ECONOMETRIC MODELLING OF THE EFFECTS OF 
INTERSECTORAL LABOUR MOBILITY ON 
DEFORESTATION IN GHANA, 1970-99.
BY
VICTOR OWUSU
THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF 
GHANA, LEGON IN PARTIAL FULFILMENT FOR THE 
REQUIREMENT FOR THE AWARD OF M. PHIL 
DEGREE IN AGRICULTURAL ECONOMICS.
DEPARTMENT OF AGRICULTURAL ECONOMICS AND
AGRIBUSINESS.
JULY 2001
< > - • 3 7 0 4 0 3
s i > V '? -  0 !uo 7
J^ X-hc c  • I
f a
DECLARATION
I, V ICTOR OWUSU, the author o f this thesis, "ECONOMETRIC  MODELLING OF 
THE EFFECTS OF INTERSECTORAL LABOUR MOBILITY ON 
DEFORESTATION IN GHANA, 1970—99" do hereby declare that, with the exception
o f references to past and present literature duly cited in this study, the entire research 
leading to this thesis was done by me at the Department o f  Agricultural Economics and 
Agribusiness, University o f Ghana, Legon. I further declare that the research has never 
been presented either in whole or in part for any degree in this University or elsewhere.
V ICTOR OWUSU
This thesis has been submitted for examination with our approval as supervisors.
K  YERFT FOSU
DANIEL BRUCE SARPONG 
(CO-SUPERVISPOR)
DEDICATION
This thesis is dedicated to my daughters
Akua Durowaa Owusu aad Adwoa Takyiwa Owusu.
First, I express my gratitude to the Almighty God whose divine guidance and 
wisdom have enabled me to complete this thesis research successfully. Second, I am 
grateful to my Major Supervisor, K. Yerfi Fosu for his immense contribution, direction 
and constructive criticism. I also appreciate my Major Supervisor’s efforts of arranging 
for me to visit Vrije Universiteit, Amsterdam to undertake the econometric modelling 
and update the literature of this M Phil thesis. I sincerely owe a debt o f gratitude to Dr 
Bruce Sarpong of Legon who was the Co-Supervisor o f this thesis. I wish to thank Prof. 
Nico Heerink of Wageningen University, The Netherlands and Prof. Awudu Abdulai, 
Department of Agricultural Economics, Swiss Federal Institute of Technology, Zurich, 
Switzerland whose invaluable input assisted in refining the thesis.
Third, I wish to extend special thanks to Dr. Kees Burger, Head o f Economics 
Division, Economic and Social Institute (ESI), Vrije Universiteit (VU), who was my 
Supervisor at ESI, VU, Amsterdam, Dr. Hidde Smit, Director o f ESI, VU, Amsterdam 
with whom I shared an office at ESI, VU, for his fatherly advise and the supporting staff 
o f ESI, VU and Department of Agricultural Economics and Agribusiness, Legon for 
availing themselves and putting their facilities at my disposal during the research work. 
Finally, I deeply appreciate the prayer, moral and financial support o f my brothers, 
Kwasi Kyere (Brooklyn, New York, USA), Kwasi Boamah (Amsterdam, Netherlands) 
and Kwasi Acheampong (Kumasi) and Mr. Brobbey Frimpong and his dear wife, 
Christie (Accra) without which I would not have been able to pursue this M Phil. Degree 
in Agricultural Economics. May God richly bless and reward them bountifully.
ABSTRACT
The study analyses the effects o f intersectoral labour mobility between agricultural and 
non-agricultural sectors on deforestation in Ghana. It is based on annual time series data
covering the period 1970—99. The study describes the basic determinants o f intersectoral
labour mobility and deforestation and estimates the magnitudes o f the effects o f their 
determinants in Ghana. The empirical results show that the effects o f intersectoral labour 
mobility between agricultural and non-agricultural sectors on the environment is weak in 
the long run but statistically strong in the short-run. Furthermore, unemployment rate in 
the non-agricultural sector has contributed to the slow down o f intersectoral labour 
mobility to the non-agricultural sector. In addition, the increased rural population 
pressure on land has tended to increase the forest area cleared for agriculture. The study 
also observed that the producer price of cocoa tended to exert a significant positive effect 
on deforestation through agricultural expansion. This result suggests, inter alia, that 
polices which tend to provide incentives to farmers and encourage the use o f improved 
technologies which stimulates increased productivity in the cocoa sector will reduce 
pressure on land and slow down the rate of deforestation in Ghana. Other policy 
recommendations and suggestions for future research are also made in the study.
LIST OF ACRONYMS
ADB African Development Bank
ADF Augmented Dickey Fuller
AERC African Economic Research Consortium
BG Breusch Godfrey
COCOBOB Ghana Cocoa Board
CSAE Centre for the Study of African Economies
DGLS Dynamic Generalised Least-squares
DF Dickey Fuller
DOLS Dynamic Ordinary Least-squares
DW Durbin Watson
ECA The European Commission for Africa
ECM Error Correction Mechanism
ERP Economic Recovery Programme
FAO Food and Agriculture Organisation of the United Nations
GCA Global Coalition for Africa
GDP Gross Domestic Product
Ha Hectares
IFS International Financial Statistics Published by the International Monetary
Fund
ILO International Labour Organisation
IMF International Monetary Fund
LM Lagrange Multiplier
MCV MacKinnon Critical Value
OAU Organisation o f African Unity
RSS Residual Sum of Squares
SAP Structural Adjustment Programme
UN United Nations
UNEP United Nations’ Environmental Programme
WRI World Resources Institute
TABLE OF CONTENTS
DEDICATION 1
DECLARATION ii
ACKNOWLEDGEMENTS iii
ABSTRACT iv
LIST OF ACRONYMS V
LIST OF TABLES X
CHAPTER 1 INTRODUCTION 1
1.1. Background 1
1.2. Problem Statement 6
1.3. Objectives o f the Study 8
1.4. Justification of the Study 9
1.5. Organisation of the Study 12
vii
CHAPTER 2. LITERATURE REVIEW 13
2.1. Labour Mobility 13
2.2. Labour Migration and Unemployment 15
2.3. Determinants of Intersectoral Labour Mobility 17
2.4. Deforestation in Ghana 22
2.5. Definition and Determinants o f Deforestation 24
2.6. Empirical Literature on Deforestation 30
CHAPTER 3. METHODOLOGY 35
3.1. Theoretical Framework 35
3.2. Statement of Hypotheses 42
3.3. Specification of the Empirical Econometric Model 43
3.4. Cointegration and Equilibrium Correction Modelling 45
3.5. Description o f Variables and Sources of Data 52
CHAPTER 4. EMPIRICAL ECONOMETRIC RESULTS 57
4.1. Stationarity Tests 57
4.2. Granger Causality Tests 58
4.3. Empirical Results for Long Run Relationship 61
4.4. Short Run Effects 65
viii
CHAPTER 5. SUMMARY AND RECOMMENDATIONS 69
5.1. Summary 69
5.2. Policy Recommendations 70
5.3. Limitations o f the Study and Suggestions for Future Research 71
APPENDICES
Appendix I. Intertemporal Maximisation Problem 73
Appendix II. Correlation Matrix 75
Appendix III. Long Run Relationships 76
Appendix IV. Short Run Relationships 78
LIST OF REFERENCES 80
ix
LIST OF TABLES
Table 1. Results of Unit Root Tests 57
Table 2. Results o f Granger Causality Test 59
Table 3. Results of Long Run Relationships 61
Table 4. Results of Short Run Relationships 65
X
CHAPTER 1
INTRODUCTION
1.1. Background
The relationship between people and the environment has changed over time. 
Nations all over the world are recognising that the environment and development are 
inextricably interlinked. Excessive deforestation, significant declines in natural biomass 
and reduction in soil quality, in many cases induced by rapid population growth, have 
been identified as key factors explaining the apparent fall or stagnation of rural income 
in Africa (African Development Bank et al., 1984; FAO, 1986). The process of 
economic development involves reallocation of resources, inter alia. One such economic 
phenomenon of resource transfer which has been accelerating over the years in 
developing countries is the mobility of labour between the agricultural and non- 
agricultural sectors. A typical dualistic economy postulated by Lewis (1954) and Ranis 
and Fei (1961) suggests the existence of unlimited supplies of labour with a zero 
marginal product in agriculture. The exodus of labour from agriculture to non-agriculture 
would continue if  an autonomous expansion in the non-agricultural sector moves in 
tandem with job creation. The universal validity of labour surplus models may be 
applicable in developed countries, but their strict adherence in developing countries has 
not been upheld (Collier, 1979). This is particularly true of tropical Africa where there
l
exist limited markets of manufactured goods and a slight increase in agricultural 
employment.
In Ghana, the transfer of labour from the agricultural sector to non-agricultural 
sector has occurred in relative terms. The total labour force in Ghana still remains 
predominantly agricultural. The share o f agriculture in total employment has declined; 
but in absolute terms, employment in the agricultural sector has actually increased 
(Ewusi et al, 1983). While there was decline in the proportionate share o f agriculture in 
total employment from 63 percent to 53 percent between 1960 and 1970; in absolute 
terms, there was an increase in the number of persons employed in the agricultural 
sector. In all regions except Greater Accra, well over 80 percent o f the male labour force
aged 15 years and over were employed in the agricultural sector during 1960—1970.
Between 1980 and 1985, formal sector employment increased by 75 percent per 
annum, while the average annual growth of labour force was 2.3 percent. With the 
implementation of the Economic Recovery Programme and Structural Adjustment 
Programme (ERP/SAP), formal sector employment decreased by 9.2 percent per annum 
between 1986 and 1991. However, the growth rate o f the labour force remained the 
same, at 2.3 percent per annum (Boateng, 1999). In the short run, more unemployment is 
created particularly among low-skilled workers through layoffs and retrenchment. A 
recent study conducted by the World Bank shows, among other things, that a significant
2
amount of labour was shed after privatisation of majority o f firms in Ghana (London 
Economics, 1996).
Policy reforms under ERP/SAP initiated in April 1983 have encouraged reversed 
migration in Ghana as urban dwellers have returned to the farm (Jaegar, 1992; Abdulai, 
1999). The primary goal of the adjustment programme was to liberalise the price system 
in order to stimulate production in sectors where Ghana possesses a comparative 
advantage. Evidence suggests major intersectoral flow of workers since 1983 from the 
non-agricultural sector to agricultural sector, this coinciding with growth in the 
agricultural sector especially the production o f cocoa as targeted by the ERP (Fosu, 
1989; Beaudry and Sowa, 1994).
Prior to 1970, 46.5 percent of net migration was directed towards Accra. 
However, the patterns o f migration were completely reversed in the 1980s (Ewusi et al., 
1983). Between 1982 and 1987, net migration out o f Accra accounted for almost 60 
percent of net outmigration in Ghana and the Western Region being an expanding cocoa 
production zone, has attracted a lot of these migrants (Beaudry and Sowa, 1994). 
Agricultural policies pursued in an attempt to achieve economic development by 
following export-led growth models seem to have stimulated extensive deforestation 
through increases in the area of cultivated land (Fosu, 1992, 1996; Fosu and Heerink,
1996).
3
The current observation on the African continent is that most countries seem 
to be caught in a vicious circle of increasing population growth and lack o f non- 
agricultural employment opportunities (Rnerr, 1992). The role o f human population 
pressures on economic development and tropical deforestation has received significant 
attention. The major dividing line in the debate on the causes o f tropical deforestation is 
the explanation emphasising poverty and population growth as the driving force on the 
one hand, and the explanation emphasising market factors such as prices, access costs 
and property rights on the other (Angelsen, 1999). Population would influence 
deforestation through its effect on the minimum acceptable level o f consumption. The 
minimum acceptable level o f aggregate consumption will increase if  population 
increases, leading to more deforestation.
Farmers make a number of farm decisions such as area cultivated, crop 
choice, labour and other inputs, cropping and fallow periods that are relevant to the 
management o f natural resources. Agricultural household, in the view o f Nakajima
(1986), is both a producer and consumer. In the former role, households act like profit 
maximising producers whose activities could lead to deforestation when output prices or 
productivity increases. Increased productivity in frontier agriculture will cause a 
downward pressure on domestic food prices, which will dampen the effects on 
agricultural expansion. Technological progress on the other hand exerts an upward 
pressure on wages, which would dampen the effects of deforestation. The net effect 
therefore will depend on price elasticites in the output and labour markets (Angelsen,
4
1999). Higher population increases the prices of agricultural products and increased 
labour supply would reduce the wage rate. If there are sufficient employment 
opportunities in the non-frontier agricultural sector or off-farm income opportunities, the 
natural tendency is to reduce pressure on forest frontier.
The process of environmental degradation is intimately associated with 
frontier expansion. If there is shortage o f land, the degradation is exacerbated as farm 
workers are forced to overexploit the land. Moreover, when other realistic alternatives do 
not exist, adjustment programmes characterised by the curtailment o f subsidies on 
agricultural inputs, inter alia, stimulate the expansion of export crop production through 
increases in area of cultivated land (Khan and Knight, 1985; Zulu and Nsouli, 1985; 
Fosu, 1989; World Bank, 1989).
The tropical rainforest, which constitutes about 41 percent o f the total tropical 
forest cover, is considered to be the richest and most valuable ecosystem on the earth. 
However, the FAO (1992) estimated that during the 1980s, about 15.4 million hectares 
of tropical forests were lost each year, while from 1990 to 1995 the annual loss was 
estimated at 12.7 million hectares (FAO, 1997). West Africa has experienced marked 
acceleration of forest loss during the last few decades o f the Twentieth Century. Food 
and Agriculture Organisation (FAO) estimates that 30 million hectares have disappeared 
from the region during the Twentieth Century whereas the World Monitoring Centre
5
suggests that only 13 percent of "original" forest cover remains (Leach and Fairhead, 
2000).
The physical destruction of Africa’s tropical forest is largely effected through 
shifting cultivators who act within a socio-economic framework which leaves them 
fewer alternatives for survival than to regularly clear new forests (Knerr, 1992). For 
example, about 70 percent o f original closed forest in Ghana have been destroyed due to 
the demand for agricultural lands, fuelwood and logging and bush fires (Ampadu-^gyez 
et al, 1994). The conversion of land from its natural climax vegetation to a more specific 
vegetative cover and land use system is driven by the human needs of today, short term 
and increasing demand for food, fuel, timber and raw material to improve living 
standards and food security (Knerr, 1992).
1.2. Problem Statement
Tropical deforestation is one of the environmental crises of our time. Historical 
evidence portrays that forests have consistently lost ground to agriculture, and while 
deforestation stabilised in Europe in the last century, the trends in tropical countries 
point towards their accelerating unequal competition (Buschbacher, 1986). According to 
the FAO (1993), the links between population pressure, poverty, non-availability o f non- 
agricultural employment, intersectoral labour mobility and deforestation are thought to 
be very strong. The Ghanaian economy derives its present strength principally from
6
agriculture, mining and forestry. Moreover, food crop and livestock production must 
expand in order to feed the ever-increasing population o f over 18 million people, 
currently growing at a rate of 2.5 percent per annum country (Ghana Statistical Service, 
Ghana Housing and Population Census, 2000).
Despite the growing evidence o f commodity crises in the 1970s in Africa, and the 
serious decline in export earnings from cocoa during the structural adjustment era in 
Ghana, the major focus o f agricultural policy has not been divulged from promoting 
rehabilitation o f cocoa production in agriculture and maintenance o f a monocultural 
export oriented economy. An attempt to diversify exports through policy initiatives of 
tax concessions to promote a new range o f non-traditional export crops such as 
pineapple, coffee, medicinal plants, cane furniture (Jebuni et al, 1992) has also 
unfortunately been a disincentive to the goal of environmental sustainability.
Labour-use patterns differ significantly in Ghana, and different crops require 
different soil types for optimum performance. The local farmer tends to scatter or rather 
distribute his farm in order to take advantage of suitable soils. The foremost factor 
influencing the existing behavioural pattern of farm labour allocation is economic. Soils 
in Ghana, being deficient in major nutrients, are generally poor (Ewusi et al, 1983). 
Small-scale farmers in the Southern sector, unlike their counterparts in the North, do not 
use farm manure or much chemical fertilisers to enrich the soils (Ewusi et al, ibid.).
7
Diminishing returns sets in as labour-use is intensified. The farmer then chooses to 
distribute his labour over a number of lands.
Ghana has over 15,000 sq. km of rain forest but its rich forest resources are 
rapidly diminishing at an alarming rate of 25,000 hectares per year (Fosu, 1996). The 
issues which arise are the following: What are the determinants o f intersectoral labour 
mobility in Ghana during the 1970 to 1999 period? To what extent has the transfer of 
labour between the agricultural and non-agricultural sectors contributed to deforestation 
in Ghana? The present study seeks to address these pertinent issues.
