size (ha)
-5.26
-16.16*
(-0.52)
(-1.70)
*
Male head (1=yes)
190.57
219.84*
(1.86)
(1.93)
Age of household head
-3.70
0.44
(-1.30)
(0.15)
Years in current settlement
-6.17**
-3.92*
(-2.48)
(-1.67)
Education of household head (base category: primary education)
No formal education (1=yes)
45.60
99.38
(0.30)
(0.73)
Secondary education (1=yes)
117.55
60.38
(1.19)
(0.62)
Tertiary education (1=yes)
669.36***
452.45**
(2.97)
(2.02)
***
Formal employment (1=yes)
-362.51
-251.40***
(-4.08)
(-2.80)
Local (1=yes)
355.77***
285.92***
(4.02)
(3.23)
Length of growing period (months)
716.73***
645.85**
(2.86)
(2.57)
Elevation (meters)
-1.06**
-1.22**
(-2.14)
(-2.53)
Slope (degrees)
28.48
31.80
(0.74)
(0.92)
Rainfall (millimeters)
6.12***
6.18***
(3.68)
(3.57)
Soil nutrient availability (base category: no or slight constraints)
Moderate constraints (1=yes)
-73.61
-109.70
(-0.72)
(-1.12)
Severe constraints (1=yes)
-244.54**
-237.22**
(-2.11)
(-2.10)

120

Table 17 (Cont’d)
Explanatory variables

District fixed effects
Operated farm size dummy (1=farms with 5ha and less)
Total inorganic fertilizer (kg/ha)
Share of area planted with:
Maize area (%)
Beans and oilseeds area (%)
Other cereals area (%)
Traditional cash crops area (%)
Roots and tubers area (%)
Family labor days per hectare (days/ha)
Hired labor days per hectare (days/ha)
Manual power (1=yes)
Animal draft power (1=yes)
Mechanical power (1=yes)
Herbicide use (1=yes)
Insecticide use (1=yes)
Total fertilizer (kg/ha) X operated farm size dummy
Maize area (%) X operated farm size dummy
Beans and oil seeds area (%) X operated farm size dummy
Other cereals area (%) X operated farm size dummy
Traditional cash crops area (%) X operated farm size dummy

121

Dependent variable
Net value of crop
production per hectare
(ZMW/ha)
Restricted
Full
Yes
Yes
-1552.36*
(-1.80)
4.12***
(3.80)
557.57
(1.25)
1646.83***
(2.79)
587.67
(0.67)
1120.95**
(2.29)
-202.46
(-0.17)
-2.47
(-1.31)
-2.31
(-0.36)
-70.64
(-0.31)
-913.59**
(-2.02)
-68.41
(-0.20)
-194.31
(-1.48)
342.72**
(2.17)
-1.95*
(-1.70)
919.11
(1.34)
572.34
(0.66)
251.35
(0.24)
1620.83**
(2.18)

Table 17 (Cont’d)
Explanatory variables

Roots and tubers area (%) X operated farm size dummy
Family labor days per hectare (days/ha) X operated farm size
dummy

Dependent variable
Net value of crop
production per hectare
(ZMW/ha)
Restricted
Full
4027.48***
(2.85)
-0.24

Hired labor days per hectare (days/ha) X operated farm size dummy
Manual power (1=yes) X operated farm size dummy
Animal draft power (1=yes) X operated farm size dummy
Mechanical power (1=yes) X operated farm size dummy
Herbicide use (1=yes) X operated farm size dummy
Insecticide use (1=yes) X operated farm size dummy
-7549.90***
(-2.81)
1479
0.15
0.14

Constant

(-0.11)
-3.17
(-0.36)
163.79
(0.58)
768.35
(1.57)
-50.36
(-0.12)
528.71***
(2.61)
-161.99
(-0.70)
-7285.18**
(-2.54)
1426
0.27
0.24

N
R2
Adj. R2
Source: Author’s computation from RALS (2012) and ACS (2013)
Note: t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
Length of growing period variable shows a positive association with our land
productivity measure. We also find evidence that when farms are located on soils with somewhat
severe constraints in terms of soil nutrient availability (e.g., soil texture, soil organic carbon, soil
pH, total exchangeable bases), this is negatively associated with net value of crop production per
hectare.
According to the full model, the results show that the same variables that are significant
in the vector of exogenous variables in the restricted specification are also important factors in

122

the full specification. For the vector of crop management and input use variables (only included
in the full specification), the notable findings are as follows. A one-unit increase in total
inorganic fertilizer applied per hectare is associated with an increase of ZMW 4 in net value of
crop production per hectare and increasing the share of land planted to beans and oilseeds as well
as traditional cash crops has a positive effect on net value of crop production per hectare.
Households that use animal draft power as a land preparation technique are less productive
relative to those not using the technique while application of insecticide has a positive effect on
productivity.
The estimates for the two measures of labor productivity are presented in Table 18. The
results show that there is a positive and significant relationship between operated farm size and
our labor productivity measures in all the specifications. These findings fail to uphold the IR
hypothesis. For net value of crop production per family labor day, the estimated coefficients
indicate that a one hectare increase in operated farm size increases labor productivity by ZMW
23 and ZMW 30 in the restricted and full model specifications respectively. For net value of crop
production per labor day (family and hired labor), an increase of one hectare of operated farm
size leads to a corresponding increase in labor productivity of ZMW 37 and ZMW 55 in the
restricted and full model specifications respectively. Our findings imply that the returns to labor
are higher on medium-scale farms relative to small-scale farms.

123

Table 18: Parameter estimates (OLS) of the relationship between operated farm size and net
value of crop production per family labor/total labor day
Explanatory variables
Dependent variable
Net value of crop
Net value of crop
production per family
production per labor
labor day (ZMW/day)
day (family and hired)
(ZMW/day)
Restricted
Full
Restricted
Full
Operated farm size (ha)
22.71***
30.32***
36.54***
55.48**
(4.00)
(3.87)
(2.88)
(1.97)
Operated farm size squared
-0.13
-0.28
-0.35
-0.73
(-0.51)
(-1.08)
(-1.41)
(-1.61)
Male head (1=yes)
12.23
21.05
-3.98
21.48
(0.83)
(1.42)
(-0.21)
(0.79)
Age of household head
-0.26
0.21
-0.40
0.19
(-0.50)
(0.46)
(-0.75)
(0.35)
Years in current settlement
-0.62
-0.36
-0.58
0.05
(-1.04)
(-0.79)
(-1.17)
(0.09)
Education of household head (base
category: primary education)
No formal education
27.82
14.56
54.40*
45.97
(1.57)
(0.82)
(1.72)
(1.32)
Secondary education
75.51**
51.14*
52.68*
52.83*
(2.08)
(1.72)
(1.69)
(1.94)
Tertiary education
140.42*
74.22
431.46
376.09
(1.65)
(1.15)
(1.63)
(1.39)
Formal employment (1=yes)
-7.59
-5.20
9.70
15.90
(-0.34)
(-0.27)
(0.41)
(0.46)
Local (1=yes)
18.52
16.81
12.69
-7.26
(0.83)
(0.87)
(0.58)
(-0.24)
Length of growing period (months)
157.01***
123.22***
138.13***
83.69**
(3.92)
(3.44)
(2.76)
(2.11)
Elevation (meters)
-0.18
-0.16
-0.18
-0.13
(-1.42)
(-1.37)
(-1.54)
(-1.18)
Slope (degrees)
18.62**
8.04
14.57*
4.78
(2.31)
(1.31)
(1.90)
(0.70)
Rainfall (millimeters)
0.56**
0.46**
0.12
0.16
(2.54)
(2.33)
(0.46)
(0.66)
Soil nutrient availability (base
category: no or slight constraints)
Moderate constraints (1=yes)
-29.11
-32.23
-79.95
-123.41
(-0.78)
(-1.07)
(-0.89)
(-1.16)
Severe constraints (1=yes)
-39.92
-31.89
-82.65
-104.27
(-1.19)
(-1.15)
(-1.13)
(-1.22)
District fixed effects
Yes
Yes
Yes
Yes
Operated farm size dummy (1=farms
1.57
-462.98
with 5ha and less)
(0.01)
(-1.06)

124

Table 18 (Cont’d)
Explanatory variables

Total inorganic fertilizer (kg/ha)

Dependent variable
Net value of crop
Net value of crop
production per family
production per labor
labor day (ZMW/day)
day (family and hired)
(ZMW/day)
Restricted
Full
Restricted
Full
**
0.74
1.59
(2.14)
(1.50)

Share of area planted with:
Maize area (%)
Beans and oilseeds area (%)
Other cereals area (%)
Traditional cash crops area
(%)
Roots and tubers area (%)
Family labor days per hectare
(days/ha)
Hired labor days per hectare
(days/ha)
Manual power (1=yes)
Animal draft power (1=yes)
Mechanical power (1=yes)
Herbicide use (1=yes)
Insecticide use (1=yes)
Total fertilizer (kg/ha) X operated
farm size dummy
Maize area (%) X operated farm size
dummy
Beans and oil seeds area (%) X
operated farm size dummy

125

256.22**
(2.02)
77.23
(0.40)
641.26*
(1.65)
142.82

464.54*
(1.66)
865.04
(1.20)
133.93
(0.25)
558.71

(1.22)
-157.90
(-0.65)
-2.42***

(1.23)
40.47
(0.10)
-1.24***

(-4.15)
-1.25

(-3.58)
-6.44

(-0.44)
79.39
(0.68)
-57.01
(-0.26)
154.78
(0.78)
-21.64
(-0.61)
85.65
(1.23)
-0.66*

(-0.93)
-79.18
(-0.50)
-1136.45
(-1.25)
-613.51
(-0.93)
46.21
(0.74)
5.48
(0.09)
-1.54

(-1.94)
-163.28

(-1.46)
-367.25

(-1.24)
26.49

(-1.30)
-716.24

(0.13)

(-1.04)

Table 18 (Cont’d)
Explanatory variables

Other cereals area (%) X operated
farm size dummy

Dependent variable
Net value of crop
Net value of crop
production per family
production per labor
labor day (ZMW/day)
day (family and hired)
(ZMW/day)
Restricted
Full
Restricted
Full
*
-654.48
-120.38

Traditional cash crops area (%) X
operated farm size dummy
Roots and tubers area (%) X operated
farm size dummy
Family labor days per hectare
(days/ha) X operated farm size
dummy
Hired labor days per hectare
(days/ha) X operated farm size
dummy
Manual power (1=yes) X operated
farm size dummy
Animal draft power (1=yes) X
operated farm size dummy
Mechanical power (1=yes) X
operated farm size dummy
Herbicide use (1=yes) X operated
farm size dummy
Insecticide use (1=yes) X operated
farm size dummy

(-1.65)
45.83

(-0.22)
-371.34

(0.34)
263.83

(-0.90)
54.95

(1.06)
1.88***

(0.13)
0.71*

(3.28)
1.89

(1.70)
5.14

(0.68)
-93.44

(0.94)
71.24

(-0.81)
64.68

(0.44)
1146.15

(0.30)
-156.69

(1.26)
578.18

(-0.79)
0.93

(0.92)
-115.57

(0.02)
-107.60

(-1.05)
-22.94

(-1.51)
(-0.39)
***
***
Constant
-1355.51
-1148.49
-883.46
-216.06
(-4.16)
(-3.19)
(-3.91)
(-0.46)
Inflection point (ha)
87.35
55.04
52.2
38.45
N
1479
1426
1479
1426
R2
0.15
0.26
0.09
0.15
Adj. R2
0.14
0.24
0.08
0.12
Source: Author’s computation from RALS (2012) and ACS (2013)
Note: t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
***

126

We test the null hypothesis that including both the linear and quadratic term for farm size
is not significantly different from zero. The null hypothesis is rejected and thus we conclude that
the positive relationship between operated farm size and labor productivity is only true for farms
cultivating in the 0-40 hectares domain depicted by the points of inflection reported in Table 18.
To demonstrate the effects of both exogenous (demographic and farm characteristics) and
endogenous (crop management and input use) variables, we focus mainly on the net value of
crop production per family labor day. According to the restricted specification for net value of
crop production per family labor day, education is an important explanatory variable as
evidenced by the positive and significant coefficients on the variables “secondary education” and
“tertiary education”. Length of growing period and rainfall received are also positively
associated with labor productivity. According to the full model, the results show that the same
variables that are significant in the vector of exogenous variables in the restricted specification
are also important factors in the full specification. For the vector of crop management and input
use variables, we find that the returns to family labor are positively associated with total
inorganic fertilizer applied per hectare, share of land planted to maize and share of land planted
to other cereals but negatively associated with family labor days per hectare.
Results for the relationship between cost of maize production per metric ton and maize
area planted, ceteris paribus, uphold the IR hypothesis (see Table 19). However, the coefficient
on the maize area planted variable is only statistically significant in the restricted model
specification.

127

Table 19: Parameter estimates (OLS) of the relationship between operated farm size and cost of
maize production per metric ton
Explanatory variables
Dependent variable
Cost of maize production per
metric ton (ZMW/MT)
Restricted
Full
Maize area planted (ha)
208.77***
226.45
(3.92)
(1.34)
Maize area only squared
-10.16***
-10.99**
(-3.80)
(-2.37)
Male head (1=yes)
-62.63
-143.74
(-0.33)
(-0.77)
Age of household head
-7.07
-8.49
(-1.25)
(-1.64)
Years in current settlement
20.27***
16.24***
(4.39)
(3.84)
Education of household head (base category: primary
education)
No formal education
38.79
5.74
(0.10)
(0.02)
Secondary education
-167.91
-241.45
(-0.63)
(-0.78)
***
Tertiary education
-879.22
-692.15***
(-3.15)
(-2.61)
Formal employment (1=yes)
474.19*
366.88
(1.93)
(1.41)
Local (1=yes)
-990.97***
-862.70**
(-2.60)
(-2.45)
Length of growing period (months)
-431.30**
-335.98
(-2.16)
(-1.39)
Elevation (meters)
1.15
2.05
(0.53)
(0.72)
Slope (degrees)
59.65
86.77
(0.55)
(0.61)
Rainfall (millimeters)
-5.23
-3.91
(-1.35)
(-0.85)
Soil nutrient availability (base category: no or slight
constraints)
Moderate constraints (1=yes)
190.84
205.88
(0.97)
(0.85)
*
Severe constraints (1=yes)
652.93
664.01*
(1.80)
(1.79)
District fixed effects
Yes
Yes
Operated farm size dummy (1=farms with 5ha and less)
2408.95
(1.01)

128

Table 19 (Cont’d)
Explanatory variables

Dependent variable
Cost of maize production per
metric ton (ZMW/MT)
Restricted
Full
-4.52***
(-2.67)

Total inorganic fertilizer (kg/ha)
Share of area planted with:
Maize area (%)
Beans and oilseeds area (%)
Other cereals area (%)
Traditional cash crops area (%)
Roots and tubers area (%)
Family labor days per hectare (days/ha)
Hired labor days per hectare (days/ha)
Manual power (1=yes)
Animal draft power (1=yes)
Mechanical power (1=yes)
Herbicide use (1=yes)
Insecticide use (1=yes)
Total inorganic fertilizer (kg/ha) X operated farm size dummy
Maize area (%) X operated farm size dummy
Beans and oil seeds area (%) X operated farm size dummy
Other cereals area (%) X operated farm size dummy
Traditional cash crops area (%) X operated farm size dummy
Roots and tubers area (%) X operated farm size dummy

129

548.98
(0.21)
-2412.33*
(-1.92)
7596.00
(1.11)
4715.75
(0.90)
5415.98
(1.53)
4.78
(1.11)
39.85
(0.95)
-1066.78
(-1.55)
814.91**
(2.12)
-603.06
(-0.74)
737.56
(0.84)
-904.52*
(-1.83)
2.42
(1.39)
-658.18
(-0.26)
1535.09
(0.84)
-7764.06
(-1.13)
-5101.43
(-1.00)
-6199.86*
(-1.66)

Table 19 (Cont’d)
Explanatory variables

Family labor days per hectare (days/ha) X operated farm size
dummy

Dependent variable
Cost of maize production per
metric ton (ZMW/MT)
Restricted
Full
-3.81
(-0.86)
-24.55

Hired labor days per hectare (days/ha) X operated farm size
dummy
Manual power (1=yes) X operated farm size dummy
Animal draft power (1=yes) X operated farm size dummy
Mechanical power (1=yes) X operated farm size dummy
Herbicide use (1=yes) X operated farm size dummy
Insecticide use (1=yes) X operated farm size dummy
Constant

7028.55
(1.64)
10.27
1407
0.04
0.03

(-0.63)
311.16
(0.42)
-1183.43**
(-2.13)
896.63
(0.80)
-784.06
(-0.95)
604.47
(0.99)
2972.63
(0.43)
9.83
1383
0.08
0.04

Inflection point (ha)
N
R2
Adj. R2
Source: Author’s computation from RALS (2012) and ACS (2013)
Note: t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
For a one-unit increase in maize area planted, the cost of maize production per metric ton
increases by over ZMW 209. This result is significant at 1% confidence level. When we test the
null hypothesis that including both the linear and quadratic term for maize area planted is not
significantly different from zero, we reject the null hypothesis but we conclude that the positive
relationship between maize area planted and cost of maize production per metric ton is true for
farms cultivating maize between 0 and 10 hectares (points of inflection reported in Table 19).
Beyond this range, the cost of maize production begins to drop.

