.:

in:

>3».

“

um".

.
a.
4‘

fin

i

,3...

”vi—J

a.

 

Em

.
1.... .

1:

dry.”

7..

.12

1.

t
c

,r.

l!!-
2 31.1: I.
t. I: . I .
vital
[:3 K.
.IO .3. .a-

“...hw". . 1w

.
:2.
5)....

STE;

 

 

L # "mt“tht‘tflflttflittw
Michigan State .
; University

 

 

This is to certify that the

dissertation entitled

IMPACTS OF TSETSE CONTROL ON
MIGRATION AND CAPITAL ACCUMULATION:
ZAMBEZI VALLEY, ZIMBABWE

presented by

JONES GOVEREH

has been accepted towards fulfillment
of the requirements for

Ph.D degree in Ag . Economics

 

MMMJM

Major professor

Date MGVCk 22, ’17?

 

MSU i: an Affirmative Action/Equal Opportunity Institution 0—12771

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IMPACTS OF TSETSE CONTROL ON MIGRATION AND
CAPITAL ACCUMULATION: ZAMBEZI VALLEY, ZIMBABWE

By

Jones Govereh

A DISSERTATION

Submitted to
Michigan State University
in partial fulfilment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

1999

ABSTRACT

ILIPACTS OF TSETSE CONTROL ON MIGRATION AND
CAPITAL ACCUMULATION: ZAMBEZI VALLEY, ZIMBABWE

By

Jones Govereh

Tsetse control is ordinarily considered as simultaneously responsible for motivating
household emigration from those districts bordering the tsetse cleared front as well as
household immigration into cleared areas. The direct contribution of tsetse control to the
process of internal migration in Zimbabwe, holding constant the historical imbalance in
land access, regional difl'erences in the development of infi’astructure services and markets
for commercial enterprises, is unknown. This study modeled annual growth in immigration
at the village level and obtained settler’s reasons for emigrating from their original
settlements. The direct and indirect contributions of tsetse and trypanosomosis controls to
livestock adoption in tsetse cleared areas is also not well understood. The study modeled
growth in livestock adoption at the village level and investigated whether adoption began
with or was accelerated by tsetse control. In addition, a production function model
evaluated the effects of animal traction use on farm level factor productivity. Finally, the
study also analyzed the distribution, extent and determinants of settler wealth

accumulation and established the relative success of settlers.

This study was conducted in Gokwe North District, Zimbabwe in 1996. The survey was
conducted in three cluster sites: an early cleared (1965 - 1976) area; mid-period

controlled (1978 - 1987) area; and a recently cleared (1982 - 1994) area. All sites had

uniform drainage, soil and vegetation characteristics. Within each cluster, two wards were
selected and in each ward, two villages were selected based on access to infi'astructure

services. A random sample of 40 households was selected from each village population.

Historical and regional land distribution patterns were responsible for household
emigration from other districts contiguous and non-contiguous to Gokwe North. The
relative unavailability of economic fanning opportunities in old and well established areas
led to emigration of young and growing families. The relatively unsettled alluvial plains of
Gokwe North made it an attractive destination for immigrants whether there was or there
as no tsetse flies. While tsetse control did not affect emigration, tsetse control directly '
afi‘ected immigration of households who moved into Gokwe North with livestock. The
introduction of cattle and donkeys in Gokwe North was partly due to trypanosomosis
control and in-migration. The pathways towards improved productivity were different
between hand tillers, draft owners and draft renters. Use of animal traction enabled
households to utilize additional amounts of land and labor. Wealth accumulation was
influenced by the life cycle phenomenon as predicted in theory but was unexpectedly not
impacted by the length of residential tenure. The presence of tsetse flies eliminated the
premiums associated with early arrival. Early arrival afier tsetse control was the best

timing for establishing residence.

DEDICATION

This study is dedicated to all those whom I love from the bottom of my heart
most of all, my parents.
I thank you for the love that you always have had for me.
I give that love to you in return.
I miss you.

Rest In Peace.

iv

ACKNOWLEDGMENTS

I give thanks and praises to the Almighty for her continued blessings. Thank you to those
rural villagers and professors alike. I benefitted immensely from the assistance of many
people, in particular, I am indebted to Dr. Allan Schmid my academic advisor and
dissertation committee chair, Dr. Brent Swallow my field supervisor and my other
committee members, Dr. Tom Reardon, Dr. Thorn Jayne and Dr. Warren Samuels. Their
counsel and support in my academic development is greatly appreciated. They are the

reason I am where I am today.

Special thanks to Dr Rob Eley for administering over the award of my research fellowship
at International Livestock Research Institute. His unwavering encouragement and
personal development is appreciated. The field research on which this study is based was
primarily financed by the International Fund for Agricultural Development through a
Technical Assistance Grant to the International Livestock Research Institute. In
Zimbabwe, the research was generously supported by the Tsetse and Trypanosomosis
Control Branch of the Department of Veterinary Services (TTCB), the Regional Tsetse
and Trypanosomosis Control Programme (RTTCP) and the Department of Agricultural
Economics at the University of Zimbabwe (UZim). Thanks to the staff at TTCB and
RTTCP in Harare and Gokwe. The author is grateful for information, assistance and

advice provided in Zimbabwe by V. Chadenga, W. Shereni, R. Connor, M. Doran, R.

Mabeza-Chimedza and K. Muir-Lereshe, at the International Livestock Research Institute

by R Reid, B. Perry, and G. d’Ieteren.

Success in survey work is owed to the diligent work of field assistants: Peter Mureveri;
Reason Majoto; Journey Sithole; Jabulani Mbano; Jane Motsi; Munyanyu Ngwenya;
Gorosi Chikubwe; Killian Farikayi; Douglas Machau Mudukuti; Edna Siziba; Moses
Muzanago and Svinurai Makoni. Thanks to the Sithole family for hosting me. I am also
grateful for the help provided in data entry and cleaning by Florence and Simba Chimedza,

Nester Chitepo and Lorraine. The research assistants handled their business well.
Finally, I owe a sincere debt to my wife, Annah Gasseler my daughter Shingi and son

Tatenda. Keep your heads up. Thank you family for being my greatest inspiration.

Everyone whom I forgot, you know I thank you.

vi.

TABLE OF CONTENTS

5991199 Page
ABSTRACT ......................................................... ii
ACKNOWLEDGMENTS ............................................... v
TABLE OF CONTENTS .............................................. vii
LIST OF TABLES .................................................... x
LIST OF FIGURES ................................................... xii
CHAPTER 1: TSETSE, TRYPANOSOMOSIS AND AGRICULTURAL
DEVELOPMENT ............................................... 1
1.1 Tsetse and Trypanosome Species ................................. 1
1.2 Tsetse and Trypanosomosis Control Profiles ........................ _ 2
1.2.1 Tsetse Control ........................................ 2
1.2.2DiseaseControl...................: ............... ....5
1.2.3 Regulating Livestock Movements ......................... 6
1.3 Impacts of Tsetse and Trypanosomosis Controls ..................... 6
1.3.1 Statement of Problem .................................. 6
1.3.2 Purpose of Study ..................................... 10
1.3.3 Objectives .......................................... 11
1.3.4 Hypothesis .......................................... 12
1.4 Dissertation Outline .......................................... 14
CHAPTER 2: RESEARCH DESIGN AND INSTRUMENTS .................. 16
2.0 Introduction ................................................ 16
2.1 Study Location and Physical Characteristics ........................ 16
2.2 Survey Design .............................................. 17
2.3 Sites and Household Selection ................................. 20
2.4 Tsetse and Trypanosomosis Control Operations in Cluster Sites ......... 21
2.5 Survey Instruments .......................................... 22
CHAPTER 3: THE IMPACTS OF TSETSE CONTROL ON IMMIGRATION ..... 24
3 .0 Introduction ................................................ 24
3.1 Modeling Immigration Stocks and Flows .......................... 25
3.1.1 Background and Relevant Literature ...................... 25
3.1.2 Foundations of Immigration Grth Model ................. 28
3.1.3 Specification of the Empirical Model ...................... 30
3.2 Extent of Immigration in Study Sites ............................. 33

vii.

3.3 Relative Impacts of Tsetse Control on Immigration .................. 41

3 .4 Importance of Tsetse Control and other Amenities ................... 44
3.5 Summary .................................................. 47
CHAPTER 4: FARMING SYSTEMS CHANGE WITH TSETSE CONTROL AND
IMMIGRATION ............................................... 49
4.0 Introduction ................................................ 49
4.1 Adoption of Animal Traction Technology .......................... 49
4.2 Estimating Rate of Livestock Adoption ........................... 52
4.3 Animal Traction and Factor Productivity .......................... 54
4.4 Estimating Impacts of Animal Traction ............................ 59
4.5 Adoption Trends and Characteristics of Farmer Adopters .............. 60
4.6 Characteristics of Households Using Different Tillage Systems .......... 63
4.7 Grth in Cattle Adoption ..................................... 66
4.8 Production function analysis .................................... 69
4.8.1 Elasticities: ......................................... 70
4.8.2 Returns to Scale: .................................... 71
4.9 Ownership versus Hiring Animal Traction ......................... 72
4.9.1 Marginal Productivity ................................. 72
4.9.2 Marginal Rate of Technical Substitution .................... 73
4.9.3 Cross-Productivity Efl‘ects .............................. 74
4.8.4 Technical Efliciency ................................... 76
CHAPTER 5: DETERMINANTS OF WEALTH ACCUMULATION AND
SOCIOECONOMIC MOBILITY .................................. 78
5.0 Introduction ................................................ 78
5.1 Settlement fi'ontiers as safety nets ................................ 78
5.2 Distribution of Wealth ........................................ 81
5.3 Association of Wealth Accumulation with Demographic Characteristics . . . 83
5.4 Direction and Magnitude of Socioeconomic Mobility ................. 88
5.4.1 Equipment mobility ................................... 88
5.4.2 Livestock Wealth Mobility .............................. 92
5.5 Determinants of Per Capita Wealth ............................... 96
5.5 Determinants of Livestock and Equipment Asset Mobility ............ 100
CHAPTER 6: DISCUSSION AND CONCLUSIONS ........................ 105
6.1 Tsetse Control and Migration .................................. 105
6.2 Farming System Change With Tsetse Control and Immigration ......... 107
6.3 Wealth Distribution and its Determinants ......................... 110
6.4 Limitations of the study ...................................... 112
6.5 Areas for Further Research .................................... 114

viii

BIBLIOGRAPHY ................................................... 1 16

Ix

LIST OF TABLES

Table 3.1: Distribution of first and second generation immigrants and indigenous residents
in Gokwe North, Zimbabwe, 1996 .................................. 34

Table 3.2: The Inflow of Immigrants into Villages 1965 - 1995, Gokwe North, Zimbabwe,
1996 ........................................................ 35

Table 3.3: Provinces where Migrants Emigrated, Gokwe North, Zimbabwe, 1996 . . . . 36

Table 3.4: Reasons Why Migrants Choose to Emigrate From Their Areas of Origin
Before, During and After Tsetse Control into Gokwe North, Zimbabwe, 1996 . 37

Table 3.5: Frequency of Sequential Migration “Within and Between Wards Before, During
and After Tsetse Control, Zimbabwe, 1996 ..................... , ..... 3.8

Table 3.6: Sequence of Establishing Services in Study Sites, Gokwe North, Zimbabwe,
1996 ........................................................ 40

Table 3.7: Determinants of the stock of first-generation migrants in 12 villages of Gokwe
North District, Zimbabwe, between 1963 and 1995 ..................... 42

Table 3.8: Reasons Why Immigrants Perceived Tsetse Challenge as or not Important
When They Chose a Settlement Site in Gokwe North Before, During and After
Tsetse Control, Zimbabwe, 1996 ................................... 45

Table 3.9: Reasons why immigrants perceived infrastructure to be important or not
important when they chose a settlement site in Gokwe North before, during and
after tsetse control, Zimbabwe, 1996 ................................ 47

Table 4.1 Description of Variables Included in Models ......................... 53

Table 4.2 The Number of Additional Cattle Adopters in Villages 1963 - 1995, Gokwe
North, Zimbabwe, 1996' ......................................... 61

Table 4.4: Percentages of pioneer and newcomer households owning livestock at time of
Survey in Gokwe North District, Zimbabwe, 1995/96 ................... 62

Table 4.5 Major sources of finance for acquiring livestock, Gokwe North District,
Zimbabwe, 1995/96 ............................................. 63

Table 4.6: Characteristics of Farming Systems by Type of Tillage, Gokwe North District,
Zimbabwe, 1995/96 ............................................. 65

Table 4.7: Results of Econometric Models of the Flow of Immigrants and Cattle
Adopters, Gokwe North District, Zimbabwe, 1965 -96 .................. 68

Table 4.8: Results of Cobb-Douglas Model of Value of Aggregate Crop Output, Gokwe
North District, Zimbabwe, 1995/96 ................................. 70

Table 4.9: Value of Marginal Products, rate of Technical Substitution and Cross
Productivity Efl‘ects for Land, Labor and Capital by Traction Use, Gokwe North
District, Zimbabwe, 1995/96 ...................................... 75

Table 5.1 Distribution of Per Capita Wealth by Wealth Source Among Settlers in Gokwe
North District, Zimbabwe, 1996. ................................... 81

Table 5.2: Demographic Characteristics of Settler Households in Gokwe North District,
Zimbabwe, 1996. .............................................. ’. 84

Table 5.3: Transitional Matrix of Equipment Wealth in Linked Settlers, (1987 - 1996)
Gokwe North District, Zimbabwe, 1996. ............................. 89

Table 5.4: Average Livestock and Equipment Wealth Positions in 1987 and 1996, Gokwe
North district, Zimbabwe, 1996 .................................... 91

Table 5.5: Transitional Matrix of Livestock Wealth in Linked Settlers (1987—1996),
Gokwe North District, Zimbabwe, 1996. ............................. 94

Table 5.6: Results of an Econometric Model for Wealth (Livestock and Equipment)
Mobility for Settlers in Gokwe North District, Zimbabwe, 1996. ........... 97

xi.

LIST OF FIGURES

Figure 1. Time line of tsetse and trypanosomosis control techniques used in Zimbabwe
1910 - 1999 .................................................... 4

Figure 2.1 Sample Stratification in the Socioeconomic Impact Assessment of Tsetse and
Trypanosomosis Control Survey ................................... 19

xii.

CHAPTER 1: TSETSE, TRYPANOSOMOSIS AND AGRICULTURAL

DEVELOPMENT

Trypanosomosis is a constraint to rural development. The rationale for tsetse and
trypanosomosis control is to develop the rural economy of affected areas by reducing the
impact and constraints which trypanosomosis pose on crop and livestock production. Tsetse
and trypanosomosis controls reduce diseases risk to livestock and people and indirectly boosts
agricultural production / incomes, ultimately providing a feasible chance for rural villagers to

accumulate wealth.

1.1 Tsetse and Trypanosome Species

In Zimbabwe, the main species of tsetse were G. morsitans submorsitans, G. pallidipes
and G. austem'. The two species common in the Zambezi Valley were, G. pallidrpes and
G. morsitans submorsitans. The real source of the disease problem is a one-celled parasite
known as a trypanosome. The tsetse fly is only the carrier, or vector of the parasite; the
bite of a fly not infected with trypanosomes is not harmful. There are different
trypanosomes present in Zambezi Valley. There is a human trypanosome called
Topanosome rhodesiense but it is rare and occurs in isolated locations along the Zambezi

Valley.

In Zimbabwe, the problem parasites are those which affect livestock. There are three
species of trypanosomes that afl‘ect cattle and other domestic animals. These are Ibrucei,
Icongolense and Iivivax. Each tsetse species is a good vector for particular trypanosome
specie. The G. m. wbmorsr’tms are good vectors for T.vivax and Icongolense. The T.
brucer' have good vectors in G. pallidr‘pes but infection rates are below 1% (Leak, 1996).
The infection rates of Ivivax increase when the major hosts are bovines. On the other

hand, an increase in infection rates of Ticongolense occurs when the major hosts are suids.

1.2 Tsetse and Trypanosomosis Control. Profiles

Zimbabwe has one of the longest and most continuous histories of tsetse and
trypanosomosis control in Afiica. Systematic tsetse control began as early as 1919 and
drug therapy began soon alter the discovery of the trypanocidal drugs in the early 19508.
Figure 1 is a time line of the techniques used for tsetse and trypanosomosis control in

Zimbabwe.

1.2.1 Tsetse Control

Vector control measures dominated the approach to control trypanosomosis until the
discovery of dirnidium bromide compound in the early 50s (Connor, 1989). In Zimbabwe,
tsetse control measures began as early as 1919. At that time, wholesale game elimination
and bush clearing were the main forms of tsetse control. This went on until the discovery

and development of pesticide science and application methods in the 60s (Shereni, 1991).

Bush clearing eliminated the tsetse habitat but vegetation loss along valleys destroyed the

ecological resilience and this technique was discontinued quickly.

There was a public outcry locally and internationally against the shooting of wild animals.
These concerns led to scaling down of these operations (Tsetse and Trypanosomosis
Commission of Inquiry, 1954). Game elimination continued through the 19608 selectively
and systematically along the tsetse front. Researchers first identified the preferred hosts of
G morsitans and then targeted hosts for elimination. The identified host animals were
kudu, bushbuck, warthog and bushpig. Elimination of these animals occurred within
controlled hunting areas or hunting corridors. A game-hunting corridor had a game fence
on one side and a cattle fence on the other side. The game fence prevented wild animals
fiom moving into hunting areas while the cattle fence prevented cattle fi'om moving into
hunting areas. Selective hunting of wild animals reduced tsetse blood meals and starved
ofl' tsetse flies in the hunting corridors. The process was slow and positive results came
alter a few years (Coulrnan, personal communication). Pesticide application increased
subsequent to the decline in bush clearing and hunting. Ground spraying of DDT began in
the 60s and was the main method of control until the 803. There were intermittent aerial
spraying operations during the same period. Control operations where disrupted by the
liberation war and some previously cleared areas, for example, part of Zambezi Valley

Area, where re-infested.

 

Restrictions on movement of animals, prophylactic
and curative treatment with trypanocidal drugs
Odor baits &
insecticide
ground & aerial spraying 1

 

 

 

 

 

 

 

clearin f
1910 1920 1930 1940 1950 1960 1970 1980 1990 00

Game and bush clearance, then selective hunting and bush l

 

Figure 1. Time line of tsetse and trypanosomosis control techniques used in
Zimbabwe 1910 - 1999

Source: Annual Reports, Tsetse and Trypanosomosis Control Branch ('I'I‘CB) 1960
- 95

Following national independence, tsetse control progressed through aerial and ground
spraying. Most recently, odor baits and insecticide application techniques have been the
main methods of vector control. Odor-bait techniques - targets, pourons - have become
the main techniques used to control tsetse since the mid-1980s (Shereni, 1991; Connor,
1989). The odor-bait techniques are considered to be cost-effective (Barrett, 1997) and to

have minimal direct environmental impacts (Grant, 1998). These technologies have helped

push back the tsetse frontier and also act as barriers against re-invasion.