1.3. Objectives of the Study
The primary objective of the present study is to analyse the effect o f migration of 
labour on deforestation in Ghana.
The specific objectives are as follows:
1. To estimate the rate of intersectoral labour mobility between agriculture and non­
agriculture sectors in Ghana and the magnitudes o f the effects o f its determinants
in Ghana during the period 1970—99.
2. To model the effects of migration of labour on deforestation in Ghana.
3. To provide policy recommendations concerning occupational labour mobility, 
and its effects on deforestation in Ghana.
1.4. Justification of the Study
The current study is very relevant for a number o f reasons. First, the factors 
underlying labour force migration are very crucial in any development process. 
Ignorance of this has resulted in several problems such as chronic urban unemployment 
and under-employment in almost every contemporary developing country (Todaro, 
1969). Second, conversion of forestland has conferred substantial economic benefits by 
allowing expansion of agriculture and sustenance for a rapidly growing population. 
However, diminishing returns limits this process. In addition, the livelihoods o f forest- 
dependent people are undermined and forest-based foreign exchange revenue is lost. 
Third, the current modelling approach views deforestation as a title establishment 
strategy or an investment decision (Angelsen, 1999). This is because forest clearing and 
cultivation give some rights or claim to farmers. This contrasts with the social planner’s 
viewpoint of deforestation as a form of disinvestment (Deacon, 1994) with regard to 
resource use decisions.
Fourth, Ghana abounds in natural resources, which have contributed significantly 
to the country’s agricultural and industrial development since the attainment of 
independence. However, a point has been reached where it is necessary to balance
9
economic growth with rational management o f resources to ensure that the resource base 
is not entirely eroded in the process of development. Ghana’s Environmental Action 
Plan also makes this point (EPA, 1994).
Fifth, not only are forests home to some important plant species on earth, but 
they also play the vital role of climate regulation and carbon sequestration. Potentially, 
one of the most damaging effects of forest clearance is its impacts on the planet’s climate 
through global warming and green house effects. Trees and other green plants absorb 
carbon dioxide and produce oxygen through photosynthesis whereas animals breathe in 
oxygen and breathe out carbon dioxide. The destruction of tropical forests therefore 
brings about a disastrous imbalance in the amount of carbon dioxide produced and
St
recycled, leading to a greater build-up in the atmosphere and increased climatic change.
Sixth, forests contribute immensely to hydrological stabilisation. Once forests are 
cleared, topsoils are eroded making the soils unusable. This in turn could lead to 
devastating floods and desertification (Anderson, 1990). Perhaps, the greatest danger 
according to Buschbacher (1986) and Sunderlin and Resosudarmo (1999) due to this 
increased soil erosion, siltation and watershed damage, is the loss of genetic diversity.
Seventh, deforestation is changing the habitats of disease-carrying insects and 
creating conditions that may help to spread malaria, river blindness and other devastating 
diseases (Ted Case Studies, 1997). The worms that cause river blindness or
10
onchocerciasis are transmitted mostly by a blackfly known as Simulium damnosum. 
These cytoforms whose habitat is in the Savanna regions are now beginning to spread 
into areas of cleared forests in Ghana.
Eighth, deforestation is threatening yet another area linked to the well being of 
Ghanaians. Most Ghanaians have always opted for herbal treatment over Western 
medicine (Ted Case Studies, ibid.). A report by the Centre for Scientific and Industrial 
Research in Plant Medicine paints a gloomy picture that more than 250 indigenous trees 
and plants of herbal extracts and healing properties are in danger o f extinction if  trees 
continue to be cut indiscriminately from the forests (Ted Case Studies, ibid.).
Nineth, the reality o f adverse impacts o f structural change and economic 
development on the environment cannot be over emphasised. Lewis (1965, 1966) stands 
out among development economists who repeatedly called the world’s attention to the 
serious problem of urban unemployment, but his analyses were qualitative rather than 
rigorous (Todaro 1969). Knerr’s (1992) effort to analyse the socio-economic effects on 
deforestation in particular, in less developed countries, was analytical in nature. Other 
researchers have looked at deforestation in general (Gomez-Pompa et al, 1972, 1980; 
WRI 1985; Vincent et al, 1992) without considering the potential linkages between 
migration of agricultural labour and depletion of forest resources as a result o f structural 
reforms and economic development.
11
Finally, the growing sustainable development literature seeks to identify 
development strategies that could lead to sustainable use o f natural resources and the 
environment (Munasinghe, 1996). This suggests the need, inter alia, to analyse 
intersectoral labour mobility and how this influences deforestation. The present study 
which employs a quantitative modelling framework to analyse intersectoral labour 
mobility and deforestation in Ghana is therefore relevant.
1.5. Organisation of Study
The study is structured into five chapters. In Chapter 2, a review o f the relevant 
literature is undertaken. The methodology formulated to generate the requisite data and 
the model required to achieve the set objectives are outlined in Chapter 3. The empirical 
econometric results relating to the effects of the determinants of labour mobility as well 
as the effect o f labour mobility on deforestation in Ghana is presented in Chapter 4. 
Conclusions and recommendations are distilled in Chapter 5.
12
CHAPTER 2
LITERATURE REVIEW
This chapter first reviews the literature on labour migration with particular 
emphasis on the types of migratory movements. Moreover, light is shed on the various 
theories of migration and the basic determinants of intersectoral labour mobility. The set 
o f literature on deforestation in Ghana is reviewed. Definitions o f forests and their 
classification are then provided, and the determinants of deforestation are also reviewed. 
The theoretical linkages between labour mobility and deforestation are also reviewed. 
The chapter also attempts to trace some of the empirical work on labour mobility and 
deforestation by paying attention to results, conclusions and recommendations from 
previous studies.
2.1. Labour Mobility
Labour migration and labour mobility have been used interchangeably in the
literature on migration. Traditionally, migration has been viewed as an activity
undertaken by an individual (Sahota, 1968; Falaris, 1979) or household (Stark, 1991;
Shields and Shields, 1993) who, as a rational decision maker (Sjaastad, 1962; Beals et
a l, 1967; Knight, 1972), moves from one location to another in search of job in response
to income incentives (Greenwood and McDowell, 1991; Faini and Venturi, 1993) or cost
of living differentials (Cebula, 1981, 1993) and location-specific amenities (Clark and
13
Cogrove, 1991). Diez-Canedo (1991) has observed that family migration in the U.S.- 
Mexico border region differs from typical international movements which portray 
migrants in the Third World cities as destitute and sometimes confronted with endemic 
political upheavals and violence (Morrison, 1993). Typical migrants therefore take 
migration as a last resort as they are forced out of their countryside where they had no 
alternatives (Breman, 1985, 1990; Chapman and Prothero, 1985; Pamreiter, 1995; Singh, 
1995).
Modem economic development has invariably been accompanied by pronounced 
migration from rural and urban sectors o f employment. Migratory movements in sub- 
Saharan Africa, apart from rural to urban, takes the form of rural to rural which has had 
implications for agricultural development in the region (Abdulai, 1999). Permanent 
migration in Ghana has been associated with movement of rural dwellers to either rural 
or urban areas for long periods of time (de-Graft-Johnson, 1974), whereas temporary or 
seasonal migration involves short-term movement o f labour from the north to the south 
to take up temporary employment on cocoa plantations.
Theories and policies of economic growth and development have cited 
intersectoral or occupational mobility o f labour from agricultural to the non agricultural 
sectors as a key instrument for promoting economic welfare (Ghatak and Price, 1996). 
Recent evidence seems to underline the case of adopting economic policies that would 
re-allocate labour from low productivity to high productivity areas, and promote factor
14
(labour and capital) mobility that would improve efficiency of the tradable sector as an 
engine of growth (World Bank, 1989; 1990).
In Ghana however, intersectoral labour mobility is also seen as an internal 
migration where domestic labour supply is reallocated between rural (agriculture) and 
urban (nonagricultural) sectors. While one mostly observes migration from rural to urban 
employment, there has been periodic brief interruptions marked by episodes of reverse 
migration from industry back to agriculture due to sharp economic downturns as it were 
in the United States during the 1920-21 recessions (Hatton and Williamson, 1994), and 
during the adjustment periods of Ghana in the 1980s (Ewusi, 1987; Fosu, 1989; 1996; 
Abdulai, 1999).
2.2. Labour Migration and Unemployment
The link between migration and unemployment has received significant research 
attention. The growth of the urban free-entry sector has been responsible for both the 
emergence o f substantial urban poverty and labour misallocation (de Hann, 1997). In the 
words of Bencivenga and Rosenweig (1997), migration is often associated with 
significant urban underemployment. The pioneering work of Todaro (1969) which 
predicted that urban job creation will increase the size o f the urban informal sector has 
been argued not to be true in some developing countries (Collier, 1979) and therefore 
provide inadequate and untested framework in which to analyse rural-urban labour
15
market interactions. Todaro (1969) believes that potential migrants have the choice 
between earning their marginal product in agriculture or of seeking employment by 
joining the urban unemployed and participating in search for wage employment.
If all jobs become vacant annually, then the average annual probability of gaining 
employment is equal to the proportion of jobs to the labour force. In the short run, the 
equilibrium expected income from job search will be determined by the existing stock of 
urban employed and the returns to job search schedule for the unemployed. Hence, the 
configuration o f the volume o f migration as a function of the level o f expected income, 
which is determined by the stock o f unemployed, is only true in the short run. In the long 
run, the equilibrium rate of migration determines the equilibrium level of 
unemployment. Since positive flow o f migration with constant level o f wage and 
unemployment will increase the stock o f unemployed as the expected income is reduced 
(Collier, 1979).
Pissarides and Wardsworth (1989) view the effects of unemployment on 
migration in three different ways. Firstly, unemployment increases the propensity of an 
individual to move (Bailey, 1993; Herzog, Schlottmann and Boehm, 1993), particularly 
if he is the breadwinner for the household (Da Vanzo, 1978). However, high 
unemployment insurance (Goss and Paul, 1990; Antolin and Bover, 1993) and long 
unem p loym en t duration (McHone and Rungeling, 1993) negate this effect. Secondly, 
regional unemployment differentials may encourage migration (Bentolila and Dolado,
16
1991) and for the risk averse worker, employment probabilities are more important 
determinants o f migration than wage rate (Treyz et al, 1993). Thirdly, increases in the 
rate of unemployment may be a deterrent to potential first-time movers (Withers and 
Pope, 1985; Bentolila and Dolado, 1991).
2.3. Determinants of Intersectoral Labour Mobility
The basic determinant of intersectoral labour mobility is the existence o f income 
differences between sectors (Mundlak and Larson, 1997). The influential Todaro (1969) 
model postulated that migration is an increasing function o f expected differential 
between sectors and the probability o f obtaining a job which is a push factor in 
migratory decisions. Emerson (1989) explains that those with comparative advantage in 
activities associated with migratory work are more likely to migrate than those without. 
The human capital paradigm o f Sjaastad (1962) assumes that an individual locates 
himself spatially so as to maximise the present stream o f returns he enjoys. The 
economic component in that particular case is the difference in the discounted present 
value o f the expected incomes between source and receiving sectors and the net o f the 
direct cost of moving. Labour therefore calculates the gross utility for remaining in a 
sector vis-a-vis the one for moving to the best alternative sector of their choice. A move 
takes place if  the cost o f migration does not exceed the gross gain o f moving. Changing 
occupations and for that matter sectors is very costly.
17
Molle and van Mourik (1989) believe that ability to migrate depends not only on 
access to financial resources to cover direct costs of movement and acquisition of 
information, but also on the psychic costs such as the unwillingness to leave family, 
friends and a familiar culture and language at the origin and cultural diversities, eating 
habits and homesickness at the destination. According to Yap (1977), Beals et al (1967) 
and Boijas (1987) distance is introduced in every migration function to proxy for these 
costs. In the opinion of Beals et al (ibid.), distance is expected to be a deterrent to labour 
mobility.
Other determinants of labour mobility are couched under the laws o f migration. 
The push factors originate mostly from rural areas such as the level o f poverty and 
income inequality in the agricultural sector and outmoded land tenure systems (Baneijee 
and Kanbur, 1994; Sahota, 1968). The pull factors on the other hand include 
employment, education and urbanization. Urbanization per se, as remarked by Baneijee 
and Kanbur (1981), may not attract migrants but reflects the existence o f amenities, 
better educational opportunities, degree of market orientation of the population and the 
quality of transport network.
The presence o f self-selectivity in migratory stream suggests that migrants sort 
themselves into advantageous groups given their characteristics (Roy, 1951; Maddalla 
1983). In support of this hypothesis, Mundlak and Larson (1997) believe that the 
younger generation and the better educated have a higher propensity to migrate since the
18
young can benefit from longer life time income as compared to the old while education 
enhances one’s chance o f getting employment in a sector.
The introduction of expectations which captures the probability o f getting a job 
in a sector reveals the importance of unemployment in migration studies (Johnson, 1948; 
Todaro, 1969). Johnson reveals further that when the unemployment rate in a sector is 
high, the rate o f labour mobility to that sector is likely to be small. Because of 
unavailability of immediate non-agricultural work, the issue o f lifetime income becomes 
important since lower income in the period after migration may be compensated by 
higher income in future (Mundlak and Larson, 1997). Pissarides and McMaster (1990) 
who maintained that low unemployment regions are characterised by negative net 
migration buttress this point. According to Gordon (1985), the response o f net migration 
to the unemployment ratio is that, in times of high unemployment, migration in general 
suffers.
Levy and Wadycki (1973) used the stock of previous migrants in a sector as an 
explanatory variable in macro migration functions to represent the family and friend 
contact. Salvatore (1981) was of the view that previous migration rate is inversely 
related to the cost of migration due to the fact that previous migrants can provide job 
information, cost of transportation, temporary shelter and subsistence to migrant 
relatives and friends.
19
Economic growth and the distribution o f its benefits depend on how well the 
labour market mobilises human resources in new activities, locations and skills 
(Kannappan, 1988). Moreover, the development process of less developed country 
hinges very much on its ability to raise the level of productivity in the various sectors of 
the economy. Unfortunately, such countries are handicapped, since a large segment of 
the working force seem to be concentrated in the agricultural sector which contributes 
little in relationship to their size to growth in total product. Their country’s population is 
therefore caught in a vicious circle of rural underdevelopment and urban unemployment. 
Gordon (1969) was o f the view that underemployment in less developed countries could 
be tackled by limiting population growth, increasing agricultural output levels and 
reducing the agricultural working force by increasing job-opportunities in the non- 
agricultural sector.
When additional labour leave agriculture, the average agricultural surplus falls. 
The terms of trade turn further against the non-agricultural sector such that unless the 
industrial real wage is reduced, its exchange value will fall (Gordon, ibid.). It is therefore 
important to note that the levels of development and industrial composition are dictated 
by the extent of mobility rate between sectors. From orthodox economic analysis, one 
would expect the exodus of labour from agriculture to non-agriculture as soon as the 
economy experiences low marginal product in the latter sector as its supply curve 
flattens out. Balough (1962) has however observed that, the characteristics of West
20
African urban unemployed seem not to make this likely, particularly even after 
prolonged spells of unemployment.
Labour force that sought job in the formal sector has three options. Migrants 
could return to the villages whence they came or stay and contribute to urban 
unemployment or they could swell the ranks of the informal sector. Todaro (1985) and 
Kannappan (1983, 1985) argued that rural workers would migrate to the cities even if 
they were unlikely to find jobs and in that particular case earning nothing but with the 
hope o f eventually securing employment in the formal sector. The chances of getting an 
urban job falls as more rural workers join the ranks o f the unemployed, thus 
discouraging additional migration in excess of the rate at which new jobs are being 
created in the formal sector. Gelb et al. (1989) have demonstrated that in most 
developing economies, the diversion of capital into unproductive public employment 
does not reduce urban unemployment.
In many developing countries especially in sub-Saharan Africa, high population 
growth rates and increasing labour force participation rates continue to accelerate 
economic growth. These countries increasingly find themselves confronted with the task 
of achieving high rates of economic growth in order to absorb their growing labour force 
in productive employment. The situation is even more dramatic in countries which 
experience declining rates of economic growth and productivity (Abdulai, 1999). 
Increased supply of and reduced demand for labour in these countries have resulted in
21
declining employment and productivity o f labour and high rates o f intersectoral labour 
mobility, rising urban unemployment, and a shift to low income jobs in the informal 
sectors outside the organised urban labour market o f the urban economy (Sabot, 1977).
Early writers such as Nurkse (1953) and Lewis (1954) cited zero or low marginal 
productivity o f labour in agriculture and concluded that there was hardly any disguised 
unemployment in agriculture in West African countries. They maintained that food 
shortages in most West African countries in the second half o f the 1960s was mainly 
caused by shortages of labour resulting from the high rate of migration o f workers from 
rural areas where food is produced to towns. Beals and Menzes (1970) have also shown 
in their empirical study on migrant labour and agricultural output that temporary 
migration improves the allocation o f resources and has contributed significantly to 
growth of output in Ghana.