130

According to the restricted specification, factors that significantly increase the cost of
maize production per metric ton include the household head’s number of years in current
settlement, whether the head had a formal job, and whether the farm is located on severely
constrained soils. On the other hand, the household head attaining tertiary education, whether the
head is local, and length of the growing period are all associated with reduction in the cost of
maize production per metric ton. For the full model, the results show that the same variables that
are significant in the vector of exogenous variables in the restricted specification are also
important factors in the full specification. The full model further shows that quantity of total
inorganic fertilizer, share of beans and oilseeds area, and whether the farmer used insecticides
are variables that are negatively associated with cost of maize production per metric ton
produced. On the other hand, share of roots and tubers and whether the household used animal
draft power are variables that are positively associated with cost of maize production per metric
ton.
To fully capture the comprehensive use of different inputs during the crop production
process, we estimate the relationship between operated farm size and total factor productivity
(Table 20). The results for both the restricted and full model specifications show a negative but
insignificant relationship between operated farm size and TFP.8 These results fail to uphold the
IR and suggest that, when the measure of productivity is TFP, there is no adequate evidence to
support the argument that small farms are more productive than large farms. This finding is
similar to that observed in China by Li et al. (2013) where they found that while the relationship
between their land size measurement and TFP was negative, it was not statistically negative.

8

Cobb-Douglas production function estimates used to compute TFP at household level are in
Table 22.
131

Table 20: Parameter estimates (OLS) of the relationship between operated farm size and total
factor productivity
Explanatory variables
Dependent variable
Total factor productivity
Restricted
Full
Operated farm size (ha)
-0.58
-1.49
(-0.80)
(-1.55)
Male head (1=yes)
39.60***
28.86**
(3.25)
(2.21)
*
Age of household head
-0.63
-0.26
(-1.86)
(-0.78)
Years in current settlement
-0.35
-0.20
(-1.20)
(-0.74)
Education of household head (base category: primary education)
No formal education
2.41
14.89
(0.13)
(0.88)
Secondary education
14.32
1.86
(1.28)
(0.17)
Tertiary education
72.64***
41.37*
(3.14)
(1.82)
***
Formal employment (1=yes)
-38.96
-27.39***
(-3.79)
(-2.76)
***
Local (1=yes)
35.31
28.27***
(3.48)
(2.81)
Length of growing period (months)
70.89**
63.19**
(2.56)
(2.44)
Elevation (meters)
0.00
-0.09
(0.01)
(-1.50)
Slope (degrees)
-0.28
3.92
(-0.06)
(0.99)
Rainfall (millimeters)
0.52***
0.59***
(2.80)
(3.32)
Soil nutrient availability (base category: no or slight constraints)
Moderate constraints (1=yes)
-14.23
-17.62
(-1.21)
(-1.53)
Severe constraints (1=yes)
-43.20***
-39.74***
(-3.19)
(-2.98)
District fixed effects
Yes
Yes

132

Table 20 (Cont’d)
Explanatory variables

Operated farm size dummy (1=farms with 5ha and less)
Total inorganic fertilizer (kg/ha)
Share of area planted with:
Maize area (%)
Beans and oilseeds area (%)
Other cereals area (%)
Traditional cash crops area (%)
Roots and tubers area (%)
Family labor days per hectare (days/ha)
Hired labor days per hectare (days/ha)
Manual power (1=yes)
Animal draft power (1=yes)
Mechanical power (1=yes)
Herbicide use (1=yes)
Insecticide use (1=yes)
Total inorganic fertilizer (kg/ha) X operated farm size dummy
Maize area (%) X operated farm size dummy
Beans and oil seeds area (%) X operated farm size dummy
Other cereals area (%) X operated farm size dummy
Traditional cash crops area (%) X operated farm size dummy
Roots and tubers area (%) X operated farm size dummy

133

Dependent variable
Total factor productivity
Restricted
Full
-165.26*
(-1.81)
0.55***
(5.70)
142.71***
(3.06)
177.99***
(2.80)
18.29
(0.19)
157.48***
(2.73)
11.57
(0.09)
0.06
(0.34)
-0.23
(-0.36)
-8.18
(-0.30)
-86.82*
(-1.82)
-8.16
(-0.22)
-19.56
(-1.39)
37.39**
(2.28)
-0.20**
(-1.99)
100.25
(1.40)
119.21
(1.30)
129.90
(1.08)
211.71**
(2.48)
410.66**
(2.57)

Table 20 (Cont’d)
Explanatory variables

Family labor days per hectare (days/ha) X operated farm size
dummy

Dependent variable
Total factor productivity
Restricted
Full
-0.01

Hired labor days per hectare (days/ha) X operated farm size dummy
Manual power (1=yes) X operated farm size dummy
Animal draft power (1=yes) X operated farm size dummy
Mechanical power (1=yes) X operated farm size dummy
Herbicide use (1=yes) X operated farm size dummy
Insecticide use (1=yes) X operated farm size dummy
-636.55**
(-2.14)
1479
0.13
0.12

Constant

(-0.03)
-0.28
(-0.31)
17.56
(0.51)
59.79
(1.13)
-7.56
(-0.16)
67.35***
(3.00)
-18.91
(-0.74)
-682.09**
(-2.38)
1426
0.28
0.25

N
R2
Adj. R2
Source: Author’s computation from RALS (2012) and ACS (2013)
Note: t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
Without going through the details of similarities or differences between the restricted and
full specifications, the pattern is similar to that observed for our first measure of productivity (net
value of crop production per hectare), in particular the signs and levels of significance of the
coefficients on the explanatory variables. Table 21 provides a summary of the key findings from
this study; whether or not the IR is upheld for each measure of productivity.

134

Table 21: Summary of key findings for the relationship between each measure of productivity
and operated farm size
Measure of productivity
Is the IR upheld?
Level of statistical
significance
Restricted Full
Net value of crop production per hectare
Yes
Yes
Only significant in full
specification at 10%
confidence level
Net value of crop production per labor
No
No
Significantly positive farm
day (family labor only)
size—productivity
relationship in both
specifications at 1%
confidence level
Net value of crop production per labor
No
No
Significantly positive farm
day (family and hired labor)
size—productivity
relationship in both
specifications at 1%
confidence level
Cost of maize production per metric ton
Yes
Yes
Only significant in restricted
specification at 1%
confidence level
Total factor productivity
Yes
Yes
Not significant in both
specifications
3.3.3. Understanding productivity differences across farm sizes
A key aspect of our study is to understand why there are productivity differences between
small- (0-5 ha) and medium-scale farms (5-100 ha). To implement this, it is vital to relax the
assumption of CRS that is imposed in a number of studies that have hypothesized the IR. As
already discussed in the methods section, we relax the assumption of CRS by including
interaction terms between a dummy variable for operated farm size (1 = farms with 5 ha or less;
0 = otherwise) and crop management practices/input use variables. The interaction terms test the
following hypothesis: the relationship between crop management practices/input use and each
measure of productivity is different between small-scale producers and medium-scale producers
in our data.
Results in Table 17 show that the interaction term between total inorganic fertilizer per
hectare and operated farm size dummy is negatively associated (-1.95) with net value of crop
135

production per hectare. This suggests that small-scale farms are less efficient relative to
mediums-scale farms in terms of inorganic fertilizer application. However, our results show that
the interaction term between herbicide use dummy and operated farm size dummy is positively
associated with net value of crop production per hectare and is statistically significant at 1%
confidence level. Applying herbicides for weed control increases net value of crop production
per hectare more for small-scale farms than medium-scale farms by ZMW 529. These results
suggest that when small-scale farmers use a labor saving technology like herbicides, the increase
in productivity is relatively higher probably because they are now able to complete weeding
tasks faster than medium-scale farms.
The other source of difference in net value of crop production per hectare is explained by
the allocation of land to traditional cash crops (e.g., cotton) and roots and tubers (e.g., sweet
potatoes). The results show that increasing the allocation of land to traditional cash crops and
roots and tubers increases net value of crop production per hectare more for small-scale farms
than for medium-scale farms by ZMW 1,620 and ZMW 4,027, respectively.
What could explain differences in net value of crop production per family labor day
between small- (0-5 ha) and medium-scale (5-100 ha) farms? Based on the estimates on
interaction terms included in our full model specification (Table 18), we find that differences in
labor productivity between the two categories of farms are driven by differences in total
inorganic fertilizer use but only by a small magnitude as demonstrated by the coefficient (-0.66).
However, other inputs like herbicides and insecticides do not drive labor productivity differences
between small- and medium-scale farms (estimates are not statistically significant). Allocating
additional land to other cereals (e.g., sorghum) reduces the net value of crop production per
family labor day by ZMW 654 for small-scale farms relative to the medium-scale farms. On the

136

other hand, allocating additional family labor per hectare is more beneficial to small- than
medium-scale farms (approximately ZMW 2 per labor day increase in net value of crop
production per family labor day).
For cost of maize production per metric ton (Table 19), we find that differences in the
cost of maize production per metric ton between the two categories of farms are not driven by
differences in fertilizer use or other inputs like herbicides and insecticides (estimates are not
statistically significant). We only find that use of animal draft power is more beneficial to smallthan medium-scale farms (approximately ZMW 1,300 reduction in the cost of maize production
per metric ton).
We also find that differences in total factor productivity between small-scale and
medium-scale farms can be explained by differences in the effects of: (1) inorganic fertilizer
application (additional units of inorganic fertilizer are negatively associated with changes in TFP
on small-scale farms relative to medium-scale farms); (2) the allocation of an additional unit of
land to traditional cash crops (e.g., cotton) which increases total factor productivity by over 200
TFP units more for small-scale farms than medium-scale farms; (3) the allocation of an
additional unit of land to roots and tubers (e.g., sweet potatoes) which increase total factor
productivity by over 400 TFP units for small-scale farms relative to medium-scale farms; (4)
herbicide use (herbicide use is positively associated with changes in TFP on small-scale farms
relative to medium-scale farms).
Lastly, we report findings from graphs (Figures 5 to 9) which depict the relationship
between operated farm size and alternative predicted values of each measure of productivity: (1)
predicted values obtained from the ceteris paribus restricted model, and; (2) predicted values
obtained from the full model when crop management practices vary across farms. This empirical

137

approach helps us to compare IR hypothesis tests between the scenarios when the CRS
assumption is still imposed (first set of alternative predicted values) and when the assumption is
relaxed (second set of alternative predicted values). Figure 5 shows that although farm size is
negatively associated with net value of crop production per hectare when using predicted values
from the ceteris paribus restricted model, the relationship is negatively steeper when predicted
values obtained from the full model are used. This finding suggests that relaxing the CRS
assumption further strengthens the IR hypothesis in the case of land productivity.

Figure 5: Operated farm size versus alternative predicted values of net value of crop production
per hectare (with and without CRS assumption)
Figures 6 and 7 illustrate the relationship between operated farm size and alternative
predicted values of the two labor productivity measures constructed in this study. Both figures
show that when the constant returns-to-scale assumption is imposed, the relationship between
operated farm size and each measure of labor productivity is positive and steeper than when the
CRS assumption is relaxed. This finding suggests that relaxing the CRS assumption actually

138

weakens the argument that large farms are more productive than small farms in terms of labor
productivity.

Figure 6: Operated farm size versus alternative predicted values of net value of crop production
per family labor day (with and without CRS assumption)

Figure 7: Operated farm size versus alternative predicted values of net value of crop production
per labor day (with and without CRS assumption)

139

Figure 8 shows that when the CRS assumption is imposed, the cost of maize production
is positively related with maize area planted between 0 - 10 hectares but the relationship reverses
beyond ten hectares. The graph depicting the relationship between cost of maize production per
metric ton produced and maize planted when the assumption of constant returns-to-scale is
relaxed reveals a similar pattern as in the case when the assumption is imposed. This means that
in the case of the fourth measure of productivity—cost of maize production per metric ton
produced—relaxing the constant returns-to-scale assumption is not that different from the case
when the assumption is imposed.

Figure 8: Operated farm size versus alternative predicted values of cost of maize production per
metric ton produced (with and without CRS assumption)
The last figure (Figure 9) demonstrates that the relationship between TFP and operated
farm size is negative in both scenarios (imposed/restricted model and relaxed/full model).
However, similar to net value of crop production per hectare, relaxing the CRS assumption
reveals a negative but steeper relationship between TFP and operated farm size, which

140

strengthens the argument for the existence of an inverse relationship between farm size and
productivity.

Figure 9: Operated farm size and alternative predicted values of total factor productivity (with
and without CRS assumption)
In sum, Figures 5 – 9 demonstrate that relaxing the CRS assumption either strengthens
the IR hypothesis (in the case of net value of crop production per hectare and TFP) or weakens
the argument that large farms are more productive than small farms (in the case of the two labor
productivity measures). This important finding points to the fact that previous studies looking at
the farm size-productivity hypothesis have probably understated the extent of the existence of an
IR because they have generally imposed the CRS assumption. Therefore, apart from extending
the methodological approach used in the literature, the current study further proposes that studies
looking at this topical issue in development should lift the CRS assumption in order to get better
insights.

141

3.4. Conclusions and implications for policy
The interest by development scholars to revisit the IR hypothesis is mainly driven by the
need to guide policy on whether smallholder-led agriculture remains a viable pathway to achieve
the much-needed changes in the economic fortunes for the majority of rural SSA. Equally
important is the need to examine the hypothesis under conditions with very different relative
factor abundance than where most of the other studies have been done. Very few studies have
been comprehensive in terms of using alternative measures of productivity other than land
productivity, to explore the relationship between farm size and productivity. Moreover, as
observed by Collier and Dercon (2014) and demonstrated by our review of past empirical
literature, a critical number of studies conducted in Africa have on one hand included very few
farms outside the zero to ten hectare range while on the other hand have extrapolated their
findings beyond this farm size range. Furthermore, past empirical studies have been modeled on
the assumption that returns to scale in agricultural production are constant. While the assumption
may have been appropriate for studying the IR across small farms, it is less relevant when
extending beyond this range to include medium- and large-scale farms.
The current study addressed these gaps and the following findings emerged. First, when
the IR hypothesis was tested using representative data with about 15 percent of farms with
landholding sizes of 20 hectares or more, the results were not uniform across the five measures
of productivity used in the study. The relationship between farm size and productivity upheld the
IR, rejected the IR or was uncorrelated depending on the measure of productivity. Results
showed that when net value of crop production per hectare was regressed on farm size ceteris
paribus, the IR was upheld at 10 percent significance level. This modest level of significance

142

could be attributed to the fact that, unlike in past studies where input use intensity has been found
to be higher on small relative to large farms, the opposite was true in the case of this study.
When the outcome variable was a labor productivity measure, we found a strong positive
relationship between farm size and labor productivity similar to Li et al. (2013). In the case of
cost of maize production per metric ton produced, we found that the cost of maize production per
ton increased with maize area planted of up to 10 hectares. This result upheld the IR because
small farms were able to produce a ton of maize at a cheaper cost than large farms, but only for
the 0 – 10 hectares domain. The estimation results for TFP showed that this comprehensive
measure of productivity was negatively associated with operated farm size albeit not statistically
significant.
Second, we found that differences in crop management practices played a key role in
driving productivity differences between small- and medium-scale farms. Management practices
that were prominent in explaining productivity differences were total inorganic fertilizer use,
herbicide use, animal draft power use and the nature of land allocation to different crop types.
Third, relaxing the CRS assumption either strengthened the IR hypothesis (in the case of net
value of crop production per hectare and TFP) or weakened the argument that large farms are
more productive than small farms (in the case of the two labor productivity measures).
Overall, the following insights were gleaned from the key findings. Research that focuses
on making comparisons of productivity across farm sizes should extend beyond idiosyncratic
indicators of productivity that only assess land productivity. As shown in this study, results
emanating from using a more comprehensive set of productivity measures are not exactly
identical. This study shows that while small farms are more productive when considering net
value of crop production per hectare and the cost of producing a staple food crop like maize,

143

large farms are productive in terms of labor productivity associated with aggregate crop
production on farm.
What do these findings imply especially at the time when smallholder-led agricultural
growth strategies have been questioned? In truth, findings on the relationship between
productivity and operated farm size, while important, should not be the decisive factor in guiding
agricultural development and land policies in SSA because there are many other important
considerations. The challenges of rural poverty that continue to bedevil SSA are better addressed
by pursuing broad based development strategies whose central focus should remain smallholder
farmers. This entails promoting policies that would enhance productivity growth among
smallholders, as this is crucial to poverty reduction and development. Moreover, an inclusive
smallholder form of agricultural growth will clearly generate employment growth in both farm
and non-farm sectors than a comparable rate of agricultural growth concentrated among a few
large farms. This is very clear from the factor-intensity graphs in Figure 3 showing how laborland and labor-capital intensities fall dramatically as farm size increases beyond five hectares. As
argued by Jayne et al. (2014, p48), “the employment effects and growth multipliers resulting
from broadly based agricultural growth are likely to contribute much more to economic growth
and poverty reduction than from any potential efficiency advantages that large farms might
have…”. Pioneering work by Mellor and Johnston (1984) contrasting employment effects from
smallholder Asia versus Latifundia Latin America as well as more recent work by Christiaensen,
Demery, and Kuhl (2011) show that inclusive forms of agricultural growth generate much
greater growth and employment multipliers than agricultural growth concentrated among
relatively few large farms. However, the challenges of achieving broad-based inclusive forms of
agricultural growth remain daunting.