While much of northern and southern Zimbabwe is climatically suitable for tsetse, today
only small pockets of tsetse remain in the north and northwest areas of the country. The
tsetse population has been gradually eliminated from large parts of the country as control

operations have systematically pushed the tsetse front further north and east. This gradual

expansion of the tsetse-free area provides an opportunity to evaluate the sequence of

impacts that occur after tsetse control.

1. 2. 2 Disease Control

The Department of Veterinary Services monitor the incidence of trypanosomosis cases in
domestic livestock. Trypanosomosis control measures are both preventative and curative.
As early as 1953, treatments were prescribed to all livestock potentially exposed to disease
risk (Connor, 1989). Administering of treatment occurred in central places and every
communal farmer had to participate. To reduce chances of re-infection, treatment was
administered to all the herds at the same time. The disease can spread from one animal to

another whenever tsetse flies are present.

The standard policy was to administer prophylactics (Samorin) every 3-4 month
depending on the degree of tsetse challenge. After diagnosis, treatment was given to all
positive cases. Treatments were administered before and afier tsetse control. Whenever
the demand for treatment over stretched the capacity of the Department of Veterinary
Services, it was economic to bring in vector control so as to reduce the need to procure
and administer drugs. Chadenga (1994) argues that drug use was effective when the
vector was eliminated or reduced to a low level. Treatments were suspending during

vector control in order to evaluation the effectiveness of tsetse control operations.

Monitoring the incidence of trypanosomosis continued after tsetse control in order to

assess the degree of re-infestation.

1. 2. 3 Regulating Livestock Movements

The general policy of the Department of Veterinary Services was to prohibit domestic
livestock presence in areas with a high degree of tsetse challenge. Free movement of
livestock in and out of fly country increased the probability of spreading the disease in fly-
free areas and increased the demand for treatments. Besides the risk of spreading disease,

domestic livestock movement restrictions prejudiced the success of hunting operations.

Despite the disease risks and fines, immigrants relentlessly brought in cattle and other
livestock (Connor, 1989). Farmers were capable of maintaining livestock under disease
risk through grazing management strategies. Regulating stock movements helped in

keeping the disease at bay.

1.3 Impacts of Tsetse and Trypanosomosis Controls

1. 3. 1 Statement of Problem
The majority of studies on the impacts of tsetse presence on niral development, including
traditional studies by Ford (1971) and Jordan (1986) focused on how the tsetse challenge

inhibited human settlement and how animal diseases made these areas unattractive for

agriculture led settlement. Proponents for tsetse and trypanosomosis control in Zimbabwe
argue that the presence of vector and disease challenges is a major constraint to livestock

production (Chadenga 1994) and retards rural development (Connor 1989).

In Zimbabwe, it is not well established whether the presence of tsetse flies impedes
immigration of families from less agriculturally endowed areas bordering the tsetse
fi'ontier. Yet tsetse flies and trypanosomosis controls are singled out as the main force
behind the uncontrolled influx of immigrants and livestock into cleared areas. The
contribution of other factors in attracting immigrants in cleared areas is unknown. One
view is that tsetse control eliminates the human disease threat that keep people away fiom
infested areas. A second view is that the construction of access roads by tsetse control
gangs makes the previously remote tsetse-infested areas accessible to immigrants. A third
view is that since the rise in world cotton lint prices in the 19603, cotton became an
attractive cash crop. Areas in Gokwe District with alluvial soils were a primary target for

the “white gold rush ” Lovemore (1997).

Part of the problem is that most attention is focused on the inflow of immigrants into
Zambezi Valley while no connections are made to address why emigration occurs in other
Communal Areas (CA). Young families who cannot secure individual arable land in their

original settlement areas spontaneously move to other areas where they can earn a

livelihood through farming. Well-established families also move in search of larger

holdings to support their growing families.

What is at work is the process of out-migration fiom areas bordering Gokwe and in-
rnigration into Gokwe. The impact on both in-migration and out-migration due to tsetse
control but separate fiom other factors are unknown at presentiln addition, the interaction
effects of tsetse control with all other factors are also unknown. Part of the problem is to

separate these efl‘ects.

Investment in animal disease control in agricultural frontier areas has potential to
transform the food systems by enabling technological progress and a shift from subsistence
towards specialized production and exchange. It is not clear what impact these
investments had in cattle adoption in Gokwe North District. Conventional wisdom
suggests that the introduction of cattle in previously tsetse-infested belts is associated with
eradication of tsetse flies. The empirical evidence to refute or support such claims is scant.
The problem is that merely clearing an area of tsetse flies does not guarantee farmer
adoption of cattle, especially when credit markets for cattle are non-existent. In addition,
the impact of other factors, including farmer-to-farrner extension, access to curative drugs
for trypanosomosis and the incentives or disincentives for commercialized production, are

not articulated in current debates.

The extent to which animal traction affects farm productivity in this environment has not
been clearly studied. In particular, no study has distinguished the impacts of cattle
ownership versus hiring custom service on factor productivity. The problem is that
oWnership of animal traction is highly skewed and improving access to traction services
carmot be completely addressed through extending ownership. Iffarmers who own animal
traction resources are not more productive than farmers who merely hire a custom service,

a minimum cost strategy may be to improve access to the traction rental market.

The role that investments towards an enabling environment play in socioeconomic mobility
of settlers has previously been neglected. Ifa presence of high tsetse challenge is a
constraint to rural development as suggested by proponents of tsetse control, the
socioeconomic mobility of settlers would be inhibited without than with tsetse and
trypanosomosis control. Traditionally, pioneers to a settlement frontier are expected to be
relatively wealthy because of a long residential tenure. However, when faced with disease
risk and limited agricultural opportunities, the early arrival premium may be non-existent.
Instead, newcomers who settled after tsetse fly eradication may be progressing up the
social ladder equally fast. The problem is that the early arrival premium cannot be taken as
a given in these circumstances. Whether first settlers and newcomers have progressed

equally over time is not known.

1. 3.2 Purpose of Study

The purpose of this study was to generate understanding on the extent to which tsetse flies
are a constraint to agriculturally led development. Such knowledge will help in the
formulation of socioeconomic policies that direct the development of land settlements and

mixed farming systems in tsetse cleared areas.

A good understanding of the motives for out-migration may help in planning alternative
ways to reduce unwanted internal-immigration. Given the limited budget government
operates with, it is critical to understand what public investments are required in order to »
attract additional private investment. An improved understanding of the sequencing of
public and private investments is a pre-requisite in rural development in general and

fanning systems development in particular.

Animal traction technology has potential to improve smallholder farm productivity. This
study identifies pathways that can stimulate adoption of this technology. Observing that
adoption of animal traction will not be complete, the study analyzed the potential impact
of improving the traction rental market access on partial factor productivity. An improved
traction rental market may be a feasible short to medium term solution to ownership of

traction animals and equipment in the long run.

10

New settlement fiontiers have been perceived to provide a safety net to citizens who have
failed to be absorbed into the labor force. This study will attempt to establish whether
new agricultural frontiers provide an equitable opportunity for socioeconomic
advancement to all settlers and whether voluntary- unassisted settlements can be tolerated

as an equitable redistribution process.

1. 3. 3 Objectives
The general objective is to investigate the efl‘ects of tsetse control on human migration and
livestock adoption and household socioeconomic mobility. The specific objectives are:
1. to identify why families choose to leave their previous residences and settle in
agricultural frontier areas;
2. to quantify the impacts of tsetse control relative to other factors that attract migrants
to fi'ontier areas;
3. to measure the impact of tsetse control, immigration and commercial production
enabling environment on the rate of cattle adoption per .village over time and space;
4. to identify the difl‘erences in factor productivity between farmers who use and own
animal traction and those farmers who do not own but hire animal traction services; and
5. to measure difl‘erences in wealth holding and socioeconomic mobility between different

types of settlers.

11

1. 3.4 Hmothesis
This study tested several hypotheses. These hypotheses are presented below in an alternate
form. The hypotheses test relationships that I believe exist between tsetse and

trypanosomosis controls and in-migration, livestock adoption and socioeconomic mobility.

1. The distribution of agricultural economic opportunities among the rural districts of
Zimbabwe is in not uniform. I expect economic factors especially access to land
rather than social ties to dominate motives for out-migration.

2. Tsetse and trypanosomosis controls positively contribute to making a site
attractive for in-migration especially when immigrants are moving in together with
their livestock. Everything else being the same, I expect immigrants to settle in
areas controlled of tsetse and trypanosomosis first before choosing tsetse-infested
areas.

3. The availability of infrastructure services including public transport, health and
education services positively contributes to making a site attractive for in-
migration. I expect immigrants to settle in villages with better infinstructure
services first before they settle in relatively remote villages.

4. Tsetse and trypanosomosis controls allow livestock to thrive in areas they would
have perished. Access to curative drugs could be suflicient for the introduction of

cattle in tsetse-infested areas but both tsetse and trypanosomosis controls may be

12

essential for widespread adoption of livestock. Subject to access to drug therapy, I
expect settlers to adopt cattle before active tsetse fly control.

There are cultural differences between indigenous and immigrant settlers in terms
of livestock adoption. Pioneer settlers had relied primarily on manual tillage in
their cropping systems. Cattle had an extremely limited farming role in their
traditional way of life. As a result, I expect newcomer settlers to be an
instrumental force in introducing mixed farming systems to tsetse cleared areas.
Access to animal traction services can be secured through custom hiring. A rental
market enables non-owners to improve their resource productivity. Owing to
excessive demand for custom service, not all those who need to hire these services
get it in time. Everything else being the same, I expect owners of animal traction
services to be much more productive than renters of animal traction service.

The presence of tsetse flies in a settlement area is viewed as an impediment to
economic progress of settlers. Despite being the pioneers, settlers who move in
before tsetse control do not gain much economically because of the adverse
physical conditions in settlement areas. I expect the premium associated with
settlement tenure to be insignificant prior to tsetse control but significant after

tsetse control.

13

1.4 Dissertation Outline

The following chapter describes the research design adopted in this study and provides a
general description of the study sites. This chapter highlights the key variables controlled
in the study and describes the general population characteristics for which results from this

study can be relevant.

Chapter 3 examines the factors that push and pull migration into the Zambezi Valley of
Zimbabwe. Specifically, the impacts of tsetse and trypanosomosis controls are separated
fi'om other factors contributing to immigration into Gokwe North District. A model of the
stock of migration is developed to study the effects on migration due to the interaction

between tsetse control and other factors.

Chapter 4 presents an annual growth model of cattle adoption from 1965 to 1996 for 12
villages in Gokwe North. The role of tsetse control in both the introduction of cattle and
the spread in cattle is evaluated. This chapter also examines the impact of animal traction
on resource productivity and measures the differential impact of animal traction use

between household who own and households who rent-in traction services.

Chapter 5 discusses the distribution and determinants of wealth among settlers. The
sources of wealth considered are livestock, land improvements and farm equipment assets.

Previous studies focussed on household wealth per se but in this study per capita wealth

14

holding is the dependent variable. Based on the evidence, the life cycle and settlement
tenure hypotheses are examined. In addition, this chapter also evaluates socioeconomic
mobility of household wealth. A sub-sample of households who settled before 1988 are
the ones considered for this analysis. The livestock and equipment wealth position of the
household in 1987 and 1996 are compared to examine the direction and extent of wealth

mobility.

Chapter 6 provides a discussion of the results, summarizes the conclusions and identifies

issues that require further research.

15

CHAPTER 2: RESEARCH DESIGN AND INSTRUMENTS

2.0 Introduction
In this chapter, a description of general physical conditions in study sites is provided. The
study design, the sampling approach and the instruments used to implement the study are also

discussed in this chapter.

2.1 Study Location and Physical Characteristics

In order to focus attention on the effects of tsetse control on migration and livestock
adoption, the study was designed to control for some of the factors that vary across space,
especially district government, soil type and rainfall. The study was confined to Gokwe North
District. Gokwe North is located in north-western Zimbabwe. Almost all the district is in the
Zambezi Valley and it is climatically suited for tsetse fly habitat. Gokwe North District was
an ideal site because it had a 25-year history of active tsetse and trypanosomosis controls.
Parts of the district were actively cleared of tsetse since the mid-1960s, while other parts were
cleared as recently as 1995. The gradual expansion of the tsetse-free area in Gokwe North
provides an opportunity to evaluate the sequence of impacts that occur afier tsetse control.
Study villages were purposively selected to represent difl‘erent lengths of time since tsetse

control.

16

Sites were selected fi'om a uniform land class unit. The land class unit had uniform parent
rock material, vegetation characteristics and drainage patterns (Anderson, 1993). Uniformity
of land class unit was important in order to make the impacts of tsetse control on the farming
systems across villages comparable. This land class unit selected was along major river
systems (Ume, Sessame, Sengwa and Gunguwe) where settlement and agricultural production
were first established and most concentrated in the Zambezi Valley. Land in those drainage
areas is comprised of flood plains and terraces of the larger river systems with alluvial and
argillaceous parent material. Soils are moderately well drained, moderately shallow, sandy
clay loam or clay. These soils have high agricultural potential and thus have been an early
destination area for immigrants interested in crop production (Anderson et al 1993). The
dominant vegetative characteristic under this land unit is the mature mopane or miombo
woodlands. The vegetation is dense woodland in unsettled areas but wooded grasslands in

sites with active settlement.

2.2 Survey Design
The survey design featured two control variables (1) the period of time since control and (2)
the degree of access to infiastmcture services. The design chose to control these variables

in order to measure their efl‘ect on performance variables.

The study chose three levels to represent the tsetse control gradient: early clearance (1970 -

1976); mid clearance (1980 - 1986); and late or recent clearance (1984 - 1990). This

17

stratification gave a tsetse control gradient with three cluster sites. The period of time since '

control was expected to have profound effects on the pace of in-migration, livestock adoption

and farming systems change.

Access to infrastructure and markets were factors that attracted immigrants and afl‘ected the
development of farming systems. I expected immigrants to prefer areas with better
infiastructure services, for example, road network, schools and clinics. I also expected
settlers in villages with better infiastructure to have a commercial oriented production system,
for example, a high percentage of cropped area under cotton. The infrastructure parameter
had two levels: villages with the best access to infrastructure and villages with the poorest

access to infi'astructure services as of 1996.

In summary, there were two design factors, one with three levels and the other with two
levels. The design gave six combinations of control period and access to infrastructure: (1)
early clearance with good infrastructure; (2) early clearance with poor infrastructure; (3)
medium clearance with good infi'astructure; (4) medium clearance with poor infrastructure;
(5) late clearance with good infrastnrcture; (6) and late clearance with poor infi'astructure.
Selecting two wards in each cluster gave each combination a replication which brought the
total study sites to 12. Figure 2.1 shows the three levels of the time of tsetse control gradient,

the replication at ward level and two levels of infrastructure gradient.

18

 

Time Gradient of WARDS VILLAGES

 

 

 

 

 

 

 

 

 

 

 

 

 

tsetse control Uniform Land Class Infrastructure service access
Unit
IGood (Tofe) I
IGoredema I
Gorederna Poor ' a
Cluster A I (Nylk ) I
Early Clearance
(1970 — 1976) R300“ (NCVW) I
I Chireya III I
rPoor (Nyahomba) I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Good (Kakwari I) I
I Gumunyu IV I
Masharne Cluster [Poor (Kakwari II) I
B Mid Clearance
(1980 - 1986) [Good (Nechinyika) q
INechinyika 1
[Poor (Mayava) I
Good (Chirudzi)
. . Madzivazvido
13:3; réazvrdo Poor (Kwaedza)
Late Clearance
(1984 —1990) Good (Ntamo 1)
Nenyunga
Poor (Chirisa)

 

 

 

 

 

 

Figure 2.1 Sample Stratification in the Socioeconomic Impact Assessment of Tsetse and
Trypanosomosis Control Survey

19

2.3 Sites and Household Selection

A fivestage stratified sampling procedure was adopted to select study sites within Gokwe
North. The sample stratification is shown in Figure 2. 1. The first stage involved the purposive
selection of the three cluster areas representing a gradient of time since tsetse control. Once
the cluster areas were identified, sites within a similar land class unit across the gradient were
targeted for firrther selection. During the second stage of sampling, two wards (local
government units) within the same land classification were selected fiom each cluster area.
Selection of two wards gave the study an opportunity to identify any differences in local
administration that could have affected migration and settlement. Two wards also improved
representativeness within the cluster. During the third stage, two villages were purposively
selected from each ward: one village had the best access to services and the other had the
poorest access to services at that time. Villages, also referred to as VIDCOs or village
development committee areas, are the smallest local administrative unit. Important services
targeted in the selection criteria included access to public transport, primary schools and
health centers. During the fourth stage, a random sample of households was drawn from the

population of each selected village.

The study randomly selected a sample of forty households from a village population list. I
used registers of village leaders and extension workers to compile a village population list.
I conducted random selection of households during village meetings that village leaders and

other villagers attended. I purposively included village heads in the sample in order to

20

guarantee popular participation but I excluded them from the analysis. As leaders, village

heads exerted much influence, and made a difl‘erence between success and failure of the study.

2.4 Tsetse and Trypanosomosis Control Operations in Cluster Sites

The study selected three cluster sites along the tsetse control gradient. Cluster A is a site
known as Goredema, Cluster B is Mashame and Cluster C is Madzivazvido. In Goredema,
active tsetse and trypanosomosis controls started around 1965. Tsetse control started with
selective game hunting. Ground spraying of tsetse with DDT commenced in 1970 until 1976.
Although there were positive infection cases in Goredema fi'om 1980 to 1983, the disease risk

was low.

In Mashame, strategies used to control tsetse were selective game hunting in the 703, ground
spraying fiom 1980 to 1986 and use of odor baits in 1990. Odor baits were planted along the
district northern border to provide a barrier against re-invasion. Trypanosomosis control

efforts started around 1983 and continued during and after tsetse control.

In Madzivazvido, control operations included use of aerial spraying fi’om 1983 to 1984.
Ground spraying followed thereafter in 1986/87 in areas where tsetse continued to present
problems. In Madzivazvido, planting of odor baits occurred in 1988 but were removed in
1990. Trypanosomosis control efforts started before 1983 and continued well after tsetse

control campaigns.

21

2.5 Survey Instruments
I used questionnaires to collect household-level data. The study had four questionnaires,
hmce multiple visits were done. I supplemented household survey data with structured

interviews with local administrative and traditional leaders, civil servants and business

entrepreneurs.

The first questionnaire focused on effects of tsetse controls on livestock adoption and human
migration. This questionnaire addressed issues pertaining to in-and-out migration, and how
tsetse control affected the numbers and species oflivestock kept by existing residents and that
new immigrants brought with them. Interviews on this questionnaire started in March until

May, 1996.

The second questionnaire addressed the indirect impacts of tsetse control on crop production.
The questionnaire focused on assessing differences and similarities in crop production among

existing residents and immigrants. Household visits happened fi'om June to July, 1996.

The third questionnaire focused on two issues. The first focus was on soliciting perceptions
about the incidence of the utility and disutility of tsetse control. The second focus of the
questionnaire determined the conditions under which households had access to land-based

resources. Interviews started in August until October 1996.