2.4. Deforestation in Ghana
The population of Ghana stood at about 18 million during the 2000 population 
census Ghana has a total area of 238,540 sq. km o f which land area is estimated to be 
about 230,020 sq. km. Trends in the population density reveal an interesting scenario. 
Between 1984 and 2000, the percentage change in population density in the Ashanti, 
Brong-Ahafo and Western Regions where the country’s remaining forest are 
concentrated were 52.5 percent, 51.2 percent and 59.2 percent respectively while in all
22
regions, the percentage change in the population density remained at about 49.7 percent 
(Ghana Statistical Service, Ghana Population Census, 2000). Land use distribution in 
Ghana as at 1997 was as follows: 5 percent arable land, 7 percent permanent crop, 15 
percent meadow and pasture, 37 percent forest and woodland and others are 36 percent 
(FAO, 1997).
The scale o f deforestation in the moist tropics has only recently been quantified 
on a worldwide basis. Data compiled by the FAO show that during 1980-85, the loss of 
tropical closed forests averaged 75,000 km2 per year, an area roughly equivalent to the 
country Panama (WRI, 1986). Over the last 40 years, Africa alone has lost one quarter 
of its tree cover (WRI, 1989). The rate of deforestation in Ghana has increased by 50 
percent over the last ten years (FAO, 2000). Since 1981, the annual rate o f deforestation 
has been two percent per year or 750 hectares each year. Ghana’s tropical forest area is 
now 25 percent o f its original size. The devastation of Ghana’ s forest is further 
underscored by The Ecologist. Between 1937/38 and 1980/81, the area o f closed forests 
in Ghana was reduced by 64 percent from 47,900 km to 17,000km and open woodland 
declined by 37 percent from 111,000km to 69,000km (The Ecologist, 50).
The Forestry Department of the Ministry of Lands and Forestry also estimated 
that in 1954, there was 1,544,000 hectares of unreserved forests constituting about 19 
percent o f the original forest cover in Ghana. By 1972, this had reduced to about 362,000 
hectares and towards 1989, less than 5 percent of the original cover of high forest was
23
left outside the reserves while clearing continued at around 20,000 hectares per annum. 
An estimate by the World Bank indicates a deforestation rate o f 75,000 hectares per 
annum in Ghana (World Bank, 1989).
According to Ruf (1997), the supply of cocoa seems very dependent on the 
clearance o f tropical forests. In an effort to open up more area for cocoa production, trees 
are cut down. In addition, cocoa farmers slash and bum forest themselves or move on to 
land which has been commercially logged. Ghana has therefore started feeling the pain 
o f the associated environmental imbalances. The impact o f deforestation in Ghana is 
seriously jeopardising the country’s future. Environmentalists believe that if  the trend 
continues unchecked, there would be no forest cover in Ghana by the year 2020 and 
conservationists also predict that if  the current wanton log extraction and reckless forest 
destruction continue, Ghana’s forest would be gone in the next 45 years (Ted Case 
Studies, 1997). There is therefore the need to understand this phenomenon. Hence, the 
theme of the present study is relevant.
2.5. Definition and Determinants of Deforestation
The term tropical forests refers to an ecosystem with a minimum of 10 percent 
crown cover of trees and/or bamboo interspersed with wild flora and fauna and natural 
soil conditions which has not been subjected to agricultural activities (Singh, 1993; 
FAO, 1997). The FAO considers forests as a vegetation cover in which the woody
24
element is mainly a tree cover with a height of more than 17 metres. Kahn and 
McDonald (1994) categorise forests into "unprotected forests", " reserved forests" and 
"extractive reserved forests". The "unprotected forests" also termed "non-reserved 
forests" are those which, having suffered heavy encroachment in the absence o f control 
over occupation, may become as Lambardini (1994) puts it, an open, resource 
collectively open to public use resulting in degradation. "Reserved forests" or 
"intensively managed forests" which exclude all types o f production activities and 
impose severe restrictions on public use, are considered to be better organised and in a 
state of maintenance (Lai, 1989). According to Chakraborty (1994), "reserved forests" 
have a high degree o f area under dense forests o f over 40 percent crown cover. 
"Extractive reserved forests" or "working forests" however allow certain types of 
production activities.
Natural forests may be grouped into virgin, primary and secondary forests. Other 
subcategories exist which make the definition of forests extremely difficult and in some 
cases country-specific. These include "logged-over forests", "undisturbed forests", "old- 
growth forests", " fragmented forests" and "clustered forests". Amelung and Diehk
(1992) define "undisturbed forests" as forests, which for the last time have been logged 
60-80 years ago. " Secondary forests" differ from "untouched forests" with respect to 
their ecological characteristics. "Closed forests" are characterised as having unbroken 
canopies whereas in "Open forests", the tree cover is broken with the trees growing 
further apart. Cropper and Griffiths (1994) refer to forests interspersed with agricultural
25
and other land use as "fragmented forests" whereas those with dense vegetative cover, 
devoid of agricultural and other land use activities as " clustered forests".
It is also important to distinguish between " deforestation" and "degradation". 
The FAO defines "deforestation" as the permanent conversion o f forestlands into other 
uses (FAO, 1993). The FAO-UNEP definition however is a restrictive one, which 
defines deforestation as a decline in a closed tropical forest area. Panayotou and 
Sungsuwan (1994) explained deforestation as an abrupt change in forestland use from 
forestry to something else. They considered land previously under forest but now under 
shifting or permanent agriculture, plantation or crop as deforested even if  the new land- 
use is more beneficial than forestry. In contrast, forest "degradation" is the gradual 
deterioration in the quality of the selective and destructive logging and uncontrolled bush 
fires.
"Degradation" may eventually lead to "deforestation" by either conversion to 
agriculture or other non-forestry uses, which may degenerate, into wastelands. The 
above definition is supported by Singh (1993) who referred to degradation as the 
changes occurring within both open and closed forests that negatively affect the stand or 
site and in particular, lower the productive capacity o f forests. According to the FAO, 
deforestation is said to have occurred when the canopy is reduced to 20 percent or less 
and for tropical countries when the crown cover is depleted to less than 10 percent 
(FAO, 1997). However, any changes within the forest class which severely affect the
26
regenerative capacity o f the forest is termed degradation instead o f deforestation even if 
the crown cover is reduced to 30 percent.
The physical destruction of Africa’s tropical forest has reached unimaginable 
proportion as the conflict between food production for subsistence survival and the 
preservation of forest resources deepens. The most seriously affected region is West 
Africa where the rate of deforestation through agricultural expansion on closed forest is 
believed to be six times higher than in Central Africa (Myers, 1989). Experience in 
Ghana shows that when the bearing life of cocoa trees is finished and the trees die as a 
result of old age, disease and damage, one of the choices available to the farmer include 
migrant forest degrading subsystem. This land use subsystem involves a search for new 
forest for new farms, which becomes unsustainable where forestlands are limited and 
population is increasing (Gyasi et al, 1993).
A number of factors have interacted to bring West Africa’s rate o f deforestation 
to the highest level. Angelsen and Kaimowitz (1999) classified the determinants of 
deforestation into immediate and underlying causes. While the immediate causes 
attribute tropical deforestation rates to agricultural prices, prices of inputs and credits, 
off-farm employment opportunities, technological progress in agriculture, accessibility 
and roads, the underlying causes link deforestation to population pressures, income level 
and economic growth, external debt, trade and structural adjustment, among others. 
Other authors look at it from the ecological perspective (Myers, 1984; WRI, 1985). The
27
political economy perspective of causes o f tropical deforestation emphasises the role of 
public and private capital rates o f deforestation (Hecht, 1985; Shane, 1986).
Higher prices o f agricultural products stimulate forest clearing in that, as frontier 
agriculture becomes more profitable, both existing population and migrants from areas 
begin to shift resources into forest clearing. Higher fertiliser prices induce farmers to 
adopt more intensive production systems that use more land and less fertiliser. This 
hypothesis is also true for higher prices of agricultural inputs such as seeds, pesticides 
and hand tools (Ruben, Kruseman and Hengsdijk, 1994). Credit expansion could reduce 
the pressure on forests if  intensive agriculture and forest management investments were 
to be pursued. However, it increases pressure on forest if  it was used to finance activities 
that promote forest clearing (Angelsen, 1997; Pfaff, 1997). In addition, the activities of 
log-exporting timber firms, road construction and mining activities may open up 
previously forested areas where farmers could move on to cultivate.
According to Southgate (1990), technological changes that increase yields 
without significantly altering labour or capital requirements can be expected to increase 
deforestation. As population grows, the need for more land for food, fuelwood, timber or 
forest products may increase deforestation (Hafner, 1990; Panayotou and Sungsuwan,
1992). In Ghana, decades of spread o f shifting cultivation promoted by large-scale 
logging operations and transformation of closed woodland into permanent tree crop 
mainly cocoa, have destroyed nearly all the tree cover o f once forest-rich country
28
(Kneer, 1992). Policies that favour rural wage and generate off-farm employment 
opportunities for rural people should reduce deforestation and simultaneously conserve 
forest. Some analysts claim that government policies which affect migration such as 
road construction, colonisation policies, agricultural subsidies and tax incentives rather 
than population growth lead to deforestation (Panayotou and Sungsuwan, 1994). In their 
view, people migrate to forestlands because clearing of forests for agriculture has 
become economically attractive so the size o f the population in that particular case 
cannot be considered as an independent variable in the deforestation model.
Evidence from several analytical and quantitative macroeconomic models reveal 
that policies which improve the terms of trade in favour of agriculture and structural 
adjustment policies in particular that tend to raise the prices received by fanners 
potentially increase deforestation (Cruz and Repetto, 1992; Jones and O’Neil, 
1994,1995; Babier and Burgess, 1996). Moreover, policies designed to promote 
agricultural and forest product exports are likely to affect deforestation more than 
policies to promote production for domestic market. The relative incentive for the 
production of export crops like cocoa (Fosu, 1992; Amanor, 1994; Fosu, 1996; Fosu and 
Heerink, 1996) and other non-traditional agricultural exports like pineapples, coffee and 
medicinal plants (Jebuni et al, 1992) have contributed significantly to environmental 
degradation during the periods of Ghana’s economic reforms.
29
2.6. Empirical Literature on Labour Mobility and Deforestation
The empirical study based on the original Todaro (1969) theoretical framework 
that migration is an increasing function of expected income differential between dual 
sectors is found to be true in most cases. Winkelmann (1966) had earlier observed in a 
study based on the United States that lower income to farmers speeds up the exodus of 
farm labour from the agricultural sector to the non-agricultural sector. Farm workers 
therefore seasonally migrate for agricultural work because o f economic choice (Lucas 
and Stark, 1985; Taylor, 1987). These results are also consistent with what Baneijee and 
Kanbur (1981) obtained for Indian data and House and Rampel (1980) obtained for 
Kenya.
Mundlak and Larson (ibid.) also observed from their study of intersectoral labour 
mobility based on cross-country data that, the rate of migration is positively related to 
the ratio of income in non agriculture to agriculture and that labour supply from 
agriculture to non agriculture is upward sloping. They found out that as labour leaves 
agriculture, labour productivity increases, leading to a decline in both income differential 
and migration respectively. However, migration stops when the income is roughly equal 
across sectors, thus validating what Bamum and Sabot (1977) had earlier on observed in 
their study on rural-to-urban migration in Tanzania.
30
The choice o f income or wage as an appropriate measure in migratory studies has 
been a bone o f contention. Sahota (1968), Bamum and Sabot (1977), Falaris (1979) and 
Pissarides and McMaster (1990) employed wages in Brazil, Tanzania, Peru and Great 
Britain respectively. Beals, et al (1967) used income in their Ghanaian study but even 
here, limited their estimates to regional averages while other researchers like Knight 
(1972) and Mundlak and Larson (ibid.) have preferred GDP as a proxy for national 
income due to unavailability of good quality wage data. Moreover, economic advantage 
is better captured by income than by wages. Wage data do not provide an adequate 
picture o f employment opportunities due to lack o f seasonal variations in agricultural 
work. Also, when wage is used, agricultural non-wage income such as rent on land and 
returns to both physical and human capital, which constitutes an important component of 
total income, are neglected.
More recently, the importance of the informal sector as a base for job-search in 
the formal sector has been stressed (Thomas, 1992; Bhattcharga, 1993). Baneijee and 
Bucci (1994) found for India internal migrants that, workers currently in low income 
employment, the landless, and the better educated are the most likely to engage in on- 
the-job search; thus, invalidating Todaro’s job-lottery and high unemployment view of 
urban labour markets in the developing world.
31
Distance sharply reduces migration (Sjaastad, 1962; Beals et al, 1967). Boijas
(1987) therefore concluded that, distance is a strong deterrent to migration and can be 
used to proxy for both monetary and non-monetary costs of migration. Distance is an 
important variable in migration studies but some studies have tended to omit it from their 
empirical estimations citing a possible correlation with the time and regional dummies 
(Mundlak and Larson, 1979). Although it is difficult to estimate psychic costs associated 
with the feasibility of migration, Molle and van Mourik (1989) found significant and 
negative influence of psychic costs on migrations within Europe.
Empirical studies by Baneqee and Kanbur (1981) based on Indian data found 
inequality of land holdings to be positive which indicates, ceteris paribus, that the more 
unequal the land distribution is in the rural areas, the higher the rate o f migration from 
that sector. Their study further reveals that the coefficient of the poverty index in the 
rural sector was negative, running counter to the idea o f rural push which is supposed to 
encourage migration. Poverty in this case seems to be hindering migration probably due 
to the inability to finance migration expenditures. Factors affecting deforestation, the 
interaction between them and the magnitudes of their effects all vary significantly from 
one location to another. Global views, however, emphasise the similarities between 
countries and regions instead of differences. Empirical studies on deforestation have 
utilised both microeconomic and macroeconomic models. The micromodels seek to 
explain how individuals allocate their resources using standard economic variables such 
as background preferences, prices, institutions, access to infrastructure and services, and
32
technological alternatives. A major distinction is between models that assume that all 
prices are market determined and that farmers are fully integrated into perfect markets 
(Southgate, 1990; Bluffstone, 1995; Angelsen and Kaimowitz, 1999) and those that do 
not (Dvorak, 1992; Holden, 1993). Production decisions are studied as a profit- 
maximising problem. The macromodels on the other hand emphasise the relations 
among underlying variables, decision parameters and deforestation. Most o f the models 
include interactions among different sectors such as subsectors o f agriculture, forestry 
and manufacturing which make them useful in analysing the underlying causes of 
deforestation.
Ruben, Kruseman and Hengsdijk (1994) and Angelsen and Kaimowitz (1999) 
found less deforestation when agricultural prices were higher while models that assume 
profit-maximising behaviour show the opposite (Monela, 1995). Empirical regression 
models on Tanzania by Angelsen and Kaimowitz (1^98) and on Thailand by Panayotou 
and Sungsuwan (1994) all found a positive correlation between higher agricultural prices 
and deforestation. The effect of changes in agricultural input prices on forest clearing 
however leads to an indeterminate conclusion and mixed empirical evidence especially 
in the case o f fertiliser. Regression models used on South Africa suggest a positive 
correlation between fertiliser prices and deforestation (Monela, 1995; Holden, 1997) due 
to intensification. Other studies on Latin America by Babier and Burgess (1996) 
indicates a reduced deforestation as fertiliser prices increase, the reasons being that 
agriculture becomes less profitable when there is a higher cost associated with the
33
acquisition of fertiliser. Moreover, higher prices of inputs such as seeds, pesticides and 
hand tools reduce forest clearing (Ruben et al, 1994; Monela, 1995). Although empirical 
studies on Africa and Asia have ignored the issue o f credit availability, Monela (ibid.) 
found a positive relationship between credit availability and deforestation in Tanzania.
According to Nghiep (1986) and Holden (1997), technologies that make 
intensive production systems more profitable, reduces the demand for forestland to be 
converted to agriculture. From analytical and empirical models, greater access to forest 
and markets accelerates deforestation (Chomitz and Gray, 1996; Mortens and Lambin,
1997). All microeconomic models suggest that higher rural wages and greater off-farm 
employment opportunities reduce deforestation by making agricultural and forestry 
activities more costly (Holden, 1993; Ruben et al., 1994; Bluffstone, 1995). Several 
multicountry regression models show a positive correlation between population density 
and deforestation but such results have been spurious and criticised as they rely on FAO 
Forest Resources Assessments based on population data (Palo, 1994; Rock, 1996). 
Although empirical evidence on the relation between population growth and 
deforestation has been weaker, other researchers like Kimsey (1991) and Rock (ibid.) 
obtained a positive relationship between them while Cropper and Griffiths (1994) found 
no effect. Studies on developing countries associate higher national per capita income 
with greater deforestation although the models have significant methodological 
weakness (Burgess, 1993; Krutilla et al, 1995; Mainardi, 1996). Empirical evidence on 
the impact of economic growth rates on forest clearing is even weaker.