144

While migration from farm to non-farm sectors, and from rural to urban areas will
provide the brightest prospects for the transformation and modernization of Africa’s economies,
it will happen only as fast as educational advances and growth in the non-farm job opportunities
will allow. The rate of growth of non-farm jobs in turn depends on the rate of income growth
among the millions of families still engaged in smallholder agriculture. So there is a symbiotic
relationship between inclusive agricultural growth, nonfarm sector growth, and poverty
reduction.

145

APPENDIX

146

Figure 10: Map of Zambia showing location of ACS and RALS households

147

Table 22: Production function estimates used for TFP estimation
Estimated results
9.16***
(28.71)
0.36***
(12.58)
0.01
(0.27)
0.79***
(28.71)
1.16

𝑙𝑛𝐴0 + 𝜂𝑡
𝛼𝐾
𝛼𝐿
𝛼𝑀
𝑅𝑇𝑆 = 𝛼𝐾 + 𝛼𝐿 + 𝛼𝑀
𝛼𝐾∗

0.31

𝛼𝐿∗

0.01

∗
𝛼𝑀

0.69

R2
0.63
F-statistic
826.18
Source: Authors’ computation from RALS (2012) and ACS (2013)
Note: t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01

148

REFERENCES

149

REFERENCES

Ali, D. A., & Deininger, K. (2015). Is There a Farm Size–Productivity Relationship in African
Agriculture? Evidence from Rwanda. Land Economics, 91(2), 317–343.
Assunção, J. J., & Braido, L. H. B. (2007). Testing Household-Specific Explanations for the
Inverse Productivity Relationship. American Journal of Agricultural Economics, 89(4),
980–990.
Barrett, C. B. (1996). On price risk and the inverse farm size-productivity relationship. Journal
of Development Economics, 51(2), 193–215. https://doi.org/10.1016/S03043878(96)00412-9
Barrett, C. B., Bellemare, M. F., & Hou, J. Y. (2010). Reconsidering conventional explanations
of the inverse productivity–size relationship. World Development, 38(1), 88–97.
Berry, R. A., & Cline, W. R. (1979). Agrarian Structure and Productivity in Developing
Countries: A Study Prepared for the International Labour Office Within the Framework
of the World Employment Programme. John Hopkins University Press.
Bhalla, S. S., & Roy, P. (1988). Mis-Specification in Farm Productivity Analysis: The Role of
Land Quality. Oxford Economic Papers, 40(1), 55–73.
Binswanger, H. P., Deininger, K., & Feder, G. (1995). Chapter 42 Power, distortions, revolt and
reform in agricultural land relations. In J. B. and T. N. Srinivasan (Ed.), Handbook of
Development Economics (Vol. 3, Part B, pp. 2659–2772). Elsevier. Retrieved from
http://www.sciencedirect.com/science/article/pii/S1573447195300198
Block, S. (2010). The decline and rise of agricultural productivity in sub-Saharan Africa since
1961. National Bureau of Economic Research. Retrieved from
http://www.nber.org/papers/w16481
Carletto, C., Savastano, S., & Zezza, A. (2013). Fact or artifact: The impact of measurement
errors on the farm size–productivity relationship. Journal of Development Economics,
103, 254–261. https://doi.org/10.1016/j.jdeveco.2013.03.004
Carter, M. R. (1984). Identification of the Inverse Relationship between Farm Size and
Productivity: An Empirical Analysis of Peasant Agricultural Production. Oxford
Economic Papers, 36(1), 131–145.
Chapoto, A., Kabaghe, C., & Zulu-Mbata, O. (2015). The Politics of the Maize Sector in
Zambia. In A. Chapoto & N. J. Sitko (Eds.), Agriculture in Zambia: Past, Present, and
Future (pp. 55–71). Indaba Agricultural Policy Research Institute (IAPRI).

150

Chen, Z., Huffman, W. E., & Rozelle, S. (2011). Inverse Relationship Between Productivity and
Farm Size: The Case of China. Contemporary Economic Policy, 29(4), 580–592.
https://doi.org/10.1111/j.1465-7287.2010.00236.x
Christiaensen, L., Demery, L., & Kuhl, J. (2011). The (evolving) role of agriculture in poverty
reduction—An empirical perspective. Journal of Development Economics, 96(2), 239–
254. https://doi.org/10.1016/j.jdeveco.2010.10.006
Collier, P., & Dercon, S. (2014). African Agriculture in 50Years: Smallholders in a Rapidly
Changing World? World Development, 63, 92–101.
https://doi.org/10.1016/j.worlddev.2013.10.001
Deininger, K., & Byerlee, D. (2012). The Rise of Large Farms in Land Abundant Countries: Do
They Have a Future? World Development, 40(4), 701–714.
Diao, X., Hazell, P. B. R., Resnick, D., & Thurlow, J. (2007). The Role of Agriculture in
Development: Implications for Sub-Saharan Africa. Intl Food Policy Res Inst.
Dorward, A. (1999). Farm Size and Productivity in Malawian Smallholder Agriculture. Journal
of Development Studies, 35(5), 141.
Eswaran, M., & Kotwal, A. (1986). Access to Capital and Agrarian Production Organisation.
Economic Journal, 96(382), 482–98.
Fan, S. (1991). Effects of Technological Change and Institutional Reform on Production Growth
in Chinese Agriculture. American Journal of Agricultural Economics, 73(2), 266–275.
https://doi.org/10.2307/1242711
Feder, G., Just, R. E., & Zilberman, D. (1985). Adoption of Agricultural Innovations in
Developing Countries: A Survey. Economic Development and Cultural Change, 33(2),
255–298. https://doi.org/10.1086/451461
Fischer, G., Velthuizen, H. V., Nachtergaele, F. O., & JernelĂśv, A. (2000). Global AgroEcological Zones Assessment: Methodology and Results.
Hazell, P. B. R. (2005). Is there a future for small farms? Agricultural Economics, 32, 93–101.
https://doi.org/10.1111/j.0169-5150.2004.00016.x
Heltberg, R. (1998). Rural market imperfections and the farm size—productivity relationship:
Evidence from Pakistan. World Development, 26(10), 1807–1826.
https://doi.org/10.1016/S0305-750X(98)00084-9
Holden, S., & Fisher, M. (2013). Can area measurement error explain the inverse farm size
productivity relationship? Centre for Land Tenure Studies, Norwegian University of Life
Sciences. Retrieved from
http://www.researchgate.net/publication/257526369_Can_Area_Measurement_Error_Ex
plain_the_Inverse_Farm_Size_Productivity_Relationship/file/3deec5255aa9f2f5ff.pdf

151

Jayne, T. S., Chamberlin, J., & Muyanga, M. (2012). Emerging Land Issues in African
Agriculture: Implications for food security and poverty reduction strategies (Stanford
Symposium Series on Global Food Policy and Food Security in the 21st Century).
Stanford University.
Jayne, T. S., Chamberlin, J., Traub, L., Sitko, N., Muyanga, M., Yeboah, F. K., … Kachule, R.
(2016). Africa’s changing farm size distribution patterns: the rise of medium-scale farms.
Agricultural Economics, 47(S1), 197–214. https://doi.org/10.1111/agec.12308
Jayne, T. S., Chapoto, A., Sitko, N., Nkonde, C., Muyanga, M., & Chamberlin, J. (2014). Is the
Scramble for Land in Africa Foreclosing a Smallholder Agricultural Expansion Strategy?
Journal of International Affairs, 67(2), 35–53.
Kevane, M. (1996). Agrarian Structure and Agricultural Practice: Typology and Application to
Western Sudan. American Journal of Agricultural Economics, 78(1), 236–245.
https://doi.org/10.2307/1243794
Kimhi, A. (2006). Plot size and maize productivity in Zambia: is there an inverse relationship?
Agricultural Economics, 35(1), 1–9. https://doi.org/10.1111/j.1574-0862.2006.00133.x
Lal, R. (1995). Erosion-Crop Productivity Relationships for Soils of Africa. Soil Science Society
of America Journal, 59(3), 661.
https://doi.org/10.2136/sssaj1995.03615995005900030004x
Lamb, R. L. (2003). Inverse productivity: land quality, labor markets, and measurement error.
Journal of Development Economics, 71(1), 71–95. https://doi.org/10.1016/S03043878(02)00134-7
Land Matrix. (2015). Zambia - LAND MATRIX. Retrieved June 19, 2015, from
http://www.landmatrix.org/en/get-the-detail/by-targetcountry/zambia/?order_by=&starts_with=Z
Larson, D. F., Otsuka, K., Matsumoto, T., & Kilic, T. (2014). Should African rural development
strategies depend on smallholder farms? An exploration of the inverse-productivity
hypothesis. Agricultural Economics, 45(3), 355–367. https://doi.org/10.1111/agec.12070
Li, G., Feng, Z., You, L., & Fan, L. (2013). Re-examining the inverse relationship between farm
size and efficiency: The empirical evidence in China. China Agricultural Economic
Review, 5(4), 473–488.
https://doi.org/http://dx.doi.org.proxy2.cl.msu.edu/10.1108/CAER-09-2011-0108
Lombard, C. S. & T., A. H. C. (1972). Agriculture in Zambia since Independence. Neczam.
Mellor, J. W., & Johnston, B. F. (1984). The World Food Equation: Interrelations Among
Development, Employment, and Food Consumption. Journal of Economic Literature,
22(2), 531.

152

Mosley, P. (2002). The African green revolution as a pro-poor policy instrument. Journal of
International Development, 14(6), 695–724. https://doi.org/10.1002/jid.912
Nachtergaele, F., & Batjes, N. (2012). Harmonized world soil database. FAO. Retrieved from
http://www.fao.org/nr/water/docs/harm-world-soil-dbv7cv.Pdf
Sen, A. (1962). An Aspect of Indian Agriculture. Economic Weekly, Annual Number 14.
Sitko, N. J., & Jayne, T. S. (2014). Structural transformation or elite land capture? The growth of
“emergent” farmers in Zambia. Food Policy, 48, 194–202.
https://doi.org/10.1016/j.foodpol.2014.05.006
Sitko, N. J., Jayne, T. S., & Hichaambwa, M. (2013). Emergent Farmer Growth Trajectories in
Zambia: Land Accumulation and its Implications for Poverty Reduction. Presented at the
Annual World Bank Conference on Land and Poverty, Washington, D.C.
Toufique, K. A. (2005). Farm Size and Productivity in Bangladesh Agriculture: Role of
Transaction Costs in Rural Labour Markets. Economic and Political Weekly, 40(10),
988–992.
Verschelde, M., D’Haese, M., Rayp, G., & Vandamme, E. (2013). Challenging Small-Scale
Farming: A Non-Parametric Analysis of the (Inverse) Relationship Between Farm
Productivity and Farm Size in Burundi. Journal of Agricultural Economics, 64(2), 319–
342. https://doi.org/10.1111/j.1477-9552.2012.00373.x
Zhang, B., & Carter, C. A. (1997). Reforms, the Weather, and Productivity Growth in China’s
Grain Sector. American Journal of Agricultural Economics, 79(4), 1266–1277.
https://doi.org/10.2307/1244283

153

CHAPTER 4: AGRICULTURAL MECHANIZATION IN ZAMBIA: TRENDS AND
EFFECTS ON HOUSEHOLD PRODUCTION
4.1. Introduction
Access to improved technologies by agricultural households in sub-Saharan Africa (SSA)
engenders increased agricultural production and productivity, which are important conditions for
reducing high poverty rates and persistent food insecurities (Minten & Barrett, 2008). Where
investment in agricultural research and development (R&D) has been made, delivery of
improved technologies to farmers has largely been successful which in turn has spurred
agricultural total factor productivity growth (Fugile & Rada, 2013). Improved seed varieties and
inorganic fertilizers have received particular attention in harnessing desired productivity
improvements in Africa. As such, research examining the African agricultural transformation
process has focused more on adoption of these technologies, and the policies needed to promote
their adoption (Houssou et al., 2013).
Yet, agricultural mechanization—the application of powered machinery to agricultural
production—and the potential role it can play to enhance agricultural production, has not been a
prominent feature in recent research on agricultural development in SSA. This lack of
prominence could be attributed to two reasons explained below.
First, most mechanization programs promoted by African governments in the 1980’s and
prior to that, failed (Binswanger & Pingali, 1988; Houssou et al., 2013). Logically, tractors were
thought to be appropriate for SSA given the region’s land abundance at the time (Binswanger &
Pingali, 1988). Practically, the transition from hand cultivation through animal traction to greater
use of tractors had to be driven by population pressure increases and gradual changes in cropping
patterns from bush or forest fallow to intensive agricultural production (Pingali, Bigot, &
Binswanger-Mkhize, 1987). Further, the transition from hand tools was only achievable if

154

alternatives (animal traction or tractors) were perceived to be cost-effective by farmers.
Unfortunately, policy makers in SSA imposed alternatives through subsidized mechanization
programs that disregarded the criteria for adoption. This led to a higher failure rate of these
programs.
As a result, mechanization was essentially pushed to the bottom of the agricultural
development priority list. This contributed to lower levels of agricultural mechanization
promotion and use across the region compared with the rest of the developing world. For
example, of the total area cultivated for crop production in SSA at the turn of this century, about
65 percent was achieved through the use of hand tools while 25 percent was by animal draft
power, and 10 percent was by tractors (Takeshima & Salau, 2010). Meanwhile, tractor use in
other developing regions was significantly higher (35 percent) than SSA during the same period.
Second, the re-emergence of input subsidy programs (ISPs) across SSA in recent years
has helped to enhance the promotion of improved seed varieties and chemical fertilizers (RickerGilbert, Jayne, & Shively, 2013). Mechanization, on the other hand, has not been part of the
package of technologies promoted alongside these ISPs. In light of the high cost of these ISPs,
policy makers and development experts in SSA have been more concerned with investigating the
impacts of associated technologies (seed and fertilizer) on livelihoods of individual agricultural
households, the rural economy and overall national welfare.
In view of these reasons, it is not surprising that agricultural mechanization has not been
given much attention by researchers concerned with agricultural transformation and development
in SSA. However, current anecdotal evidence indicating that farm labor costs in SSA have been
increasing (Takeshima et al., 2013) and recent empirical estimates in countries like Zambia
showing that labor (family and hired labor) accounts for more than 60 percent of the cost of

155

production of maize (Burke et al., 2011) suggests that mechanization deserves more attention.
Coincidentally, there has been a rise in the number of agricultural mechanization initiatives in
parts of SSA (including Zambia) that involve the public sector, the private sector and farmer
organizations aimed at promoting mechanization among small- and medium-scale farming
households. In light of the rise of domestic medium-scale land investors in parts of SSA (Jayne
et al., 2016) who might be more inclined to demand mechanization services, the need for
research that reassesses the role of agricultural mechanization in enhancing agricultural
development has become imperative. The aims of this paper is to examine agricultural
mechanization trends, effects on agricultural household production and the factors contributing
to recent developments aimed at enhancing mechanization in SSA based on the case study of
Zambia.
This study contributes to the emerging research aimed at guiding policy on
mechanization in SSA (Mrema et al., 2008; Houssou et al., 2013; Takeshima et al., 2013; Diao et
al., 2014). Further, it provides useful insights to private sector actors, farmers and farmer
organizations on the sustainability of agricultural mechanization initiatives that have recently
been promoted in parts of SSA and Zambia in particular. Consequently, this research investigates
three main questions as stated below using the case study of Zambia:
1. How has mechanization use for agricultural production across small- and

medium-scale farming households evolved in recent years? What is the level of
tractor-hiring service provision and use among small- and medium-scale farming
households for agricultural production?