22

The fourth and final questionnaire addressed issues pertaining to households’ aspirations and
desires to expand and improve their arable and residential plots. Household visits were from

November 1996 to January 1997.

I hired enumerators to administer these questionnaires. Enumerators were high school
graduates and were resident in the village they worked. I shared the responsibility for
selecting enumerators with village leaders. Village leaders and agricultural extension workers
participated in the selection so as to increase the chances of respondent cooperation during
interviews. By engaging local leaders in this process, they accepted the responsibility of solve
potential problems between enumerators and respondents. There was one case of enumerator

turnover during the survey.

23

CHAPTER 3: THE IMPACTS OF TSETSE CONTROL ON IMMIGRATION

3.0 Introduction

Migration is one of the most important factors shaping processes of change and development
in agricultural fiontier areas. In this chapter, the factors that push people from older settled
areas and pull migration into the Zambezi Valley of Zimbabwe are examined. Specifically,
the impacts of tsetse and trypanosomosis control from other factors contributing to
immigration into the Gokwe North District of the Zambezi Valley are separated. A model
of the stock of migration was developed and estimated using survey data collected fi'om 482
households in 12 villages having different experiences with tsetse control. The results suggest
that emigration from source areas is primarily motivated by desires to gain greater access to
agricultural land. Although it started before tsetse control was undertaken, immigration into
destination areas was accelerated by public investments in tsetse control. The establishment

of roads and markets also had significant impacts on the rate of immigration.

This chapter seeks to accomplish two objectives. The first objective is to understand why
families choose to leave their previous residences and settle in agricultural fi'ontier areas. The
second objective is to quantify the impacts of tsetse control relative to other factors that

attract migrants to fiontier areas.

24

3.1 Modding Immigration Stocks and Flows

3. 1. 1 Background and Relevant Literature

Economic models of migration assume that potential migrants will move fi'om one location
to another in order to maximize the expected net present value of their earnings, subject to
budget constraints and the costs associated with migration. An implication of this individual
behavior is that there will be net immigration into areas with high expected earning potential
and net emigration from areas with low expected earning potential. The Hecksher, Ohlin and
Samuelson trade model suggests that a region with relatively abundant labor will expert that
factor to a labor-deficit region until the wage rate equalizes across regions. Land-abundant
areas with low population densities and labor-abundant areas with high population densities
stand to mutually benefit fiom the flow of immigrants to the land-abundant area (Borjas

1988)

There are at least four additional factors that have been shown to affect migration patterns:
(1) the monetary and psychological costs associated with migration; (2) the quantity and
quality of the public amenities available in both source and destination areas (Diamond and
Tolley 1982); (3) the institutional and legal procedures affecting migration (Pessino, 1989);
and (4) civil unrest and war. High monetary and / or psychological costs associated with
migration will reduce expected net earnings and utility in potential destination areas and

potential source areas.

25

Public amenities that are purely consumptive can also affect the comparison of source and
destination areas; potential migrants may be willing to accept lower earnings to increase their
access to superior consumption amenities (Carlino & Mills, 1987; Knapp & Graves, 1989;
Mathur, 1988). Public amenities that are more productive — for example, tsetse control,
markets, roads, communication infrastructure - will affect migration through expected
earnings. Here the study hypothesizes that the main impact of productive amenities like tsetse
control and roads is to raise earning potential and thus make the area more attractive to
potential migrants. Tsetse control will affect expected earnings for households that currently
own livestock, for households that intend to own livestock, or for households that intend to
employ livestock in their farming operations. A priori, however, there is no clear causal
relationship between immigration, tsetse control and the establishment of other public
amenities. There are at least four non-exclusive possibilities: 1) Public amenities may be
developed in anticipation of immigration and the prior location of public amenities may make
an area attractive for new immigrants; 2) The agencies that plan the development of public
amenities may respond to the presence of large numbers of new immigrants; 3) The agencies
that plan tsetse control may respond to the needs of large numbers of new immigrants with
infected animals; and 4) Tsetse control may be a keystone public investment that leads both

private migration and the public development of other public amenities.

War can affect migration in several ways. What is commonly observed is the involuntary

displacement of war victims who are resettled outside battle-fronts. Rohrbach’s study of the

26

determinants of growth in smallholder maize production in Zimbabwe found lower than
average growth of smallholder farmers (2.5% per annum) during the peak of violence (1976
to 1979). As the war wound down in late 1979, households, especially young ones, “took
advantage of this period of flux to move away from their parents and establish independent
farms voluntary resettlement was substantially greater than that made necessary by the
war” (Rohrbach 1989). The growth in maize production, dubbed Zimbabwe’s Green
Revolution by Eicher and Rukuni (1995), was due to an expansion of total maize acreage as
abandoned land, new land and part of the former grazing land were planted with maize

(Rohrbach 1989).

Rural instability associated with war increases mobility and adaptation costs. During the civil
war in Zimbabwe, the provision of all public amenities including tsetse and trypanosomosis
control was hindered by the increasing security risk for public omcers. Thus we expected a
large increase in immigration into the Zambezi Valley after 1979 due to the transition from

war to peace.

Prior to this study it was proposed that sequential migration — movement from a source area
to a destination area and then on to another destination area — was an important phenomenon
in the Zambezi Valley of Zimbabwe. Pessino (1989) argues that the more sound the basis of
information for the initial move, the lower the propensity for subsequent migration. In a study

he conducted in Peru, Pessino found very little sequential migration. The cost of establishing

27

a new residential site and clearing land for cropping may have been the major disincentive.
To avoid losses associated with the abandonment of such sites, immigrants invested a great

deal of time gathering information about potential settlement sites fi'om fiiends in those sites.

The conventional wisdom on internal rural migration in Zimbabwe is that tsetse control causes
migration into the Zambezi Valley. Lovemore (1997) points to population growth in rural
areas outside the tsetse belt as the cause of immigrant inflow to tsetse cleared areas. The
Government of Zimbabwe does not have a clear-cut policy on spontaneous and unassisted
internal migration. While the Government does not actively encourage voluntary internal
migration, it does indirectly support migration through its policy of granting 5 hectares of
arable land to new immigrant families. Immigrants are, however, perceived by the local media
to be “looters” of land resources (1mg, 4 May 1996) and contributors to leadership

wrangles in local politics.

3.1.2 Foundations of Immigration Growth Model

Immigration into a particular area can be conceptualized as a stock variable. The stock of
immigrants is the total number of immigrants in a particular village at a particular time. Here
the study denotes the variable M“, as the cumulative number of immigrant families in a village
i, at time t. For example, when t = 1990 and i = 10, 114,, was 14. The number of immigrant

families who settle in the village during a particular year is the flow of immigrants. The stock

28

and flow variables were determined by aggregating the responses immigrant settlers gave

when asked “Which year did you establish you first settlement in this village?”

Empirical research on the determinants of migration involves estimation of an exponential
equation with the stock of migrants as the dependent variable and independent variables such
as time as exponents (Schutlz, 1982). The generalized form of an exponential function is
given by equation (3.1):
y = ab“ (3.1)

where a and c are ”compressing” and "extending" agents, respectively.
With base e in exponential function, the function of cumulative stock of migrant settlers
becomes equation (3 .2),

M, = Ae’“" (3.2)
Where r is a constant - the instantaneous rate of growth of migrants in the village at any given
time and u is the random error term. In logarithmic form, the model is given by equation
(3 .3).

lnM“ = InA + rt+u. (3.3)
The variables that either compress or extend this function are described in the following

subsection.

29

3. 1. 3 Specification of the Empirical Model

The following variables were included in the model as independent variables for explaining
the stock of migrants in each village per year. Some of the other variables that might also be
important, e. g. soil fertility and potential yields, were held constant by the design of the

sample selection procedure.

T setse Control: This variable measures the periods with tsetse control, denoted by a value
of one, and periods without tsetse control, denoted by a value zero. A continuous measure
of the prevalence of both tsetse and trypanosomosis was hard to capture for each site. I
expected that the growth of the immigrant population would be higher during periods of
tsetse control than during periods without tsetse control. If immigrants anticipated tsetse
control in a particular area, the rate of immigration into that area could have been affected
before active control began. However, if immigrants were only reacting to actual control, the
flow would be significantly affected only after tsetse control. It is also possible that
immigrants were reacting to disease control in anticipation of disease control. Tsetse control
is a productive public amenity that has the potential to increase livestock productivity, and

through draft power, the crop productivity for individual farmers.

Cotton / maize price ratio: The ratio of the real price of cotton relative to the real price of
maize (1970 = 100) for maize and cotton traded in Zimbabwe’s central market was included

in the model. This variable is a proxy for expected earning potential in both destination and

30

source areas. In both the source and destination areas, cotton is the main cash crop and maize
is the main subsistence crop. Everything else equal, an increase in the cotton / maize price-
ratio indicates an increase in potential earnings for food-secure households that produced
cotton and enough maize to meet household subsistence requirements and a decrease in
potential earnings for households that did not produce cotton and produced more maize than
they consumed. A positive relationship between the price-ratio and the rate of migration into
destination areas would imply that food-secure households moved in response to expectations
of better opportunities to produce cotton, while a negative relationship would imply that
food-insecure households moved in response to expectations of better opportunities to

produce maize.

Access to Schools: This variable measures access to primary schools. When a school was
accessible within an eight-kilometer radius, the variable was given a value of one, otherwise
it was given a value of zero. The study expected migrants to be more attracted to areas with
good access to schools. Eight kilometers is an estimate of the maximum distance that people

might walk each way to a school, clinic or market and return home within the same day.
Access to Clinics: When a clinic was accessible within an eight-kilometer radius, this variable

was given a value of one, otherwise it was given a value of zero. Migrants are expected to

be more attracted to areas with good access to clinics.

31

Marketing Costs: This variable measures access to a central marketing depot; one denotes
the presence of a marketing depot within 60 kilometers, zero indicates no marketing depot
within 60 kilometers of the village. Better access to marketing depots reduces marketing
costs through reductions in fi’eight charges and creates greater market stability. The study

expected immigrants to be more attracted to areas with better access to markets.

T ransportation Costs: Roads and public transportation facilities make rural areas more
accessible for settlement by migrants. The effect of a road is to reduce costs the costs
associated with migration into an area as well as the costs of purchasing inputs and selling
outputs. Areas with bus stations within a radius of eight kilometers were considered
accessible and denoted by a value one; otherwise they were given a value of zero. The study

expected migrants to be more attracted to areas with lower transportation costs.

Time Trend: The flow of migrants into villages was different at different times. The dependent
variable and one of the independent variables, purchasing power, have trends. A time variable
was included as an independent variable to account for the trend efl‘ect in the other variables.
The study hypothesized that the growth rate of migrant settlers would be positive and

accelerated by tsetse control.

32

The estrmat' ed rmmr' 'gration function is equation (3.4):
rT +aL +bY +cD

Mir = Pi! i j 1: (3-4)
where:
ithe ward = l,...,6; t the period = l,...,35 years; j the amenity type = 1 (roads and
transport), 2(markets), 3(schools), 4(clinics); and kexperience with tsetse control (1
with and 0 without tsetse control)
M, = the stock of immigrants in village i during year t
P, = the earnings ratio
T = time trend
r = rate of instantaneous grth in immigration
a = percentage difference in immigration stock in different locations
L = ward location
b = percentage difference in immigration stock with and without access to amenity
Y= type of amenity

c = percentage difference in immigration stock with and without tsetse control
D= with and without tsetse control

3.2 Extent of Immigration in Study Sites

Of the 482 sample households in 1996, 43% were indigenous residents, 28% were second-
generation immigrants and 29% were first-generation immigrants (Table 3.1). Significantly,
all of the village heads were indigenous residents. Indigenous residents are households in
which the parents of the household head grew up in the area and established their first
homesteads in the study sites. Second-generation immigrants are households in which the
parents of the household head immigrated into the area and established households. First

generation immigrants are households that left earlier settlements outside of the study site to

33

establish new settlements in the study sites. The results presented in this chapter focus

primarily on first-generation immigrants.

Table 3.1: Distribution of first and second generation immigrants and indigenous
residents in Gokwe North, Zimbabwe, 1996

 

 

 

 

Household Group No. households % of Total
First generation immigrants 142 29
Second generation immigrants 133 28
Indigenous residents 207 43
Total sample 482 100
Village heads - indigenous residents 55

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North, Zimbabwe, 1996-97

Table 3.2 indicates the number of first-generation immigrants in each village and the time
period during which they moved into the village. The results indicate relatively slow and
steady immigration to the area between 1966 and 1980 when 18% of the first-generation
immigrants moved to the area. Total immigration peaked between 1981 and 1985, that is,
immediately after Zimbabwe achieved political independence afier several years of civil war.
Twenty-eight percent of the first-generation immigrants indicated that they moved to the area
during that 5-year period. Immigration remained fairly high thereafier with 21% of first-
generation immigrants arriving between 1986 and 1990 and another 21% arriving between

1991 and 1995.

34

Table 3.2: The Inflow of Immigrants into Villages 1965 - 1995, Gokwe North,
Zimbabwe, 1996
Period
Village <196 66-70 71-75 76-80 81-85 86-90 91-95 Total

 

    
 

Tofe 0 1
Goredema l 14
Masvingo 3 21
Nyahomba 0 5
Kakwari I 3 14
Kakwari II 3 20

Nechinyika O 0 l 9
Mayava 0 0 0 0 0 8 '
Chimdzi 0 O 0 0 l4
Kwaedza 1 0 l 0 19

Ntarno 2 0 0 0 2
Chirisa 1 1 0 0 14
TOTAL 1 l 9 6 l l 45 30 29 141

 

(Note: The shaded cells indicate periods of active tsetse control in each village.)
Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

Table 3.3 presents information on the provinces from which people emigrated to Gokwe
North. All but one of Zimbabwe’e eight provinces were source areas for migrants to Gokwe
North. The exception, Manicaland, is on the eastern border and is the furthest province fi'om

Gokwe North District. Otherwise there does not appear to be a strong relationship between

number of migrants and distance from source to destination. About 45% of the migrants

35

were fi’om Masvingo Province, about 450 kilometers from Gokwe North, and another 45%

were from Midlands Province, about 270 kilometers fi'om Gokwe North.

Table 3.3: Provinces where Migrants Emigrated, Gokwe North, Zimbabwe, 1996
No. of Immigrant Families From Each Province

 

 

Settlement timing Mashonaland Matabeleland Masvingo Midlands
Before tsetse control 2 1 13 11
During tsetse control 4 3 23 17
Afier tsetse control 11 4 22 31
All time periods 17 8 58 59

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

Table 3. 4 presents the reasons given by first-generation immigrants as to why they moved.
Most moved voluntarily. Thirty-five percent said that they moved because their farms were
too small,17% had no farms, and 26% said that the soils in their original locations were

infertile and needed large inputs of fertilizer and manure.

36

Table 3.4: Reasons Why Migrants Choose to Emigrate From Their Areas of Origin
Before, During and After Tsetse Control into Gokwe North, Zimbabwe, 1996

 

 

 

Reason for moving fi'om original settlement Before During After
(Number in parenthesis is %)

My arable plot was too small for my needs 10(3 8) 15(3 2) 25(4)
I did not have an arable plot ofmy own 4(15) 9(19) 11(17)
The soil had poor fertility 6(23) 15(32) 16(25)
I was forced to move by government 2(8) 1(2) 1(2)
The area was highly susceptible to droughts 0 2(4) 5(8)
It was difficult to get services 0 2(4) 0
I was running away from tsetse 1(4) 0 3(5)
I got married here 1(4) 0 0
I divorced and moved here 0 0 2(3)
I was running away from witchcraft 1(4) 0 0
I was accused of being a spy during the war 1(4) 3(7) 0

MISSING CASES 1 0 5
TOTAL SAMPLE 27(100%) 47(100%) 68(100%

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97.

A minority of households felt that they were forced to move. Four of the 141 households
were evicted fi'om original settlement. Two female-headed families moved back to their
parents’ villages after divorcing their husbands. Four households that immigrated before
and during tsetse control were expelled from their former villages because they were
accused of being spies during the war of liberation. Four more households migrated to

tsetse controlled areas to escape tsetse flies.

3‘7

Table 3. 5 shows the amount of sequential migration, distinguishing between immigrants
who moved within the same village or ward in which they originally settled, and

immigrants who moved between wards.

Table 3.5: Frequency of Sequential Migration Within and Between Wards Before,
During and After Tsetse Control, Zimbabwe, 1996

 

Radius of sequential migration Before During After
Within present ward Yes 1(4) 0 3(5)
No 26(96) 47(100) 64 (95)

Total 27(100 47(100) 67(100)

 

 

Between wards Yes 5 (23) 5 (11) 5(8)
No 2207) 42(89) 62(92)
Total 27000 47(100) 67(100)

 

Primary reason for moving

Wanted more land

Poor relations with neighbours

Dispute in land ownership

Problem getting water for domestic use
Could not keep cattle due to tsetse
Wild animals caused havoc

Inadequate pasture land

Services were far away

Total

 

WO~OOOHO~
Urn—or—tv—Iu-aooo—o

OOOOOONhl—or—s

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Controls,
Gokwe North, Zimbabwe, 1996-97

38

Of the 141 first-generation migrants, only four reported that they had moved within the
same village or ward, while 15 reported that they had moved between wards within
Gokwe North. Availability of land and disputes over available land were the main reasons
given by respondents for their sequential migration. Most of the migration within ward
occurred after tsetse control had been established in the ward area but most migration

between occurred before tsetse control.

Table 3 .6 shows the time period during which different types of public amenities were
established in the study sites relative to when tsetse control occurred and the period of
greatest immigration. Generally, the initiation of tsetse control predated public

investments in infrastructure for villages in which tsetse control began before 1985.

In the four villages in which tsetse control began in 1985, tsetse control began in the mid-
range of all public infi'astructure. In general, most public amenities were in place before

the period of greatest immigration.

The order in which amenities were established varied somewhat across villages. In the
villages with the earliest tsetse control, most of the public amenities were established at
about the same time, six to ten years after the initiation of tsetse control. This reflects the
strong central planning approach to development adopted by the pro-independence '

government of Zimbabwe. Few amenities were established between 1972 and 1982,

39

probably due to the civil war that dominated events during that period. Since 1982, the
pattern of development has been somewhat more haphazard. At the time of the survey,
fourofthevillageshadnoclinicsandtwohadnoschools.