34
CHAPTER 3 
METHODOLOGY
The long run and short-run effects of intersectoral labour mobility on 
deforestation are analysed by employing an econometric model comprising a system 
of simultaneous equations. The order of integration of the relevant variables are first 
established using the Augmented Dickey-Fuller (ADF) test. The Dynamic 
Generalised Least-Squares (DGLS) estimator, introduced by Stock and Watson
(1993), is applied in the specification of the models after which the Iterative Three- 
Stage Least squares procedure is used to estimate the parameters o f the simultaneous 
equations. The long-run causal flows from the explanatory variables to labour 
mobility and deforestation are examined with the Granger causality test. Having 
determined the existence of cointegration between the variables, the DGLS approach 
is finally applied to give the models error correction representations so that the short- 
run relationships could be analysed. The theoretical framework and the hypotheses 
validated in the study are first presented in this chapter. Next, the empirical 
econometric model is specified followed by the cointegration and error correction 
modelling procedures. Finally, the description of variables and the sources of data 
employed are indicated.
3.1. Theoretical Framework
The impact of intersectoral labour migration on the rate of forest decline in 
Ghana is analysed in the context of aggregate demand for forestland for conversion 
and aggregate supply of labour by agricultural households. Supply and demand 
functions of agricultural households in Ghana are specified under the assumption of a
35
competitive rural labour market (Abdulai and Delgado, 2000). The aggregate demand 
for forestland for conversion is formulated following Panayotou and Sungsuwan
(1994), Lopez (1997, 1998), Cropper et al. (1999) and Barbier (2000). Here, 
agricultural production behaviour of a representative rural household with access to a 
stock o f non-agricultural land such as forest area is assumed, following the existing 
literature (Barbier, 2000; Lopez 1997, 1998), in order to determine the derived 
demand by households for forestland conversion to agriculture. In Ghana, especially 
the indigene households in most rural areas happen to be the original landowners of 
family lands inherited through paternal or maternal lineage or sometimes inherited 
from the husband’s family (Besley, 1993; Sakyi-Dawson, 2000). In addition, rural 
migrant households have access to stool lands or community lands through lease 
(rented lands), sharecropping and in some cases, through outright purchase (Sakyi- 
Dawson, ibid.).
Furthermore, the representative rural household is assumed to be a price taker 
in all markets for commodities and services it buys, consumes and produces, and 
profit and utility maximising decisions of households are determined. The optimal 
household production function is determined independently o f consumption and 
leisure choices. The agricultural production function (Y) o f the household is 
expressed as the product of the stock of arable land (in ha) A and agricultural 
productivity or yield per hectare, s. If  agricultural productivity is assumed to be 
determined by the amount of purchased inputs X and household labour allocated to 
agriculture L*, then the basic agricultural production technology is given by
36
Y = A. s (X, L*), sx > 0, sxx< 0, sL* > 0, s l*l* < 0 (1)
Where sx denotes the first order partial derivative of s (•) with respect to X, sxx denotes 
the second order partial derivative of s (•) with respect to X, and so on. The stock of 
agricultural land is not constant but grows by the amount of new land n(L) brought 
into production by allocating some household labour L to convert forestland to 
agriculture. It is assumed that a proportion of agricultural land is taken out of 
production at each time in period t at a constant rate k. The net expansion of 
agricultural land is therefore given by 
P) A •
—  = A , = n (L t) - k A t (2)
9t
where n L > 0 ,  n LL < 0 and A ( o ) = A 0 and k =£ 0.
Maximising aggregate profits n  of the agricultural sector over an infinite time 
horizon, the relevant optimisation problem can be written as
Max J = J FI, e~ndt (3)
0
subject to A, =n(Lt) - k A t ,
where r is the discount rate. The total endowment of household or hired labour D is 
used either for agricultural production L* or land conversion L. Given an aggregate 
output price p, input price w, and cost function for the conversion of forestland c (L), 
aggregate profits II for households is given by
IT =pY -  wX - c(L) (4)
where cL>0 and cu <0 and the time subscript t is suppressed for ease of exposition.
By definition,
L* = D -  L (5)
37
Substituting (5) into (1), the production function of the household becomes 
Y= A .s (X, D -  L) (6)
By substituting (6) into (4), the aggregate profit function becomes 
n  = PA s ( X , D - L ) - w X - c ( L )  (7)
Therefore, the Hamiltonian for the solution o f the intertemporal maximisation 
problem stated in (3) is given by (8):
H = pAs (X, D -  L) -  wX -  c (L) + O [n (L) -  k A] (8)
where X and L are control variables, A is the state variable and <5 is the 
corresponding co-state variable representing the shadow value o f land in terms of 
additional agricultural profits.
The optimal path to steady-state equilibrium can be solved from the first order 
conditions which show how the prices o f other factors influence both the rate o f rural 
land use change (<?A/dt) and demand for land A. The relevant first order conditions 
from (8) yield
pA sx — w (9)
a T 7 1  =  “ “  ( 1 0 )0(w/p) Sx
An equation for the optimal level o f land use in each period A(t) is derived by 
differentiating (9) with respect to time, thus giving
n ( L ) + w _ ^ - X
P(s , )
, where k *  0 (11)
d A
< 0  o r - 0 o r > 0 i f X > 0  or = 0 o r < 0  (see Appendix I for the derivation
d(w/p) 
of equations 9 -12). Notably,
5A s~ .X  (12)
5(w/p ) ( s j 2
Equation (11) shows that the demand for land for agriculture depends not only 
on the rate o f land conversion n(L) and the proportion o f land fallowing k, but also on 
the level of agricultural returns (p /w ) ,  the rate o f marginal productivity of
^  3X  . „  x
purchased inputs — , and the rate o f input use over time ----- . Equation (12)
s x dt
d k
determines the impact of an increase in agricultural returns on land use ——•—r . If
j d { p /w)
input use is growing overtime, then agricultural land use will increase with rise in
agricultural returns and vice versa. Mathematically, if  —  > 0 then . > 0 , or
di 3(p/ w)
if — <0  then . < 0. Equation (11) cannot be empirically estimated but 
8t 9(p/w)
provides useful insights into how input use and prices influence the demand for land 
(Barbier, 2000).
The direct increase in the relative prices o f agricultural products and decrease 
in the relative price of agricultural inputs stimulate farmers to increase output 
resulting in increased demand for land to be converted to agriculture (Panayotou and 
Sungsuwan, 1994; Barbier, ibid.). An increase in the price o f agricultural food 
products on the other hand could lead to a temporary decline in the food purchasing 
power of agricultural wage (Abdulai, 2000). The indirect effects of agricultural on
39
area cleared for agriculture is captured through the increased profitability o f frontier 
farming which, inter alia, stimulates labour displacement through migration to the 
agricultural sector. This on-farm migration could lead to increased population 
pressure on land resulting in increased production, extensification of agriculture and 
extensive deforestation (Boserup, 1965; Rothenberg, 1980). On the other hand, an 
increase in availability o f off-farm employment and income will unambiguously 
reduce deforestation. Burger (2000) argues that population pressure can be met by 
enhanced migration out of the agricultural sector where there need not be more 
pressure on land. The analysis above suggests that land cleared for agriculture is a 
function o f input and output prices, population pressure and a labour migration 
variable. The cleared land equation is thus specified as
AC=A ( f . C f . P * ,  M an)  (13)
where Ac is the area cleared for agriculture or area deforested, f  is a vector of prices 
of agricultural inputs like fertiliser, insecticides, fungicides and so on, C f is a vector 
of prices of agricultural products such as cocoa and coffee, M *  is migration from 
agricultural sector to non-agricultural sector and PR is population pressure. Notably,
8 A jd Ip > Q ,dA jdOp > 0,dAc/dP R > 0 and BAc/dM AN < 0.
In formulating the supply of agricultural labour, some aspects o f the Harris 
and Todaro (1970) framework are incorporated where intersectoral labour mobility 
occurs when expected real income in non-agriculture (urban sector) exceeds that in 
agricultural sector (rural sector). The change in out-migration in response to short-run 
increases in the income differential depends on the unemployment rate in the non­
40
agriculture sector. Migration stops when wages are equal across sectors (Mundlak 
and Larson, 1997). Because of general excess supply of urban labour in Ghana, urban 
wages are assumed to be exogenous to rural labour markets whereas rural wages are 
assumed to be endogenous to rural-urban migration (Abdulai, 2000).
The model is therefore postulated by endogenizing labour inputs (Angelsen, 
1999). There is also interplay of the Malthusian effect which postulates that higher 
wages could lead to higher fertility and lower mortality rates. Natural growth rate of 
population due to high fertility or low mortality rates could stimulate migration. A 
dummy variable is included in the model to quantify the effect o f the Structural 
Adjustment Programme on labour mobility rates (Barbier, 2000). The labour mobility 
model is specified as
M an = m( WA ,PR, l f<EM S°) (14)
where M u is migration of labour from the agricultural sector to the non-agricultural 
sector, is the expected income in agriculture relative to that in the non-
agricultural sector, PR is population pressure, l f EM is rate of unemployment in non- 
agricultural sector and S°  is structural adjustment dummy ( S°  =1 for 1983 -  99 and 
zero otherwise)
Notably, 3MAN/ dWA < 0 ,8MAN /dPR > 0 , dMA / dUREM < 0, dMA/ d SD < 0.
41
3.2 . Statement of Hypotheses
The null hypothesis is represented by Ho  and the alternative hypothesis by HA. 
The hypotheses to be validated are:
I. H0 Rural population pressure exerts no effect on intersectoral labour
mobility, versus
Ha: Rural Population pressure exerts a positive effect on intersectoral
labour mobility (that is, from agriculture to non-agriculture).
II. Ho- Income in agriculture relative to that in non-agriculture exerts no
effect on intersectoral labour mobility, versus 
HA. Income in agriculture relative to that in non-agriculture exerts a 
negative effect on intersectoral labour mobility from agricultural to 
non-agricultural sector.
EH. Ho: Unemployment rate in non-agricultural sector exerts no effect on
intersectoral labour mobility, versus 
HA: Unemployment rate in non-agricultural sector exerts a negative effect
on intersectoral labour mobility from agriculture to non-agriculture.
IV. Ho'. Structural Adjustment Programme exerts no effect on the rate of
intersectoral labour mobility from agricultural sector to non- 
agricultural sector, versus 
Ha '. Structural Adjustment Programme exerts a negative effect on the rate
of intersectoral labour mobility from agricultural sector to non- 
agricultural sector.
42
V. Ho. Intersectoral labour mobility in favour o f the agricultural sector exerts
no effect on the rate of deforestation, versus
Ha . Intersectoral labour mobility in favour of the agricultural sector exerts 
a positive effect on the rate of deforestation.
VI H0'. Population pressure exerts no effect on the rate o f deforestation,
versus
Ha Population pressure exerts a positive effect on the rate o f deforestation.
VII. Ho'. Producer price of cocoa exerts no effect on the rate o f deforestation,
versus
Ha: Producer price o f cocoa exerts a positive effect on the rate of
deforestation.
V m  H 0'. Producer price o f maize exerts no effect on the rate o f deforestation,
versus
HA: Producer price of maize exerts a positive effect on the rate of
deforestation.
X. H0'. Price o f insecticides exerts no effect on the rate o f deforestation, versus 
Ha: Price o f insecticides exerts a negative effect on the rate of
deforestation.
3.3. Specification of the Empirical Econometric Model
The empirical simultaneous equation econometric model in two endogenous
variables with the relevant a priori parameter signs is given by (15) and (16)
43
M f  = a0+a lWIAN + a 2R™pdn +a iU«EM +aASAPl +e„ 
(-)  (+) (-)  (-)
(15)
D area =b0+ b,M fN + b2R p°PDN + k  P,C0C0A + KP,ma,ZE + bsP,,NSECT + e2, (16)
(-) (+) (+) (+) H
where M AN denotes labour mobility from the agricultural sector to the non- 
agricultural sector, W AN denotes income in agriculture relative to that in non­
agriculture, R  !'0PDN denotes population pressure in the rural areas, UREM denotes 
unemployment rate in the non-agricultural sector, D AREA denotes area deforested, 
p i n s e c t  (jgjQo^eg t]ie price index of insecticides, p COCOA denotes real producer price 
of cocoa, p MAJZE denotes real producer price of maize, SAP denotes structural 
adjustment dummy ( SAP= 1 for 1983 — 99, 0 = otherwise), t denotes current time
period and e denotes error term satisfying the classical normal regression assumptions 
(Judge et al, 1984).
The intercept terms capture factors affecting the rates o f migration and 
deforestation which changes slowly overtime (Cropper and Griffith, 1994). In the 
deforestation model, such fixed factors may include closeness o f forests to cities and 
rivers, the sizes and distribution of forests and other land uses (Cropper and Griffith, 
ibid.). Error terms are included to take account of possible omission o f variables from
44
the functions and errors in the measurement and aggregation of the variables. The 
following approach is employed in the estimation of the parameters o f the model.
3.4. Cointegration and Equilibrium Correction Modelling
Before any regression analysis on a time series is carried out, the order of 
integration of the variables has to be determined. This is because in practice, most 
economic time series tend to be generated by non-stationary stochastic processes 
(Tambi, 1999). Cointegration analysis begins with the determination o f the order of 
integration o f each o f the relevant time series (Baneijee et al, 1993; Alogoskoufis and 
Smith, 1995). This is essential if  spurious regression, which is normally associated 
with non-stationary series, is to be avoided (Granger and Newbold, 1974). A time 
series is stationary if  its mean, variance and autocovariances are independent of time 
(Engle and Granger, 1987). Any series that is not stationary is said to be non- 
stationary and its variance is such that any stochastic shock may not return it to its 
mean level.
The necessary conditions for variables to be cointegrated are that the variables 
exhibit similar statistical properties of being o f the same order o f integration and that 
linear regression involving the levels o f the series must be stationary. Many series are 
non-stationary in their levels but stationary in their first differences. This procedure of 
first differencing tends to reduce potential multicolinearity problems (Holden and 
Permon, 1994). A non-stationary series X, which can be transformed into a stationary 
series by differencing d times is said to be integrated o f order d\ and this is denoted
45
by x, ~  I  (d). The order of integration is the number o f unit roots contained in the 
series or the number of differencing operations required to make it stationary. A 
stationary series is said to be integrated of order zero, that is 1(0).
Many procedures exist for identifying the order of integration to ascertain 
whether a given time series has unit root (Campbell and Perron, 1991; Kwiatkowski 
et al., 1992; Silvapulle and Jaysuriya, 1994). The present study, however, employs 
the frequently used Augmented Dickey-Fuller (ADF) unit root tests to identify the 
order of integration of the economic variables. Given a time series Z„ the Dickey- 
Fuller procedure involves the estimation o f the regression 
AZ, = a  + yZ,_1 + U U (17)
where y = 5 -  l ,  a  is a constant drift, 8  is the coefficient to be tested and U„ is a 
stationary random disturbance term. The ADF test however ensures that a lag length k 
is chosen which makes U2l a white noise term:
AZ, = os + y ZM + Z  Pi AZ<-/ + u2, (18)
i=l
The Augmented Dickey-Fuller test statistic is the f-value for y = 8  - 1  in
equation (18) under a null hypothesis of H 0 : y = 0 implying non-stationarity
against H A : y  < 0 implying stationarity, I  (0). When y =  0 , implying non-rejection
of H0, a unit root test on AZt is run to verify whether Z, is integrated of order one.
When H0 is rejected for AZt then Z, is 1(0). The statistical significance is arrived at
by comparing the computed ADF test statistic to the appropriate critical value from 
Fuller (1976), Dickey-Fuller (1981), Engle and Granger (1987) and Phillip and
46
Perron’s (1988) non-parametric method for controlling higher order serial 
autocorrelation in the series. In recent times, the statistical table o f Mackirmon (1991) 
which permits the calculation of the Dickey-Fuller critical values for any sample size 
is frequently applied.
The null hypothesis of non-stationarity is rejected if  ADF statistic exceeds the 
Mackinnon critical value at the 5 percent level of significance. The asymptotic 
distribution o f the t-statistic under the null hypothesis depends on the assumption 
regarding the inclusion of deterministic terms like a trend or linear drift or both in the 
test regression. Although this complicates the testing procedure by increasing the 
number o f cases to consider, Hylleberg and Mizon (1989) have shown that the 
standard normal critical values are likely to lead to frequent rejection of the null 
hypothesis even in large samples, unless the constant term is very large. As observed 
by Hamilton (1994), the general principle is to choose a specification that plausibly 
describes the data under both the null and alternative hypotheses.