156

2. What is the effect of mechanization on the following indicators: (i) area

cultivated, and; (ii) net value of crop production per hectare? Is mechanization
likely to have a positive effect on both of these indicators?
3. What are the factors contributing to the rise of agricultural mechanization

initiatives in Zambia? Are changes in factor price ratios inducing demand for
these services? Alternatively, what opportunities on the supply side are
encouraging the rise of these services?
The rest of this paper is organized follows. The following section reviews the background
literature on the historical perspectives and recent developments of agricultural mechanization
within the context of SSA and Zambia. Next, the data and methods used in the current study are
described while the results are presented and discussed in the penultimate section. The last
section concludes and provides implications for agricultural mechanization policy.
4.2. Background
4.2.1. Agricultural mechanization in SSA and Zambia: historical perspective
Promotion and use of agricultural mechanization is not new to SSA. During the early
colonial period (before the second world war), tractor use was mainly among European settler
farms and government-run farms (Pingali, 2007). After the Second World War, African farmers
also started using tractors because financing of tractor imports was made possible by a fund for
farm machinery established through the Marshall Plan (ibid). When African countries gained
political independence from the 1960’s through the 1970’s, some of the governments were
directly involved in the supply of tractors during the early stages of independence until the
1980’s (Houssou et al., 2013). Overall, there were three significant waves during which
agricultural mechanization use was actively promoted in Africa from the colonial period up to
157

the 1980’s (Pingali, 2007). The three waves could be classified as follows: (1) colonial
influenced agricultural mechanization initiatives (1945-1955); (2) state sponsored agricultural
mechanization initiatives promoted during the initial stages of political independence (19581970), and; (3) state sponsored agricultural mechanization initiatives promoted as a consequence
of gains from mineral resource exports (1970-1980).
Interestingly, the number of tractors in SSA during the post-independence period (second
wave) was higher than other developing regions of the world (for example Asia and Latin
America), which suggested at the time that SSA was on a trajectory of enhancing agricultural
mechanization at a faster rate than the rest of the developing world. To put this in perspective,
SSA had approximately two, three and six times more tractors in use than in Brazil, India and
China respectively (Mrema et al., 2008). However, the level of mechanization in these three
countries steadily increased and surpassed SSA in part because the increase in factor prices for
labor and animal draft power was relatively higher in the aforementioned countries. By the year
2000, Brazil, India and China now had consecutively, four, seven, and four times more tractors
in use than the entire SSA region (ibid). As a consequence, some agricultural indicators such as
profitability in other parts of the developing world have steadily improved relative to SSA
because of increased use of improved technologies including mechanization.
In China, for example, productivity has increased over time, in part because of the rapid
increase in farm mechanization ownership and provision of farm mechanization services for
labor-intensive activities such as land preparation and harvesting (Yang et al., 2013). In India, it
has been shown that farm mechanization has led to increased use of other productivity enhancing
inputs, increased cropland, improved labor productivity and enhanced profitability of agriculture
(Verma, 2008).

158

The situation in SSA has, however, been different. Despite the fact that mechanization
has several benefits—timeliness of field operations, reducing the drudgery associated with farm
work and can be used for numerous tasks on and off the farm—the level of agricultural
mechanization use in most of SSA has remained low. Moreover, data from countries where
tractor use showed promise during the three waves discussed above, suggests that levels of
mechanization have remarkably reduced (Pingali, 2007). The reason for the persistently low
levels of agricultural mechanization has been attributed to low agricultural intensification and
lack of incentives for increasing productivity growth in SSA (ibid). Where countries showed
promise in terms of enhancing agricultural mechanization (countries that experienced the three
waves discussed above), the initiatives largely failed due to direct involvement by governments
in the supply of mechanization services, which undoubtedly contributed to low efficiency and
high cost without sustainability (Pingali, Bigot, & Binswanger-Mkhize, 1987).
Zambia, to a larger extent, presents an interesting case study because the country
experienced the three waves of agricultural mechanization use discussed above. Pingali (2007)
identifies four countries in SSA, including Zambia, where promotion of tractor use could be
traced back to the colonial period (first wave). After independence (second and third wave), the
Zambian government wanted to modernize the agriculture sector by establishing various tractor
schemes that distributed tractors, implements and tools to small-scale farmers (Jonsson, 1985).
While there were a few exceptions, most tractor schemes established in Zambia, like in the case
of other SSA countries, performed poorly and became very expensive to sustain. By mid-1980’s,
government policy had shifted away from active promotion of tractors to promotion of animal
draft power as summarized below:

159

“Animal [draft] power will form a basis for mechanization in Zambia because animals
involve very little foreign exchange and they are appropriate for most of the farm
operations required by the majority of the farming sector in Zambia who own between
five to 40 hectare[s] of land” (Tembo, 1985, p. 32).
4.2.2. Agricultural mechanization in SSA and Zambia: recent developments
Although agricultural mechanization has remained low on the policy agenda of most SSA
countries till today, some governments have begun to reemphasize the need to establish
mechanization initiatives that are funded either publicly, privately or both. A good example of a
country in SSA where mechanization initiatives have resurfaced is Ghana. According to Houssou
et al. (2013), the Government of Ghana (GoG) initiated the Agricultural Mechanization Service
Enterprise Centers (AMSECS) concept as a response to increasing demand for mechanization
services by farmers who could not afford required machinery for land preparation.
Support to the AMSECs started in 2007 which involved subsidies to the privately run
centers for the purchase of tractors, implements such as plows and harrows, and haulage
equipment (trailers). The GoG subsidized 30 percent of the purchase price of the equipment
while each AMSEC made an up-front payment of 20 percent with the remainder paid in
installments for two to three years. Other countries that have begun to support agricultural
mechanization initiatives in the last five to ten years include Nigeria, Tanzania, Mali and the
Democratic Republic of Congo (DRC) with financial support from emerging economies such as
Brazil, China and India (Diao et al., 2014).
Similarly, there has been a recent drive to enhance agricultural mechanization promotion
and use among small- and medium-scale farms in Zambia through initiatives spearheaded by
agribusiness firms, farmer organizations, financial institutions and the Zambian government. In

160

2012, three pilot initiatives were operational in Zambia, one promoted by Dunavant Cotton (now
NWK Agri-services), the second by AFGRI with the support of Conservation Farming Unit
(CFU) and Musika Development Initiatives Limited, and the third by the Ministry of Agriculture
(MoA) and Food and Agriculture Organization of the United Nations with a total of 58 Tractor
Service Providers (TSPs) operating at the time (Aagaard, 2012). These initiatives have since
been remodeled to include other partners and products on offer, and have expanded the number
of TSPs; about 448 TSPs are operating in the field (Kasanda, 2016).
Although each initiative may vary from the other, they typically involve selection of
suitably qualified farmers—those with good entrepreneurial skills and credit history—to access
available credit facilities intended for TSPs. The organization with presence in the farming
areas—usually a farmer organization or non-governmental organization—is responsible for
selecting beneficiaries (identified TSPs) and providing technical and business training to them.
Financing the purchase of equipment proceeds as follows. First, financial institutions provide
loans once it is established that selected beneficiaries meet the general terms and conditions of
the bank. Second, selected beneficiaries are expected to pay a deposit (over 20 percent in most
cases) towards the purchase of the equipment from a tractor supplier. Third, selected
beneficiaries are expected to repay the loan, with interest, within a period not exceeding three
years. Fourth, most of these initiatives include a provision of first loss guarantee, which is
managed by one of the parties to the agreement.
Existing mechanization initiatives have been designed in such a way that selected TSPs
choose their preferred combination of equipment needed. Equipment includes tractors (various
sizes between 37 to 100 horsepower), rippers (most mechanization initiatives are promoting
minimum tillage), planters and trailers for haulage. Because these initiatives have been designed

161

to facilitate the growth of TSPs, one of the key indicators of whether TSPs are performing
according to expectations is the number farmers/hectares serviced for land preparation.
According to a model developed in 2012, a single TSP can reasonably provide land preparation
services (ripping) to a total of 180 farmers (180ha) (ACF, 2012). In addition, the same model
estimates that a single TSP can provide services for planting, spraying and transportation to 165,
60 and 78 farmers respectively. Therefore, a TSP meeting the aforementioned targets could be
considered to be operating optimally.
In sum, recent developments aimed at enhancing agricultural mechanization in SSA
demonstrate that countries in the region recognize the important role that mechanization plays in
improving agriculture. Whether these new initiatives represent the right business model that
would address challenges of mechanization programs of yesteryear is an issue that is of empirical
significance and hence the relevance of this study.
4.3. Data and methods
This research combines qualitative and quantitative methods to examine agricultural
mechanization in Zambia in line with three research questions. First, how has mechanization use
for agricultural production (including the level tractor-hiring service provision) evolved in recent
years? Second, what has been the effect of mechanization on area cultivated and net value of
crop production per hectare? Third, what are the factors contributing to the rise of agricultural
mechanization initiatives in Zambia?
An explanatory sequential mixed-methods approach was used during this research and
this involved the following steps. The first step involved collecting both secondary (national
level data) and primary (six districts) household survey data. The survey data were initially
analyzed and the findings from this analysis informed the subsequent development of interview

162

guides for the qualitative phase of the study. Each of these approaches is discussed in turn and
the section concludes with a detailed description of data analysis techniques used to answer each
research question.
4.3.1. Phase 1: quantitative data collection
The quantitative data were collected using secondary and primary data sources. The
secondary data sources were drawn from two nationally representative surveys conducted in
Zambia in the last decade or so. The first nationally representative source of secondary data that
the study relied on was various years of the Crop Forecast Surveys (CFS) of small-, mediumand large-scale farms in Zambia beginning from 2001. The Central Statistical Office (CSO) of
Zambia conducts the CFSs on an annual basis in collaboration with the Ministry of Agriculture
and Livestock (MAL). The CFSs are designed to collect data from an average of 12,000
households on key agricultural production parameters during each agricultural season that runs
from October 1 to September 30 the following year.
Because it is not practical to collect all policy-relevant information in this annual survey,
this study relied on another secondary source of nationally representative data—the Rural
Agricultural Livelihoods Survey (RALS) of small and medium-scale farming households in
Zambia collected in 2012 and 2015. The RALS is a longitudinal survey conducted by the Indaba
Agricultural Policy Research Institute (IAPRI) in conjunction with the two aforementioned
public offices. The RALS used the new sampling frame derived from the 2010 population census
of Zambia. A total of 8,839 and 7,934 households were interviewed during the 2012 and 2015
surveys respectively. The survey collected data on a number of questions related to the following
main themes: demographic characteristics of household members; farmland and use; crop sales
from own production; input and credit acquisition; livestock ownership and marketing; off-farm

163

income sources; food security indicators, and; other themes such as kinship ties of the household
head. In addition, soil samples were collected from largest maize fields of sub-sampled
households during the 2012 survey.
The final source of quantitative data used in this study was primary data collected
through the Agricultural Commercialization Survey (ACS). The ACS mainly targeted mediumscale farms in Zambia and had a more elaborate module dedicated to agricultural mechanization
among this group of relatively land-endowed farmers.
The ACS of 2013 was a follow-up to a survey of medium-scale farms conducted in
Zambia in 2011 by Sitko & Jayne (2012) with assistance from the author of this paper. The main
objective of the ACS was to understand the history, characteristics, land use decisions and
patterns, and level of commercialization of medium-scale farmers observed to be a growing class
of farmers in Zambia. Similar to the first survey, the 2013 ACS survey purposively selected
research sites (administrative districts) based on the concentration and number of medium-scale
farms as reported in the 2010/2011 CFS.
Two main criteria were used to select the districts included in the survey. First, at least 3
percent of farming households had to be classified as medium-scale farms in order to ensure a
reasonable population from which to sample. Second, the research sites were selected along a
continuum of concentrations to ensure geographical diversity. But as noted by Sitko &
Chamberlin (2015), because medium-scale farms are concentrated along what is referred to as
the ‘line of rail’ in Zambia, the selected districts were in close proximity to this region. The line
of rail is a region of Zambia that is served by the railway linking the Copperbelt province with
Lusaka, and with the border town Livingstone that is south of the country.

164

Therefore, the six selected districts were Chibombo, Choma, Chongwe, Kalomo,
Mpongwe and Mumbwa. A total of 482 households were randomly selected from a list of
medium-scale farms generated in consultation with the Zambia National Farmers’ Union
(ZNFU) and the Ministry of Agriculture and Livestock (MAL) district offices. The Ministry of
Agriculture and Livestock (MAL) block areas were identified as the sampling units with help
from local district offices. Whereas the sampling procedure was supposed to ensure reasonable
representativeness of farmers in the 5-100 hectare category within the selected districts, the
sample may have not been statistically representative of all medium-scale farmers in Zambia.
This is because medium-scale farmers were not sampled from the other 65 districts. But we can
say that the chosen districts are understood to contain the highest proportions of farms in this size
class.
4.3.2. Phase 2: qualitative data collection
Qualitative data were collected through key informant interviews of 48 purposively
sampled respondents. These interviews targeted: (1) farmers (small- and medium-scale farmers);
(2) community leaders and relevant district officials, and; (3) national level stakeholders. The
study also conducted additional in-depth interviews with specific stakeholders involved in
mechanization service provision in Zambia.
For each interview, detailed notes were taken and where the interviewee consented, an
audio recorder was used to capture the full interview. At the end of each day, the field notes and
audio recordings were fully transcribed in readiness for analysis. Each source for qualitative data
is described here in turn.

165

4.3.2.1.

Farmer in-depth interviews

The in-depth interviews with farmers allowed the researcher to get further insights on
issues not adequately addressed during the survey such as the following: (1) why they became
farmers and what motivated them to acquire land; (2) agricultural land use and farmers’
perception of critical drivers; (3) agricultural technology choices, and; (4) relationship between
farm structure change and prevailing institutions and policies. The component of the interview
that asked questions on agricultural technology choices primarily focused on sources of power
for land preparation used by the farmers and the role that agricultural mechanization plays in
enhancing farmers’ production levels.
A total of 24 farmer in-depth interviews were conducted, with 12 small-scale farmers and
12 medium-scale farmers. The in-depth interviews were conducted in the six districts were
household survey data was collected using the ACS. The study first identified two medium-scale
farmer in-depth interview participants per district using the ACS list. Selection of the participants
was based on the trajectory the household followed to achieve medium-scale farmer status:
organic growth through small-scale farming or lateral entry through non-farm income. Using
snowball sampling, the study then selected small-scale farmers who were neighbors to the
medium-scale farmers who participated in the in-depth interviews.
4.3.2.2.

Community leader in-depth interviews

While a significant part of the interviews was dedicated to land acquisition patterns in
respective communities, the community leaders selected for this study also had to have extensive
knowledge on agricultural mechanization development in the community. The community leader
interviews targeted two community leaders in each of the six districts. In each district, the study
undertook interviews with one village headman from the agricultural blocks visited during the

166

survey and one high level official working for one of the key government ministries at district
level with knowledge about the evolution of farm ownership and agricultural mechanization use
in the district. In total, the study interviewed 12 community leaders.
4.3.2.3.

Interviews with national-level stakeholders

The national level stakeholders helped to give a national perspective on agricultural
mechanization development in Zambia. A total of twelve national level stakeholder interviews
were conducted with senior officials at relevant government ministries (Ministry of Agriculture,
Ministry of Lands etc.), civil society organizations dealing with land issues in Zambia such as
the Zambia Land Alliance (ZLA), farmer organizations such as the Zambia National Farmers’
Union (ZNFU), private sector firms and experts on agricultural development issues in Zambia.
4.3.2.4.

Interviews of mechanization service providers

To supplement the researcher’s understanding of opportunities and constraints associated
with agricultural mechanization on the supply side, the study conducted additional interviews
with main actors involved in recent mechanization initiatives in Zambia. These interviews were
implemented to get a much more detailed picture of what is driving mechanization initiatives,
successes realized so far, constraints likely to undermine mechanization efforts and sustainability
of these initiatives as perceived by those implementing them. A total of six interviews were
conducted with key actors in mechanization service provision in Zambia. The actors were
identified after a thorough review of program evaluation reports on mechanization availed to the
author by the Zambia National Farmers’ Union (ZNFU).