Table 3.6: Sequence of Establishing Services in Study Sites, Gokwe North,
Zimbabwe, 1996

 

 

 

 

Village by Year public amenity was established Peak Flow

Name Clinic' School‘ Roads” Market“ Clearance ..... 1 ofMigrants
Goredema 1972 1972 1972 1972 3.5;;g1966-75 1966-70
Tofe 1972 1972 1972 1972 32‘7‘7'1966.757"7,i 1981-85
Masvingo 1982 1972 1972 1982 1972-76}. 1976-80
Nyahomba 1982 1981 1987 1982 :1 {-31197217621;+5: 1981-85
Kakwari I 1982 1972 1967 1992 1975-881..,f...,:;-,: 1981-85
Kakwari 11 None 1987 1972 1992 ,_1975-88ffj;. 1981-85
Nechinyika 1983 1987 1967 1982 ~"5:"1981-88f?%i‘f 1986-90
Mayava None None 1982 1992 i_;17,199.81s-88329.f. 1981-85
Chirudzi 1987 1963 1993 1982 77I.1976:943519fi 1991-9s
Kwaedza None None 1987 1982 ' .71976e9455y’- 1991-95

Ntamo 1987 1963 1987 1982 1983—921'! ------ <1966

Chirisa None 1994 1987 1982 1983-92 1986-90

'Clinics and schools are accessible within an 8-km radiusb anary roads open to buses
and trucks" Marketing depots are within a 60 km radius‘I Years of active tsetse control

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

40

3.3 Relative Impacts of Tsetse Control on Immigration

The migration data discussed above were used to develop a variable measuring the stock
of first-generation migrants in each village for each year (Mm) between 1963 and 1995.
The dependent variable had 396 observations, representing 33 periods in the 12 villages.
Similar variables were specified for all of the explanatory variables presented above. All
of the explanatory variables were included with interaction terms to allow for difi‘erent

efl‘ects before and after tsetse control.

The results presented in Table 3 .7 show that the model has good overall explanatory
power, explaining 68% of the variation in the stock of migrants. The strong F test shows
the significance of the overall model. The results for each explanatory variable are

discussed in turn hereafter.

Constant: The estimate of the constant shows the expected stock of first-generation
immigrants prior to 1966 was 3.3. Across the 12 villages, this represents 28% of all first-

generation immigrants in the villages at the time of the study.
Ward eflects: Villages in the ward that had experienced tsetse control more recently had a

higher proportion of immigrants than other villages prior to tsetse control. The preference

for newer settlement fiontiers suggests immmigrants anticipated control in the areas they

41

moved into. There were no systematic differences between the stock of immigrants in

villages. This reflects an absence in control gradient preference after tsetse control.

Table 3.7 : Determinants of the stock of first-generation migrants in 12 villages of
Gokwe North District, Zimbabwe, between 1963 and 1995

 

 

Explanatory Variables Parameter Estimates
Without Tsetse Control With Tsetse Control

Constant (Ward 6: recent 3.3014" -.3828
Ward 5 (less recent control)LS -.4485" .0432
Ward 4 (mid control)L4 -.7036" -.2243
Ward 3(mid control)L3 -.4810** .6978
Ward 2 (early control)L2 -.2977** .6546
Ward 1 (early control) L1 -.5422*"‘ -.0657
Cotton/Maize Price Ratio (R) .0231 -.3311**
Time trend (T) .0148 .0366”
Access to Markets -.0359 .3183"
Access to Public Transport .8600" .5457"
Access to Schools -.4593" -.2479"'*
Access to Health Centers - -. 1386

 

“p<0.05 R2=0.68 F=39.1

Source: Socio-Economic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

Growth rates: The instantaneous growth rate of immigrant households was positive with
or without tsetse control. With tsetse control, the effect of an additional year on the stock

of migrants was to increase the stock by 3.7%. Without tsetse control, the effect of an

additional year on the stock of migrants was to increase the stock by 1.5% per year.

42

While the effect of time was statistically significant with tsetse control, it was not
statistically significant without tsetse control. Overall it can be concluded that, everything
else equal, tsetse control was associated with an increase in the annual rate ofirnmigration

by about 150%.

Cotton / Maize Price Ratio: The results indicate no relationship between the price ratio
and migration before tsetse control but a negative relationship occurred after tsetse
control. Following the logic presented above, this implies that most of the migrant
households who moved to the study sites after tsetse control were food insecure in their

previous locations.

Amenities: Access to public transport was a significant immigrant attractant with and
without tsetse control. Immigrants settled in villages that had public transport despite the
presence of tsetse flies. Provision of roads and public transport increased the stock of
immigrants by 80% without control and 55% with tsetse control. Access to crop markets
was an important amenity at the time of immigration. With tsetse control, access to
markets increased immigration by 32%, while it had no effect without tsetse control.
Apparently, the ability to take advantage of markets depended on producing a crop

surplus.

43

The relationship between migrant accumulation and access to schools was significantly
negative. Migrants moved into villages with poor access to schools even after tsetse
control. Apparently access to schools was not a significant attractant at the time of
migration The relationship between access to health services and migration after tsetse
control was negative but insignificant. Again, it appears that immigrants may have failed to
settle in villages with health centers but the health centers were not an essential service at

the time of settlement.

3.4 Importance of Tsetse Control and other Amenities

The quantitative assessment of factors affecting immigration was augmented by a study of
immigrants’ perceptions of the importance of tsetse control and public amenities on their
decision to move into their village in Gokwe North. In particular, first-generation
immigrants were asked whether or not they thought that tsetse control or public amenities
had important effects on their decision to migrate. The questionnaire indicates that the
majority (88%) of immigrants who moved in prior to tsetse control felt that the degree of
tsetse infestation did not afl‘ect their decision regarding where to settle within Gokwe
North. The data summarized in Table 3.8 show a third of the immigrants were not even

aware of the presence of tsetse when they settled in tsetse infested areas.

About two thirds of the migrants appeared to be aware of tsetse presence at the time they

settled. Halfof these migrants discounted tsetse presence because they had no livestock

44

when they moved in and perceived that the flies would not be detrimental to human health.
The other half discounted the presence of tsetse because their only interest was to gain
access to more land. Based on assessments from migrants who moved in before tsetse
control, it appears that tsetse infestation most deterred migration of households that
wanted to move with their cattle.

Table 3.8: Reasons Why Immigrants Perceived Tsetse Challenge as or not
Important When They Chose a Settlement Site in Gokwe North Before, During and
After Tsetse Control, Zimbabwe, 1996

 

 

 

 

Importance of tsetse/trypanosomosis Before During After
I had many cattle which would have died 2 11 9
I wanted to avoid tsetse bites and sleeping sickness 0 2 12
I knew tsetse control was ongoing l 8 0
Total number in sub-group 3 21 21
% Subgroup in valid sample 12.0 46. 7 33. 9
Insignificance of tsetse/trypanosomosis

I only wanted land 8 6 11
I knew nothing about tsetse 8 9

There were settlements there and people survived 4 2

Tsetse flies had been cleared 0 1 18
I had no livestock 2 6 3
Total number in subgroup 22 24 41
% Subgroup in valid sample 88.0 54. 3 66.1
Missing cases 1 0 5
Total sample 27 47 68

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97.

45

The questionnaire indicates that over half of the migrants who moved in during tsetse
control considered the degree of tsetse challenge as important in their decision of where to
settle. One of the reasons (53%) this group of migrants gave a premium to tsetse
controlled areas was because they owned cattle which could have been lost due to
trypanosomosis (Table 3.8). The other migrants wanted to keep cattle and chose areas
that they knew were controlled or were about to be controlled. A few of these migrants
wanted to avoid the nuisance of tsetse bites. Those migrants who moved in during tsetse
control but discounted the importance of tsetse control did so because of ignorance of

tsetse (36%), non-ownership of cattle (33%), and interest only in obtaining land (31%).

Table 3.9 shows respondents’ assessments of the importance of public amenities to the
inflow of migrants. When migrants selected potential sites for settlement, about two thirds
of them felt they gave a premium to those villages that had good access to public
amenities. The most important amenities, in terms of the number of times respondents

referred to them, were schools, retail shops, roads and health centres.

The group of migrants who discounted the importance of the infrastructure did so for
difl‘erent reasons. Forty-five percent believed that new settlement would stimulate the
development of public amenities. The remainder indicated that they were primarily

interested in gaining access to more land and that everything else was of minor

46

importance. Villages that were abundant in land were sparsely settled and poorly serviced
with amenities.
Table 3.9: Reasons why immigrants perceived infrastructure to be important or not

important when they chose a settlement site in Gokwe North before, during and
after tsetse control, Zimbabwe, 1996

 

 

 

Reason infi'astructure was important Before During After
Good roads required for public transport 3 5 13
Health centers needed for disease treatment 4 2 9
Infant schools within walking distance convenient 2 13 17
Short distances to grocery stores are convenient 3 6 15
No reason given 1 l 0
Total sample 13 27 54
Reason infrastructure not important

There was no infiastructure to think of 2 1 3
The infrastructure is attracted by settlement 4 15 4
I only wanted land 7 4
No reason given 1 3

 

Total sample 14 20 14

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97.

 

3.5 Summary

Abandonment of old settlements was occurring due to overcrowding in Masvingo and
Midlands provinces. Emigration fiom those areas was inspired by people’s desires to
access more farm resources in order to increase household food security and produce cash

crops. The overriding desire behind emigration was getting access to more land.

47

Tsetse control had a large and significant impact on the rate of immigration; everything
else equal, migration increased by 150% after tsetse control. It is not clear whether
immigrants were reacting to tsetse control or other broad but concurrent national policies.
Nonetheless, immigration started well before the areas were cleared of tsetse flies.
Whether immigrants anticipated tsetse control when they moved into settlements with
tsetse control is a moot point. Settlers who did not own cattle at the time of migration
appear to have been less concerned about the presence of tsetse and trypanosomosis.
Other settlers who moved in with cattle settled in areas with disease control in anticipation

of vector control.

Besides tsetse control, primary roads were important public infrastructure. Good roads
reduced the transportation and communication costs that were important in market
transactions. Many of the primary and secondary roads in the sites started as tracks
established by tsetse control gangs as they moved supplies around the area. Good roads
and services also facilitated migration. Migrants stated their preference for villages with
roads, schools, clinics and convenient stores nearby. However, such a desire was mostly
unmet because it was secondary and only relevant alter arable land had been secured.
Failure to find villages with nearby schools and clinics was not critical to migration.
Immigrants may have expected that the amenities would follow immigration after the areas

became more populated.

48

CHAPTER 4: FARMING SYSTEMS CHANGE WITH TSETSE CONTROL AND
IMMIGRATION

4.0 Introduction

The focus of this chapter is on (1) the effect of increased immigration and investment in
tsetse control on the rate of adoption of cattle per village and (2) the impacts of
introducing animal traction technology on farm productivity. In addressing the first
objective, this chapter will seek to explain how tsetse eradication affects the rate of cattle
adoption at the village level. One of the study’s hypotheses is that cattle adoption was
feasible with disease control but the impact of tsetse control was to enhance the returns to
cattle adoption. To address the second objective, the study will provide answers to
whether animal traction use and ownership contribute to difference in the stage of
operation on the production function. The study hypothesizes that renters and owners

operate within the same stage along the production filnction.

4.1 Adoption of Animal Traction Technology

Adoption can be considered a problem of reallocating resources in response to
opportunities provided by the introduction of new techniques (Schultz 197 5). The
probability of a farmer adopting a particular technique can be influenced by several factors
including farm size, quantity of family labor, proximity to market, human capital, capital
availability, input and output prices, agricultural information, production uncertainty,

nonfarrn income earnings and risk (Rauniyar & Geode 1996). Wozniak (1986) examined

49

the extent to which difi‘erences in human capital and knowledge of the new technology
explain differences in the speed with which individuals adopted new techniques. The
evidence shows that education and information enhanced the ability to reduce costs of
adoption. Cattle adoption in Gokwe North, is hypothesized to be endogenously driven by
the process of in-migration. Immigrants into Gokwe North had training and experience in
cattle breeding prior to in-migrating and were instrumental in spreading adoption among
indigenous settlers. In Chiawa, Zambia, B1inn(1996) found that the introduction of cattle
into a tsetse control zone was due to in-migration. Immigrants were experiencing a
shortage of pastures in adjacent tsetse cleared areas. In Chiawa, Brinn found that 65% of
the immigrants moved into the control zone with cattle and the remaining immigrants
acquired cattle alter settling-in. Cattle acquisition by indigenous settlers of Chiawa was

well below the levels of immigrant settlers.

Immigration was one among several driving forces for adoption. The reduction in the risk
of contracting trypanosomosis was an important element in cattle adoption in Gokwe
North. Although tsetse flies presented an enormous challenge to cattle breeding and made
it a risky venture, Reid (1994) found that tsetse fly control was not necessary for the
introduction of cattle breeding in the Zambezi Valley. The study hypothesizes that the
introduction of cattle in the Gokwe North was possible due to the availability of
preventative and curative drugs. According to findings by Reid (1994) and Pender and

Rosenburg (1995), the acceleration in livestock productivity growth in herd size and

SO

increase in number of owners was strongly depended on tsetse control. Tsetse control

reduced the risk and health management costs and boosted returns to adoption.

Animal traction is a relatively costly package or lumpy investment for most smallholder
finners. Non-farm income, therefore becomes a crucial liquidity source for investments in
capital formation (Savadogo et a1, 1998) substituting for formal or informal credit. This
alternative source of income makes firmers more amenable to bearing the risk of firm
investments including purchase of cattle. Savings fiom wage earnings, remittances and to
a lessor extent gifts/grants positively afl’ect the probabilities of adopting animal traction

technology.

Farmers’ ability to invest in farm capital is also contingent on making a significant
production of a marketable surplus. Households who are less commercial oriented are
expected to have liquidity problems limiting their ability to invest in capital formation.

The relationship between commercial crop production and farm capital investments is
synergistic. Farmers who rely on hand tillage are expected to market a small proportion of
their output and unless these farmers have non-farm income, they struggle to invest in
capital formation and fail to increase the proportion of their output they market as surplus.
Savadogo et a1 (1998) studied the links fi'om commercialization to animal traction
investment and subsequently to technical and allocative efficiency. This study

hypothesizes that during periods when cash crop prices are attractive relative to

51

subsistence crops, finners take advantage of the opportunity to invest in capital formation.
Relatively high cash crop prices and low food prices are important incentives for farm

investment.

4.2 Estimating Rate of Livestock Adoption

To explore the efl’ects of tsetse control and immigration on cattle adoption at the village
level, a two-stage least square estimation was used. Immigration and cattle adoption
occurred simultaneously within a village and it is therefore necessary to use an
instrumental variable approach. This required developing a conditional cattle adoption
function and an immigration function. The immigration function developed in the previous

chapter is adopted to complete this task.

The cattle adoption variable is an aggregate annual measure of the number of settlers
adopting cattle for the first time in a village. These data are used in order to capture the
shape of the time dependent diffusion process. The pattern of adoption is modeled as a
logistic-shaped exponential function. The theoretical framework of the determinants of

immigration on cattle adoption is presented in the two equations below.
The conditional cattle adoption function is:
lnC" = 1M, +A,lnX,, +A,InM,, + rt + 11,, (i = 1,... 12 villages; t = l,...33 years); (4.1)

And the immigration function is:

52

lnM, = InB, + B,InX, + BJnF, + rt + v, (4.2)
where C, is the stock of cattle adopters in a given year and village; X, is a vector of
exogenous village and macroeconomic variables; M, is a vector of endogenous variables;
and F, is a vector of instrumental variables used to predict C, and u, and v, are residual
terms and r is a constant — the instantaneous rate of growth of cattle adopters in a village

at any given time. The specific exogenous, endogenous and instrumental variables and

their expected signs are found in Table 4.1.

Table 4.1 Description of Variables Included in Models

 

Anticipated Sign
Stock of Stock of
Immigrants Cattle

 

Adopters
Exogenous Variables (X I)
Location with recent tsetse control =1 otherwise = 0 ‘7 ?
Location with mid tsetse control =1 otherwise = 0 '7 7
Location with early tsetse control =1 otherwise =0 ‘7 ?
Time trend (number) + +
Cotton to Maize Price Ratio + +
Period with tsetse Control = 1 otherwise =0 + +
Endogenous variables 014,)
Stock of immigrants in village in a given year +
(number)
Instrumental variables (F ,)
Access to public transport good =1 otherwise =0 + IV
Access to schools good =1 otherwise = 0 + IV

 

53

The selection of instrumental variables was driven by the hypothesis that these variables
are strongly correlated with stock of immigrants and weekly associated with the stock of
cattle adopters. The construction of roads was hypothesized to be critical in attracting
immigrants who had to move in with their belongings. Vlllages with good access to
schools were also expected to attract more immigrants than those villages without
schools. The majority of the immigrant firnilies were young and their children were
attending school.

4.3 Animal Traction and Factor Productivity

Animal traction studies in Sub-Saharan Afiica have filled to distinguish the impacts of
ownership and use of animal traction. Ownership of draft animals and ploughs does not
necessarily translate to use while use of animal traction services does not require
ownership if a market for custom service exists. Previous animal traction studies focus on
technical and allocative efl'lciency by examining impacts on cropped area, yields, cropping
pattern and labor productivity. In Sub-Saharan Afiica, evidence on the direction and

magnitude of the impacts is not entirely consistent.

A study on animal traction conducted by Panin (1989) in Ghana reported that animal
traction users held larger land holdings than hoe users. However, Panin could not
establish causality and failed to substantiate whether larger land endowment was a product

or cause of animal traction use. Savadogo’s (1998) study suggested that large farm

54

holdings doubled the probability of adopting animal traction. A study in Upper Volta by
Barrett et al (1982 & 1984) and another by Famham (1996) in Togo, found animal
traction users cropping larger areas per family than non uses. In Ethiopia, Gryseels et a1
(1984) found a related pattern between the number of oxen owned and the area planted
per fimily. It is not clear whether the differences in cropped area are due to larger family
sizes or larger numbers of active workers. For example, Famham (1996) found that
firnilies of animal traction users grew fister than that of hoe users because the former
married additional spouses but Famham could not attribute this fister growth to use of

oxen.

Neither is it easy to attribute use of animal traction to size of family but Savadogo’s study
(1998) did confirm results from other studies that families with four or less adult
equivalence found it difficult to use animal traction. Barrett (1984) found insignificant
difi’erences in cropped area per capita between users and non-users. He argued that ’
difi’erences are better reflected by area cropped per active worker than per capita. As
Barrett (1984) expected, Panin (1989) found the cropped area per active worker higher
for animal traction users than hoe users. Evidence of how animal traction use could cause
the area cropped to increase is found in Ethiopia where a study by Astatke and
Mohammed Saleem (1996) confirmed that use of animal traction technology enabled
improved surface drainage of vertisols and allowed farmers to utilize such soils. It is also

clear that in agro-ecological zones with a relatively short rainy season, timing of land

55

preparation and planting operations are critical for overall production success. \Vlthin that

short planting period, plough users and ox owners in particular, might be able to plant

more hectares/day than hoe users.

The impact of animal traction ownership on crop yields is identified as positive but the
significance of this impact is doubtful. Animal traction owners obtain higher yields than
non-owners and evidence fi'om Zimbabwe by Shumba (1994) confirms this. Animal
traction owners had access to manure which improved soils nutrient status. However,
Savadogo (1998) and Sanders et al (1996) found the effects of manure use insignificant
because of low application levels. While this efl‘ect can be observed for owners, the efl‘ect
on land productivity of using animal traction is mixed or insignificant at best. A case study
on productivity determinants in Burkina Faso by Reardon et al (1996) reported strong
impacts on land productivity. Panin (1989) and Ramisch (1996) reported increases in land
productivity due to use of animal traction on average but particularly for major crops. In
contrast, Barrett (1984) and Savadogo (1998) only found modest increases in land
productivity. The causality between animal traction use and crop yields is diflicult to
identify and quantify without controlling for numerous other factors that affect yields,

including soil nutrient status, crop variety, planting date and others.