Since lagged dependent variables are used in the ADF specifications, it is 
important to note that the reported Durbin Watson statistic is inappropriate for 
identifying the presence o f serial autocorrelation in each series (Steward and Wallis, 
1981). The Breusch-Godfrey (BG) test which is a Lagrange Multiplier (LM) test is 
used to test for higher order autocorrelation in the model (Maddalla, 1992). The null 
hypothesis of no autocorrelation is tested against the alternative hypothesis o f a p//z 
order autocorrelation. The Breusch-Godfrey LM test statistic which is asymptotically
X  (P) distributed is computed by conducting the following regression using the
47
Ordinary Least Squares approach:
(19)
where the null hypothesis is identical to the hypothesis that O, = ®2 = • • • Op = 0 and
explanatory variables, (3j* denote estimated coefficients, e, are the residuals, n is the 
number o f observations and p  is the order of autocorrelation.
Once the stationarity properties of the individual series are established, linear 
combinations of the integrated series are tested for cointegration. Cointegration 
analysis purports to describe long run or equilibrium relationships (Tambi, 1999). An 
equilibrium relationship exists when variables in the model are cointegrated. Engle 
and Granger (1987) emphasise that if  there is a long run relationship between two 
non-stationary time series, then the cointegration test verifies whether deviations from 
the long run path are stationary.
Time series X, and Z, are said to be cointegrated of order (d, b) where 0<b<d, 
denoted by X, ,Z, ~CI(d,b), i f  X, ~I(d) and Z, ~I(d) and (X, fiZ, )~I(d-b). The two 
series cointegrated if  their cointegration regression as specified in equation (20) 
produces a residual series U, which is stationary:
Various tests for evaluating the cointegrating properties o f a pair o f non-stationary 
series exist (Hendry, 1986; Engle and Granger, 1987; Johansen, 1988 and Johansen
48
the test statistic of the BG-test is computed as nR2 ~  %2(p), X, and Z, denote the
X, =  a  + P  Z, + U, (20)
and Juselius, 1990). Johansen’s (1991) test procedure for example determines the 
number of cointegration relations or cointegration rank. In the Johansen framework, N  
endogenous variables each integrated of order one produce (N-l) linearly independent 
cointegrating equations. If  there are N  cointegrating equations, it means that none of 
these series is actually integrated. Specification errors might be responsible if  the test 
shows that the cointegrating rank is N. When such contradiction occurs, the relevant 
Vector Autocorrelation Regression (VAR) is reformulated in terms of the levels of all 
the series (Johansen and Juselius, 1990).
The Johansen framework also allows for consideration where cointegrating 
equations are known to have intercepts and deterministic trends. In the words of 
Ravallion (1986), the cointegration relation is useful because the short run and long 
run integration can be tested as nested hypothesis within a multivariate model. 
Boswijk and Franses (1992) emphasise that the results from VAR models are sensitive 
to the choice o f the lag-length. They therefore suggest the application o f the Johansen 
procedure for the determination of the different lag-lengths and to base the final choice 
of parameters on the absence of serial correlation in the residuals and the significance 
of parameters at higher lags using LR-tests.
Engle and Granger (1987) show that if  two series are cointegrated, then there 
exists an error correction representation. In other words, models with data series, 
which have long run equilibrium relationship but have significant short run 
divergences can be given an error correction representation. The error correction
49
mechanism captures the short run dynamics while making them consistent with long 
run dynamics (Engle and Granger, 1987; Holden and Pennon, 1994). Whilst various 
authors have applied the classical Engle-Granger approach in analysing the existence 
o f long-run and short-run relationships between economic variables, the present study 
employs the Dynamic Generalised Least-Square (DGLS) estimator o f Stock and 
Watson (1993). Recent studies have shown that the DGLS estimator provides 
efficient estimates of long-run parameters (Burger, Smit and Vogelvang, 2001). 
Saikkonen (1991) was the first to develop this method but he used the Dynamic 
Ordinary Least-Square (DOLS) estimation. Another advantage o f using the DGLS 
estimator is that its formulation eliminates and corrects for serial correlation in the 
model (Burger, Smit and Vogelvang, 2001). The general notation o f the model is 
specified as
where y ,  is the dependent variable, a  is the intercept, A Z , are lags and leads
short-run effects a n d / i ', vector of long-run parameters. Following the DGLS 
estimator proposed by Stock and Watson (1993), the simultaneous equation models 
explaining the long-run relationships between deforestation and intersectoral labour 
mobility are specified as
y , = a  + p ' Z ,  +  2 > , 'A Z ,_ ,+ S , (21)
specified in such a manner to eliminate residual autocorrelation f i '  is a vector of
P
(22)
i=0
(23)
where /?,„ and /320 are intercepts and Z„ is the vector [ /, WrAN R f0PDN u f EM) , Zt2
50
■ r  ,  n POPDN INSECT COCOA MAIZE  1  '  w ,  •is the vector I M  R P P  P  with /,=1 Vt  mL ' ' ' 1 1 J
equations (22) and (23) respectively. /?,' and /?' are the vectors o f long-run 
coefficients, /?.' and  /5'r. are vectors of short-run effects that are included to enhance 
the efficiency o f the estimator whereas sufficient lags of AZ(_;j and AZ,_i 2 are 
chosen to estimate equations efficiently (Burger, Smit and Vogelvang, 2001).
The short-run dynamics is considered in line with the error correction 
framework postulated by Engle and Granger (1987). For estimating the error 
correction models that converge to the equilibrium relationships in equations (22) and 
(23), the error correction terms are estimated as
n  a
e c tn = M At N- a , - Y a Z , jX (24)
j=l
a  n  a
e c ^ D T - K - i m ,  <25>
j=i
where aQ and b0 and a; and b- are estimated constant terms and vector of estimated
coefficients. Z,; j and Z;;2 are vectors of explanatory variables in equations (15) and
(16) respectively and n is number o f explanatory variables. The error correction 
models estimated be specified are specified as follows:
=So + Z  Sh/^Z t-i,i + + u4, (26)
/=0
ADfREi = 5, + Z  S'n AZ,_,2 + y2ectt_ia + u5l (27)
51
Where k is the chosen lag lengths, S0 and 5X are constant terms, 5'n and S'i2 are vectors 
of estimated coefficients, AZ,_n and AZ( i2 are vectors of explanatory variables, 
ectt_u  and ectM 2 are the error correction terms and yl and y2 are the coefficients of 
the error correction terms.
3.5. Description of Variables and Sources of Data
The data used in this study cover the period from 1970 to 1999. They are 
annual time series data on Ghana. In what follows, the definition and measurement of 
each variable in the present study are provided. The sources of all the data employed 
are also indicated.
Labour Mobility
Assuming that agriculture is the largest sector and the residual employer, 
labour mobility rate is measured as the magnitude of labour contributed by the 
agricultural sector to the rest of the economy. The Cownie (1974) structural 
transformation coefficient as quoted in Johnston and Kilby (1975) is employed in this 
study. This coefficient measures the absolute increase in the non-agriculture labour 
force per annum as
where m(t) is the labour mobility rate in period t, ( L, - L t ) is known as the Folke
Dovring coefficient of differential growth (Dovring, 1959), L,  is the growth rate of
(28)
N
52
labour force in the non-agricultural sector in period t and L t denotes growth rate of 
total labour force. Total labour force represents all persons aged 15 years and above 
engaged in or seeking employment in an economic activity in the agricultural sector 
or the non-agricultural sector. Labour force in agriculture refers to all economically 
active persons aged 15 years and above principally engaged in agriculture, forestry, 
hunting and fishing. The time series data on labour employed in agriculture and the 
whole economy are obtained from various issues of the UN Food and Agriculture 
Organisation (FAO) Production Yearbook and the Quarterly Digest o f  Statistics 
published by the Ghana Statistical Service.
Deforestation Rate
Various authors have relied on the availability of time series data on forest 
stock in the measurement of deforestation rate. Ehui and Hertel (1989) computed 
deforestation rate as the difference between the remaining forest stock for 
consecutive years. The present study however, discarded Ehui and Hertel’s (ibid.) 
method because complete set of reliable quality annual time series data on remaining 
forest were not available during the time of the study (personal communication with 
the relevant personnel at the Department o f Forestry at the Ministry o f Lands and 
Forestry in Kumasi). The deforestation rate used in this study represents deforestation 
through agricultural activities. The variable was computed by finding the difference 
between the area under agricultural cultivation and other industrial activities like 
logging and mining for consecutive years. Notably, FAO defines deforestation as the 
permanent conversion of forestland to other uses. Data on deforestation are obtained
53
from various issues of the FAO Production Yearbook, the data files o f the Ghana 
Forestry Department in Kumasi and the Ministry o f  Food and Agriculture.
Income
Total Gross Domestic Product (GDP) is used to proxy national income. Total 
GDP is made up of GDP in agriculture and GDP in non-agriculture. Non-agricultural 
GDP is the difference between total GDP and GDP in agriculture. These are used to 
proxy agricultural and non-agricultural incomes, respectively. The income variables 
used in the estimations are income per capita. Agricultural income per capita is the 
ratio of agricultural income to agricultural population and non-agricultural income per 
capita is the ratio o f non-agricultural income to non-agricultural population. The 
income in agriculture relative to that in non-agriculture was derived as the ratio of 
agricultural income per capita to non-agricultural income per capita. The agricultural 
population and non-agricultural population data are obtained from the various issues 
of FAO Production Yearbook. The time series data on gross domestic product (GDP) 
in constant 1985 prices originating in the agricultural and non-agricultural sectors are 
obtained from the Ghana Statistical Service and the various issues of the International 
Financial Statistics (IFS) published by the International Monetary Fund (IMF). Both 
GDP in current and constant 1985 prices are obtained. Dividing the nominal GDP 
(current) by the real GDP (constant) gave the GDP deflators (indexed to 1985=100).
Rural Population Density
Many writers, including Boserup (1965) have tended to equate population 
density with population pressure in rural areas. Gleave and White’s (1969) analysis
54
of how agricultural systems in West Africa varies with population density also arrived 
at similar conclusions as that of Boserup. This variable is then derived as:
Rural Population (millions) . . . .
Rural Population Pressure  ------------------------------------- (29)
Area cultivated {ha)
Notably, the deflator is area under cultivation in forest regions. However 
complete reliable time series data were not available so area under maize cultivation 
was used in the empirical analysis as a proxy since maize is one of the major staple 
foods cultivated extensively in Ghana annually by rural farmers. Data on rural and 
urban populations in Ghana are taken from the various issues o f the FAO Production 
Yearbook and International Financial Statistics (IFS) o f the IMF. Time series data on 
annual cultivated land for various crops are obtained from the various issues of the 
FAO Production Yearbook.
Rate o f  Unemployment in Non-agricultural Sector
The rate o f non-agricultural unemployment is defined as the ratio of the 
difference between total employment and agricultural employment to the difference 
between the economically active population and agricultural employment. This 
assumes that economically active population in agriculture is equal to the number 
employed in that sector. The sources o f the employment data are the ILO Yearbook o f  
Labour Statistics and the FAO Production Yearbook.
55
Price o f  Insecticides
Annual time series data on the price o f insecticides measured in 0 / litres of 
Gammalin 20 are employed. The relevant data are obtained from the Cocoa Services 
Division o f Ghana COCOBOD.
Real Producer Price o f  Cocoa
Real producer price o f cocoa is used in the present study. In view o f the high 
significant correlation coefficient of 0.92 between the prices o f cocoa and coffee (see 
Appendix II) and the coffee producer price is excluded from the model. Moreover in 
Ghana, farmers tend to clear more land for cocoa than coffee because o f its relative 
greater export potential. The real producer price o f cocoa is obtained by deflating the 
nominal price of cocoa to 1985 prices using the CPI. Annual time series data on 
producer price o f cocoa in i  I bag were obtained from various issues o f Quarterly 
Digest o f  Statistics (QDS) published by the Ghana Statistical Service and Ghana 
COCOBOD.
Real Producer Price o f  Maize
The real producer price of maize was similarly obtained by deflating the 
producer price o f maize to 1985 prices with the rural CPI. Data on nominal producer 
price o f maize were obtained from the Ministry o f Food and Agriculture, Ghana.
56
CHAPTER 4
EMPIRICAL ECONOMETRIC RESULTS
The empirical results of the study are presented in this chapter. First, results of 
the relevant stationarity tests are presented, followed by the results o f causality tests. 
The cointegration analyses are then performed. Finally, the results o f the error 
correction modelling exercise are presented.
4.1. Stationarity Tests
Having applied the Augmented Dickey-Fuller test to the variables under 
investigation, the empirical results are presented in Table 1. The tests reveal that
Table 1. Results of Unit Root Test.
SERIES LEVEL
ADF MCV LAG-
LENGTH (S)
DIFFERENCE 
ADF MCV LAG-
LENGTH (S)
CONCLUSION
M -3.5218 -3.5867 2 -4.5320 -3.5943 2 M ™  ~ I(1 )
j j AKEA -2.2342 -2.9750 1 -4.2993 -2.9798 1 D AREA ~l(\)
POPDN -2.4108 -3.5867 2 -4.7865 -3.5943 2 R P0PDN~l{\)
fflAN -3.2657 -3.5796 1 -4.1351 -3.5943 2 W M~ 1(1)
j j  REM -2.2075 -3.5943 3 -6.6206 -3.5867 1 U***-  i(l)
p  INSECT -0.7624 -3.5796 1 -4.3885 -3.5867 1 p  INSECT ^
pCOCOA -3.0828 -3.5796 2 -3.8110 -3.5943 2 P COCOA~ 1(1)
p  MAIZE -3.4228 -3.5867 2 -4.8609 -3.5943 2 p  MAIZE ^
Note: All variables are in their raw form. The ADF technique tests H0:Xt ~ 1(1) against H,:Xi ~ 1(0). MCV are the
asymptotic critical values or the Mackinnon critical values for rejection of hypothesis of a unit root and they 
are at the 5 percent level of significance. The ADF critical values are obtained from Mackinnon (1991). Each 
of the ADF equations includes a drift and a trend term.
M M denotes labour mobility from agricultural to non-agricultural sector, D denotes area deforested,
POPDN . ANR denotes population pressure in rural areas, W denotes income in agricultural sector relative to that
in non-agricultural sector, U denotes unemployment rate in non-agricultural sector, P >mECT denotes price
of insecticides, P denotes real producer price of cocoa and P MAIZB denotes real producer price of maize. 
Source: Author’s computations.
57
while the levels of each of the series are non-stationary, the first differences are 
stationary. All the series became stationary when deterministic trend and drift term 
were included. The hypothesis of unit root could not be rejected at the levels for all the 
series because as the results indicate, none of the ADF t-statistics for each of the series 
exceeds its asymptotic critical value at the levels. The hypothesis o f a unit root was 
however rejected at the 5 level of significance for all the series since their ADF t- 
statistics exceeded the relevant Mackinnon critical value (Mackinnon, 1991) after the 
first difference. The Breusch Godfrey Lagrange Multiplier test for second order 
autocorrelation shows the absence of serial correlation in the error term. The lag-length 
selected is the smallest without autocorrelation.
4.2. Granger Causality Tests
The existence of cointegration suggests Granger causality between the variables 
in one direction (Engle and Granger, 1987). The procedure for establishing causality as 
proposed by Granger (1969) involves regression of a variable on its lagged values and 
another variable. The general model for Granger causality test o f two non-stationary 
variables y and z after their first difference is represented as
Ay = Z A Ay,-, + 1 4 ^ - ,  +et (30)i=\ 1=1
where k is the number of lag-lengths. The null hypothesis H 0 : Aa = 0 versus 
alternative hypothesis H A : X;i *  0 is tested. If  the relevant computed F-statistic is 
greater than the critical F-value from statistical tables then the null hypothesis that z 
does not cause y is rejected in favour of the alternative hypothesis that z causes y, and
58
vice versa (see for instance, Pindyck and Rubinfeld, 1991). The results o f the causality 
tests are shown in Table 2.
Table 2. G ranger Causality Test Results.
Null Hypothesis (H0) Lag-length F- statistic p  - value
AN
M f * =£> W, 
W f  M f 2
1.25497 
6.59363 ***
0.30389
0.00545
M f  ^  U fM 
U?™ =t> M  f 6
2.33435 
16.1081 ***
0.10584
0.00007
M f  =f> R f PDN 
R f PDN M f 7
0.80471 
7.80916 ***
0.60629
0.00477
D f EA =f> M f  
M f  D f EA 6
2.32965 
2.83780 *
0.11370
0.07006
j j AREA £  POPDN 
POPDN j j AREA 3
2.23009 
3.24646 *
0.13126
0.05815
1 
i 
*
*
 
-
V
i
i 
ft
2
0.25215 
8.06767 ***
0.77934
0.00236
j-^AREA ^  pCOCOA  
pCOCOA j-y  AREA 4
2.66600 
7.11050 **
0.17957
0.03770
j j AREA pMAIZE  
pMAlZE  j j AREA 1
0.36517 
5.96879 **
0.55109
0.02197
Note: *** significant at the 1 percent level, ** significant at the 5 percent level, * significant at the 10 
percent level, means " does not Granger cause". All values are in their raw form (natural logarithms
were not taken)
M  denotes labour mobility from agricultural to non-agricultural sector, D AREA denotes area deforested,
POPDN , . . AN
R denotes population pressure in rural areas, W  denotes income in agricultural sector relative to that
in non-agricultural sector, U  denotes unemployment rate in non-agricultural sector, P /ra£CT denotes
price o f insecticides, P  denotes real producer price o f  cocoa and P  denotes real producer price o f  
maize.