167

4.3.3. Data analysis
The quantitative methods used to analyze the data assembled included descriptive and
regression analysis using Stata 14 software. Analysis of the secondary data accounted for the
survey design by using the svy command available in Stata. Details of all the quantitative
analysis are presented separately by each research question. The qualitative data analysis
methods employed in the study are collectively described at the end of the section.
4.3.3.1.
services

RQ1: Quantitative analysis of trends in mechanization use and tractor hiring

In order to quantitatively explore the trends in agricultural mechanization use and tractor
hiring services in Zambia and how they have evolved in recent years, the study analyzed both
secondary and primary sources of survey data available. Specifically, the following analyses
were conducted: (1) a comparison of the proportion of small- and medium-scale farming
households using mechanization for land preparation by province over a three-year period; (2) a
cross sectional comparison of the proportion of medium-scale farming households using
mechanization for agricultural land preparation in selected districts; (3) estimation of the number
of tractors available per thousand hectares cultivated in Zambia; (4) a comparison of the
demographic and agricultural production characteristics of households using mechanization
versus non-users over a three year period, and; (5) a characterization of tractor hiring services
access and provision among medium-scale farming households in selected districts of Zambia.
4.3.3.2.
production

RQ2: Quantitative analysis of effects of mechanization on household level

To analyze the contemporaneous effects of mechanization on area cultivated and net
value of crop production per hectare (a measure of crop productivity), this research used and
extended the analytical framework adapted by Houssou & Chapoto (2014). Their study assessed

168

the impacts of agricultural mechanization on cropland expansion and farm input intensification
in Northern Ghana. To make the assessment, the aforementioned study recognized that
agricultural mechanization is driven by various factors and that it requires a comprehensive
framework to understand the ways through which these factors interact. The factors broadly
include agricultural policies, the market environment, farming systems and farmers’ own
behavioral strategies.
In this study, a similar approach was used to assess the effects of mechanization on area
cultivated in Zambia and extended the analysis by assessing the effects of mechanization on crop
productivity. The study only assessed contemporaneous effects using a cross sectional dataset—
ACS (2013). The empirical procedure used in this study was an endogenous switching regression
model (Maddala, 1983). The model accounts for potential challenges—endogeneity and sample
selection—that may arise in assessing the effects of a technology such as tractor use on area
cultivated and net value of crop production per hectare cultivated. To illustrate these challenges,
I begin by stating the underlying relationship of interest.
Suppose the following regression is estimated at household level:
(1)

𝑦 = 𝑋 ′ 𝛽 + 𝛿𝐼 + 𝑢

where 𝑦 represents an outcome variable of interest such as area cultivated or net value of crop
production per hectare, 𝑋 is a vector of exogenous characteristics, 𝐼 is a dummy variable defined
as 𝐼 = 1 if the household used a tractor for land preparation and 0 otherwise, and 𝑢 is the error
term. In this study, area cultivated is defined as the total number of hectares prepared by the
farmer for crop production. Net value of crop production per hectare cultivated is the monetary
value of crop production net of variable input costs divided by the number of hectares cultivated.

169

The regression model outlined above assumes that 𝐼 is uncorrelated with the error term
(exogeneity condition). This approach might lead to biased parameter estimates (𝛽, 𝛿) because of
this strong exogeneity assumption. In reality, some unobserved characteristics that influence the
probability that a farmer decides to use a tractor for land preparation could also influence the
other outcome variables of interest. The endogenous switching regression model is designed to
circumvent these challenges.
The model consists of two parts. The first part is a selection equation, which models the
decision to use or not to use tractors for land preparation using a standard limited dependent
variable method. The second part presents the production outcome equations estimated for each
group (users and non-users of tractors). Thus, the model is specified as follows:
1⁡⁡⁡⁡⁡𝑖𝑓⁡⁡⁡⁡𝐼𝑖∗ > 0
0⁡⁡⁡⁡⁡𝑖𝑓⁡⁡⁡⁡𝐼𝑖∗ ≤ 0

(2.1)

Selection equation

𝐼𝑖∗ = 𝑍𝑖 𝛾⁡ + ⁡ 𝜀𝑖 , 𝐼𝑖 = {

(2.2)

Regime 1 (users)

𝑦1𝑖 = 𝑋1𝑖 𝛽1 ⁡ + ⁡ 𝑢1𝑖 ⁡𝑖𝑓⁡𝐼𝑖 = 1

(2.3)

Regime 2 (non-users)

𝑦2𝑖 = 𝑋2𝑖 𝛽2 ⁡ + ⁡ 𝑢2𝑖 ⁡𝑖𝑓⁡𝐼𝑖 = 0

where, in the selection equation, 𝐼 ∗ is a latent variable for choice of land preparation technology,
𝐼 is what is observed (𝐼 = 1 if the household used tractor for land preparation and 0 otherwise),
𝑍 is a vector of observed farm and non-farm characteristics determining the decision to use or
not to use tractors for land preparation, and 𝑢 is the error term. For the second part, 𝑦 represents
the two outcome variables (area cultivated and net value of crop production per hectare
cultivated), 𝑋 is a vector of variables that capture household characteristics, market environment,
district fixed effects and the agricultural policy environment.

170

Based on Lokshin & Sajaia (2004), this study implements the full information maximum
likelihood method (FIML) to simultaneously estimate the selection equation and the two regime
equations outlined above. This is done to correct the inefficiencies inherent in other estimation
methods like the two-step least square or maximum likelihood estimation. The FIML yields
consistent standard errors.
The empirical strategy for modeling the effect of mechanization on the two outcome
variables of interest proceeded as follows. First, in order to identify the model, a test for
exclusion restrictions was conducted; isolating variables that go into the selection equation but
are excluded from the outcome equations. Based on Di Falco, Veronesi, & Yesuf (2011), the test
for exclusion restrictions involved running a probit model for the whole sample on variables
conjectured to determine whether or not a household uses mechanization for land preparation.
Next, an ordinary least squares (OLS) regression was estimated for each outcome variable (area
cultivated and net value of crop production per hectare cultivated) where the right hand side
variables only included the significant variables from the probit model estimated earlier. The
OLS was only estimated for sample households that did not use mechanization. The variables
that were not statistically significant in the second stage (OLS) constituted the possible
candidates for the exclusion restrictions.
The second step of the empirical strategy involved estimating the endogenous switching
regression model using the movestay command in Stata 14. We also did counterfactual analysis
by comparing the expected values for the two outcome variables under the actual and
counterfactual cases that farmers used or did not use mechanization for land preparation (ibid).
Further, treatment and heterogeneity effects were computed to understand the differences in area
cultivated and net value of crop production per hectare between farm households that were users

171

and those that were non-users of mechanization. The counterfactual and treatment effects
analysis relied on post-estimation procedures available in Stata. We provide a detailed
description of this analysis when presenting our findings.
4.3.3.3.

RQ3: Quantitative analysis of the rise of agricultural mechanization services

Although the induced innovation hypothesis may not be a perfect fit, this paper relied on
this widely used concept of technical change in development economics to explain recent
developments related to agricultural mechanization in Zambia. Formalized by Hayami & Ruttan
(1970, 1985), the hypothesis explains that development and adoption of technology to save an
input, is induced by the input’s relatively higher price. Put another way, the hypothesis posits
that “alternative agricultural technologies are developed to facilitate the substitution of relatively
abundant (hence, cheap) factors for relatively scarce (hence, expensive) factors” (Ruttan, 2002,
p. 163).
Hayami & Ruttan (1985) illustrate the process of induced technical change in Japan and
the United States using historical data (1880-1980) of the following variables: price of fertilizer,
price of land, amount of fertilizer applied per hectare of land, price of draft power, price of labor
and the amount of draft power per worker. For both countries, the price of fertilizer declined
relative to the price of land and the price of draft power declined relative to the price of labor. As
a consequence, fertilizer use per hectare and power per worker rose in both countries. Based on
this illustration, fertilizer would be regarded as “land saving” because increased fertilizer use
substitutes for land, while draft power or mechanization is “labor saving” because it substitutes
power and machinery for labor (Binswanger, 1978a; Ruttan, 2002).
Binswanger (1978b) demonstrated that a shift to tractor use could only be economically
induced under rapidly rising prices of factors of production that tractors have the potential to

172

replace—labor or animal draft power. A more recent application of the induced innovation
hypothesis is by Diao et al. (2014) who assessed the rise of mechanization services in Ghana.
Their study found that rising wage rate and increased opportunities for rural workers in nonfarm
sectors has induced the demand for labor-saving technology and demand for certain mechanized
farming operations, particularly plowing, has emerged even among small-scale farmers.
Equipped with the induced innovation hypothesis, the current study assessed the changes
in two factor prices—agricultural wage labor and rental prices for mechanization—in Zambia
and whether or not these changes were likely to induce a shift to labor-saving technologies.
Using the CFS data (2012 – 2015), the following steps were followed to estimate the factor price
ratios. First, median mechanization rental rates and agricultural wage rates were computed by
province and agricultural activities (for example, land preparation, planting, harvesting). Only
the factor prices for land preparation were used for subsequent analysis because they directly
reflect tractor use, the focus of this study. Second, the factor prices were adjusted for inflation
using available CPI indices. Third, factor price ratios were generated for each year by province
by dividing the median rental rate of mechanization by the median agricultural wage rate. Graphs
were generated to further interpret the findings.
4.3.3.4.

Qualitative data analysis

The data collected using key informant and case study interviews were used to gain
further insights for this research. Transcripts were developed for each of the interviews captured
through either audio recordings or handwritten notes. Based on grounded theory approach
(Strauss & Corbin, 1994), the study used an iterative process to develop a coding scheme in
NVivo 10 software. The coding scheme specified the concepts and themes, their definition and
the rules for applying the codes. The codes developed enabled the researcher to easily retrieve

173

similar information across individual transcripts. Summary statements were written to represent
the diversity of responses and where possible, quotes from the respondents were used to clearly
represent their perspectives.
4.4. Results and discussion
4.4.1. Trends in agricultural mechanization use and tractor hiring services
The following analyses were conducted to help with understanding the trends in
agricultural mechanization use and tractor hiring services in Zambia. First, a comparison of the
proportions of small- and medium-scale farming households using mechanization for land
preparation was done by province over a three-year period to give both a spatial and overall
picture. Second, the study also compared the proportions of medium-scale farming households
using mechanization for agricultural land preparation by selected districts in Zambia. Third, the
study estimated the number of tractors available per thousand hectares cultivated in Zambia.
Fourth, the study compared demographic and agricultural production characteristics of
households using mechanization versus non-users over a three-year period. Fifth, this research
also characterized access to and provision of tractor-hiring services by medium-scale farming
households in selected districts of Zambia. The results are presented and interpreted in turn.
4.4.1.1.

Mechanization use of small- and medium-scale farmers by province
Figure 11 presents results for the proportions of small- and medium-scale farming

households that used mechanization for agricultural land preparation by province between 2012
and 2015. First and foremost, these results clearly show that agricultural mechanization use
among small- and medium-scale farmers has remained low in Zambia. The aggregate proportion
of farming households using mechanization for land preparation stood at 1.5 percent in 2012 and
marginally increased to 1.6 percent in 2015. Second, except for Western province, the other
174

provinces where the proportion of farming households using agricultural mechanization was
close to or above the national average of 1.5 percent were provinces—Central, Copperbelt and
Lusaka—located along Zambia’s ‘line of rail’. This finding only goes to confirm that
development of mechanization has primarily happened in parts of the country where there is
better infrastructure and modest availability of agricultural support services.

10.0%
9.0%
8.0%
7.0%
6.0%
5.0%
4.0%

3.3%

3.0%

3.0%
2.0%

1.7%

1.2%

1.0%

0.4%

0.3%

0.6%

0.4%

1.6%

0.9%

0.0%

0.0%

% using mechanization for land preparation (2012)
% using mechanization for land preparation (2015)

Figure 11: Proportion of small- and medium-scale farming households using mechanization for
agricultural land preparation by province between 2012 and 2015
Source: Author’s computation from RALS (2012, 2015)
4.4.1.2.

Mechanization use of medium-scale farmers by selected districts

Reported in Figure 12, we make a cross sectional comparison of the proportion of
medium-scale farming households that used mechanization for agricultural land preparation by
selected districts. The aggregate proportion of medium-scale farm households using
mechanization was above eight percent. This is above the 1.5 percent and 1.6 percent national
175

average reported among small- and medium-scale farms in 2012 and 2015 respectively. This
result suggests that medium-scale farms (specifically those located in the selected districts) were
more likely to use mechanization for land preparation relative to both small- and medium-scale
farms at national level. Further, the highest proportion of mechanization users was reported in
two districts located in Central and Copperbelt provinces; Mumbwa and Mpongwe where about
11 percent and approximately 14 percent respectively of medium-scale households stated that
they used mechanization for land preparation. Nevertheless, these numbers still fall short of
levels of mechanization reported in other parts of SSA such as Ghana where farm power use
rates in some regions of the country have been reported to be above 90 percent of farming
households (Houssou & Chapoto, 2014).

20.0%
18.0%
16.0%
13.7%

14.0%
11.3%

12.0%
10.0%

8.3%

8.2%
8.0%
6.0%
4.0%

3.7%
2.6%

2.4%
2.0%
0.0%
Chibombo

Mumbwa

Mpongwe

Chongwe

Choma

Kalomo

Aggregate

% using mechanization for land preparation (2013)

Figure 12: Proportion of medium-scale farming households using mechanization for agricultural
land preparation by selected districts
Source: Author’s computation from ACS (2013)

176

4.4.1.3.

Tractor use per area cultivated in Zambia

The study also assessed the level of tractor use relative to the area cultivated in Zambia.
Table 23 below shows the number of tractors per one thousand hectares cultivated in Zambia
based on the RALS (2012 and 2015). To compute this statistic, the analysis relied on available
sampling weights in the two secondary datasets that made it possible to estimate the national
level number of tractors available for use, the number of hectares cultivated and thereafter
compute the number of tractors per one thousand hectares cultivated. Results show that tractor
use in Zambia over the three-year period remained relatively the same; about 1.3 and 1.2 tractors
per one thousand hectares cultivated in 2012 and 2015 respectively. This finding is consistent
with estimates presented by Pingali (2007) that indicate that tractor use in SSA was around 1.3
tractors per one thousand hectares cultivated in 2002. In the same period, however, the number
of tractors per one thousand hectares was around 9.1 in South Asia and 10.4 in Latin America
(ibid). Findings from the current study imply that tractor use per area cultivated in Zambia has
remained stagnant when compared to other developing regions of the world. However, we
recognize that these differences in tractor use per one thousand hectares have to be interpreted
cautiously since this outcome might be a function of social, economic and environmental
conditions (conditions not controlled for in this analysis) that are regional specific.
Table 23: Tractor use per thousand hectares cultivated in Zambia
Year
2012
Total number of tractors (weighted)
Total number of hectares cultivated (weighted)
Tractor use/1000 hectares cultivated
Source: Author’s computation from RALS (2012, 2015)

177

2,990
2,282,021
1.3

2015
3,612
3,079,826
1.2

4.4.1.4.
Demographic and production characteristics of users versus non-users of
mechanization
Table 24 presents difference in means comparisons of demographic and agricultural
production characteristics of users versus non-users of mechanization for land preparation in
2012 and 2015. The table clearly shows that the number of users of mechanization in both years
was significantly lower relative to non-users. Only 134 out of 8839 households (1.5 percent) and
122 out of 7747 households (1.6 percent) interviewed in 2012 and 2015 respectively, used
mechanization for agricultural production. The key findings worth noting from these results are
the following. First, farmers who used mechanization were more likely to be older than those
who did not and this was evident across both years. In 2012, for example, the average age of
household heads who used mechanization was 50 years while those who did not was 45 years
and the result was significant at one percent level. This finding could be attributed to the fact that
older farmers have accumulated enough experience and enhanced their status in order to adopt
use of mechanization for land preparation.
Second, the results show that the level of education of the household head is positively
related with use of mechanization both in 2012 and 2015. This result is not surprising. This
indicates that households with better education were able to easily access and understand
information that pertains to the relative benefits of mechanization. Third, households that used
mechanization for land preparation had a higher likelihood of having a member earning income
from wages than those who did not use mechanization. For instance, in 2015, about 44 percent of
mechanization using households had someone who earned income from wages while only 29
percent of those not using mechanization had such income sources. The interpretation of this
result is that wage income probably helped farmers enhance their ability to pay for the cost of
either hiring tractor services or acquiring their own equipment for use on the farm.

178

Table 24: Comparison of demographic and agricultural production characteristics of households
using mechanization versus non-users over a three-year period
Variable
2012
2015
Users
Nont-test
Users
Nont-test
users
users
Number of observations
134
8705
***
122
7626
***
Age of household head (years)
50.3
45.4
***
51.9
48.6
**
Male headed households (% yes)
85.1
80.7
82.8
78.7
Education of head (years)
9.4
6.1
***
8.6
5.9
***
Household size (count)
6.5
5.8
***
6.6
6.2
Did the household earn income
53.7
45.9
*
54.9
48.2
from business activities? (% yes)
Did the household earn income
44.0
20.6
***
44.3
28.8
***
from wages? (% yes)
Did the household also use
15.7
38.3
***
27.0
41.2
***
animal draft power for land
preparation? (% yes)
Did the household also use
25.4
67.6
***
38.5
67.7
***
manual labor for land
preparation? (% yes)
Did the household use its own
33.6
0
***
33.6
0
***
tractor for land preparation (%
yes)
Did the household use tractor
64.4
0
***
67.2
0
***
hiring services for land
preparation (% yes)
Area cultivated (hectares)
3.0
2.3
***
3.7
2.4
***
Source: Author’s computation from RALS (2012, 2015)
Significance levels for t-test are as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
Fourth, the average area cultivated by users of mechanization was significantly higher
than that cultivated by non-users in 2012 and 2015. In 2012, the average area cultivated by users
of mechanization was 3 hectares while the area cultivated by non-users was 2.3 hectares. The
average area cultivated by both users and non-users of mechanization increased slightly in 2015
with the former cultivating 3.7 hectares and the latter 2.4 hectares. Although the average area
cultivated by those using mechanization does not seem to be much higher than those not using
mechanization in absolute terms, this finding is the first indication of the likely positive effects of

179

mechanization on area cultivated. This essay returns to this discussion when presenting the
econometric results in section 4.4.2.
4.4.1.5.