Evidence on the impact of animal traction use on crop diversification is relatively

consistent. Plough users commercialize to a higher degree than hoe users. Gryseels

56

(1984), found animal traction use associated strongly with cropping patterns. Barrett
(1984) and Panin (1989) found very slight shifts fi'om staple food to cash crops. Dionne
(1989) and Famham (1996) found that animal traction users grew more food and cash
crops. The debate over the shift towards commercialization cannot be adequately
addressed by looking at food versus cash crops. Instead, the proportion of output that is
marketed as surplus is a more comprehensive measure and provides a better indicator of
cropping system changes. An important finding from Dionne (1989) and supported by
Famham (1996) is that animal traction users were also self-sufficient in grain. Hoe users

were potentially vulnerable from food and income insecurity.

Studies on the impacts of animal traction use on labor productivity show overall positive
effects. By examining the average value product, Savadogo (1998) found the effect of
animal traction use on labor productivity to be higher than the effects on land productivity.
Panin (1989), Famham (1996) and Ramisch (1996) found that use of animal traction
reduced labor requirements per family during land preparation and planting but disagree
on the impacts on weeding. Ramisch (1996) contends that the requirements per family for
weeding will be reduced but Panin (1989) and Famham (1996) suggest that labor required
for weeding increases. The qualifier could be that adoption of secondary tillage reduces
labor required for weeding per hectare but requirements per family might increase when
the cropped area increases as well. However, use of animal traction causes a labor

bottleneck for the family during harvesting and the evidence for this is consistent. Overall,

57

the labor demand per hectare cropped is significantly reduced (Barrett 1984) suggesting

that the impacts on labor productivity are strong as shown by Reardon (1996).

Comparisons of “with and without” animal traction have dominated the literature while
comparisons of “before and alter” are very few (Petheram et al 1989). The explanation
advanced by Petherarn (1989) is that “with and without” scenario comparison tends to
overestimate the impacts of animal traction unlike “before and after” studies. He alleges
that the former approach confounds other factors like family size with the impact of animal
traction (Petheram et al, 1989). The “with and without” comparison is the preferred
method because it holds other factors constant. Historical accounts of before and after, '
however, assume that these other fictors e.g., family and firm size are constant, but they
also change. This methodological challenge requires a longitudinal approach in order to
capture changes of other factors over time and separate their effects from the impacts of

animal traction use.

While one can perform comparisons of plough and non-plough users, establishing
causality due to animal traction use is difficult. The performance difference between
plough and non-plough users is a firnction of other variables apart from animal traction.
Barrett (1984) found that plough users had higher education, were innovative and more
successful than hoe farmers. To determine the impact of animal traction on crop

production requires the elimination of the impacts due to other variables first.

58

Production function analysis has been the common approach for estimating the impact of
animal traction apart from other factors. The magnitude and direction of the coemcients
in the estimated production function model indicate the influence of each production
factor. The values of the estimated parameters are the production elasticities of the
respective production factor. For each production factor, the elasticity indicates the
percentage change in total output that would occur by a unit percentage change in that
particular factor holding all other factor levels, constant. The factor with the highest

elasticity value is the factor that will increase output by the highest margin.

4.4 Estimating Impacts of Animal Traction

To analyze the responsiveness of factor productivity to investment in capital formation at
the household level, an econometric model is developed to model the relationship between
gross value of total crop output with investment in animal traction, farm size, family size
and education of decision makers. Given that all the explanatory variables are exogenous
to the dependent variable, an OLS estimation was used to explore the effects of these
factors on aggregate crop production. Owing to lack of crop specific input data, an

aggregate total farm production function approach was adopted.

The production function was of the Cobb-Douglas form:

Q, = AK'L‘NM. (i=l,....,453 households) (43)

59

Where Q represents the aggregate value (ZS) of crop output, Ki represents the value of
investment in animal draft and equipment (ZS), Li is the household size; D, is the farm size
(ha); M, is the level of education of husband and wife (years in school) and A represents
the state of technology. This Cobb-Douglas firnction form is linear in logarithms (4.4).
That is:

LnQ, = 1M + char, + bInL, + cInN, + dInM, + n, (4.4)

4.5 Adoption Trends and Characteristics of Farmer Adapters

Of the 452 households considered in this study, 61% had adopted cattle during their
period of residence in the study sites. Cattle adoption represents a lumpy type of
investment. As expected, a sizable proportion of the settlers had not generated adequate
resources to make this investment. The absence of a well-fimctioning livestock credit

market means cattle adoption depends on one’s personal savings.

Table 4.2 indicates the number of households and the period when they adopted cattle for
the first time in each village. About 25% of the adoption occurred before tsetse flies were
controlled. It is not well known whether the cattle were resistant to the disease. The
conventional wisdom is that such adopters had to rely on drugs to prevent the risk of
contracting disease and to cure trypanosomosis. Adoption of cattle during this period was
discouraged because it stretched out the resources of the department of veterinary services

responsible for administering the drugs fi'ee of charge. Another 41% of the sample

60

households adopted cattle during the period when tsetse flies were being actively
eradicated. Adoption of cattle during tsetse control operations was restricted to avoid

prejudicing the success of tsetse control campaigns.

Table 4.2 The Number of Additional Cattle Adopters in Villages 1963 - 1995,
Gokwe North, Zimbabwe, 1996'

 

Village by Period of Adoption
Name

 

<196 1966— 1971- 1976- 1981- 1986- 1991- Total
6 7O 75 80 85 9O 95

 

     

  

       

   

 

 

 

Tofe 4 3 2 5 6 5 28
Goredema 4 4 2 10 2 l 29
Nevana 2 7 o 4 l 0 20
Nyahomba l 4 2 3 2 3 l8
Kakwari I 0 1 l 13 7 8 3O
Kakwari H O l l ‘ 6 8 10 29
Nechinyika 0 O 2 2 I: 7 23
Mayava O 3 3 4 8 2 23
Chirudzi O 2 7 4 7 4 27
Kwaedza 4 2 1 5 6 o 26
Ntamo 0 0 0 0 O (3 6
Chirisa 2 2 3 6 2 ‘1 0 ___§_Q_____
TOTAL 28 17 29 3O 62 61 62 289
Total # of 11 9 6 11 45 30 29 141
Immigrants

 

‘the shaded cells indicate periods of active tsetse control in each village

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

Many migrants brought livestock capital with them to the Valley, Table 4.3. This is
particularly so for the migrants who moved into areas that had tsetse control relatively

recently. Among the migrants into the late clearance areas, 45% brought some cattle,

40% brought some goats, 18% brought some donkeys, and 7% brought some sheep.

61

At the time of the survey, 23% more of the migrant households held cattle, 8% more held
donkeys, 3% more held goats and 6% more held sheep. Indeed, migrant households were
more likely to hold cattle at the time they migrated as indigenous households were overall,
Table 4.4.

Table 4.3: Migrants’ livestock holdings at the time that they migrate into Gokwe
North District, Zimbabwe, 1995/96

% households holding species at time of immigration

Early clearance Mid clearance Late clearance

 

Cattle 21 26 45
Donkeys 10 l6 l8
Goats 25 16 40
Sheep 6 8 7

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

Table 4.4: Percentages of pioneer and newcomer households owning livestock at time
of Survey in Gokwe North District, Zimbabwe, 1995/96

 

 

Pioneer Settlers Newcomers
Cattle 38 61
Donkeys 37 45
Goats 76 79
Sheep 18 24

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

62

Sources of finance to acquire livestock suggest the existence of a possible synergistic
relationship between livestock and other economic activities, Table 4.5. Production of
cotton was a major activity generating cash to purchase livestock. Other crop sales
included maize. Besides crop surplus, off-farm employment generated savings used in

livestock acquisition.

Table 4.5 Major sources of finance for acquiring livestock, Gokwe North District,
Zimbabwe, 1995196

% of households with species

 

Cattle Donkeys Goats Sheep

 

Cotton sales 53 49 38 38
Cotton & maize sales 14 16 13 12
Other crop sales 5 10 12 12
Gifis 7 13 12 19
Salary savings 18 9 18 17
Non-agric. Sales 3 3 6 2

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Zimbabwe, 1996-97

4.6 Characteristics of Households Using Different Tillage Systems

Table 4.6 presents descriptive information about production practices and resource
characteristics of smallholder producers using difi‘erent tillage systems. There are large
difi‘erences between ox owners and ox renters and between ox renters and those who

cultivate by hand. These trends are especially strong for farm size, grain output per capita,

63

total income per capita, degree of self-suficiency, marketing of grain, and degree of

commercialization.

These trends do not hold for grain and total crop yields. On the contrary, yields are higher
for hand cultivators than for ox owners. In sum, however, multiplying yields by area, ox
owners generated much higher revenues per capita than either ox renters or hand
cultivators. Furthermore, ox renters generated twice as much revenue per capita than hand

cultivators.

The commercialization index measured by the proportion of output that is marketed shows
that ox owners were most commercialized marketing about 78% of their output as
compared to 51% for hand cultivators. The proportion of cropped land placed under
cotton was 175% higher for ox owners than hand cultivators. The proportion of ox
owners who were self-suficient in grain was 62% as compared to 49% for hand
cultivators. Animal traction facilitated a higher degree of both commercialization and self-
sufiicient production of food crops than hand cultivation. There was no difl‘erence in the

grain commercialization index between these smallholder groups.

Ox owners invested most in farm equipment than the other two groups. Farm equipment
made crop production and carting inputs and products less laborious. Farmers without

access to transport resorted to head-loading.

64

Table 4.6: Characteristics of Farming Systems by Type of Tillage, Gokwe North

 

 

District, Zimbabwe, 1995/96

Characteristics

Oxen Oxen Hand Total

Owners Renters Tillage
Grain Yield (kg/ha) 1218 1070 997 1136
Grain Output (kg/capita) 286 368 468 386
Gross Grain ZS/ha 1523 1337 1246 1348
Gross Grain ZS/capita 357 461 585 483
Total Crop ZS/ha 2440 2272 23 52 2341
Total Crop ZS/capita 962 1570 2237 1676
% Cotton to total output 38.77 57.84 66.96 56.89
%Grainsurplus to total output 11.93 11.13 10.76 11.17
% Sales to total output 50.70 68.97 77.72 68.06
% Area under Cotton 28.0 41.6 46.0 40.2
% Cotton Growers 57.5 86.9 92.8 82.6
% Food self-sufficient 48.9 58.2 61.8 57.3
% Selling grain 14.9 27.7 34.9 27.3
Land Size (ha) 4.24 5.51 7.67 6.1
Fallow Area (ha) 1.13 .81 .73 .85
Family Size (No) 6.3 6.2 6.9 6.5
Animal Traction Investment (ZS) 1720 3718 12090 6241
Years Farming (yrs) 13.3 12.1 13.5 12.8
Household head’s age (yrs) 45.7 43 47.6 45.3
Total years in school for husband 7.23 10.71 11.22 10.1
and wife
Distance to Market (km) 26.9 25.9 22.6 25.0
Late tsetse control (%) 41 36 24 33
Mid tsetse control (%) 1 31 58 34
Early tsetse control (%) 58 33 18 33

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North District, Zimbabwe, 1996/97.

65

4.7 Growth in Cattle Adoption

Table 4.7 shows results of the OLS estimation of immigration flows and an instrumental
variable model of cattle adoption. The results of the immigration model, equation 3.4 in
Chapter 3 have been discussed already. This section will show how the number of new
cattle adopters in each year for each village changes when immigration increases while
holding other factors constant.

Constant: The estimate of the constant predicts that there were 1.5 cattle adopters per
village prior to 1963. Although this result was least expected, it was insignificant. The
degree of tsetse challenge made cattle adoption very risky. According to this model, there
is no evidence to suggest that ownership of cattle in Gokwe North can be entirely

attributed to in-migration and tsetse control.

T setse control gradient: Adoption of cattle in those villages cleared of tsetse flies earlier
was higher than the levels of adoption in recently cleared areas. The difi‘erences in the
levels of cattle adoption between villages along the tsetse control gradient were significant
(p=.01). Although all villages are free from tsetse, the adoption levels difl‘er. Besides the
gradient of time of tsetse control, other factors may play a role in determining the levels of
adoption. Recently cleared areas have newer settlements and settlers may still be adapting

to their new situations and not ready to make investments.

66

Period Wrth or Without Tsetse Control: The result shows an unexpected 40.15% more
adoption of cattle without tsetse control than with control. This result was significant at
p=.044. The adoption of cattle by settlers appeared to be a step ahead of active tsetse
control efforts. Such levels of adoption prior to tsetse control could have been induced by
in-migration into these villages. Immigrants illegally drove cattle into these villages prior
to tsetse control. Settlers in tsetse infested villages could adopt cattle as long as they used

preventative and curative drugs against trypanosomosis.

Growth rates: Although insignificant, the uniform rate of grth of the cattle adoption
function r, was -2.23% per year between 1965-96. Ifthis rate r prevailing in 1996 is
allowed to continue undisturbed, then C“ will have grown by the amount rC“ by the end of
1997 . The stock of cattle adopters in each village was increasing at a decreasing rate each
year but the absolute amount of increment of C, increases as time goes on. This negative
growth rate is not a surprise. Innovative settlers with the resources to invest have already
done so. Potential adopters may be laggards struggling to break the poverty cycle. Lack
of formal livestock credit markets may explain why there is a negative growth rate in cattle

adoption over time.

67

Table 4.7: Results of Econometric Models of the Flow of Immigrants and Cattle
Adopters, Gokwe North District, Zimbabwe, 1965 -96

 

Immigrants in Each Cattle Adopters Per Village

 

Village Each Year Each Year (IV)
(OLS)

Endogenous Variables Coeficient T-stat Coefficient T-stat
Immigrants/village/year 1 .9024 6.307”
Exogenous Variables

Cotton/Maize Price -.288 -2.756** .4530 2.060”
Ratio (1 year lag)

Tsetse Clearance .379 3.651" -.5134 -2.020"
Time trend .0457 8.508" -.0226 1.347
Early Cleared Ward 2 .455 4.482" -1.1496 4.938"
Mid Cleared Ward 3 .0208 .176 -1.0550 4.786“
Mid Cleared Ward 4 .183 1.531 -1.2674 -5.481“
Late Cleared Ward 5 .321 3.030“ -2. 1661 -8.843"
Late Cleared Ward 6 -.354 -2.938** -.2730 -1.270
Constant .652 3.766" .4096 1.147
Instrumental Variables

Availability of schools -.393 -5.728"

Availability of public .392 4.200“

transport

SE. .5540 1.0806

N 383 383

Adjusted R-square .607 .4577

 

" denotes statistical significance (one-tailed test) at the 5% level
Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North District, Zimbabwe, 1996/97.

Elasticity of Cotton to Maize Price Ratio: The adoption of cattle increased significantly
as the ratio to real cotton to maize price increased. The results reveal a less than unitary

measure of responsiveness, that is, a 1% increase in the ratio was associated with a .453%

68

increase in the level of cattle adoption. This result was as predicted. A shift in the balance
of cropping incentives in favor of cash crops made livestock adoption more likely.
Whenever the macroeconomic environment makes food cheap relative to farm incomes,

farmers are presented with an incentive to make farm investments.

Elasticity ofimmigrant Stock: The results in Table 4.6 highlight the complementary
relationship that exists between immigration and cattle adoption. The immigration
elasticity of cattle adoption was elastic achieving 1.9% increase in cattle adoption for each
1% increase in immigration. Immigrants were instrumental in the spread of cattle

adoption in the study sites.

4.8 Production function analysis

The estimated model is presented in equation (4.5) and other statistics are in Table 4.7.
Q, = 709.1K”’L""N”7M"” (4.5)

The model explains 32.4% of the variation in the aggregate value of crop output between

the sampled households. The overall model is statistically significant. The efl‘ect of all the

variables on crop output was significantly different from zero at the 1% level of

confidence.

69

Table 4.8: Results of Cobb-Douglas Model of Value of Aggregate Crop Output,
Gokwe North District, Zimbabwe, 1995/96

 

Dep. Var. is Total Gross Crop Value
(ZS)

 

Coefiicient T-statistic Mean

 

Exogenous Variables

Farm Size (ha) .527 7.856" 6.04
Family Size (adult equivalence) .214 2.767“ 6.46
Value of Animal Traction (ZS) .075 . 5.364" 6241
Level of Education (years in school) .101 2.386" 10.1
Constant 6.564 35.095”

8.12 .853

N 425

Adjusted R—square .324

 

” denotes statistical significance (one-tailed test) at the 5% level
Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North District, Zimbabwe, 1996/97.

4. 8. 1 Elasticities:

The elasticity of total crop value with respect to firm size was statistically significant fiom
zero. An increase in firm size by one percent increases output by .527%. An increase in
firm size by one percent from 6.04 to 6.1 hectares costs $14.82 in the rental market and

led to an increases in output of S41 .66, all other input levels constant.

The elasticity of output with respect to labor was statistically significant. An increase in

labor by one percent leads to an increase in output of .214%. This responsiveness means

70

that an increase in crop value of S16.92 is achieved with an increase of S500 in labor

costs, ceteris paribus.

The elasticity of output with respect to investment in animal traction and equipment was
statistically significant fiom zero. For every increase in capital of 1%, output increased by
.075%. In value terms, an increased investment in animal traction of S6241 was
associated with an increase in output of S593. On fice value, this magnitude of
responsiveness appears unattractive. However, investment in animal traction has multiple
purposes and lasts for several production seasons. A careful estimation is required to
assess the profitability of adopting animal traction at the household level. The results also
show that the quality of management, measured by the education level of both decision
makers, significantly afi‘ect crop output. Crops such as cotton demand a high level of
discipline in input management. Decisions such as determining economic injury due to

pest infestation are very technical.

4. 8.2 Returns to Scale:

The returns to scale exhibited by the estimated production function shows that there is less
than proportionate increase in output when all inputs are increased. This result indicates
that there are substantial problems with increasing firm size above the average of 6.04
hectares. There are logistical problems in managing a larger pool of workers (especially if

they are hired), additional farmland (especially if the plots are in different locations) and

71

additional firm animals (especially with limited pasture). Hence careful calculations are
required in deciding to tradeoofi‘ the advantages and disadvantages of large smallholder
firms.

4.9 Ownership versus Hiring Animal Traction

The returns to factor productivity between ox owners and ox renters are not necessarily
uniform. What is also certain is that neither the strategy for intensifying production nor
approaches to increasing technical efficiency at the firm level are uniform across farmers
with different capacities. The following results reveal important insights that can help
increase understanding of various alternatives for increasing firm productivity for firmers

with diverse resource capacities.

The results fi'om the production firnction analysis are used to estimate fictor productivity,
technical emciency, rate of technical substitution and cross-productivity effects. Table 4.8
reports estimates of these indicators for three types of farm households: firrners who use

and own oxen; farmers who rent-in oxen; and farmers who use hand tillage.

4. 9. 1 Marginal Productivity
Differentiating equation (4.5) with respect to each factor generates estimates of the value

of marginal products. The value of marginal product of a production factor is the value of

72

firm output that results from a unit increase in the production fictor, all other factors

constant.