Source: Author’s computations.
59
The test is conducted first to determine the causality between M AN and each 
of the variables WAN, jjrbm and R P0PDN using equation (24) and finally another test is 
conducted to examine the causal relationship between D AREA and the variables M  AN, 
R P0PDN, p w sE C T ' pcocoAm d  reSp ectiveiy. The results indicate that the
hypothesis that a change in either WAN , U REM ox r popdn does not cause a change in 
M  is strongly rejected in each case, while the hypothesis that M ANchanges do not 
cause changes in either WAN, \JREM or R P0PDN cannot be rejected at the 1 percent 
significance level respectively at lag-lengths indicated in Table 2. This empirical result 
also suggests that the differences o f WAN , U MM or R F0PDN are strongly exogenous at 
the given lags.
Similarly, the null hypothesis that a change in either M AN, R P0PDN, 
p  co c o a  or p m a ize  ^ oes not cauge a change jn d AR£A js rejected in favour o f the 
alternative hypothesis and that they do cause DAliEA. On the contrary, a change in 
D AREA does not cause a change in either M AN, R P0PDN, p msECT > p  c o c o a  Qr p  m a ize  at 
the 1 percent, 10 percent, 1 percent, 5 percent and 5 percent respectively at the chosen 
lag-lengths. These imply that the differences of M AN, R P0PDN, p mBCT; p C0C0A an(|  
p ,m ahe  at tjie gjven jagS aj.e strongly exogenous. These results further show that
WAN, U KEM and RP0PDN are Granger prior to M AN, and that M AN, R P0PDN, PmECT, 
pcocoA  ancj pmaize are Grariger prior t0 j j  AREA
60
4.3. Empirical Results of Long Run Relationship
The results from the Iterative Three Stage Least-Squares estimation o f the long- 
run relationship are shown in Table 3 (It should be noted that, the explanatory variables 
in the general models were specified at the beginning with three lag lengths. See
Table 3. Results of Long Run Relationships
Dependent Variable =M AN
Variable Coeff. Std. Error t-Stat. Prob.
WAN 0.0137 0.0073 1.8827* 0.0676
1^ POPDN 0.1274 0.0579 2.2016 ** 0.0340
j j  REM -0.0653 0.0225 -2.8975*** 0.0063
SAP -0.0073 0.0088 -0.8252 0.4146
Constant -0.0484 0.0308 -1.5715 0.1246
R2 =0.6995 
Adj. R2 =0.5729 
RSS =0.0052
Std Deviation = 0.0254
Mean = -0.0016
S.E. of Regression = 0.0166
BG (2) = 1.4721(0.4791)
Dependent Variable = D AREA
Variable Coeff. Std. Error t-Stat. Prob.
M an -26.2866 33.5285 -0.7840 0.4380
POPDN 19.5032 9.6098 2.0295 ** 0.0496
pCOCOA 0.0002 6.24E-05 2.5409 *** 0.0038
p  MAIZE 0.0012 0.0035 0.3505 0.7280
p  INSECT - 0 . 0 0 1 1 0.0004 -3.0840 *** 0.0038
Constant -11.9106 7.0664 -1.6855 0.1003
R2=0.6017 
Adj. R2 =0.4025 
RSS =349.2977 
BG (2)=1.7934 (0.1810)
Std Deviation = 5.6992
Mean = 4.0231
S.E. of Regression = 4.4052
Note: *** Significant at 1 percent, ** significant at 5 percent, * significant at 10 percent. The variables 
are in their raw form.
M  M  denotes labour mobility from agricultural to non-agricultural sector, D  '**”* denotes area deforested,
R f° ' denotes population pressure in rural areas, W A denotes income in agricultural sector relative to that
in non-agricultural sector, U "tM denotes unemployment rate in non-agricultural sector, P msECT denotes
price o f  insecticides, P  denotes real producer price o f  cocoa and P "A'ZE denotes real producer price o f 
maize, SAP  denotes structural adjustment programme.
Source: Author’s computations.
61
Appendix III for the detailed specifications and empirical results of the long run 
relationships). Convergence was achieved after 26 iterations. The estimated coefficient 
o f determination R2 of 0.699 indicates that about 70 percent o f the variation in 
intersectoral labour mobility from the agricultural to the non-agricultural sector are 
explained by the independent variables. Similarly, the R2 o f 0.60 for the deforestation 
equation implies that the independent variables explain about 60 percent of the 
variation in the area deforested. The DW-statistics cannot be used to determine if  serial 
correlation exist in the residuals because of the inclusion of lagged-dependent variables 
in the models. The Q-statistics from the correlogram o f the residual o f  the labour 
mobility model are not significant even at 10 percent. Further serial correlation test with 
the Breusch-Godfrey LM  test gave a BG (2) value o f 1.472 with a probability of 47.9 
percent which is not significant so we do not reject the null hypothesis of no 
autocorrelation in the model. Similarly, serial correlation test with the Breusch-Godfrey 
LM  test indicated a BG (2) value o f 1.472 with a probability o f 47.9 percent in the 
deforestation model and since this is not significantly different from zero, we do not 
reject the null hypothesis of no autocorrelation in the model.
In the labour mobility model, the estimated coefficient o f income in agriculture 
relative to that in non-agriculture does not exhibit the postulated sign but is 
significantly different from zero at the 10 percent level. The coefficients of rural 
population density and unemployment rate in non-agricultural sector have the expected 
signs and are significant at the 5 percent and 1 percent levels, respectively. The results 
suggest that in the long run, intersectoral labour mobility increases by 0.0137 units and 
declines by 0.07 units when the rural population density increases by lunit and the
62
unemployment rate in the non-agricultural sector increases by 1 unit, respectively, 
ceteris paribus. The positive coefficient of agricultural income relative to non- 
agricultural income runs counter to the Todaro (1969) hypothesis where an increase in 
the relative agricultural income is expected to encourage people to stay in the 
agricultural sector. The rejection of the Todaro hypothesis in the empirical model could 
be explained as follows: perhaps more recently, non-agricultural opportunities probably 
have come to dominate mobility decisions (Booth and Sundrum, 1984). The negative 
and insignificant coefficient (even at 10 percent level) o f the structural adjustment 
dummy reveals that its long-run effect on intersectoral labour mobility from agricultural 
to non-agricultural sector is minimal. Future research should examine these unexpected 
results in greater detail.
In the deforestation model, the coefficient o f the intersectoral labour mobility 
variable bears the postulated sign but it is not significantly different from zero even at 
the 10 percent level. The rural population pressure, producer price o f cocoa and 
insecticide price variables all carried the postulated signs and are significantly different 
from zero at the 5 percent, 1 percent and 1 percent levels, respectively. Although the 
producer price o f maize has the expected positive sign, it is not significantly different 
from zero even at the 10 percent level. The empirical negative coefficient of 
intersectoral labour mobility from agricultural sector to non-agricultural sector seems to 
support what other studies (Jaegar, 1992; Abdulai, 1999) have observed. However, it is 
not statistically significant even at the 10 percent level. Substantial evidence in the 
relevant literature suggests that major intersectoral flow o f workers since 1983 has been 
towards the agricultural sector especially due to the production of cocoa as targeted by
63
the Economic Recovery Programme (Fosu, 1989; Beaudry and Sowa, 1994). A unit 
increase in rural population density leads to a 19.5 unit increase in the area deforested 
through agricultural expansion in the long run. This empirical result confirms the 
theoretical underpinnings of the neo-Malthusian belief about the effect of growing 
human populations on natural resources (Sunderlin and Resosudarmo, 1999). With the 
increasing population pressure on land, as the present study indicates, the natural 
tendency is for people to clear more land as a result o f food scarcity which indirectly 
leads to environmental problems (Brown, 1995).
The positive cointegration coefficient of 0.0002 o f producer price o f cocoa 
indicates that in the long run, a unit increase in the producer price o f cocoa results in 
about 0.0002 unit increase in the area deforested. Although this change is small, the 
empirical finding is not surprising because between 1983 and 1987, the prices received 
by cocoa farmers have increased about sevenfold (Abdulai and Rieder, 1995). This 
profit margin o f farmers might have encouraged them to increase output even though 
pesticides and fertiliser subsidies were curtailed during the stabilisation period of 
Ghana’s ERP. The estimated coefficient of price o f insecticides has the postulated 
negative sign and is significantly different from zero at the 1 percent level. This agrees 
with the empirical studies by Ruben et al., (1994) and Monela (1995) in South Africa 
that reduced prices o f inputs such as seeds, insecticides and hand tools increase forest 
clearing in the long run.
64
4.4. Short Run Effects
In this section, the analysis of the results of the short-run relationship between 
intersectoral labour mobility and its determinants and short-run relationship between 
area deforested and its determinants are undertaken. The short-run effects are captured 
by the estimated coefficients. The models which are estimated by applying Hendry’s 
general to specific modelling start from a general model from economic theory and are 
then narrowed down to an acceptable specification, given the data (Hendry et al., 1984).
Table 4 presents results of the short-run relationships after application of 
Iterative Three Stage-Least Squares to the simultaneous equations (26) and (27). A (■)
denotes the first difference of the respective variable at the current time t and it would 
be noted from Table 4 that lagged first differences at the current time are reported (The 
independent variables in the general model of the short run relationships were specified 
with three lag lengths before arriving at the results presented in Table 4. See Appendix 
HI for detailed information on the final specifications and results from the estimation of 
the short run relationships). In the short-run relationship, convergence was achieved 
after 36 iterations.
2The R o f 0.56 in the labour mobility model implies that 56 percent o f the
variation in the intersectoral labour mobility is explained by its independent variables
2whereas the R of 0.78 in the deforestation model indicates that about 78 percent o f the
65
Table 4. Results of Short Run Relationships
Dependent Variable =AMan
Variable Coeff. Std. Error t-Stat. Prob.
a W an 0.0021 0.0129 0.1659 0.8693
a R P0PDN 0.1136 0.5066 0.2242 0.8242
A U mM -0.2373 0.0489 -4.8542 *** 0.0000
ASAP 0.0073 0.0249 0.2948 0.7702
ectt-i, i -1.1051 0.3718 -2.9719 *** 0.0059
Constant 0.0106 0.0103 1.0320 0.3106
R2 = 0.5641 Std Deviation = 0.0388
Adj. R2 = 0.3945 Mean = 0.0002
RSS=0.0164 S.E. of Regression = 0.0302
BG (2) =1.7194(0.1898)
Dependent Variable =A D  AREA
Variable Coeff. Std. Error t-Stat. Prob.
a m an -28.4015 14.2044 -1.9995 * 0.0550
a R P0PDN -33.0634 85.9614 -0.3846 0.7033
-pCOCOA -0.0003 0.0001 2.1738 ** 0.0380
p  MAIZE 0.0005 0.0019 0.2674 0.7911
^ p  INSECT -0.0007 0.0004 -1.9875 * 0.0564
o -0.8303 0.2445 -3.3958 *** 0.0020
Constant 2.2429 1.4445 1.5528 0.1313
R2= 0.7763 Std Deviation = 3.7097
Adj. R2 =0.4915 Mean = 0.2215
RSS=76.9783 S.E. of Regression = 2.6454
BG (2) =1.6987 (0.4267)
Note: *** significant at 1 percent level, ** significant at 5 percent level, * significant at 10 percent level. 
All values are in their raw form.
A M M denotes difference of labour mobility from agricultural to non-agricultural sector, A  D ASEA denotes 
difference of area deforested, A P denotes difference of population pressure in rural areas, A W M denotes
difference of income in agricultural sector relative to that in non-agricultural sector, Al/"™  denotes difference 
of unemployment rate in non-agricultural sector, A P ms 01 denotes difference of price of insecticides,
A P C0C°A denotes difference of real producer price of cocoa, A P MMZe notes difference of real producer price of 
maize, ASAP denotes difference of structural adjustment programme and ect denotes error correction term. 
Source: Author’s computations.
variation in the dependent variable is explained by the independent variables. Because
of the inclusion of lagged-dependent variables in the model, the DW-statistics cannot
be used to determine if serial correlation exist in the residual. Serial correlation test
with the Breusch-Godfrey LM  test shows that the BG (2) value o f 1.7194 with a p-value
66
of 0.1898 is not significant so the null hypothesis of no autocorrelation cannot be 
rejected in the labour mobility model. Similarly, the Breusch-Godfrey LM  test for the 
deforestation model gave a BG (2) value of 1.6987 with a p-value o f 0.4267 which is 
also not significantly different from zero so we do not reject the null hypothesis of no 
autocorrelation in the model.
The empirical results show that the income in agriculture relative to that in non­
agriculture again has the wrong positive sign in the short-run but unlike the long-run 
relationships, it is statistically not significant. The rural population density variable in 
the intersectoral labour mobility model has the expected positive sign but in this case 
failed to be significantly different form zero in the short-run. The coefficient of 
unemployment rate in the non-agricultural sector variable has the hypothesised negative 
sign and is statistically significant at the 1 percent level. This empirical finding lends 
support to the view that as the unemployment rate in a sector increases, labour 
migration to that sector decreases (Collier, 1979; Bencivenga and Rosenweig, 1997). 
Similar to the long run effects, the Structural Adjustment dummy is statistically 
insignificant. Surprisingly, this quantitative empirical result seems to contradict the 
effects of structural adjustment in some analytical studies on Ghana (Jaegar, 1992; 
Abdulai, 1999). One of the reason for the reverse labour migration to the agricultural 
sector during the structural adjustment period then may be attributed to the expansion 
of the cocoa production zones as asserted by Beaudry and Sowa (1994).
In terms of postulated signs, the short-run dynamics did not differ very much 
from the long run dynamics in the deforestation model. However, the intersectoral
67
labour mobility variable has the hypothesised negative sign and it is significantly 
different from zero at the 10 percent level in the short-run. The price o f maize and 
producer prices of cocoa again carried the correct signs in the short-run model but only 
the producer price of cocoa is statistically significant at the 5 percent level as it 
happened in the long-run estimations.
The inclusion of inputs in deforestation models highlights the fundamental 
difference that has existed between population and market approaches in explaining 
their effects on deforestation (Angelsen, 1999). The population approach contends that 
lower input prices such as price of insecticides will reduce deforestation, as insecticides 
are substituted for land in agricultural extensification. Under the market approach 
however, lower input prices will augment the profitability o f frontier farming and 
therefore increase deforestation under intensification. The negative coefficient of
0.0007 o f insecticide prices at the 10 percent significant level implies that in the short- 
run, a proportionate decline in the price o f insecticides leads to 0.0007 percent increase 
in the area deforested. Although this effect is minimal because o f the small coefficient, 
the latter thesis seems to give credence to the present empirical finding that reduced 
price o f agricultural inputs encourages area expanded for agriculture. The coefficients 
of the error correction terms have the expected negative signs. Though the sizes o f the 
coefficients of the error correction terms in both the intersectoral labour mobility and 
deforestation models are comparatively small, they are statistically significant at the 1 
percent respectively, indicating that last periods equilibrium error of intersectoral labour 
mobility from agricultural to non agricultural sectors has a significant impact on 
subsequent changes in the amount of area deforested through agricultural expansion.
68
CHAPTER 5
SUMM ARY  AND  RECOMM ENDAT ION S  
5.1 Summary
This study has analysed the long ran and short-run relationships between
intersectoral labour mobility and area deforested in Ghana over the period 1970—99.
The Dynamic Generalised Least Squares Estimator (DGLS) is used to specify a 
system o f simultaneous equations involving labour mobility and its determinants and 
area deforested and its determinants in the framework o f supply o f agricultural labour 
and demand for land for agricultural expansion. Having checked the identifiability 
status of the two equations and found them to be overidentified, the Iterative Three 
Stage Least Squares approach is used to estimate the model. With the presence of 
cointegration established, the relevant Error Correction Models which incorporate the 
long-run equilibrium and short-run dynamics are estimated.
The results of the study show that rural population pressure and the rate of 
unemployment in the non-agricultural sector exert a strong positive and negative 
effects respectively, on intersectoral labour mobility from agricultural to non- 
agricultural sector in the long run. These results seem plausible since policies which 
tend to increase rural income would encourage more people to stay in the rural areas 
but in the long-run, if  rural population density continues to increase, then this benefit
69
would be neutralised. On the other hand, policies which suppress urban incomes and 
lead to high unemployment rate where the bulk o f non-agricultural work is 
concentrated in Ghana would discourage labour from staying in that sector, as it is 
believed to have happened in the late 1980s and early 1990s.