Access to and provision of tractor hiring services in selected districts

Both rounds of the RALS (2012 and 2015) did not go into the details of tractor-hiring
services available in the respective communities surveyed across the whole of Zambia. The ACS,
on the other hand, had a specific module that asked respondents questions that included
information about the household’s proximity to tractor-hiring services, whether the household
used nearby tractor-hiring services, and whether the household provided tractor-hiring services in
case of those who owned tractors. Table 25 presents information about access to and provision of
tractor-hiring services by medium-scale farming households in six selected districts of Zambia.
Except for Choma where less than 40 percent of the respondents indicated that they had
tractor-hiring services within or nearby their villages, more than 50 percent of the respondents in
each of the other five districts indicated that they had such services nearby. This result generally
suggests that access to tractor-hiring services was relatively good and in close proximity to
farmers to use conditional on their ability to pay for the service.

180

Table 25: Access to and provision of tractor hiring services among medium-scale farmers in selected districts of Zambia
Variable
District
Chibombo
Choma
Chongwe Kalomo Mpongwe
Tractor hiring services nearby? (% yes)
61.1
37.7
64.3
61.5
51.5
Proportion of households with nearby tractor hiring services
7.6
40.9
3.7
4.2
20
using the services? (% yes)
Distance to the nearest tractor hiring service (km)
5.6
10.3
6.3
5.8
5.1
Owned tractor in the past that is no longer used? (% yes)
3.7
4.9
2.4
5.1
11.1
How was this tractor that is no longer used acquired?
Inheritance (%)
25.0
33.3
100
0
23.1
Purchased using income from sales of own farm
75.0
0
0
50
61.5
produce (%)
Purchased using income from off-farm sources (%)
0
33.3
0
50
15.4
Purchased using a loan obtained through a bank (%)
0
33.3
0
0
0
Own tractor that that is currently used? (% yes)
1.0
3.3
0
2.5
16.2
How was this tractor that is currently used acquired?
Inheritance (%)
0
0
0
0
0
Purchased using income from sales of own farm
0
50
0
1
94.7
produce (%)
Purchased using income from off-farm sources (%)
0
0
0
0
5.3
Purchased using a loan obtained through a bank (%)
100
50
0
0
0
If you own a tractor, do you hire it out? (% yes)
100
50
0
0
42.1
Number of observations
108
61
42
39
117
Source: Author’s computation from ACS (2013)

181

Mumbwa
69.6
30
8.6
15.7
22.2
72.2
5.6
0
5.3
14.3
85.7
0
0
66.7
115

From the perspective of most respondents interviewed during the qualitative field
research, the consensus was that the steady increase in the number of actors providing
mechanization services since 2011 had played a critical in improving availability of and access to
mechanization for farmers. They gave examples of mechanization service providers available in
their communities. One of the community leaders explained it this way: “[access to
mechanization] is on the rise. For example, the ZNFU have started a credit facility where they
give tractors to emergent farmers. The program is working very well.” However, what clearly
came out of the interviews especially with the farmer respondents was that use of mechanization
services was still very low despite improved availability and access. This observation was also
confirmed by the quantitative analysis reported in Table 24. Three of the six survey districts
reported that less than 10 percent of medium-scale farming households who said they lived
within or nearby tractor-hiring services actually used these services.
On further inquiry of why the majority of respondents did not use available tractor hiring
services, the main reason given during the in-depth interviews was that tractor-hiring services
were very expensive. One farmer explained, “Tractor hiring is expensive ….. oxen are readily
available in the area and cheaper than tractor.” In other cases, the respondents stated that the time
they had to wait in order to be serviced by available providers jeopardized their chances of
planting on time given the rain fed nature of agriculture production in Zambia. Therefore, most
decided not to use available tractor-hiring services for fear of losing out on doing other
subsequent farming activities on time.
With the exception of Mumbwa district where about 16 percent of sampled households
stated that they owned a tractor that they currently used, less than 5 percent did so in the other
five districts. On the other hand, hiring out of own tractors did not follow a similar pattern. In

182

some districts, a large proportion of tractor owners hired out their tractors while in others, very
few to none hired out tractors. The study further asked those who did not hire out to explain why
they did not do so. The main explanations given had to do with past experiences of tractors
breaking down when hired out or that the tractors were not in very good condition to be hired
out.
4.4.2. Effects of mechanization on area cultivated and net value of crop production
Before presenting the results for the econometric model used to estimate the effects of
mechanization on the two outcome variables, the explanatory and dependent variables used are
briefly highlighted. Table 26 presents the comparison of means between users and non-users of
mechanization among medium-scale farmers in selected districts of Zambia. The results show
that when users of mechanization are compared with non-users, they were statistically more
likely to have: (1) a higher level of education; (2) a member earning wage income; (3) owned a
tractor in the past that they no longer use; (4) title to their land; (5) hired labor for agricultural
production; (6) relied less on animal draft power for land preparation; (7) more land; (8) more
area under cultivation, and; (9) a higher net value of crop production per hectare.

183

Table 26: Comparison of means between users and non-users of mechanization among mediumscale farmers in selected districts of Zambia
Variable
Means
Users
Non-users
t-test
Age of household head (years)
51.0
50.0
Male headed households (% yes)
86.4
91.8
Education of head (years)
10.5
8.5
***
Number of years settled in the area
32.8
33.2
Household earned income from business activities?
63.6
58.4
(% yes)
Household earned income from wages? (% yes)
43.9
33.7
*
Household owned tractor in the past that it no longer
22.7
6.3
***
uses? (% yes)
Household has title to land? (% yes)
30.3
12.2
***
Household used hired labor for agricultural
86.4
60.6
***
production? (% yes)
Household used animal draft power for land
39.4
96.1
***
preparation? (% yes)
Proportion of area planted to maize1 (%)
75.8
73.9
Proportion of area planted to beans and oilseeds2 (%)
12.2
13.5
Proportion of area planted to other cereals3 (%)
0.2
0.7
4
Proportion of area planted to traditional cash crops
11.0
10.6
(%)
Proportion of area planted to roots and tubers5 (%)
0.8
1.3
Total landholding size (hectares)
92.1
32.6
***
Area cultivated (hectares)
16.9
9.1
***
Net value of crop production per hectare cultivated
1568.5
695.2
***
(Kwacha/hectare)
Number of observations
66
416
Source: Author’s computation from ACS (2013)
Significance levels for t-test are as follows: * p < 0.10, ** p < 0.05, *** p < 0.01
Notes: Crop category/categories: 1 = Maize; 2=beans and oilseeds (groundnuts, soya beans,
mixed beans, Bambara nuts, cowpeas and velvet beans); 3=other cereal grains (sorghum, rice,
millet, popcorn); 4= traditional cash crops (seed cotton, tobacco, coffee and sugarcane); 5=roots
and tubers (sweet potatoes, potatoes and cassava).
4.4.2.1.

Effects of mechanization on area cultivated

Table 27 reports the full information maximum likelihood estimates of the endogenous
switching regression for area cultivated of users of mechanization versus non-users. The
selection equation column reports the estimated coefficients on variables that determine the
household’s decision to use or not to use mechanization. The column for the outcome variable,

184

area cultivated, presents estimated coefficients on variables conjectured to influence the level of
area cultivated conditional on whether or not the farmer used mechanization or not. Moreover,
the dependent variable underwent logarithmic transformation while all the explanatory variables
were in levels.
The main interest of this study was to estimate the effect of mechanization on area
cultivated. For brevity, we do not focus on explaining the ceteris paribus relationships reported
in Table 27 but on the expected value estimates reported in Table 28 as these help to show the
effects of mechanization on area cultivated. Table 28 presents the expected log of area cultivated
in hectares for users versus non-users of mechanization under the actual and counterfactual
conditions. The values of expected log of area cultivated observed in the sample for
mechanization users and non-users are represented by cells (a) and (b) respectively. Cell (c)
represents the area that users of mechanization would cultivate if they did not use mechanization
while and cell (d) represents the expected log of area cultivated for non-users if they had decided
to use mechanization.

185

Table 27: Full information maximum likelihood estimates of the endogenous switching
regression for area cultivated of users of mechanization versus non-users
Explanatory variables
Selection equation
Users
Non-users
Dep. Var: 1 = used
Dep. Var: Log of hectares
mechanization; 0 =
cultivated
otherwise
Age of household head
-0.017
0.012
-0.004
(-1.59)
(0.64)
(-0.90)
Male head (1=yes)
-0.029
0.261
0.827***
(-0.08)
(0.48)
(4.12)
Years in formal education of the
0.022
-0.039
-0.002
household head
(0.73)
(-0.89)
(-0.11)
Formal employment (1=yes)
0.089
0.179
-0.107
(0.38)
(0.45)
(-0.86)
Involved in business activity
-0.105
0.045
0.212*
(1=yes)
(-0.50)
(0.14)
(1.88)
*
***
Total landholding (hectares)
0.002
0.005
0.007***
(1.79)
(3.06)
(5.81)
Owned tractor before that is no
-0.064
0.232
0.378
longer used
(-0.19)
(0.59)
(1.64)
Has title deed (1=yes)
0.559*
0.269
-0.240
(1.65)
(0.62)
(-1.23)
***
Hired labor for agricultural
0.745
0.701
0.294**
production (1=yes)
(2.86)
(1.42)
(2.48)
Annual expected rainfall
-0.005
0.005
-0.011**
(millimeters)
(-0.72)
(0.49)
(-2.54)
Maize area (%)
0.000
0.036
0.006
(0.01)
(0.60)
(0.62)
Beans and oilseeds area (%)
-0.006
0.044
0.006
(-0.29)
(0.73)
(0.62)
Other cereals area (%)
-0.027
-0.024
0.027*
(-0.78)
(-0.22)
(1.73)
Traditional cash crops area (%)
0.006
0.042
0.008
(0.26)
(0.70)
(0.76)
Cost of land preparation per hectare
0.107***
0.030
0.081***
using animal draft power
(ZMW/ha)
(3.12)
(0.61)
(5.15)

186

Table 27 (Cont’d)
Explanatory variables

Cost of land preparation per hectare
using mechanical draft power
(ZMW/ha)
Years in current settlement
Animal draft power (1=yes)
District fixed effects
Constant

Selection equation
Dep. Var: 1 = used
mechanization; 0 =
otherwise
-0.057***

Users
Dep. Var: Log of
hectares cultivated

Non-users

-0.014

-0.046***

(-2.92)
0.012*
(1.85)
-2.373***
(-8.41)
Yes
0.326
(0.07)

(-0.48)

(-5.51)

Yes
-6.139
(-0.75)
0.119
-0.140
66

Yes
8.077***
(2.69)
0.067*
-0.569*
416

Sigma
Rho
N
Source: Author’s computation from ACS (2013)
Note: (1) Wald test of independence of equations: Chi2(19) = 34.64 Prob > chi2 = 0.0154 (2) t
statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01

Table 28: Expected log of hectares cultivated of users of mechanization versus non-users
Decision stage
To use
Not to use
Mean difference
mechanization
mechanization
(within)
Users of
(a) 4.47
(c) 3.60
(TT) 0.87***
mechanization
Non-users of
(d) 3.74
(b) 3.47
(TU) 0.27***
mechanization
Mean difference
(BH1) 0.73
(BH2) 0.13
(TH) 0.60
(across)
Notes: Significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01. TT: the effect of the
treatment (i.e., use of mechanization) on the treated (i.e., farm households that used
mechanization); TU: the effect of the treatment (i.e., use of mechanization) on the untreated (i.e.,
farm households that did not use mechanization); BHi : the effect of base heterogeneity for farm
households that used mechanization (i =1), and did not use mechanization (i =2); TH = (TT TU), i.e., transitional heterogeneity.

187

A typical way of assessing effects of an intervention is to compare the outcome variable
for the actual cases, in this case cell (a) for the adopters and cell (b) for the non-adopters.
However, such a simple comparison could lead to overestimating the effects of an intervention.
Here, it suggests that the area cultivated by farmers using mechanization was 100 percent more
than that for non-users (cell (a) – cell (b)).9 A better approach is to use the mean differences
(treatment effects) reported in the last column of Table 28. Suppose users of mechanization
decided not to use mechanization, they would have cultivated about 87 percent less area (TT =
(c) - (a)). On the other hand, if farmers that did not use mechanization decided to use
mechanization, mechanization would have had a positive effect of 27 percent increase in area
cultivated (TU = (d) - (b)). These results imply that use of mechanization significantly increases
area cultivated by farmers. In addition, the transitional heterogeneity effect is positive (60
percent difference), that is, the effect of mechanization on area cultivated is significantly higher
for those who used mechanization relative to those that did not use it.
One important issue that a researcher needs to account for is the potential for
heterogeneity that makes those farmers using mechanization cultivate larger areas compared to
those not using, irrespective of the availability of mechanization. The results show that
households that used mechanization would significantly increase area cultivated by 73 percent
(BH1 = (a) – (d)) relative to non-users had they (non-users) used mechanization. Moreover, had
the users decided not to use mechanization, they would still cultivate more area (about 13 percent
difference) relative to non-users. These results (BH1 and BH2) suggest existence of unobserved
heterogeneity; there are other unaccounted for factors that make users of mechanization more
likely to cultivate larger areas.
9

Since the expected values are in logs, the difference between two values can be interpreted as
percentage differences.
188

4.4.2.2.

Effect of mechanization on net value of crop production per hectare

Table 29 reports the full information maximum likelihood estimates of the endogenous
switching regression for net value of crop production per hectare of users of mechanization
versus non-users. The selection equation column reports the estimated coefficients on variables
that determine the household’s decision to use or not to use mechanization. The column for the
outcome variable, net value of crop production per hectare, presents estimated coefficients on
variables conjectured to influence the level of net value of crop production per hectare cultivated
conditional on whether or not the farmer used mechanization or not. The dependent variable
underwent logarithmic transformation while all the explanatory variables were in levels.
Similar to the previous estimation of the effects of mechanization on area cultivated, the
focus of discussion in this section is on the expected value estimates. Table 30 presents the
expected log of net value of crop production per hectare cultivated for users versus non-users of
mechanization under the actual and counterfactual conditions. The values of expected log of net
value of crop production per hectare cultivated observed in the sample for mechanization users
and non-users are represented by cells (a) and (b) respectively. Cell (c) represents the net value
of crop production per hectare cultivated that users of mechanization would realize if they did
not use mechanization while and cell (d) represents the expected log of net value of crop
production per hectare cultivated for non-users if they had decided to use mechanization.

189

Table 29: Full information maximum likelihood estimates of the endogenous switching
regression for net value of crop production per hectare of users of mechanization versus nonusers
Explanatory variables
Selection equation
Users
Non-users
Dep. Var: 1 = used
Dep. Var: Log of net value of crop
mechanization; 0 =
production per hectare cultivated
otherwise
Age of household head
-0.017
-0.011
-0.002
(-1.52)
(-0.43)
(-0.18)
Male head (1=yes)
-0.232
0.294
0.305
(-0.67)
(0.41)
(0.74)
Years in formal education of the
0.030
0.032
-0.034
household head
(0.94)
(0.54)
(-1.00)
Formal employment (1=yes)
0.086
-1.422***
-0.020
(0.36)
(-2.70)
(-0.08)
Involved in business activity
-0.081
0.197
-0.019
(1=yes)
(-0.38)
(0.45)
(-0.08)
Total landholding (hectares)
0.003**
-0.002
0.000
(2.08)
(-1.07)
(0.19)
Owned tractor before that is no
-0.083
-0.593
0.234
longer used
(-0.25)
(-1.11)
(0.49)
Has title deed (1=yes)
0.367
-0.474
0.418
(1.15)
(-0.83)
(1.02)
Hired labor for agricultural
0.816***
-0.493
0.530**
production (1=yes)
(2.96)
(-0.72)
(2.18)
Annual expected rainfall
-0.004
0.009
-0.007
(millimeters)
(-0.54)
(0.75)
(-0.77)
Maize area (%)
-0.003
0.144*
0.020
(-0.13)
(1.77)
(1.02)
Beans and oilseeds area (%)
-0.007
0.145*
0.005
(-0.32)
(1.79)
(0.23)
Other cereals area (%)
-0.043
0.110
-0.009
(-1.19)
(0.78)
(-0.29)
Traditional cash crops area (%)
0.001
0.112
-0.017
(0.04)
(1.39)
(-0.79)
Years in current settlement
0.011*
(1.70)
Animal draft power (1=yes)
-2.369***
(-8.17)

190

Table 29 (Cont’d)
Explanatory variables

Cost of land preparation per hectare
using animal draft power
(ZMW/ha)
Cost of land preparation per hectare
using mechanical power (ZMW/ha)
District fixed effects
Constant

Selection equation
Dep. Var: 1 = used
mechanization; 0 =
otherwise

Non-users

0.109***

Users
Dep. Var: Log of net
value of crop
production per
hectare cultivated
-0.122*

(3.04)
-0.058***

(-1.87)
0.061

(-1.46)
0.031*

(-2.84)
Yes
-0.377
(-0.07)

(1.60)
Yes
-1.488
(-0.14)
0.405***
-0.148
66

(1.82)
Yes
16.871***
(2.69)
0.795***
-0.200
416

-0.047

Sigma
Rho
N
Source: Author’s computation from ACS (2013)
Note: (1) Wald test of independence of equations: Chi2(19) = 64.92 Prob > Chi2 = 0.000
(2) t statistics in parentheses, significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01

Table 30: Expected log of net value of crop production per hectare cultivated of users of
mechanization versus non-users
Decision stage
Used mechanization Did not use
Mean difference
mechanization
(within)
Users of
(a) 13.50
(c) 12.28
(TT) 1.22***
mechanization
Non-users of
(d) 14.51
(b) 12.47
(TU) 2.04***
mechanization
Mean difference
(BH1) -1.01
(BH1) -0.19
(TH) -0.82
(across)
Notes: Significance levels as follows: * p < 0.10, ** p < 0.05, *** p < 0.01. TT: the effect of the
treatment (i.e., use of mechanization) on the treated (i.e., farm households that used
mechanization); TU: the effect of the treatment (i.e., use of mechanization) on the untreated (i.e.,
farm households that did not use mechanization); BHi : the effect of base heterogeneity for farm
households that used mechanization (i =1), and did not use mechanization (i =2); TH = (TT TU), i.e., transitional heterogeneity.