The results show that ox owner’s labor is most productive at the margin Table 4.8. This
group of firmers can get the highest returns to additional investment in labor. Ox owner’s
labor is 19% more productive than ox renters, but 57% more productive than hand tillage
firmers. Farmers using hand tillage are most productive in their use of land and capital at
the margin This is the case because these firrners have limited capital investment. Hoe
tillers can obtain highest returns by investing in additional farm capital than the return ox
owners and ox renters would get by additional investment in animal traction. The
productivity of firm capital investment of ox renters is more than double that of ox
owners suggesting that ox renters could get higher returns for marginal investment in firm
capital than ox owners. Ownership of oxen appeared to generate important differences in

partial factor productivity.

4. 9.2 Marginal Rate of Technical Substitution

Table 4.8 also shows estimates of the rate of technical substitution between labor, capital
and land. The marginal rate of technical substitution shows the rate at which labor can be
substituted for capital while holding output constant. At the average level of input
combinations, labor substitutes for nine times the amount of capital of extensive users than

hoe-tillers. This result implies that extensive plough users are nine times more capital

73

extensive than hoe users. Among ox renters, the value of capital that a unit of labor can
substitute for is less than a third more than what labor substitutes for among hand tillage
farmers. Overall, ox owners had a higher marginal rate of substitution for the three inputs
implying that they were also land and labor extensive than either ox renters or hoe-tillers.
Unlike access, ownership of oxen appeared to generate differences in factor

extensification.

4. 9. 3 Cross-Productivity Eflects

Table 5.8 also reports estimates of the cross-productivity efi‘ects between labor, capital
and land for the three groups of firmers. Cross-productivity effects are derived from the
cross-partial derivative and are a measure of how an increase in one factor effects the
marginal productivity of another factor. If hand tillers had more capital, they would
increase the marginal productivity of labor and land by a magnitude four times higher than
that achieved by ox owners. By increasing capital at the margin, ox renters would
increase marginal labor productivity by a factor two times greater than the increase
achieved by ox owners. Additional amount of land increased marginal labor productivity
to ox owners by the greatest margin but the difi‘erence in the increase in marginal labor
productivity between ox owners and ox renters is very small. Additional levels of labor
increased marginal productivity of capital to hand tillers by the greatest margin. The
increase in marginal productivity of capital by increasing labor at the margin was twice

greater for ox renters than owners.

74

Table 4.9: Value of Marginal Products, rate of Technical Substitution and Cross
Productivity Effects for Land, Labor and Capital by Traction Use, Gokwe North

 

 

District, Zimbabwe, 1995/96
Extensive Casual plough Hoe Tillage All firms
plough users users
Average Product Value
Land 1375.2 1318.2 1087 1308.7
Labor .875 1.99 2.67 1.27
Capital 1533.5 1197 728.4 1224.1
Marginal Productivity
Land 557.10 574.43 599.26 578.67
Labor 252.39 211.31 161.73 218.06
Capital .05 .12 .21 .08
Technical Efficiency
Land .405 .436 .551 .442
Labor .058 .06 .079 .063
Capital .165 . 176 .221 .178
Rate of Technical Substitution
RTS “who,“ 5047.8 1056.55 770.14 2725.75
RTS mum .453 .368 .270 .377
RTS mum 11142 4789 2853 7233
Cross-Productivity Effects
Mvpmwarcw .00318 .0079 .01232 .00506
MVP WW,“ 50.73 49.24 43019 49.06
MVP cw “Tm .00235 .00612 .0105 .00392
MVP ”“7“,“ .00701 .0215 .0457 .0134
MVP MWTM .01005 .02796 .05607 .018
MVP ”“7”,“ 26.184 29.311 29.963 28.355

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control

Surveys, Gokwe North District, Zimbabwe, 1996/97.

Increasing labor units at the margin generated almost equal impacts on marginal

productivity of capital across farm types. The results also indicate that hand tillers

increase their marginal productivity to capital the most by increasing land at the margin.

'75

The marginal productivity of capital for ox renters for marginal increases in land is also
greater than that of ox owners. However, hand tillage firmers and ox-renters cannot
increase their land given constraints to their production process. It is clear, that to
increase the marginal productivity of land and labor, intensifying capital investment was
the best option for both hand tillers and ox renters. As for ox owners, their advantage was
in increasing labor productivity by employing additional land in the production process.
Unlike access to animal traction services, ox ownership made a difl‘erence in the strategy

for increasing fictor productivity.

4. 8. 4 Technical Efliciency

By synthesizing marginal and average products into a single measure (Savadogo 1998),
we obtain an indicator of returns to fixed factors. The ratio of marginal to average
products identifies the stage on the production firnction the farmer operates. The ratios
reported in Table are between 0 and 1 suggesting that all the firm types are in stage 2 of
the production process which is necessary for technical efficiency but not sufficient for

economic efliciency. The ratios for hand tillage firms are the largest.

The results suggest that the production process of hand tillage firmers is pushing toward
the limit of productivity relative to the animal traction firms which means that
intensification investment is needed to maintain factor productivity. This evidence also

suggests that hand tillage firms are relatively land and labor abundant. The differences in

76

ratios for all factors (capital, land and labor) between ox owners and ox renters were very
small. These results suggest that ownership of oxen did not make much difi‘erence in

technical efficiency. What matters most is access to animal traction services.

77

CHAPTER 5: DETERMINANTS OF WEALTH ACCUMULATION AND

SOCIOECONOMIC MOBILITY

5.0 Introduction

Throughout the 20" Century, the Zambezi River Valley has had an economy characterized
by settlement of new land by immigrants each year, (Govereh 1998). A major
preoccupation of researchers has been studying the pace of wealth accumulation and the
extent to which the settlement fi'ontier presents equal opportunity to all immigrants. One
common hypothesis is that a fi'ontier such, as the Zambezi Valley, provides a safety valve
to immigrant households. The safety valve thesis assumes the frontier has a leveling
influence that promotes equality in both origin and destination areas (Turner 1920). The
paper hypothesizes that destination areas provide the basis for upward mobility even for

those immigrant households who are poor at the time of arrival.

5.1 Settlement frontiers as safety nets

The majority of studies (Soltow 1971; Kearl et al 1980 and Gregson 1996) examined
wealth distribution of cross-sectional samples at specific periods of time. A few (Steckel
1990) took the extra step to analyze mobility or change in wealth holding longitudinally.
This study uses the former approach to analyze wealth distribution of livestock, farm
equipment and land improvements in 1996 and the latter approach to study accumulation

of livestock and farm equipment wealth between 1987 and 1996.

78

Empirical evidence (Soltow 1971;Kear1 et al 1980 and Gregson 1996) acknowledge that
success of immigrant families in frontier settlements is not uniform. There is no widely
accepted economic theory to explain the determinants of wealth holding or its distribution.
The conventional approach is the rent analysis (Ricardo 1817). Ricardo’s analysis of
wealth distribution under conditions of population growth suggests a less equal
distribution to occur because of the rise in rent for those production factors with inelastic
supply, such as, land. More equality in wealth holding would be expected in the initial
periods of immigration rather than later when the economy is mature. Iffollows that new
settlements would be expected to have a more equal distribution of wealth than
settlements established earlier. The standard model of wealth accumulation allows income
Y to vary with age of household head (Gregson 1996). Assuming constant consumption
C, at age a, the individual household’s wealth is:
W, = Integral (Y, — C)e"‘”’dt (5.1)
where the constant r, is the interest rate at time t. The instantaneous growth rate of

wealth at age a and time t is the first derivative of wealth Wwith respect to age.

Other variables that have been associated with variation in household wealth are
household characteristics including time of immigration, place of birth, age, income,
literacy, household size and occupation. The socioeconomic advancement of immigrant
settlers goes through several stages (McMillan et al 1995) largely driven by the length of

residential tenure. Soon afier arrival, a settler goes through an adaptation or settling-in

79

stage. Settlers adapt for three to five years and during that period, they are preoccupied
with establishing food security. Settlers with more than five years of tenure are considered
as open to undergo a transition of economic and social development (McMillan et al
1993). The relative wealth for any settler would be a fimction of where they are on the

settlement cycle.

What is given less attention in the literature is the variation in socioeconomic status of
settlers at the time of settlement (Sewastynowicz 1986). The amount of capital held at the
time of settlement determines how rapidly the household advances on the socioeconomic
ladder. The prevalence of adverse physical conditions such as vector-borne diseases
negatively affects the pace of socioeconomic mobility. This issue has been partially
addressed through studies on settlements in the river basins of West Afi'ica infested with
onchocerciasis or river-blindness (McMillan 1995). In the Zambezi Valley, tsetse flies and
trypanosomosis make it extremely difficult to raise livestock. Immigrant settlers who
moved in well ahead of vector and disease control efforts would not be expected to be on
a significantly different socioeconomic standing with settlers who arrived after the vector
had been eradicated. Thus, the presence of tsetse flies and trypanosomosis wipes OR the

early arrival premium.

80

The following sections discuss the distribution and determinants of wealth among
immigrants. A sub-sample of only those immigrants who settled before 1987 is used later

to estimate the rapidity of wealth accumulation between 1987 and 1996.

5.2 Distribution of Wealth
Of the three sources of wealth studied, livestock assets provided the largest contribution

of about 42%, followed by land improvements and then equipment assets Table 5.1.

Table 5.1 Distribution of Per Capita Wealth by Wealth Source Among Settlers in
Gokwe North District, Zimbabwe, 1996.

 

 

 

Source of Per Capita Wealth
Land Livestock Equipment Total
Improvements Assets Assets
Average (S/capita) 1675 2084 929 4684
Percentage with S0 (%) 0 18.7 21.8 0
% Wealth Held by
Bottom 25% 10.73 .073 .31 6.24
Bottom 50% 27.95 4.32 7.01 17.81
Top 25% 46.94 79.48 74.29 61.66
Top 5% 17.58 41.84 32.38 28.73
Top 1% 7.56 20.04 15.37 12.39
Gini Coeflicient .32337 .72456 .64627 .62055

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North, Zimbabwe, 1996-97.

Despite the largest contribution of livestock investments to total wealth, about 18.7% of

the sampled firmers had neither small nor large nrminants. Every household in the sample

81

had made some form of investment in land improvements. Construction of residential
buildings is a necessity and a priority especially for recent immigrants. About 22% of the
sampled households had not made any investment in farm equipment. Households
without animal traction were of course not expected to make investments in animal drawn
equipment. The issue of what investment sequence farmers follow is debatable but simple

logic suggests that firmers without animal traction do not own much equipment.

Overall the distribution of wealth is leaning towards inequality with a gini coefiicient of
.6206. The main source of this skewed distribution is in terms of livestock and equipment
investments. With a gini coefficient of .7246, livestock investments represent the most
unequally distributed asset among the three sources studied. The top 25% of the farmers
in terms of livestock investments, owned about 80% of the total livestock assets and the
top 1% owned about a fifth of all the livestock wealth. Similarly, investments in farm
equipment were also concentrated. The top 25% farm equipment investors owned 75% of

the total assets.

In contrast, land improvements are more equally distributed with a gini coefficient of
.3234. This distribution suggests that such investments are essential as it includes
investments in residential buildings. The nature of the distribution also suggests that these

investments are a top priority particularly for recent settlers.

82

 

5.3 Association of Wealth Accumulation with Demographic Characteristics

The study established the patterns of association between independent sources of wealth

and demographic characteristics of the household. The results are presented in Table 5.2.

Age of Household Head: Evidence in this study supports the life-cycle hypothesis that
suggests an inverted U-shaped curve when wealth is plotted against age. Wealth
accumulation by settlers in Gokwe North increased with age and approached a peak
between 41-50 years of age and then wealth holding declines during the period of
retirement. This pattern is particularly evident in the case of livestock and equipment
investment. Notably, the 41-50 age group was significantly (p=.05) more resourceful than

younger fimilies and those with household heads above 60 years old.

Residential Tenure: When per capita wealth is plotted against years of residence, a pattern
similar to the inverted U shape is shown. This is a departure from conventional wisdom
suggesting that wealth is directly related with years of tenure. The view suggested in the
literature is that residential tenure is associated with knowledge of local circumstances,
information and social capital important in reducing transaction costs and prosperity.
While this may be the correct pathway by which tenure afi‘ects wealth accumulation, the

association revealed in this study was not sustained.

83

Table 5.2: Demographic Characteristics of Settler Households in Gokwe North
District, Zimbabwe, 1996.

 

 

 

Livestock Equipment Total Wealth
%N % Mean % MeanS % Mean
Sum (S) Sum Sum (S)
AGE (years)
<31 18.5 10.4 1173 14.4 721 14.6 36833
31-40 25.2 19.7 1628 21.0 773 21.7 4038
41-50 23.9 37.3 3242 30.4 1180 32.5 6373
51-60 14.9 17.5 2456 17.2 1074 16.0 5024
>60 17.5 15.0 1791 17.0 905 17.5 4074
Total/Average 100 100 2082 100 929 100 4684
TENURE (Years)
<5 24.4 13.7 1171 18.1 689 18.0 3553
5-10 19.4 18.2 1952 17 813 17.6 4256
11-15 21.0 37.2 3708 32.4 1438 32.6 7304
16-20 9.1 7.2 1661 8.8 900 7.6 3944
>20 26.1 23.7 1888 23.8 848 23.6 4238
LOCATION
Early 32.3 15.3 991 25.5 734 22.4 3624
Mid 34.0 61.8 3800 53.0 1451 53.7 7418
Recently 33.8 22.9 1421 21.5 596 23.9 3328
TIMING
Before Control 28.0 26.5 1974 26.5 880 25.7 4301
DuringControl 23.5 42.9 3799 33.9 1339 35.7 7116
AfierControl 48.5 30.6 1312 39.6 758 38.6 3726
SETTLERSTATUS
Indigenous 42.5 19.2 941 27.4 598 26.6 2930
2"“Gmigrants 33.4 31.4 1958 36.3 1009 34.3 4810
1‘Gmigrants 24.1 49.3 4259 36.3 1399 39.1 7594
GENDER
Female 20.5 25.5 2887 20.8 949 23.4 5752
Male 79.5 74.5 1950 79.2 925 76.6 4510
FAMILYTYPE
Polygarnous 23.5 13.6 1203 17.2 681 14.7 2932
Monogarnous 76.5 86.4 2352 82.8 1005 85.3 5222
NON-FARM
Yes 47.3 59.3 2590 49.5 966 53.9 5299
No 52.7 40.7 1619 50.5 895 46.1 4124

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North, Zimbabwe, 1996-97.

84

The association of tenure with wealth accumulation may be undermined by age. The
group of settlers who had been settled 11-15 years held significantly the greatest per
capita wealth. The difi‘erence in per capita wealth between recent settlers and early
settlers was not significant. The expected impact of residential tenure on per capita wealth

was not significant.

T setse Control Gradient: Locations that were along the mid-gradient of tsetse control had
significantly the greatest per capita wealth holding. Recently cleared areas were still
developing while early cleared areas might be experiencing pressure on resource access _
because of high settlement densities. There were no significant differences between early
cleared areas and recently cleared areas. The tsetse control gradient is, therefore, not a

good predictor of wealth holding.

Timing of Movement: Settlers who moved in during tsetse control had significantly
accumulated the greatest per capita wealth. The differences in per capita wealth holding
between settlers who moved in before and those who moved in after tsetse control were
not significantly. Moving in before tsetse control meant that one could not invest in
livestock while moving in afier tsetse control limited the settlers to less suitable residential
sites. Settlers who moved in during tsetse control minimized the risks of livestock
investment losses but they also had the opportunity to select some of the best locations for

siting residential as well as arable plots.

85

Settler Status: The settlers in Gokwe North could be divided into three distinct groups of:
indigenous settlers; 2" generation of newcomers; and 1" generation of newcomers. The
1" generation of newcomers was comprised of settlers who moved in from other Districts
in Zimbabwe while the second generation of settlers was comprised of children of the 1"
generation of newcomers. The 1" generation of newcomers held the most per capita
wealth especially livestock assets. While 1" generation newcomers comprised 24% of the
sample, they owned about half of the livestock wealth. The indigenous settlers were the
poorest and their wealth holding was significantly lower than the other two groups.
Observing that the 1" generation of settlers was self-selected in terms of immigration, their
access to capital and firming experience was above average. Indigenous settlers were
going through a learning process of capital accumulation but appeared to lag behind the

newcomers.

Gender: Gender appeared to have no association with per capita wealth holding. Female
headed households had higher per capita wealth than male headed households, but the
difl‘erences were not significant at p=.05. Gender literature suggests that female managers
are poorer because of institutions that are not sensitive to the plight of female managers.

In terms of per capita wealth, female managers were holding as much wealth as male

managers.

86

Family T me: It is commonly perceived that polygamous families are wealthier than
monogamous fimilies based on the ability of the husband to attract additional spouses.
The evidence from this study shows that polygamous families held significantly less
livestock, equipment and land investment wealth per capita. According to data fi'om this

study, the conventional perception is only a myth.

Non-Farm Employment: Access to non-firm income is considered important in the ability
of households to finance firm capital accumulation (Savadogo, et al 1998). Households
without access to non-farm income are hypothesized as less likely to invest in animal
traction equipment. In table 2, families with household heads who at one time had been '
engaged in formal employment had greater wealth holding per capita than those fimilies
with household heads who never had such an opportunity. This overall difference is
attributed to difi‘erent levels of investment in livestock wealth. Investments in equipment
and land improvements were not significantly different. The effect of non-farm income on

wealth holding was reflected in livestock assets.

87

 

5.4 Direction and Magnitude of Socioeconomic Mobility

5. 4. 1 Equipment mobility

Equipment investments by farmers are permanent in the sense that once farmers adopt a
particular equipment set, they will keep the equipment over time replacing it as and when
needed. As shown in Table 5.3 the mobility over the 10-year period is exclusively
positive, suggesting that firmer’s equipment asset holdings never decline. For example,
all the farmers in the top decile in 1987 maintained that position in 1996. Depreciation in
asset value is inevitable with or without use but the adoption of a piece of equipment
continued over time. Of the 111 farmers who had zero equipment investments in 1987, b
about 40% remained in that position in 1996 while 13% made it to the 5" decile and
above. Only 1% of the households made it to the top decile. While there were 84
households in the 1" decile in 1987, 53% of these household remained in that position and

23% improved to the fifth decile or higher.

Mobility in equipment investments appeared to be gradual. Big leaps in equipment
investments were not common within a ten-year period. Overall, ten years were not
adequate to have a filll-blown investment in equipment assets. This is particularly the case
because farmers self finance these investments given the absence and failure of equipment

credit markets.