In addition, the study observes that intersectoral labour mobility from the 
agricultural sector to the non-agricultural sector has had a significant effect on the 
area deforested through agricultural expansion in the short-run. The empirical result 
is consistent with appropriate macroeconomic and sector-specific policies. Exodus of 
labour to the agricultural sector due to non-existence of off-farm job opportunities 
would encourage people to clear more land for subsistence and survival which would 
inadvertently lead to adverse environmental effects. Finally, area deforested tends to 
increase when the real producer prices o f cocoa increases. The results also reveal that 
lower price o f insecticides could increase the area cleared for agricultural expansion.
5.2 Policy Recommendations
The study suggests that when unemployment rate in the non-agricultural 
sector increases, the intersectoral labour mobility from the agricultural sector to the 
non-agricultural sector slows down. Therefore rural development policies such as 
establishment of cottage industry and small-scale industries, which encourage the 
creation of jobs in the rural sector, must be pursued. Smale-scale and affordable 
irrigation projects must be given a further boost in the drier parts o f the country in 
order to absorb the surplus agricultural labour from the rural sector. Moreover, rural
70
population pressure could be reduced if policies that remove the bottlenecks in the 
rural and urban markets are initiated so that there would not be more pressure on 
land. The empirical results also suggest that lower input prices of insecticides will 
augment the profitability of frontier farming in line with the market approach 
framework. The implication of this negative relationship between agricultural 
deforestation and price of insecticides requires that land conversion for agriculture 
must be pursued in a manner, which encourages agricultural intensification.
The study further reveals that increased price of cocoa leads to a significant 
increase in the area deforested due to extensification. To prevent deforestation 
through agricultural activities, there is the need to promote productivity increasing 
technologies in the cocoa sector, which would slow down the increasing pressure on 
land by fanners. This policy initiative should be applied to maize production as well 
since it is one of the staple food crops mostly grown by farmers in Ghana.
5.3. Limitations of the Study and Suggestions for Future Research
The causality between unemployment rate in the non-agricultural sector and 
population pressure was not investigated and since population pressure is assumed to 
be exogenous in the study, it is recommended that future studies examine this 
linkage.
Due to data construction problems, the study looked only at area deforested 
through agricultural expansion. It would be useful if future studies examine
71
deforestation as a whole, to include the activities of log-exporting timber firms, road 
construction, mining and other industrial activities.
The present study used the Granger causality tests to find out whether the 
empirical model is characterised by strong exogeneity in the regressors. The literature 
(Johnston and Dinardo, 1997; Judge et. al, 1988) suggest further that it is possible for 
the model to be characterised by weak and/or super exogeneity. Therefore, future 
studies should address the issue of weak and super exogeneity in dealing with themes 
similar to the present study.
The overall performance of the model suggests that Ghanaian secondary time 
series data despite its measurement errors could be useful in the analysis of labour 
migration and deforestation rates provided adequate specification and estimation 
procedure are used. However, it is recommended that future investigations should go 
beyond this and utilise micro surveys of rural employment and demographic 
characteristics of agricultural labour in particular to perform such an exercise.
72
INTERTEMPORAL MAXIMISATION PROBLEM
The Hamiltonian for the intertemporal maximisation problem is given by 
H = pAs (X, D -  L) -  wX -  c (L) + <D [n (L) -  k A] (A. 1)
The first order conditions are as follows:
3H
—  = pA sx -  w = 0 => w = pA sx (A .2)
—  = -p A sL - c L + O n L =0  =><E>nL = cL +pA sL (A.3)
g j-j
—  = p s (x ,D -L )  +  (D(-k) =  - 0 ^ > p s ( x ,D -L ) -® k  =  -<i) (A.4)
APPENDIX I
—  =  n ( L ) - k A  =  A  
8 0  K ’
d A  kw
 =  ^ — > 0
dp  p s x
(A.5)
From (A.2)
A = w /psx (A. 6)
Substituting (A.6) into (A.5) gives 
I / \ kw
A  =  n(L)  (A. 7)
PSx
8 A  k
■ -= - = --------- < 0  (A. 8)
ow  p sx
(A.9)
73
d  A  k  
5 (w /p )  s x < 0
(A. 10)
The optimal use of land in each period A (t) is derived by differentiating (A.6) with 
respect to time:
un- w /  ,V  \-2 <7A
3t p at
From (A.5) and (A. 11)
at P (sx ):
2 XX (A. 11)
/  \  w S„Y •
n (L ) -k A  = ------- J ^ - x
P (s x )
/ \ w SYY • 
kA = n(L) + — —^ - X
P (sx)
d(l )  ■
w S,
P (Sx)'
-X (A. 12)
aA J XX
a (w /P)  ( s x ) :
■X<0 o r = 0  or> Oif X > 0 or = 0 or < 0 (A13)
74
APPENDIX II
CORRELATION MATRIX
d area M p COCOA p  COFFEE pFERT
^  AREA 1.0000 -0.0168 0.5555 0.7875 0.6293
M AN M -0.0169 1.0000 -0.0227 0.0123 0.0063
p COCOA 0.5555 -0.0227 1.0000 0.9199 0.9873
p  COFFEE 0.7874 0.0123 0.9199 1.0000 0.9281
pFERT 0.6293 0.0062 0.9873 0.9281 1.0000
p  INSECT 0.3445 -0.0872 0.9422 0.8286 0.8841
p  MAIZE -0.5013 0.0696 -0.6761 -0.6750 -0.6906
SAP 0.3898 -0.1693 0.4196 0.4320 0.4186
-jjREM 0.7001 -0.2121 0.6900 0.7510 0.7210
-0.3492 0.1889 -0.4549 -0.4460 -0.4387
p  INSECT p  MAIZE SAP |jREM W AN
0.3445 -0.5012 0.3898 0.7001 -0.3492
-0.0872 0.0696 -0.1690 -0.2122 0.1889
0.9422 -0.6760 0.4196 0.6900 -0.4549
0.8286 -0.6754 0.4320 0.7510 -0.4460
0.8841 -0.6906 0.4186 0.7210 -0.4387
1.0000 -0.5884 0.3504 0.5943 -0.3671
-0.5884 1.0000 -0.3450 -0.5873 0.4855
0.3504 -0.3453 1.0000 0.4502 -0.3707
0.5943 -0.5872 0.4502 1.0000 -0.0639
-0.3671 0.4855 -0.3710 -0.0639 1.0000
75
APPENDIX III
LONG RUN RELATIONSHIPS
1. System of Simultaneous Equations
Migration Equation
Observations
M
28
AN . -C(l)+C(2) * WAN +C(3)* Rpopdn +C(4)* Um 
+C(5)*SAP +C(6)* D (UREM)
+C(7) * D { U (-1)) +C(8)* D{Rpopdn (-1)) 
+C(9)* D(Wan( -  1)) (1)
Deforestation Equation DAREA=C(J0j+C(ll)* M AN +C(12)* Rpopdn
+C(13)* PC0C0A +C(14)* PMAJZE +C(15)* PmSECT
(-1))+C(16)* 1)) +C(17)* d (PC0C0A'
+C(18)* D{Pma,ze{-\)) 
+C(19)* D(PmECT) (2)
Observations
Instruments
28
POPDNC  WAN U SAP D(UREM) D(U (-1)) D(R
D (W AN ( - I J )  p C0C0A pMAIZE plNSECT
Pcocoa d {Man{-  1)) D(Pcocoa(-\))  D(Pmajze(-1))
p^ pINSECT ^
(-i))
Estimation Method 
Sample
Iterative Three-Stage Least Squares 
1970-1999
76
Continuation of Appendix III
2. Empirical Results of Long-run Relationships
Determinant residual covariance 0.001308
Convergence 26 iterations
Dependent Variable
XII
Variable Coefficient Std. Error t-Statistic Prob.
Constant
WAN
-0.048374
0.013745
0.030781
0.007301
-1.571516
1.882650
0.1246
0.0676
gPOPDN 0.127367 0.057851 2.201647 0.0340
yREM -0.065288 0.022533 -2.897450 0.0063
SAP -0.007271 0.008812 -0.825193 0.4146
d (u rem ) -0.174929 0.026525 -6.595005 0.0000
D{UREM(r  1)) 0.046329 0.024170 1.916794 0.0630
D{Rw pdn ( -  1)) -0.454806 0.263071 -1.728833 0.0922
D(IVAN(-1)) 0.002725 0.006263 0.435038 0.6661
R-squared = 0.699464 
Adjusted R-squared = 0.572922 
S.E. o f regression = 0.016607 
Durbin-Watson stat = 2.047442
Mean dependent var. = - 0.001634 
S.D. dependent var. = 0.025411 
Sum squared resid. = 0.005240 
BG (2) = 1.4721 (0.4791)
Dependent Variable _j-)AREA
Variable Coefficient Std. Error t-Statistic Prob.
Constant
M an
-11.91066
-26.28664
7.066402
33.52852
-1.685534
-0.784008
0.1003
0.4380
pPOPDN 19.50320 9.609848 2.029501 0.0496
pCOCOA 0,000159 6.24E-05 2.540919 0.0154
pMAlZE 0.001235 0.003523 0.350469 0.7280
pINSECT -0.001114 0.000361 -3.084027 0.0038
D{Man{ -\)) -10.53657 16.65397 -0.632676 0.5308
D (pCOCOA (—J)) 0.000151 0.000106 1.422699 0.1632
d (p mmze( - i)) -0.005110 0.003035 -1.683362 0.1007
^pINSECT  ^ 0.000324 0.000212 1.530292 0.1345
Rsquared = 0.601699 
Adjusted R-squared = 0.402548 
S.E. o f regression = 4.405159 
Durbin-Watson stat =  1.200074
Mean dependent var. = 4.023056 
S.D. dependent var. = 5.699152 
Sum squared resid. = 349.2977 
BG(2) =1.7934 (0.1810)
77
SHORT RUN RELATIONSHIPS
1. System of Simultaneous Equations
Migration Equation D(MAN)=C(1) + C(2)* D{WAN)
+C(3)* D(Rp o p d n ) +C(4)* D(U■R a t )
+C(5)* D(SAP) +C(6)* D(WAN( - 1))
+C(7)* D(Rpopdn ( - 1))
+C(8)*ectx{ -1) (1)
Observations 26
Deforestation Equation D(DAREA)=C(9) +C(10)* D{MAN) +C(11)* D(Rpopdn)
+C(12)* d (p cocoa ) +C(13)* d (pMA]ZE) 
+C(14)* D(PmBCT) +C(15) * d {Man( - l ) )  
+C(16)* D(Rpopdn ( - 1 ) )  +C(17) *  D(PmECT( - 1 ) )  
+C(18J* d (Pmajze ( - 1 ) )  +C(19) * ect2 ( - 1 )  
+C(20)* D{Darea(-\)) +C(21)* D (Man(-2)) 
+C(22)* D(_DAREA (-2))
+C(23)* D(Pcocoa{-2)) (2)
26
C  D{Wan) D (RP0PDN) D(Urem) D(SAP) D(W an (~\)) 
D(RP0PDN (—i)) ect] (-1) £>(PC0C0a) d (PMA,ZE) D (PmSECT) 
D(MAN(-\) )  D(DABEA(-\)) D (DAREA( -2)) D(MAN(-2)) 
D(Pcocoa( -2)) ect2{ -1)
Iterative Three-Stage Least Squares 
1970-1999
APPENDIX IV
Observations
Instruments
Estimation Method 
Sample
78
Continuation of Appendix IV
2. Empirical Results of Short Run Relationships
Convergence 36 iterations
Determinant residual covariance 0.001677
Dependent Variable = D (M AN)
Variable Coefficient Std. Error t-Statistic Prob.
Constant 0.010630 0.010300 1.032005 0.3106
D(WAN) 0.002138 0.012884 0.165975 0.8693
D(RP0Pm) 0.113592 0.506645 0.224204 0.8242
D(UR m ) -0.237339 0.048894 -4.854172 0.0000
D(SAP) 0.007343 0.024905 0.294840 0.7702
D(WAN (-1)) 0.011464 0.012865 0.891101 0.3802
D(Rpopdn(_  1)) -0.664088 0.494022 -1.344249 0.1893
ectr( - 1) -1.105070 0.371836 -2.971930 0.0059
R-squared = 0.564060 Mean dependent var. = 0.00016
Adjusted R-squared = 0.394528 S.D. dependent var. = 0.03882
S.E. o f regression = 0.030207 Sum squared resid. = 0.01643
Durban-Watson stat. = 2.138634 BG (2) = 1.7194 (0.1898)
Dependent Variable = D{DAREA)
Variable Coefficient Std. Error t-Statistic Prob.
Constant 2.242947 1.444459 1.552794 0.1313
D (M AN) -28.40153 14.20435 -1.999495 0.0550
D(RP0PDN) -33.06335 85.96141 -0.384630 0.7033
pCOCOA^ 0.000275 0.000127 2.173789 0.0380
D ^pM A IZE  ) 0.000511 0.001911 0.267352 0.7911
D (pm*cT  ^ -0.000722 0.000363 -1.987485 0.0564
43.37855 14.83109 2.924840 0.0066
D(RP0PDN (-1)) -138.6542 89.33927 -1.551996 0.1315
D(PmSECT (-1)) -0.000119 0.000251 -0.473725 0.6392
D(PMAJZE (-1)) 0.000853 0.002054 0.415437 0.6809
ec t2( - 1) -0.830368 0.244532 -3.395752 0.0020
D(DMEA (-1)) 0.665320 0.276324 2.407757 0.0226
D (M AN (-2)) 53.86831 12.33478 4.367190 0.0001
D(DAREA (-2)) 0.028389 0.159706 0.177756 0.8602
D(PC0C0A( -2)) -0.000181 9.52E-05 -1.907088 0.0665
R-squared = 0.776258 Mean dependent var. = 0.22153
Adjusted R-squared = 0.491496 S.D. dependent var. = 3.70972
S.E. of regression = 2.645379 Sum squared resid. = 76.9783
Durban-Watson stat. = 2.012534 BG (2) = 1.6987 (0.4267)
79
LIST OF REFERENCES
Abdulai, A. (1999). International Migration and Agricultural Development in Ghana, 
Scandinavian Journal o f  Development Alternatives and Area Studies, Vol. 18, No. 
1, March.
Abdulai, A. and C.L. Delegado (2000). An Empirical Investigation of Short-run and 
Long-run Agricultural Wage Formation in Ghana, Oxford Development Studies, 
Vol. 28, No.2: 169-185.
Abdulai, A. and P. Rieder (1995). Agriculture’s Contribution to Overall Economic 
Growth: An Application of Kaldor’s First Growth Law to Ghana, African 
Development Review.
Africa Development Bank (ADB), The Economic Commission for Africa (ECA), 
Organisation o f African Unity (OAU) (1984). "The African Food Crisis and the 
Role of African Development Bank in Tackling the Problem". A Paper presented 
jointly by the staff o f ADB, ECA and OAU to the Symposium on the Future of 
ADB and the Food Crisis of Africa, Tunis, Tunisia (May).
Alogoskokoufis, G. and R. Smith (1995). On Error Correction Models: Specifications, 
Interpretations, Estimations, In: L. Oxley, D.A.R. George, C.J. Roberts and S. 
Sayer (eds.) Survey in Econometrics. Oxford, U.K: Basil Blackwell.
Amanor, K.S. (1994). "Political Economy in the Frontier o f Ghana’s Forest Zone". In The 
Frontier, Farmers Response to Land Degradation: A West African Study. 
UNRISD (1994), London, U.K: Zed Books Ltd.
Amelung, T. and M. Diel (1992). Deforestation of Tropical Rainforests: Economic 
Causes and Impact on Development. Tungingen: Mohr.
Ampadu-Agyei, O., G.A. Manful, J.E. Fleischer, M. Opam, and A.M. Ameyaw (1994). 
Climate and Africa: Conference on Policy Options and Responses to Climate 
Change, Nairobi, Kenya, December 5-8, 1994.
Anderson, A. (1990). Alternatives to Deforestation, Columbia University Press.
Angelsen A. (1997). "State-Local Community Games of Forest Land Appropriation". 
Working Paper 1997, No. 7, Chr. Michelsen Institute, Bergen, Norway. 
Processed.
Angelsen A. (1999). Agricultural Expansion and Deforestation: Modelling the Impact of 
Population, Market Forces and Property Rights. Journal o f  Development 
Economics, Vol. 58: 185-218.
Angelsen A. and D. Kaimowitz (1999). Rethinking the Causes o f Deforestation: Lessons 
from Economic Models. The World Bank Research Observer, Vol. 14, No. 1, 
(February): 73-98.