191

The last column of Table 30 presents the mean difference or treatment effects within each
category of users or non-users of mechanization. Suppose users of mechanization decided not to
use mechanization, they would have realized about 122 percent less of the net value of crop
production per hectare cultivated ((c) – (a)). On the other hand, if farmers that did not use
mechanization decided to use mechanization, their net value of crop production per hectare
would have increased by 204 percent ((d) – (b)). As a consequence, the transitional heterogeneity
effect is negative (82 percent difference). This implies that the effect of mechanization on net
value of crop production per hectare is significantly lower for those who used mechanization
relative to those that did not use it.
The last row of Table 30 shows results after adjusting for potential heterogeneity in the
sample. We observe that households that used mechanization would realize a lower net value of
crop production per hectare (101 percent less) relative to non-users had they (non-users) used
mechanization. Moreover, had the users decided not to use mechanization, they would realize a
lower net value of crop production per hectare (19 percent difference) relative to non-users.
These results suggest existence of unobserved heterogeneity; there are other unaccounted for
factors that make non-users of mechanization more likely to enjoy net productivity gains.
Overall, our findings point to what has been observed in the literature that mechanization on its
own does not have an unequivocal positive yield effect. There are other important factors that are
clearly linked with productivity enhancement such as use of improved seed, soil nutrient
improvement and others.
4.4.3. Explaining the rise of agricultural mechanization services in Zambia
Results for the third research question are presented in two sections. The first section
highlights the changes in factor price ratios and shows whether these changes could have

192

potentially induced a shift from labor-using to labor-saving technologies. The second section
reports key insights from actors involved in promoting mechanization initiatives and questions
the extent to which the supply side has played a pivotal role in the rise of these initiatives. The
findings are discussed in turn.
4.4.3.1.
Factor price ratios: are they changing enough to induce demand for
mechanization?
The current drive to mechanize has mainly been targeted towards farmers in three
provinces—Central, Eastern and Southern. Thus, this study only reported the changes in factor
price ratios of mechanization rental rates and agricultural wage rates for land preparation over a
four-year period (2012 – 2015) in the three provinces. Figure 13 illustrates the changes in factor
price ratios (adjusted for inflation) computed by dividing median mechanization rental rates by
median agricultural wage rates. Each line represents changes in factor price ratios for each of the
three provinces.
A monotonic downward sloping curve implies that the cost of labor is increasing by a
greater margin relative to mechanization rental rates. This would in turn favor a shift to labor
saving technologies such as tractor use. The opposite scenario is also true that a monotonic
upward sloping curve implies that the cost of renting mechanization is increasing by a greater
margin relative to labor costs. This would maintain the status quo of labor using technologies
such as hand tools.
According to Figure 13, the following findings emerged. First, factor price ratios in the
three provinces did not remain the same over the four-year period. Second, Central province
experienced a monotonic downward sloping trend in factor price ratios suggesting that this could
have induced demand for mechanization in the province. Third, Southern province saw a
marginal increase in the mechanization/labor factor price ratio in the earlier years (2012 – 2013),
193

which would have made mechanization less attractive. However, from 2013 to 2015, the
province experienced a downward trend in factor price ratios, which could have induced demand
for mechanization. Fourth, Eastern province had price ratios favorable to mechanization from
2012 to 2014 but conditions changed in favor of labor using technologies from 2014 to 2015.
In sum, these results suggest that there have been changes in mechanization/labor factor
price ratios that may have worked in favor of a shift to labor saving technologies especially in
the case of Central and Southern provinces. However, one needs to be cautious when interpreting
these results because most small- and medium-scale farming households rely on family labor and
less on hired labor. And since it is difficult to account for the opportunity cost of family labor, it
is unclear whether the computed changes in factor price ratios would bring about an unequivocal

Mechanization rental rate/agricultural
wage rate (inflation adjusted)

shift to mechanization.

3.5
3
2.5
2
1.5
1
0.5
0
2012

2013

2014

2015

YEAR
Central

Eastern

Southern

Figure 13: Changes in factor price ratios for mechanization and agricultural labor (adjusted for
inflation) by selected provinces in Zambia
Source: Author’s computation from CFS (2012 - 2015)

194

4.4.3.2.

Insights from supply side actors and their role in the rise in mechanization

The study interviewed six key actors involved in mechanization service provision
identified through interaction with agricultural industry actors in Zambia and available
evaluation reports availed to the researcher. Three thematic areas were covered during each of
these interviews: (1) drivers of current mechanization initiatives; (2) performance of these
initiatives, and; (3) constraints likely to affect sustainability of these initiatives. The key issues
arising from the interviews are discussed by theme in turn.
4.4.3.2.1.

Drivers of current mechanization initiatives

There was general consensus among the respondents that development works in trends.
One key concept that has been embraced by the private sector, development agencies, equipment
suppliers and financial lending institutions is the shared value initiative. They have adapted this
initiative from Porter & Kramer (2011) who define it as “policies and operating practices that
enhance the competitiveness of a company while simultaneously advancing the economic and
social conditions in the communities in which it operates.”
These actors have recognized that in order for them to generate gains from working with
farmers, they need to be involved in promoting technologies that enhance farmer productivity (a
win-win situation). One of the respondents (representing a cotton agribusiness) indicated that
their motivation for getting into mechanization stems from the fact that alternative methods
(hand tools and animal draft power) introduce inefficiencies: late completion of tasks and high
labor costs. With increased mechanization targeted towards their farmers, the company
envisaged that this would circumvent the aforementioned inefficiencies, increase farmer
productivity and hence increase the company’s profitability.

195

Another important factor that was mentioned by two of the respondents was that smalland medium-scale farmers in parts of Central, Eastern and Southern provinces have begun to
look at farming as a business through the trainings they have received. According to these
respondents, this exposure has contributed to farmers’ expansion of area cultivated therefore
enhancing the need for mechanization. Other factors that were highlighted during the interviews
included the issue of availability of financing options to purchase mechanized equipment, and
increased demand for farm products produced in Zambia both local and internationally.
4.4.3.2.2.

Performance of these initiatives

The interviews with supply side actors revealed divergence of opinions on the
performance of these initiatives, which were evenly split between the six respondents. On the
one hand, some respondents pointed to increase in the number of farmers accessing
mechanization services, improved yields for maize in the areas they operate and the increase in
number of farmer TSPs. On the other hand, some respondents argued that adoption of
mechanization has been low, primarily because the farmers identified to be TSPs do not have a
strong business background to stimulate demand for these services. This latter view conforms to
the findings from the quantitative analysis that showed that mechanization use has remained low
among small- and medium-scale farmers.
4.4.3.2.3.

Constraints

The interviews revealed two main constraints. First, the respondents argued that a number
of TSPs lack business skills needed to provide the service. Given that most of these TSPs receive
the mechanization equipment through lease agreements, the poor business skills directly have an
impact on the TSPs profitability, which in turn affects their ability to pay back the loan for the
equipment.

196

Second, the instability in important economic indicators such as exchange rates and
lending rates has affected demand for equipment. Because tractors and associated equipment are
imported, the pricing structure is denominated in U.S. dollars. From, June 2015 to May 2015, the
Zambian Kwacha depreciated by more than 40 percent against the U.S. dollar. Interest rates
almost doubled from about 23 percent to 40 percent. The respondents strongly thought that these
constraints, if not addressed, would threaten the sustainability of mechanization initiatives.
4.4.3.2.4.

Summary

Overall, the insights from supply side actors shed more light on the developments in
mechanization service provision in Zambia. It was clear from the extensive discussions with the
different actors and review of the literature that those on the supply side are primarily driving
mechanization initiatives in Zambia. In 2012, the total number of TSPs stood at 58 (Aagaard,
2012) while the number of TSPs in 2015 increased to 448 (Kasanda, 2016). Based on
conservative estimates that show that one TSP can provide land preparation services of up to 180
hectares per farming season, the increase in the number of TSPs should have translated to an
increase in area cultivated by TSPs alone of more than eight times. The evidence from the
analysis of household level data suggests otherwise; the level of mechanization use has not
increased correspondingly. It is crucial for their survival that mechanization initiatives should
critically look at how demand for land preparation services could be stimulated besides looking
at only the supply side constraints.

197

4.5. Conclusions and recommendations
Research examining the African agricultural transformation process has focused mainly
on adoption of productivity enhancing inputs such as improved seeds and fertilizers. Yet,
agricultural mechanization and the complementary role it plays in enhancing agricultural
production and productivity has received little research attention in SSA. Recent evidence has,
however, shown that agricultural labor costs in the region have been increasing (a condition that
is necessary for a shift to labor saving technologies). Moreover, development agencies, the
private and public sectors have initiated collaborative efforts designed to enhance mechanization
service provision for agricultural production. Therefore, this essay examined the trends and
effects on agricultural household production of agricultural mechanization and explored the
factors contributing to recent developments aimed at enhancing mechanization in Zambia. The
following were the key findings from this study.
First, the evidence across a three-year period (2012-2015) showed that agricultural
mechanization use by small- and medium-scale farming households has remained low and
stagnant with only 1.5 percent of these households reporting that they used mechanization for
land preparation. When the data where spatially analyzed, provinces with better infrastructure
had a relatively higher proportion of households using mechanization. Other results examining
trends in agricultural mechanization use showed that tractor use in Zambia ranged between 1.2
and 1.3 tractors per one thousand hectares cultivated; a result that demonstrates low use of
mechanization and is consistent with estimates presented by previous studies on tractor use in
SSA. The study found that availability of tractor hiring services was generally good in some
communities but the low use of mechanization was attributed to the high cost of hiring available
services.

198

Second, the study found that use of mechanization for land preparation increased
farmers’ area cultivated by about 60 percent relative to those who did not use mechanization.
This was not a surprising result because mechanization is expected to break the power constraint
associated with more labor-intensive technologies such as hand tools and animal draft power to a
lesser extent. However, the study found that mechanization use did not necessarily translate into
net gains in productivity—measured as net value of crop production per hectare cultivated—for
farmers. The study attributed this to low levels of intensification by farmers in Zambia of
productivity enhancing inputs such as improved seed, fertilizer and others.
Third, from the demand side, the results indicated that the price ratios for rental of
mechanization relative to labor costs had changed over time in favor of a shift to labor saving
technologies. This was particularly true in two of the three regions where mechanization
initiatives have been actively promoted. But because most small- and medium-scale farming
households rely on family labor and less on hired labor, this could explain why demand for
mechanization had not correspondingly increased. The study established that mechanization
initiatives in Zambia have mainly been driven by the supply side. The study found that
constraints identified during the implementation of existing mechanization initiatives and the low
demand for mechanization services are likely to impinge on the sustainability of these initiatives.
Based on these key findings, we offer four sets of recommendations. First, mechanization
initiatives should be targeted in regions where there is scope for expansion. From our findings,
Central and Copperbelt provinces are the two leading provinces in terms of mechanization use.
The two regions have also demonstrated that mechanization use has increased over the three-year
period between 2012 and 2015. Therefore, promoters of mechanization should streamline their
activities by targeting such regions with relative potential.

199

Second, promoters of tractor services need to promote them alongside productivity
enhancing inputs. This is because mechanization does not necessarily translate into improved
productivity.
Third, training of farmers through private driven mechanization initiatives should be
facilitated alongside the widespread government extension service system. It is commendable
that the Zambian government has not played a leading role in terms of providing mechanization
services to small- and medium-scale farmers. Experiences from the past have shown that
government driven mechanization services failed and could not be sustained. But training is one
area where public intervention could be vital.
As noted by one of the key informants, farmers continue to grapple with completing
activities on a timely basis, are not adequately trained on the aspects of minimizing costs of
production and rarely get updates on improved technologies likely to enhance productivity.
Therefore developing a mechanization-training program that involves the government extension
service is likely to stimulate demand among farmers for existing mechanization services.
Research in Zambia has shown that close to 60 percent of the government’s allocation to
agriculture goes to supporting input and output subsidy programs that are maize centric.
Unfortunately, very little is allocated for strengthening of the extension service (an important
department for farmer training), research and development. Reallocating some of these funds to
farmer training might help to stimulate farmer demand and engender improvements in
production (stemming from cropland expansion) and ultimately help reduce poverty.
The fourth recommendation arose from the discussions with different actors involved on
the supply side of mechanization initiatives. From the perspective of most of these actors, they
acknowledge that they have not done enough to generate demand for mechanization services by

200

small- and medium-scale farmers in the areas they operate. They argued that unlike programs
related to promotion of other productivity enhancing inputs such as improved seeds and
fertilizers that are quickly embraced by a reasonable proportion of farmers, mechanization
initiatives and therefore use of available services take time. To ensure effectiveness of these
initiatives, efforts by supply side actors should: (1) target potential tractor service providers that
already have a business background; (2) continue to enhance business training of identified
service providers, and; (3) work closely with the government extension system to stimulate
demand among farmers who are likely to benefit from mechanization initiatives.

201

REFERENCES

202

REFERENCES

Aagaard, P. (2012). Small and medium scale agriculture and mechanisation. Lusaka, Zambia:
Conservation Farming Unit (CFU).
ACF. (2012). Tractor mechanization of conservation agriculture - Remodelling study report.
Agricultural Consultative Forum (ACF).
Binswanger, H. P. (1978a). Induced Technical Change: Evolution of Thought. In H. P.
Binswanger & V. W. Ruttan, Induced Innovation: Technology, Institutions, and
Development. (pp. 13–43). Johns Hopkins University Press.
Binswanger, H. P. (1978b). The economics of tractors in South Asia. Retrieved from
http://oar.icrisat.org/id/eprint/630
Binswanger, H., & Pingali, P. (1988). TECHNOLOGICAL PRIORITIES FOR FARMING IN
SUB-SAHARAN AFRICA. The World Bank Research Observer, 3(1), 81–98.
https://doi.org/10.1093/wbro/3.1.81
Burke, W. J., Hichaambwa, M., Banda, D., & Jayne, T. S. (2011). The Cost of Maize Production
by Smallholder Farmers in Zambia (Working Paper No. No. 50). Lusaka, Zambia: Food
Security Research Project. Retrieved from
http://saipar.org:8080/eprc/handle/123456789/207
Di Falco, S., Veronesi, M., & Yesuf, M. (2011). Does Adaptation to Climate Change Provide
Food Security? A Micro-Perspective from Ethiopia. American Journal of Agricultural
Economics, 93(3), 829–846. https://doi.org/10.1093/ajae/aar006
Diao, X., Cossar, F., Houssou, N., & Kolavalli, S. (2014). Mechanization in Ghana: Emerging
demand, and the search for alternative supply models. Food Policy, 48, 168–181.
https://doi.org/10.1016/j.foodpol.2014.05.013
Fugile, K., & Rada, N. (2013, May 6). USDA Economic Research Service - Research Raises
Agricultural Productivity in Sub-Saharan Africa. Retrieved March 27, 2014, from
http://www.ers.usda.gov/amber-waves/2013-may/research-raises-agriculturalproductivity-in-sub-saharan-africa.aspx#.UzRXIF660oY
Hayami, Y., & Ruttan, V. W. (1970). Factor prices and technical change in agricultural
development: The United States and Japan, 1880-1960. Journal of Political Economy,
78(5), 1115–1141.
Hayami, Y., & Ruttan, V. W. (1985). Agricultural development: an international perspective.
Johns Hopkins University Press.