88

Neona—
eoinaafimu .532 95.90 .2355 .8230 gueaemenaobh a... 0888 ..e «noEm8um< noun—n— u_Eeneooe_oem ”ooh-em

 

 

 

 

whN : v o_ 0— MN em m we o No 3 Z
o oo.~ o_ m
N oom. oom. o n.—
N ooo. ooo. oo _ w A
h m: . m3. omN. «NV. 5 m
o_ ooo. oo_. ooo. oom. oov. o O
o ooo. ooo. :_. _:. NNN. mmm. m n.—
: Nw_. ooo. ooo. ooo. ooo. th. MBN. v n—
on ooo. ooo. nmo. Mao. whN. om: . ooo. 3%. m
o ooo. ooo. ooo. ooo. ooo. ooo. ooo. ooo. ooo. N b
3 ooo. NS. N9 N5. avo. _n_. VNo. :N. ooo. omw. _ a
:_ ooo. ooo. ooo. 3o. «8. ooo. “No. we. ooo. mmm. com. o a
Z o— o w h o n v m N — o u
mas—HUM:— 9am—
.ooa— Jinan—EN

8.5.5 .....ez 255 Ge: 1 Sec .523 cone: ... 5.83 oceans—cm he nan: 183.35. "on uses

89

Table 5.5 shows the degree of association between mobility in equipment investments and
demographic variables. Settlement tenure appeared not to have any association with mobility in
equipment investments. On average, recent and early settlers had all moved one decile fiom the 3"
to the 4'II decile during the period 1987-96. In terms of the tsetse control gradient, areas cleared
of tsetse recently were behind in terms of equipment investments. Settlers in mid gradient had
moved 2 deciles. The degree of association suggests that investment in equipment may have been
impeded by the presence of tsetse flies. There was some association between mobility and
settlement status. Indigenous settlers appeared to lag behind other settlers and had only moved
one decile while 1" generation migrants had moved two deciles within the same period. In general,
newcomers held more equipment to start with and had moved twice as much as the first settlers.
While pioneer settlers were catching up with the rest, they were less agriculturally inclined ten or
more years ago. Again, settlers who settled before and after tsetse control had achieved a similar
magnitude of mobility in equipment investment. Settlers who moved in during tsetse control stood

above the rest.
Age appeared to have little association with equipment mobility. Although younger households

were in a disadvantaged position in 1987, they had caught up with the more mature households in

1996. The 1996 equipment investment levels had little association with age at that time.

90

Table 5.4: Average Livestock and Equipment Wealth Positions in 1987 and 1996, Gokwe
North district, Zimbabwe, 1996

 

 

 

Group Equipment Livestock
1987 1996 1987 1996
Sample 278 1.74 3.23 2.34 2.85
Settlement Tenure
5-10 18 1.06 . 3.50 1.89 2.72
11-15 97 1.76 3.81 2.10 3.55
16-20 42 1.95 3.26 2.36 2.60
>20 121 1.74 3.23 2.34 2.85
Gradient of Control
Early 105 2.07 3.02 3.16 1.94
Mid 100 2.06 4.31 1.82 4.08
Late 71 .83 2.09 1.93 2.45
Settlement Status
Pioneer settlers 136 1.04 2.21 1.40 1.68

2“" generation immigrants 77 2.39 3.71 2.82 3.37
1" generation immigrants 65 2.43 4.80 3.74 4.68

Period of Settlement
Pioneers 147 1.06 2.20 1.50 1.71
Newcomers 131 2.50 4.38 3.28 4.12
Timing of Movement
Before Tsetse Control 129 1.65 2.80 2.26 2.61
During Tsetse Control 90 1.96 4.10 2.50 3.79
After Tsetse Control 59 1.61 2.85 2.27 1.91
Age in 1996
Under 31 10 .80 3.4 .40 1.8
31 — 40 54 1.33 2.98 1.56 2.70
41—50 95 1.64 3.18 2.31 3.11
51 - 60 51 2.29 3.73 2.47 2.86
61 + 68 1.93 3.10 3.21 2.67
Family Type
Monoganrous 199 1.81 3.15 2.45 3.06
Polyganlous 79 1.57 3.43 2.06 2.33
Employment Experience
Yes 148 1.82 3.20 2.37 3.16
No 130 1.65 3.26 2.31 2.50

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control Surveys,
Gokwe North, Zimbabwe, 1996-97.

91

Whether a household type was monogamous or polygamous did not afi’ect mobility in equipment
investment during the period 1987 to 1996. This was also the case with wage experience.
Household heads with formal work experience had advanced equally well in equipment

investment as households who had no such experience.

5. 4. 2 Livestock Wealth Mobility

Livestock assets, unlike equipment, fluctuate in size over time due to sales, deaths (due to
drought and diseases) and culls. The results in Table 5 show both positive and negative
fluctuations in cattle assets for the two periods under consideration. About a third of the sub-
sample used in this analysis had not made any investment in livestock in 1987. By 1996, about
12% had moved to the fifth decile or better. However, 33% had not made any livestock
investments in 1996. Half of the households who were in the 1" decile in 1987 remained in that
position ten years later and about 8% moved to the fiflh decile or better. About 6% of the
households who were on the 1" decile in 1987 had not invested in livestock as of 1996.
Approximately 40% of the households who were on the 2'“I decile in 1987 had advanced to the
fifth decile or better in 1996 but a quarter of this group faced declines in cattle wealth. The
proportion of declines increased directly with investments. About two thrids of households in the
third deciles shified to lower deciles in 1996. Households who had large investments in livestock

had higher chances of facing a decline in livestock assets. Thirty-seven percent of households in

92

the top decile in 1987 remained in that position while the rest of them declined. Six percent of the
households dropped all the way to no investment at all in 1996. The loss could have been

voluntary‘or due to natural causes.

Overall, there was no significant change in cattle holding (Table 5.5) between 1987 and 1996.
While cattle herds are expected to grow over a ten-year period, losses were inevitable. Typically,
this agro-ecological environment experiences mild to severe droughts at least 4 out of ten years.
Besides, cattle are sold to finance large expenditures, for example, school fees and goats provided

much needed proteins and ready cash.

Settlement tenure had no obvious association with mobility in livestock investments. While all the
small groups of tenure achieved mild boosts in livestock investment in the ten-year period, the
tenure groups with less than 15 years had the greatest positive shift. Residents who have been
settled for more than 15 years had marginal changes in livestock investment. The association

between age of household head and mobility in livestock investments shows a similar pattern.

93

Sarca—
6BaanEmN .532 9.3er .9925 .9550 amen—Sonata. one 3.0:. ..e «aficionado 82.3— omEeneuoSoem 886m

mvN m Z v m w E 2 MN 5N on on Nv Z

 

 

 

o— mum ooo. ooo. nN—. mN—. So. So. ooo. 53. ooo. moo. o—
v omN. ooo. ooo. ooo. ooo. omN. ooo. oom. ooo. ooo. ooo. o
v mN—. ooo. ooo. ooo. ooo. ooo. omN. omN. omN. ooo. ooo. m m
w omN. ooo. mN—. ooo. ooo. ooo. mN_. mNZ. omN. mN_. ooo. b H
w mN_. ooo. ooo. ooo. mN—. mum. mN_. mNZ. ooo. mN_. ooo. o a” %
N— mwo. ooo. ooo. ooo. ooo. ooo. ooo. >3. oom . 52. So. m U
o— ooo. ooo. mN—. mN—. mN—. mNZ. moo. mN_. moo. NE. moo. v H
w— nmo. ooo. ooo. ooo. mmo. ooo. _:. :_. _:. v34. _:. m A—
ov mNo. who. ooo. who. o2. who. o3. omo. om _. o2. o2. N
5 ooo. o—o. ooo. ooo. ovo. o—o. Nwo. >3. 2 _. wow. moo. _ M
an :o. ooo. ooo. :o. No. woo. So. So. mN_. w— m. an. o a
Z o— o w h o n v n N _ o —
mums—HUNG oom—

 

.oaa— coffins—Ha 485nm: Suez 2.5—emu Acme—Iona: mac—30m Boo—:3 am 5:33 3938::— ue 5.5a: 1:33:95. "no 033—.

Young households advanced by more than one decile between 1987 and 1996 while
household heads greater than fifty years advanced by less than half a decile. Household
heads above 60 years de-invested in liveuock assets. This is expected as their period of
retirement sets in.The pattern shown by the gradient of tsetse control is surprising.
Households in early cleared areas experienced a decline of more than a decile in livestock
investments in 1996 when compared to their investments in 1987. Early cleared areas
could be more drought prone than other locations where pastures are relatively abundant.
In contrast, the other areas experienced an increase in livestock investments especially in

the mid control gradient.

The results exhibited a clear pattern of livestock asset positions across settler types. In
1987, indigenous settlers were at the 1" decile, 2"d generation newcomers were at the 2M
decile while 1" generation newcomers were on the 3" decile. The 1" generation migrants
were at an advantage in 1987 because some of them brought livestock at the time they
moved in while others brought capital and purchased livestock once they settled-in. Each
of these groups moved upwards by one decile maintaining the relative difi‘erence among

each other.

95

Variables including household type, and timing of movement did not show patterns of
association with livestock assets between 1987 and 1996. Work experience had the
expected effect of boosting livestock assets. While all the two groups were in the same
decile in 1987, households with formal work experience had shified upwards to the third
deciles but those households without experience remained in the same position ten years

later.

5.5 Determinants of Per Capita Wealth

Several associations of per capita wealth with household demographic variables have been
discussed. A quantitative assessment of wealth accumulation was performed by regressing
per capita wealth on demographic and location variables. The results are presented in
Table 5.6. The individual efl‘ect of variables afier partialing out the effect of other

variables is discussed next.

Delay in cattle adoption: The longer the period without cattle adoption in one firmer’s
tenure the lower was their per capita family wealth, everything else constant. An
additional year of settlement tenure spent without cattle significantly reduced per capita
wealth by S117. Adoption of cattle was critical to the supply of animal draft needed to

boost production fictor productivity. Households who had adopted cattle immediately

96

afier settlement had a greater opportunity of accumulating wealth than households who

adopted cattle several years after settling-in.

Table 5.6: Results of an Econometric Model for Wealth (Livestock and Equipment)
Mobility for Settlers in Gokwe North District, Zimbabwe, 1996.

 

Explanatory Variables Per capita total wealth

Wealth Mobility 1987-96

 

 

Coeff T-value Mean Coeff T-value Mean
Year's settled before - 4.841” 7.88 -138.97 -l.234 12.17
adopting cattle 1 17.051
Age at settlement (yrs) -80.381 -2.625” 29.79 427.70 -3.697" 28.47
Off-firm (yes =1) 515.567 .832 .48 2594.54 1.032 .53
# of family members 432.45 2.509” 1.00 -1554. 15 -1.842" .89
completed high school
Gender (female =1) 192.096 .225 .14 - -
Adopted cattle before - -- -9405.61 -2.768** .48
1986 = 1
Farm size (acres) - -- 413.35 2.943M 16.07
No. of female spouses —1045 -3.2'° 1.34 -1956.82 -1.235 1.37
1" Gen. immigrant =1 2088.63 2.665” .27 3049.85 .960 .27
Settlement before 654.88 .707 .28 -7764.29 -2.441“ .46
tsetse control = 1
Settlement afier tsetse ~1916 -2.502 .48 -- -- .21
control = 1
Early Cleared Areas 657.97 .912 .32 -9587.98 -2.729" .38
Mid cleared Areas 3983.95 5.614” .34 12032.69 3.641" .36
Constant 9406.91 6.334” 24528.58 3.672"

Adjusted R2 .195 .302
SE. of estimation 6069 19428
Sample size 458 271

 

 

Source: Socioeconomic Impact Assessment of Tsetse and Trypanosomosis Control
Surveys, Gokwe North, Zimbabwe, 1996-97.

97

Age of Settler at time of settlement: Household heads who had waited for longer periods
prior to settling-in were significantly poorer. For each year old delay, per capita wealth
decreased by $80.38 ceteris paribus. Household heads spending longer time periods in
other locations were disadvantaged. The experiences obtained elsewhere prior to

settlement appeared irrelevant to wealth accumulation in the current settlement area.

Employment experience: The impact of non-farm work experience had the expected
impact in wealth accumulation. Household heads with formal work experience were
wealthier than household heads without such an experience. Non-firm work experience
positively contributed $515.57 per capita, everything else constant. While crop surpluses
provide the primary source to finance capital accumulation, non-farm income is critical for
financing lumpy investments, for example, cattle and draft equipment. However, the extent

of the impact on per capita wealth holding was insignificant.

Education: Households with higher levels of education were wealthier than their
counterparts. An additional household member who had completed secondary education
made the household per capita wealth increase by $432.45, everything else constant.
Individuals with secondary education certificates were employable and could remit part of

their earnings or help finance seasonal inputs and farm capital accumulation.

98

Gender: Gender did not have the expected impact on wealth holding. Female-headed
households had higher per capita wealth than male headed households but the difl‘erences

were insignificant. The evidence does not support the general conception that female-

headed households are vulnerable and poor.

Polygamy: Families practicing polygamy were significantly poorer than monogamous
households. A family with more than one female spouse had $1045 less wealth per capita,
everything else constant. This evidence challenges the conventional wisdom that
polygamous households are wealthier. The results show that, on a per capita basis, single

spouse families were wealthier.

Immigrants: 1" generation immigrants were wealthier than all other settlers, everything
else constant. Settlers who abandoned their previous settlement held S2088 more wealth
than settlers who had established their first settlement site in Gokwe North district. Since
the migration of the first generation of immigrants was voluntary and unsupported by
government, the group was self-selected in terms of wealth holding. Only those

immigrants with ability and potential to succeed migrated.

Timing of settlement: Settlers who moved in prior to tsetse being controlled were not

wealthier as expected. Generally, settlers with longer tenure are expected to be wealthier

99

but the results partly contradict that thesis. Settlers who moved-in during tsetse control
were not significantly different in per capita wealth holding with settlers who moved in
prior to tsetse control. However, settlers moving in after tsetse control significantly held
$1916 less per capita wealth than settlers moving in during tsetse control. This part ofthe
results gives support to the conventional thesis. The presence of tsetse flies appeared to

take away the premium of early arrival.

T setse control gradient: The tsetse control gradient was an important predictor of per
capita wealth holding. Settlers in recently cleared areas held significantly less wealth. As
mentioned above, the presence of tsetse flies limit the ability of the household to

accumulate firm capital.

5.5 Determinants of Livestock and Equipment Asset Mobility
A sub-sample of households who settled in Gokwe North before 1987 was used in
measuring the physical change in asset position between 1987 and 1996. Econometric

results of livestock and equipment asset mobility are shown in Table 5 .6.

Years Settled Before Adopting Cattle: Lengthy delays in adopting cattle afier settlement

had a negative impact on mobility. On average settlers adopted cattle 12 years afier

100

settlement. Further delay by an additional year reduced household mobility by S139. The

effect of this impact was not significant.

Age at Settlement: Individual household heads who settled at a more mature age did not
change their asset their asset position as much as those who settled when they were
young. The average age at settlement was 29. Settling at an age one additional year older
than average contributed a reduction in household mobility of S427, everything else
constant. This impact was significant at p= .05. More mature settlers were not expected
to change their asset‘position significantly once they had passed the stage of wealth ‘

acquisition in their life cycle.

Ofl-farm employment experience: While formal employment experience contributed
positively to asset mobility, the impact was not significant. Non-firm income gives the
household an additional source of finance for capital investment but these activities may
reduce the amount of labor available for farm activities. The net effect did not make

significant impact on household asset mobility but the direction of impact was as expected.

Education: A higher number of household members who completed secondary education
surprisingly contributed to less asset mobility. The trade-ofi‘ between education and firm

assets was significant. Everything else constant, an additional member who completed

101

secondary education reduced household mobility by S1554. Households ofien liquidate
their livestock assets to raise cash to pay school fees. Unless the candidates do well in

school and get employed, the payoff to such de—investment is negative.

Cattle Adoption before 1986: About 48% of the settlers had adopted cattle before 1986.
However, adoption of cattle prior to 1987 was associated with de-investment in assets.
Between 1986 and 1996, there were three or four years of drought. It is expected that the
livestock asset position in 1996 was much worse than was the case in 1986. This
contradicts conventional understanding that smallholder herd sizes will increase
uncontrollably putting unbearable pressure on pastures. Drought seasons especially in

1994/95, decimated cattle herds and as of 1996/97, pastures were abundant.

Farm Size: The size of the farm played a significant role in asset mobility. Settlers with
an additional hectare of land advanced their asset position by $413, everything else
constant. Settlers Operating a larger than 6.5 ha firm were inclined to mechanize in order
to manage larger cropping areas. Resourcefill settlers were expected to advance rapidly

and in a bigger way than less resourceful settlers.

Number of female spouses: Households with more than one female spouse advanced

much less than households with one female spouse. However, the effect of polygamy on

102

asset mobility was insignificant. The ability of polygamous fimilies to invest in productive
capital is limited by the excessive demand for food and other basic needs associated with

bigger households.

Settler Status: While 1" generation immigrants advance their asset position higher than
other settlers, the efl‘ect was not significant. The opportunity to advance was open to all
types of settlers. This result suggests that, agricultural frontier areas are equally open to

the advance of all types of settlers.

Settlement timing. Settlers who moved in before active tsetse control began did not
advance as much as those households who moved in with tsetse control. Moving in prior
to tsetse control led to a reduction in asset position of S7764, everything else constant.
Despite having much advanced knowledge of local conditions tsetse settling in tsetse

infested areas eliminated the benefits associated with early settlement.

T setse Control Gradient: The gradient of time since tsetse control was an important
predictor of asset mobility. Settlers who occupied areas cleared of tsetse flies early had
the least change in mobility. Other settlers who moved into areas cleared in the mid-
period had the greatest asset change. Advancement was expected to be greater during the

1" ten years after tsetse fly control than later on. The first ten years after tsetse control,

103

 

firming systems change substantially as households make livestock and equipment
investments for the first time. Afler the first ten-year period, the only additional change is

expected to be incremental.

104

CHAPTER 6: DISCUSSION AND CONCLUSIONS
6.1 Tsetse Control and Migration
Tsetse control had a direct efl'ect on immigration but did not motivate emigration. The main
motivation for emigrating was to acquire adequate arable land. In addition to tsetse control,
other factors including the end of war and access to public transport positively impacted on

immigration. The results are summarized as follows:

1. Rural -to-rural movements of households was a common occurrence over
the past few decades. The historical racial imbalance in land distribution is
responsible for instituting regional imbalances in land access leading to

overcrowding and subsequently emigration.

2. Tsetse control helped in the immigration of households who were moving in
with their cattle. Motivated primarily by the desire to acquire good and

abundant land, immigrants without cattle even settled in tsetse infested areas.

3. Irmnigration into Gokwe North occurred in trickles prior to Zimbabwe’s
Independence. In the five years following the end of the war, the largest flow
of immigrants into Gokwe North occurred. Apparently there was a strong

feeling at that time that access to land was one of the rights people fought for.

105

4. Immigrants considered primary roads as important physical infrastructure.
Good roads attracted private investments in commuter and commodity
transport and allowed communication with other trading centers in the
country. Other services including schools, clinics and consumer stores were

desirable but were not necessary at the time of immigration.

Nineteen years afier independence, the Zanu(PF) government had not committed itself to a
comprehensive program of rural resettlement. The existing structural features of land tenure
gave the best land to Whites and a few well connected Blacks. Much of the land givenito
Blacks could not support them without urban remittances. So Blacks were always looking for
better land. Newly available land was attractive even if it had tsetse flies. Whether immigrants
anticipated tsetse control when they moved into settlements with tsetse flies is a moot point.
Settlers who did not own cattle at the time of migration appear to have been less concerned
about the presence of tsetse and trypanosomosis. Those settlers who moved in with cattle

settled in areas with tsetse and trypanosomosis control.
Unless government redistributes land and continues to tolerate voluntary unassisted

settlement, rural households are predicted to continue moving into areas with relatively better

firming opportunities.

106

Immigration into the Zambezi Valley reflects on problems encountered elsewhere in non-
contiguous districts. To successfirlly manage immigration in the Zambezi Valley, land-use
problems and lack of opportunities in other communal areas need attention. Attention to
economic development problems in the sending areas will help in balancing economic
opportunities across districts and dampen migration. To tackle development and
environmental problems in the Zambezi Valley requires a wider perspective beyond the

location of the valley.

6.2 Farming System Change With Tsetse Control and Immigration
Another story emanating from this study is that immigrants and tsetse control were strong
positive forces for agricultural development in Gokwe North. Results fi'om Chapter 4 can be

summarized as follows:

1. Immigrants can be considered as innovators in breeding cattle for animal
traction within Gokwe. This group of settlers wanted to pursue the type of
mixed firming system they practiced in their areas of origin. So some
immigrants in-migrated with their cattle while others purchased cattle once
they had settled. Pioneer settlers in Gokwe North were catching up on

operating mixed firming enterprises.

107

 

Despite the disease risk and livestock movement restrictions, there was as
much cattle adoption in each village with and without active tsetse control.
Cattle adoption in tsetse infested areas was possible to the extent that cattle
owners could access and use drugs provided fiee by the Department of
Veterinary Services. When cattle were introduced in Gokwe North, tsetse

flies were present but drugs were accessible.

The average growth rate in cattle adoption per village between 1960 and 1996
was negative and significant. The negative grth suggests that non-adopting
households had limited financial resources and access to livestock credit
markets. Commercialization of production was critical to overcome liquidity
constraints and allow self-financed investment in capital formation. A shift of
producer price balance in favor of cash crops provided incentives for cattle

adoption.

Ox owners and renters had different pathways available to them for improving
productivity. The avenue for intensifying productivity for ox owners was
through employing additional units of labor and land. The avenue for
intensifying productivity for ox renters and hand tillers was through additional

investment in physical capital. Hand tillers were close to their productivity

108

 

limit. To improve fictor productivity among hand tillers, investment in firm
capital intensification was needed. Access instead of ownership of animal

traction technology was all that mattered.

The presence of tsetse flies had made cattle adoption dificult but the fly presence did not
preclude the introduction of cattle into the study sites. This lays credence to one suggestion
that tsetse control operations were targeted at locations with the highest incidence of
trypanosomosis in cattle instead of the degree of tsetse fly infestation per se. This strategy

was adopted to keep the cost of administering drugs low for the exchequer.

Given the absence or underdevelopment of well functioning livestock credit markets, cattle
adoption was predominantly self- financed. The linkages between cotton production and
adoption of animal traction were more expansive than what is generally acknowledged in the
literature. Without ofllfarm employment, non-cotton growers were less likely to adopt cattle.
Extension programs that encourage use of animal traction cannot be implemented without
targeting the technology on high value crops like cotton. The least cost and short term
approach to improve capital intensification among hand tillers was to develop a market for

custom ploughing service and introducing minimum tillage practices.

109

6.3 Wealth Dktribution and its Determinants
Chapter 5 examined wealth distribution in 1996 and wealth accumulation fiom 1987 to 1996,

focusing on the influence of household demographic and regional fictors. The results can be
summarized as follows:

1. The aggregate distribution of wealth was fairly even while livestock and
equipment wealth were unevenly distributed. Investment in livestock was not

risk-free. Unlike equipment investment, livestock investment was risky.

2. The age profile of wealth accumulation supported the life cycle hypothesis.
Young households had the least wealth as they were still settling in and
striving to establish food security. Middle-aged household heads were the
wealthiest. The elderly households were nearing or already in retirement and

their asset position was declining afier years of accumulation.

3. The length of residential tenure did not have the expected effect on wealth
accumulation. Early arrival before tsetse control did not result in the expected
premium in wealth accumulation. The presence of tsetse flies eliminated the

premium associated with early arrival.

110

4. The conventional wisdom suggesting that polygamous households are
wealthier than monogamous ones was a myth. When looked at on a per capita
basis, monogamous households were wealthier than polygamous households.

In addition, female-headed households were among the better off households.

5. The greatest wealth improvement was observed in those areas that had recent
control relative to the period under scrutiny. There was more wealth acquired
within the first 10 years afier tsetse control than 10 - 20 years alter tsetse
control. The first ten years after tsetse control featured substantial changes in
the farming systems associated with livestock and equipment investment. The
fimiing systems only changed incrementally after the first ten years of

settlement.

The school of thought linking agricultural frontier areas with socioeconomic equality is not
rhetoric in an aggregate sense. The opportunities for socioeconomic advancement in the
settlement frontiers were fiirly equal among all settlers. Agricultural frontier areas, therefore,
have potential to contribute positively to redistribution of wealth throughout the country. The
dificulty of achieving more equal wealth distribution over time could be eased through a
managed program of internal migration or resettlement. The government can propel such a

process in high potential areas in order to improve the balance of socioeconomic welfare

111

throughout the country. This is in line with government efforts of resettlement. The

opportunities for acquiring wealth appeared to be gender sensitive.

6.4 Limitations of the study

While some attempt was made to control for access to infi'astructure services at the time of
the survey, there was need to accurately determine the actual time of development. Since
more than 90% of the households settled before 1996, the infi'astructure services available in
1996 did not affect their choice of settlement site before that year. A time-line of service
development of each village was required to comprehensively measure the impact of these
services on the flow of immigrants. Besides, the physical locations of settlement sites and
service sites could have been captured with a geographical position system (GPS) and data

included in the model.

Additional attention could have been given to the process of emigration from Gokwe North
District rather than focusing entirely on immigration. The records would have been difficult
to capture at the village level since local leaders do not keep physical records of who left.
With some help and time, most village-heads can construct a settlement profile for their
jurisdictions with reasonable accuracy. While aggregate data is available at the district level,

its usefillness is limited for local assessments.

112

In attempting to measure the impact of tsetse and trypanosomosis control, it is important to
have a good measure of the variable under scrutiny. Getting a good measure of the degree of
tsetse challenge and trypanosomosis risk representative for the study sites was a daunting
challenge. Ideally, sites with good data should have been targeted for selection. The Tsetse
and Trypanosomosis Control Branch conduct pre and post-control surveys of tsetse fly specie
and their prevalence. In addition, a sister branch conducts trypanosomosis surveys. Data on
positively identified cases of trypanosomosis could provide a reasonable measure of the
degree of trypanosomosis risk for a particular location and time. However, the data were not
collated and the data formats changed over time making it not readily useable. Livestock
census data over the years could also provide a mirror image of the impacts of
trypanosomosis. The department of Veterinary Services collects livestock census data
including productivity parameters. The challenge in handling this data is that the unit of

observation has consistently changed over time.

Household socioeconomic mobility can be defined in many ways besides accumulation of
physical assets. This study failed to capture the investment made in human capital in terms of
education for the parents and children. Resources spent in paying school fees and attending
school have a high opportunity cost and need to be accounted for in measuring wealth

accumulation. In addition, the study failed to assess how many household members were

113

employed elsewhere or what was the degree of non-firm participation for each household and

how much net cash savingsthey had accumulated.

6.5 Areas for Further Research

Tsetse control efforts were very successful in removing the vector and ficilitating the
development of mixed farming systems. Yet tsetse control affect other issues. Tsetse control
agencies have neglected land-use and environmental efi‘ects of activities occurring during and
after tsetse control. Apart from direct ecosystem impacts, there are concerns about resource
degradation due to uncontrolled and haphazard settlement and land-use. Coordination of
human and livestock movement into tsetse controlled areas is lacking. With a slow paced
government resettlement program, there are few choices of where people can settle. Making
voluntary-unassisted settlement illegal may keep people from emigrating but may lead to
serious problems of starvation. At the same time, uncontrolled immigration and land-use can
lead to short term gains but may cause a long term environmental disaster. What is crucial is
recognizing that tsetse control on its own is only a partial development vision. How can
public and private investments be sequenced in order to promote development of the

agricultural sector.

Often, migration studies conduct surveys in destination areas and neglect the areas of origin.

When assessing the welfare impacts of migration, the incidence of impacts could be

114

broadened to include source areas. Such an approach enables a more holistic assessment of
the impacts of migration to the nation as a whole. As was the case in this study, a significant
proportion of the immigrants where fiom neighboring districts. Future studies could devote

resources to study related processes in areas adjacent to the primary survey site.

A lot of controversy still revolves around the profitability and impact of animal traction on
labor and land productivity. Capturing the profitability of animal traction investment requires
capturing the input and output data of all the activities where animal labor is deployed.
Focusing on one particular enterprise for a given season would not provide a comprehensive
measure of the returns to investment. The most challenging aspect of such work is capturing
the animal and human labor hours and purchased inputs for a particular activity. The

resources required to do this work are immense but necessary to answer the questions.

115

BIBLIOGRAPHY

116

BIBLIOGRAPHY

Anderson, I. P., D]. Brinn, M. Moyo, and B. Nyarnwanza (1993). Inventory of the
Communal Lands of Zimbabwe - An Overview. NR! Bulletin No. 60 Chitharn, UK:
Natural Resources Institute

Annual Report of the Branch of Tsetse and Trypanosomosis Control, Department of
Veterinary Services, Ministry of Agriculture, Rhodesia 1963 - 1980; Zimbabwe 1981-
89

Astatke, A and M.A Mohammed Saleem. 1996. “Draught Animal Power for Land-use
Intensification in the Ethiopian Highlands.” Unpublished Report. International
Livestock Research Institute. Nairobi, Kenya .

Barrett, V; G. Lassiter; D. Wilcock; D. Baker; and E. Crawford. 1982. “Animal Traction in
Eastern Upper Volta: A Technical Economic and Institutional Analysis.” MS U
International Development Paper No. 4. Michigan State University

Barrett, J. C. 1992. “The Economic Role of Cattle in Communal Farming Systems of
Zimbabwe.” Pastoral Development Network Set 32, ODI, London

Barrett, 1. (1997) Control strategies for Afiican trypanosomiases: their sustainability and
efl'ectiveness. In G. Hilde, J.C. Mottram, G.H. Coombs and PH. Holmes (eds),
Trypanosomiasis and Leishmaniasis, CAB International, pp. 347-359.

Bealby, KA; RI. Connor and GI. Rowlands. 1996. Trypanosomosis in Goats in Zambia.
ILRI, Nairobi, Kenya.

Borjas, GJ. (1988) Economic Theory and International Migration. International Migration
review, Vol 33, No. 3, pp 457-485

117

Bloom, RA 1979. “A Review of Technical Evidence on the Use of Animal Traction in
Sahelian Farming Systems.”MSU International DevelopmentPaper No. 7. Michigan
State University

Brinn, RI. 1996. “Cattle in Chiawa 1994 - 96.” RTCCPLand UseMiscellaneousReportNo.
20. Ministry of Agriwlture, Food and Fisheries, Republic of Zambia

Carlino, G.A. & E.S. Mills (1985) Do Public Policies Afi‘ect County Growth7. Business
Review, July-August, 3 - l6.

Carlino, G.A & E.S. Mills (1987) The Determinants of County Growth. Journal of
Regional Science, Vol. 27, NO 1, 1987

Chadenga, V. 1994. Epidemiology and Control of Trypanosomosis. 0nderstepoort Journal
of Veterinary Research, 61: 385-3 90

Cockbill, GE 1972. A Revised Handbook for the use by Tsetse Field Stafl‘. Department of
Veterinary Services, Branch of Tsetse and Trpanosomosis Control

Connor, RI. 1989. Regional tsetse and Trypanosomosis Control Progranune Malawi,
Mozambique, Zambia and Zimbabwe: Final Report of the Regional Trypanosomosis
Expert, FGU

Connor, RI. 1994. The Impact of Nagana. 0nderstepoort Journal of Veterinary Research,
61: 379 - 383

Coulman. 1997 . Training Oficer. Personal Communication

118

Dione, J. 1989. “Informing Food Security Policy in Mali: Interactions Between Technology,
Institutions and Market Reforms.” Vol 2. Unpublished PhD Dissertation, Michigan
State University.

Eithier, W.J. (1986). International Trade Theory and International Migration. Research in
Human Capital and Development, 4.

Farnham,D.A1996. at . -- . _
-'-w 01:. pvrnmn- ”nor-o a0 '11'1' 'r-aaln orm To:-
Associated University press, Cranbury, NJ

 

Ford, J. 197 1. The Role of the Trypanosomosis in Afiican Ecology: Oxford, Clarendon Press

Grant, I. (1998). Tsetse Control and the Environment: Environmental Impacts of Insectrcrde

Technologies. Paper prepared for the Advisory Committee of the Programme Against
Afiican Trypanosomosis, Harare, Zimbabwe, October.

Graves, PE. (1983) Migration with a Composite Amenity: The Role of Rents. Journal of
Regional Science, 23, pp 541-546

Greenwood, M.J. (1975) Research on Internal Migration in the United States: A Survey.
Journal of Economic Literature, 13(2): 397-433

Gregson, M.E. “Wealth Accumulation and Distribution in the Midwest in the Late Century,”
Explorations in Economic History . Vol 33, 524 -538 (1996) Article No. 0023

Jordan, AM. 1986. Trypanomosis Control and Afiican Rural Development. Harlow and New
York: Longrnan Group Limited

Kearl, JR; C. Pope and LT. Wlmmer. “Household Wealth in a Settlement Economy: Utah,
1850 -1870,” TheJournal ofEconomic History, Vol XL, No.3, Sept 1980. 470-496.

119

Knapp, TA and PE. Graves (1989) On the Role of Amenities in Models of Migration and
Regional Development. Journal ofRegional Science, Vol. 29, No. 1, 1989, pp 71-87

Lovemore, F. 1995. The Influence of Tsetse Distributions on Settlement and Land Utilization
in Zimbabwe with Reference to the EC-funded Regional Tsetse and Trypanosomosis
Control Program (RTCCP) of Malawi, Mozambique, Zambia and Zimbabwe. In. A
Systematic Approach to tsetse and trypanosomosis Control. F.AO. Animal
Production and Health Paper 121

Lovemore, F. 1997 . The Significance of the Matusadonha National park Tsetse Fly
Infestation: Preliminary Environmental Assessment of the Impact of Tsetse Control
Operations in Matusadonha. Consultant Report Acc No. 7.ACP.ZIM. 002

Mathur, V.K; S.H., Stein & R Kumar (1988) A Dynamic Model of Regional Population
Growth and Decline. Journal of Regional Science, Vol. 28, NO 3, 1988,

McMillan, DE. (1995). Sahel Visions: Planned Settlement and River Blindness Control in
Burkina Faso, The University of Arizona Press.

McMillan D.E., J .B. Nana and K. Savadogo. 1993. Settlement and Development in the River
Blindness Control Zone. Case Study — Burkina Faso. World Bank Technical Paper
No. 200 Series on River Blindness Control in West Afiica.

Pannin, A 1989. “Hoe and Bullock Farming Systems in Northern Ghana: A Comparative
Socio-Economic Analysis.” Nyankpala Agricultural Experiment Station, Crops
Research Institute, Special Report No. 1

Pessino, C. (1991) Sequential Migration Theory and Evidence fi'om Peru. Journal of
Development Economics 36, 55-87.

120

 

Pethemn. 1U; M R CoeandA Astatke 1989 W
W Graduate School of Trovical Vetefl'nary
Science, James Cook University, Townsville, Australia.

Ramisch, J. 1996. The Role of Livestock in the Intensification of Sahelian Agriculture:
Sustainable Mixed Farming or Overgrazing and Degradation? Unpublished Paper

Reid, RS. and E. Ellis (1995) The Environmental Implications of Controlling Tsetse
Transmitted Tryapanosomosis. Unpublished report. International Livestock Research
Institute (ILRI), Nairobi, Kenya.

Ricardo, D. 1817. P' ' l P lii n i .London Press

Roback, J. (1982) Wages, Rents and Quality of Life. Journal of Political Economy, 90,
1257-1278.

Rohrbach, D. 1989. “The Growth of Smallholder maize production in Zimbabwe: Causes and
Implications for Food Security”. Ph.D. Dissertation, Michigan State University.

Savadogo, K.; T. Reardon and K. Pietola. “Adoption of Improved Land Use Technologies
to Increase Food Security in Burkina Faso: Relating Animal traction, Productivity and
Non-farm Income,” In Press, forthcoming, Agricultural Systems, October 1998
(Final version accepted Nov 1997

Savadogo, K. ; T. Reardon and K. Pietola. , “Mechanization and Agricultural Supply Response
in the Sahel: A Farm Level Profit Function Analysis,” Journal of A fi'ican Economies,
4(3):336-377, December 1995.

Savadogo, K.; T. Reardon and K. Pietola., “Farm productivity in Burkina Faso: Effects of
Animal Traction and Non-firm Income,” American Journal of Agricultural
Economics 76:608 - 612, August 1994.

121

Schachter, J. and PG. Althaus (1989). An Equilibrium Model of Gross Migration. Journal
of Regional Science, Vol. 29, No. 2, pp 143-159

Schultz, P. 1989. Lifetime Migration within Educational Strata in Venezuela; Estimates of a
Logistic Model. Journal of Economic Development and Cultural Change. Vol 45,
1989.

Shereni, W. 1990. Strategic and tactical developments in tsetse control in Zimbabwe (1981-
1989). Insect Science and its Application 1 1(3): 399-409.

Shereni, W. 1995. Tsetse and Trypanosomosis Control Program in Zimbabwe; Current
Operations and Future Plans. Paper presented to the FAO seminar on Tsetse and
Trypanosomosis Control, ILRI, Nairobi, October 1995

Shumba, E.M. “The Farming System” In eds. 1992. E. Whingwiri; M. Rukuni; K.
Mashingaidze;andC. M. Matanyaire. 11 S r 1 w F '
WWW Rockwood Publishers. Harare

Soltow, L. 1971. Eggerns of Wgalth Holding in Wimgsin Sing 1850. University of

Wisconsin Press, Madison

Steckel, RH. 1990. “Poverty and Prosperity: A Longitudinal Study ofWealth Accumulation,
1850-1860,” Review of Economics and Statistics Vol 74. 275-285.

The Zimbabwe Herald, 14 February 1995. “Rocky Desert Looms in Omay and Kanyati”.

Thomas W.E.; T.H. Davey; W. H. Potts and G.F. Cockbill. 1955. Report of the Commission
of Inquiry on Human and Animal Trypanosomosis in Southern Rhodesia, federation
of Rhodesia and Nyasaland, Government Printer, Salisbury.

122

Turner, F.J. 1920. W- New York: H Holt and Co.

123

 

nICHIan STATE UNIV. LIBRARIES
IIHIIIIIIHHIIIHIIHIIIIIll1er1111IIHHHIIlIlHI
31293017700521