203

Houssou, N., & Chapoto, A. (2014). The changing landscape of agriculture in Ghana: Drivers of
farm mechanization and its impacts on cropland expansion and intensification. Retrieved
from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2539595
Houssou, N., Diao, X., Cossar, F., Kolavalli, S., Jimah, K., & Aboagye, P. O. (2013).
Agricultural Mechanization in Ghana: Is Specialized Agricultural Mechanization Service
Provision a Viable Business Model? American Journal of Agricultural Economics, 95(5),
1237–1244. https://doi.org/10.1093/ajae/aat026
Jayne, T. S., Chamberlin, J., Traub, L., Sitko, N., Muyanga, M., Yeboah, F. K., … Kachule, R.
(2016). Africa’s changing farm size distribution patterns: the rise of medium-scale farms.
Agricultural Economics, 47(S1), 197–214. https://doi.org/10.1111/agec.12308
Jonsson, L.-O. (1985). Agricultural Mechanization. In Structural readjustments and Zambia’s
self-reliance in farm equipment (pp. 25–29). Lusaka, Zambia: International Labour
Office.
Kasanda, E. (2016). Email message to author.
Lokshin, M., & Sajaia, Z. (2004). Maximum likelihood estimation of endogenous switching
regression models, 4(3), 282–289.
Maddala, G. S. (1983). Limited-dependent and qualitative variables in econometrics. New York:
Cambridge University Press.
Minten, B., & Barrett, C. B. (2008). Agricultural Technology, Productivity, and Poverty in
Madagascar. World Development, 36(5), 797–822.
https://doi.org/10.1016/j.worlddev.2007.05.004
Mrema, G. C., Baker, D. C., & Kahan, D. (2008). Agricultural mechanization in Sub-saharan
Africa: time for a new look. Rome: Food and Agriculture Organization on the United
Nations.
Pingali, P. (2007). Chapter 54 Agricultural Mechanization: Adoption Patterns and Economic
Impact. In R. Evenson and P. Pingali (Ed.), Handbook of Agricultural Economics (Vol.
Volume 3, pp. 2779–2805). Elsevier. Retrieved from
http://www.sciencedirect.com/science/article/pii/S1574007206030544
Pingali, P. L., Bigot, Y., & Binswanger-Mkhize, H. P. (1987). Agricultural mechanization and
the evolution of farming systems in sub-Saharan Africa. Baltimore: Published for the
World Bank, Johns Hopkins University Press.
Porter, M. E., & Kramer, M. R. (2011). Creating Shared Value. Harvard Business Review,
89(1/2), 62–77.
Ricker-Gilbert, J., Jayne, T., & Shively, G. (2013). Addressing the “Wicked Problem” of Input
Subsidy Programs in Africa. Applied Economic Perspectives and Policy, 35(2), 322–340.
https://doi.org/10.1093/aepp/ppt001

204

Ruttan, V. W. (2002). Productivity Growth in World Agriculture: Sources and Constraints. The
Journal of Economic Perspectives, 16(4), 161–184.
Sitko, N., & Chamberlin, J. (2015). The Anatomy of Medium-Scale Farm Growth in Zambia:
What Are the Implications for the Future of Smallholder Agriculture? Land, 4(3), 869–
887. https://doi.org/10.3390/land4030869
Sitko, N. J., & Jayne, T. S. (2012). The Rising Class of Emergent Farmers: An Effective Model
for Achieving Agricultural Growth and Poverty Reduction in Africa? (Working Paper No.
69). Lusaka, Zambia: Indaba Agricultural Policy Research Institute (IAPRI). Retrieved
from http://www.aec.msu.edu/fs2/zambia/wp69.pdf
Strauss, A., & Corbin, J. (1994). Grounded Theory Methodology: An Overview. In N. K. Denzin
& Y. S. Lincoln (Eds.), Handbook of Qualitative Research (pp. 273–285). SAGE
Publications.
Takeshima, H., Nin--Pratt, A., & Diao, X. (2013). Mechanization and Agricultural Technology
Evolution, Agricultural Intensification in Sub-Saharan Africa: Typology of Agricultural
Mechanization in Nigeria. American Journal of Agricultural Economics, 95(5), 1230–
1236. https://doi.org/10.1093/ajae/aat045
Takeshima, H., & Salau, S. (2010). Agricultural mechanization and the smallholder farmers in
Nigeria (Policy note No. 22). Abuja, Nigeria: International Food Policy Research
Institute (IFPRI). Retrieved from
http://www.ifpri.org/sites/default/files/publications/nssppn22_0.pdf
Tembo, G. (1985). Agricultural Mechanisation Policy. In Structural readjustments and Zambia’s
self-reliance in farm equipment (pp. 31–33). Lusaka, Zambia: International Labour
Office.
Verma, S. R. (2008). Impact of agricultural mechanization on production, productivity, cropping
intensity income generation and employment of labour. Status of Farm Mechanization in
India, Punjab Agricultural University, Ludhiana, 133–153.
Yang, J., Huang, Z., Zhang, X., & Reardon, T. (2013). The Rapid Rise of Cross-Regional
Agricultural Mechanization Services in China. American Journal of Agricultural
Economics, 95(5), 1245–1251. https://doi.org/10.1093/ajae/aat027

205

CHAPTER 5: CONCLUSIONS
This dissertation explored an important issue that has re-emerged in the development
discourse—farm structure change—and whether this change represents an opportunity for
countries in SSA to address the myriad of challenges confronting agricultural development. The
general focus of scholars studying farm structure change in the current agricultural development
literature has largely been on the development implications of large-scale land acquisitions of
foreign origin (typically own thousands of hectares of land but cultivate more than 100 hectares)
and less on farm structure change associated with the rise of domestic medium-scale farmers
(landholding between 5 and 100 hectares). While foreign driven large-scale land acquisitions
have played a role in altering the agricultural landscape in SSA, the increase in pace at which
land is being acquired by domestic land investors has completely changed the farm structure
outlook. As a result, the domestic land investor dynamic has ignited interest in the development
discourse to understand whether or not this new phenomenon signals an alternative pathway for
achieving agricultural development in parts of SSA.
In this light, this dissertation examined farm structure change associated with domestic
medium-scale investor farms and its implications for agricultural development in sub-Saharan
Africa. The study had three specific objectives: (1) to investigate the causes and consequences of
the rise of domestic medium-scale investor farms; (2) to re-examine the inverse farm sizeproductivity (IR) relationship hypothesis in the context of farm structure change, and; (3) to
explore the trends and effects on agricultural production of agricultural mechanization use and
determine what is driving the rise in mechanization initiatives. The study was conducted in
Zambia because the country has been experiencing major changes in farm structure, the most

206

salient of which is a major increase in cultivated area under the control of farms between 5 and
100 hectares.
This dissertation makes three key contributions to literature on agricultural development.
First, it adds to the emerging literature on farm structure change by extending the depth of
understanding related to the causes and consequences of farm structure change associated with
medium-scale investor farms in SSA. This is achieved through a mixed-methods analytical
framework that combines political economy insights and the new institutional economics
paradigm. Second, it proposes and empirically implements a methodological approach not fully
exploited in studies examining the inverse farm size-productivity hypothesis in SSA. The
definition of productivity is extended beyond land productivity to include alternative measures
and the hypothesis is tested on a sample with farm size data outside the zero to ten hectare range;
the latter is done to address recent criticism of the IR literature that has extrapolated findings
beyond the ten-hectare range. The constant-returns-to-scale assumption upheld in most IR
empirical work is also relaxed in this study. Third, this dissertation research contributes to the
emerging literature aimed at guiding policy of mechanization in SSA.
We discuss the key findings of this research in line with each main objective. Five key
findings emerged from the first objective. First, the study identified three main drivers of farm
structure change: (1) a positive change in society’s perception of agriculture; (2) a change in
enforcement of informal rules of customary land governance; and; (3) unintended consequence
of public spending in agriculture. In terms of change in society’s perception of agriculture, for
example, the consensus among the key informants was that perception had changed from looking
at agriculture as a domain for those who are poor to a domain that is viewed as a profitable
venture for anyone with interest in agriculture and has adequate financial resources. In terms of

207

change in the governance of customary land, the main issue that was raised by the key
informants was that informal land markets had been allowed to take root contrary to established
norms. They indicated that the development of the informal market for land had significantly
contributed to individual land acquisitions especially between 10 and 50 hectares in rural areas
closer to urban centers of Zambia. With respect to the unintended consequence of public
spending in agriculture, the general consensus among the key informants was that the two main
subsidy programs in agriculture—the input support program (FISP) and the output subsidy
initiative (FRA)—had continued to operate under numerous implementation challenges and
hence distorting the incentive structure. Put together, all the three drivers identified by this study
had contributed to the rise of medium-scale farms and change in farm structure in Zambia.
Second, the study reinforced the finding by previous studies that the change in farm
structure had been characterized by expansion of medium-scale farms and land concentration in
Zambia. In particular, the study found that the change in the number of farms of five hectares
and above had been more rapid than farms below five hectares between 2001 and 2015. Further,
the study also showed that average landholding decreased the most for the smallest category of
farms examined (less than 2 hectares) while the largest category (20-100 hectares) saw an
increase in average landholding of more than 10 percent at national level.
In addition to the characterization of farm structure change, the study found that
expansion of medium-scale farms in Zambia had not only been by individuals who are local to
the community but also individuals migrating from other parts of the country. Moreover,
individuals with access to wage income and urban-based households had also driven the
expansion in medium-scale farms and that this surge in land acquisitions had gathered
momentum in the last 20-25 years.

208

Third, while the analysis showed that medium-scale farms cultivated relatively large
portions of land, the study on the other hand showed that they left significant portions of land
idle. This has implications for prospects of small-scale agricultural expansion especially when
recent evidence has clearly shown that small-scale farmers face serious land constraints. Further,
the results from both the quantitative and qualitative analysis indicated that the agricultural
sector had become maize centric in the context of farm structure change.
Fourth, the study found that titling of agricultural land had remained relatively low and
was more likely to be taken up by farmers who fall within the medium-scale farmer category.
Moreover, where traditional leaders had put in place traditional landholding certification as a
way of reducing land disputes in customary land, households with relatively large landholdings
were more likely to obtain these certificates. The study linked the low levels of titling to lack of
awareness of the process of obtaining title, traditional leaders unwillingness to allow conversion
of customary land to titled land and the bureaucracy associated with processing titles. The study
also found that land rental and sales markets have begun to emerge in customary land areas and
had been a consequence of the involvement of those with the financial capacity to induce
transactions reminiscent of a market based system.
Fifth, the study found that although more than 80 percent of farmers in Zambia fall within
the small-scale farming category, medium- and large-scale farms had steadily increased their
share of national crop production especially non-maize crop production. In terms of crop
marketing (maize and non-maize crop sales), small-scale farms continued to contribute to
national marketed output. However, the high level of participation in marketing of all crop types
by medium- and large-scale farming households suggested that relatively large farms had a more
important role in Zambia’s agricultural growth and development.

209

The key findings from the second objective were as follows. First, when the IR
hypothesis was tested using representative data with about 15 percent of farms with landholding
sizes of 20 hectares or more, the results were not uniform across the five measures of
productivity used in the study. The relationship between farm size and productivity upheld the
IR, rejected the IR or was uncorrelated depending on the measure of productivity. Results
showed that when net value of crop production per hectare was regressed on farm size ceteris
paribus, the IR was upheld at 10 percent significance level. This modest level of significance
could be attributed to the fact that, unlike in past studies where input use intensity has been found
to be higher on small relative to large farms, the opposite was true in the case of this study.
When the outcome variable was a labor productivity measure, the study found a strong
positive relationship between farm size and labor productivity. In the case of cost of maize
production per metric ton produced, the study found that the cost of maize production per ton
increased with maize area planted of up to 10 hectares. This result upheld the IR because small
farms were able to produce a ton of maize at a cheaper cost than large farms, but only for the 0 –
10 hectares domain. The estimation results for TFP showed that this comprehensive measure of
productivity was negatively associated with operated farm size albeit not statistically significant.
Second, the study found that differences in crop management practices played a key role
in driving productivity differences between small- and medium-scale farms. Management
practices that were prominent in explaining productivity differences were total inorganic
fertilizer use, herbicide use, animal draft power use and the nature of land allocation to different
crop types. Third, relaxing the constant returns to scale assumption strengthened the IR
hypothesis (in the case of net value of crop production per hectare and TFP) or weakened the

210

argument that large farms are more productive than small farms (in the case of the two labor
productivity measures).
The last objective had three main findings. First, the evidence across a three year period
(2012-2015) showed that agricultural mechanization use by small- and medium-scale farming
households had remained low and stagnated with only 1.5 percent of these households reporting
that they used mechanization for land preparation. When the data where spatially analyzed,
provinces with better infrastructure had a relatively higher proportion of households using
mechanization. Other results examining trends in agricultural mechanization use showed that
tractor use in Zambia ranged between 1.2 and 1.3 tractors per one thousand hectares cultivated; a
result that demonstrates low use of mechanization and is consistent with estimates presented by
previous studies on tractor use in SSA. The study found that availability of tractor hiring services
was generally good in most of the communities but the low use of mechanization was attributed
to the high cost of hiring available services.
Second, the study found that use of mechanization for land preparation increased
farmers’ area cultivated by about 60 percent relative to those who did not use mechanization.
This was not a surprising result because mechanization is expected to reduce the drudgery
associated with more labor-intensive technologies such as hand tools and animal draft power to a
lesser extent. However, the study found that mechanization use did not necessarily translate into
net gains in productivity—measured as net value of crop production per hectare cultivated—for
farmers. The study attributed this to low levels of intensification by farmers in Zambia of
productivity enhancing inputs such as improved seed, fertilizer and others.
Third, on the demand side, the results indicated that the price ratios for rental of
mechanization relative to labor costs had changed over time in favor of a shift to labor saving

211

technologies. This was particularly true in two of the three regions where mechanization
initiatives have been actively promoted. But because most small- and medium-scale farming
households rely on family labor and less on hired labor, this probably explains why demand for
mechanization has not correspondingly increased. The study established that mechanization
initiatives in Zambia have mainly been driven by the supply side. This is because the different
actors on the supply side have embraced the principles of shared values that entail a win-win
situation for both their clients (farming community) and themselves as service providers.
What do these findings imply for policy? The first study demonstrates that there are a
number of political economy factors that have contributed to the change in farm structure. The
consequences of this change are that agricultural growth and subsequently development have not
been widespread. Policy strategies should address not only the inequalities that exist in terms of
land access and ownership by small-scale farmers but also address the bottlenecks associated
with obtaining title or indeed traditional landholding certificates. Although there is considerable
debate on the link between agriculture productivity and tenure status, this paper demonstrates
that the process of obtaining title is mainly being exploited by farmers who are categorized as
medium-scale farmers. If small-scale farmers are not part of this process, there is a danger that,
with time, some households may become landless. To enhance small-scale farmers’ role in
agricultural markets, policy should also endeavor to promote research and development
initiatives that would uplift their production levels and generate surplus for markets. Obviously,
this strategy should not be maize centric but should extend to other crops or livestock suitable for
small-scale agriculture.
Findings on the relationship between productivity and operated farm size, while
important, should not be the decisive factor in guiding agricultural development and land policies

212

in SSA because there are many other important considerations. The challenges of rural poverty
that continue to bedevil SSA are better addressed by pursuing broad based development
strategies whose central focus should remain smallholder farmers. This entails promoting
policies that would enhance productivity growth among smallholders, as this is crucial to poverty
reduction and development. Moreover, an inclusive smallholder form of agricultural growth will
clearly generate employment growth in both farm and non-farm sectors than a comparable rate of
agricultural growth concentrated among a few large farms.
Based on the findings of the third study that demonstrate that mechanization use has
remained low and that the supply side has identified constraints that are likely to undermine the
sustainability of recent efforts to enhance mechanization, this research offers the following
recommendations. First, the study recommends policy action that is geared towards enhancing
provision of training to farmers on the benefits of using technologies such as mechanization for
land preparation. Farmers continue to grapple with completing farming activities on a timely
basis, are not adequately trained on the aspects of minimizing costs of production and rarely get
updates on improved technologies likely to enhance productivity. Therefore developing a
mechanization-training program that involves the government extension service could assist with
stimulating demand among farmers for existing mechanization services. Research in Zambia has
shown that close to 60 percent of government’s allocation to agriculture goes to supporting input
and output subsidy programs that are maize centric. Reallocating some of these funds to farmer
training might help to stimulate farmer demand and engender improvements in production
(stemming from cropland expansion) and ultimately help reduce poverty.
Second, to ensure effectiveness of mechanization initiatives, efforts by supply side actors
should: (1) target potential tractor service providers that already have a business background; (2)

213

continue to enhance business training of identified service providers, and; (3) work closely with
the government extension system to stimulate demand among farmers who are likely to benefit
from mechanization initiatives.

214