Antolin, P. and Bover, O. (1993). Regional Migration in Spain: The Effect of Personal 
Characteristics and o f Unemployment, Wage and House Price Differentials Using 
Pooled Cross-Section, Documento de Trabajo No. 9318, Servicio de Estudios, 
Banco de Espana.
Bailey, A.J. (1993). Migration History, Migration Behaviour and Selectivity, The Annals 
o f  Regional Science, Vol. 27: 315-26.
Balogh, T. (1962). Misconceived Educational Programmes in Africa, Universities 
Quarterly.
Baneijee, A.D., Rayner, A.J. and D. Greenway (1993). Cointegrating, Error Correction 
and Econometric Analysis o f Non-stationary Data. New York: Oxford University 
Press.
81
Baneijee, B. and S.M.R, Kanbur (1981). On the Specification and Estimation o f Macro 
Rural-Urban Migration Functions: With an Application to India Data, Oxford 
Bulletin o f  Economics and Statistics, Vol. 43, No. 1: 7-29.
Baneijee, B. and G.A. Bucci (1994). On the Job-Search after entering Urban 
Unemployment: An Analysis based on Indian Migrants, Oxford Bulletin o f  
Economics and Statistics, Vol. 13, No. 2: 17-32.
Barbier, E.B. (2000), Links Between Economic Liberalisation and Rural Resource 
Degradation in the Developing Regions, Agricultural Economics, Vol. 23: 299- 
310.
Barbier, E.B. and J.C. Burgess (1996). Economic Analysis o f Deforestation in Mexico, 
Environment and Development Economics, Vol. 1, No. 2: 203-39.
Bamum, H.N. and R. H. Sabot (1977). Education, Employment Probabilities and Rural- 
Urban Migration in Tanzania, Oxford Bulletin o f  Economic and Statistics, Vol. 
39, No, 2: 109-26.
Beals, B.E. and C. F. Menezes (1970). Migrant Labour and Agricultural Output in Ghana. 
Oxford Economic Papers, Vol. 22, No.l: 109-127.
Beals, B.E., Levy, M.B. and L.N. Moses (1967). Rationality and Migration in Ghana, 
Review o f  Economics and Statistics, Vol. XLIX (November): 480-486.
Beaudry, P. and N.K. Sowa (1994). Ghana: Labour Markets in an Era o f Adjustment, In, 
Horton, S., R. Kanbur and D. Mazumdar (eds.): Vol. 2, Washington, D.C.: EDI, 
The World Bank.
Bencivenga, V.R. and M. Rosenweig (1981). Contractual Arrangements, Employment 
and Wages in Rural Labour Markets in Asia. New Haven: Yale University Press, 
1982.
82
Bentolila, S. and J.J. Dolado (1991). Mismatch and Internal Migration in Spain, 1982-86, 
Padoa-Schioppa: 182-234.
Besley, T. (1993). Property Rights and Investment Incentives: Theory and Microevidence 
from Ghana, Quarterly Journal o f  Economics 105(5), 903-37.
Bhattacharya, P.C. (1993). Rural-Urban Migration in Economic Development, Journal o f  
Economic Surveys, Vol. 7, No.3: 243-81.
Bluffstone, R.A. (1995). The Effect o f Labour Performance on Deforestation in 
Developing Countries Under Open Access: An Example from Rural Nepal, 
Journal o f  Environmental Economics and Management, Vol. 29:42-63.
Boateng, K. (1992). Quarterly Jobs or Mass Employment. African Research Economic 
Consortium (AERC), Research Paper 98 (October), Nairobi, Kenya.
Booth, Anne and R.M. Sundrum (1984). Labour Absorption in Agriculture: Theoretical 
Analysis and Empirical Investigations (Oxford University Press, Delhi).
Boijas, G J. (1987). Self-Selection and the Earnings of Immigrants, American Economic 
Review, Vol. 74, No.4: 531-326.
Boserup, E. (1965). The Conditions o f  Agricultural Growth. London; George Allen and 
Unwin, Chicago: Aldine Publishing Company.
Boswijk, H.P. and P.PI. Franses (1992). Dynamic Specification and Cointegration, Oxford 
Bulletin o f  Economics and Statistics, Vol. 54, No. 3: 369-81.
Breman, J.C. (1985/ O f Peasants, Migrants and Paupers: Rural Labour Circulation and 
Capitalist Production in West India, Oxford: Oxford University Press.
Breman, J.C. (1985/ Labour Migration and Rural Transformation in Colonial Asia, 
Comparative Asian Studies, Vol. 5, Amsterdam: Free University Press.
Brown, L. (1995). "Nature’s Limits". In: State o f  the World 1995, Brown et al. (eds.),
83
New York and London: Norton.
Burgess, J.C. (1993). Timber Production, Timber Trade and Tropical Deforestation, 
Ambio, Vol. 22, No.2-3: 136-43.
Burger, K (2000). Labour Markets, Migration and The Environment in Ghana. 
Unpublished Manuscript, Vrije Universiteit, Amsterdam.
Burger, K., Smit, H. and B. Vogelvang (2001). The Effect of Exchange Rates on 
Commodity Prices-The Case of Natural Rubber. Paper Presented at the 21st 
International Symposium on Forecasting, Callaway Gardens, Pine Mountain, 
Georgia, USA, (June 17-20,2001).
Buschbacher, R.J. (1992). Tropical Deforestation and Pasture Development, Bioscience 
Vol. 36, No. 1:22-28.
Campbell, J.Y. and P. Perron (1991). Pitfalls and Opportunities: What Macroeconomists 
Should Know About Unit Roots, Econometric Research Memorandum No. 360, 
Princeton University.
Cebula, R. J. (1981). Money Illusion and Migration Decisions: An International 
Comparison o f the United States and Canadian Experiences, Regional Studies, 
Vol. 15, No. 4: 241-6.
Cebula, R. J. (1993). The Impact o f Living Costs on Geographic Mobility, The Quarterly 
Review o f  Economics and Finance, Vol. 33, No. 1: 101-5.
Chakraborty, M. (1994). An Analysis of the Causes o f Deforestation, In: The Causes o f  
Tropical Deforestation: The Economic and Statistical Analysis o f  Factors Giving 
Rise to the Loss o f  the Tropical Forests, Katrina Brown and David W. Pearce 
(eds.), University College, London (UCL) Press, 1994.
84
Chomitz, K.M. and D.A. Gray (1966). Roads, Lands, Markets and Deforestation: A 
Spatial Model of Land Use in Belize, The World Bank Economic Review, Vol. 10, 
No.3: 487-512.
Clark, D. E. and J. C. Cosgrove (1991). Amenities versus Labour Market Opportunities: 
Choosing the Optimal Distance to Move, Journal o f  Regional Science, Vol. 31, 
No. 3: 311-28.
Collier, P. (1979). Migration and Unemployment: A Dynamic General Equilibrium 
Analysis Applied to Tanzania. Oxford Economic Papers, No. 2 (July): 205-236.
Cownie, J. (1974). Agriculture, Domestic Manufacturing, and Structural Transformation: 
Assessing Economic Development, African Studies Review, Vol. 17, No. 1 
(April).
Cropper, M. and C. Griffiths (1994). The Interaction of Population Growth and 
Environmental Quality, American Economic Review, Vol. 84, No. 2, (May): 250- 
54.
Cropper, M. and C. Griffiths (1999). Roads, Population Pressures and Deforestation in 
Thailand, 1976-1989. Land Economics, Vol. 75, No. 1: 58-73.
Cruz, W.D. and R. Repetto (1992). The Environmental Effects o f  Stabilisation and 
Structural Adjustment Programmes: The Philippines Case. Washington D.C.: 
World Resources Institute.
Da Vanzo, J. (1978). Does Unemployment Affect Migration? Evidence from Micro-Data, 
Review o f  Economics and Statistics, Vol. 65: 552-59.
Deacon, R. (1994). Deforestation and the Rule of Law in a Cross-section o f Countries. 
Land Economics 70: 414-430.
85
deGraft-Johnson, K.T. (1974). Population Growth and Rural-Urban Migration, with 
Special Reference to Ghana, International Labour Review, Vol. 109: 471-485.
de Hann, A. (1997). Rural-Urban Migration and Poverty: The Case o f India, IDS Bulletin, 
Issues Focussing on, Urban Poverty: A New Research Agenda. Vol. 28, No. 2: 
35-47.
Dickey, D.A. and W.A. Fuller (1981). Likelihood Ratio Statistics for Autoregressive time 
Series with a Unit Root, Econometrica, Vol. 49, No. 4: 1057-72.
Diez-Canedo, J.R. (1991). The Effect of Policy Restrictions on Capital and Labour Flows 
in Mexico, In: Abowd and Freeman (1991): 101-19.
Dovring, F. (1959). The Share of Agriculture in a Growing Population FAO Monthly 
Bulletin o f  Agricultural Economics and Statistics, Vol. 8, No.8/9: 8 
(August/September).
Dvorak, K.A. (1992). Resource Management by West African Farmers and the 
Economics of Shifting Cultivation, American Journal o f  Agricultural Economics, 
Vol. 74: 809-15.
Ehui, S.K. and T.W. Hertel (1989). Deforestation and Agricultural Productivity in the 
Cote d’Ivoire. American Journal o f  Agricultural Economics, Vol. 71, No. 3: 703- 
711.
Emerson, R.D. (1989). Migratory Labour and Agriculture, American Journal o f  
Economics, Vol. 71: 17-29.
Engle, R.F. and C.W.J. Granger (1987). Cointegration and Error Correction: 
Representations, Estimations and Testing, Econometrica, Vol. 55: 251-76.
EPA (1994). Ghana Environmental Action Plan, Laing, E. (ed), Environmental Protection 
Agency (EPA) Publication, 1994, Vol. 2
86
Ewusi, K. (1987). Structural Adjustment and Stabilisation Policies in Developing 
Countries: A Case Study of Ghana’s Experience in 1983-1986, Tema: Ghana 
Publishing Corporation.
Ewusi, K,, S.Y. Atsu and L.O. Gyekye (1983). Land Utilisation and Labour Productivity 
in Agriculture in Ghana. A Report to the Jobs and Skills Programme for Africa 
(JASPA), International Labour Organisation, Addis-Ababa, Technical Publication 
No. 46.
Faini, R. and A. Venturi (1993). Trade, Aid and Migrations: Some Basic Policy Issues. 
European Economic Review, Vol. 37: 435-42.
Falaris, M.E. (1979). Determinants of Internal Migration in Peru: An Economic Analysis. 
Economic Development and Cultural Change, Vol. 27, No.2 (January): 327-341.
FAO (1986), Natural Resources and Human Environment for Food and Agriculture in 
Africa. Environment and Energy Paper No. 6, Rome.
FAO (1992). FAO Production Yearbook: Food and Agricultural Organisation of the 
United Nations, Rome.
FAO (1993). FAO Production Yearbook: Food and Agricultural Organisation of the 
United Nations, Rome.
FAO (1997). State o f the World’s Forest 1997. Food and Agricultural Organisation of 
the United Nations, Rome.
FAO (2000). FAO Production Yearbook: Food and Agricultural Organisation of the 
United Nations, Rome.
Fei, J.C.H. and G. Ranis (1964). Development o f the Labour Surplus Economy, 
Homewood, Illinois; Irwin, Inc., 1964
87
Fosu, K. Yerfi (1989). "Agriculture in Ghana’s Post-1970 Development Process" Paper 
Presented to the 5th Biennial Conference o f the West African Economic 
Association, Lome, Togo, 1989.
Fosu, K. Yerfi (1992). The Real Exchange Rate and Ghana’s Agricultural Exports, 
AERC 9. Centre for the Study of African Economies (CSAE), University of 
Oxford, Oxford, U. K.
Fosu, K. Yerfi (1994). "Determinants and Effectiveness of Government Expenditure 
Policy in Ghana’ Agricultural Sector", Issues in African Rural Development 2. 
Winrock International Institute for Agricultural Development, Arlington, USA.
Fosu, K. Yerfi (1996). Supply Response Under Recent Economic Policy Reforms in 
Ghana. Technical Report submitted to the Global Coalition for Africa (GCA). 
High-Level Meeting and Policy Reform Forum of GCA. Washington D.C; USA.
Fosu K. Yerfi and N. Heerink (1996). Policy Impact on Agricultural Supply Response in 
Ghana. Network of European Agricultural-Tropically and Sub-Tropically 
Oriented-University and Scientific Complexes Related with Agricultural 
Development (NATURA). Catholic University o f Leuven, Leuven-La-Neuve, 
Belgium.
Fuller, W.A. (1976). Introduction to Statistical Time Series, Wiley, New York.
Gelb, A. Knight, J.B. and R.H. Sabot (1989). Public Sector Employment, Rent-Seeking 
and Economic Growth, Unpublished Manuscript, 1989,
Ghana Statistical Service, Ghana Population Census (2000). Population Census, 
Preliminary Reports.
Ghatak, S., Levine, P. and S.W. Price (1996). Migration Theories and Evidence: An 
Assessment, Journal o f  Economic Surveys, Vol. 10, No. 2: 159-198.
Gleave, M.B, and H.P. White ‘s (1969). * Population Density and Agricultural Systems in 
West Africa’ in Thomas, M.F. and G.W. Whittington eds. Environment and Land 
Use in Africa, London: Methuen.
Gomez-Pompa, A., C. Vazquez-Yanes, S. Guevera (1972). The Tropical Rainforest: A 
Non-renewable Resources Science 117:762-65.
Gordon, J.B. (1969). Labour Mobility and Economic Growth: The Central American 
Experience-Costa Rica and El Salvador, Economic Development and Cultural 
Change, Vol. 17, No. 3 (April): 319-37.
Goss, E. and C. Paul (1990). The Impact of Unemployment Insurance Benefits on the 
Probability o f being Unemployed, Journal o f  Regional Science, Vol. 30, No. 3: 
349-58.
Granger, C .W J. (1969). Investigating the Causal Relationships by Econometric Models 
and Cross-Spectral Methods, Econometrica, 37: 540-52.
Granger, C.WJ. (1988). Some Recent Development in a Concept o f Causality, Journal o f  
Econometrics, Vol. 39: 199-211.
Granger, C.W.J.and P. Newbold (1974). Spurious Regressions in Econometrics, Journal 
o f  Econometrics: 452-64.
Greenwood, M.J. and J.M. McDowell (1991). Differential Economic Opportunity, 
Transferability o f Skills and Immigration to the United States and Canada, Review 
o f Economics and Statistics, Vol. 73, No. 2: 612-23.
Gyasi, E., G.T. Agyapong, E. Ardayfio-Schandorf, L. Enu-Kwesi, J.S. Nabila and E. 
Owusu-Bennoah (1995). Production Pressure and Environmental Change in the 
Forest-savanna zone of Southern Ghana, Global Environmental Change, Vol. 5, 
No. 4: 355-366.
89
Hafiier, J. (1990). Forces and Policy Issues affecting Forest Use in Northeast Thailand 
1900-1985. In: Keepers o f  Forest. Land Management Alternatives in Southeast 
Asia, M. Poffemberg (ed.): 69-94. West Hartford: Conn., Kumarian Press.
Hamilton, J.D. (1994). Time Series Analysis, Princeton University, Princeton: USA.
Harris, J.R. and M.P. Todaro (1970). Migration, Unemployment and Development: A 
Two-Sector Analysis, American Economic Review, Vol. 60: 126-142.
Hatton, T J . and J.G. Williamson (1994). (Eds.) Migration and the International Labour 
Markets 1850-1939, London, UK: Routeledge.
Hecht, S.B. (1985). Environment, Development and Politics: Capital Accumulation and 
Livestock Sector in Eastern Amazonia. World Development, Vol. 13, No. 6: 663- 
84.
Hendry, D.F. (1986). Econometric Modelling with Cointegrated Variables: An Overview, 
Oxford Bulletin o f  Economics Studies, Vol. 48. No.3: 201-12.
Hendry, D.F., Pagan, A.R. and J.D. Sargan (1984). "Dynamic Specification", In: Z. 
Griliches and M.D. Intriligator (eds.) Handbook o f  Econometrics, Vol.2, 
Amsterdam: Elsevier.
Herzog, H.W. Jr., Schlottmann, A.M. and T.B. Boehm (1993). Migration as Spatial Job- 
Search: A Survey of Empirical Findings, Regional Studies, Vol. 27, No. 4: 327- 
40.
Holden, D. and R. Permon (1994). "Unit Root and Cointegration for the Economist", In: 
Cointegration fo r  the Applied Economist, Bhaskara Rao, R. (ed.), New York: 
USA, St Martin’s Press.
Holden, S.T. (1993). Peasant household Modelling: Farming Systems, Education and 
Sustainability in Northern Zambia, Agricultural Economics, Vol. 9: 241-67.
90
Holden, S.T. (1997). Adjustment Policies, Peasant Household Resource Allocation, and 
Deforestation in Northern Zambia: An Overview and some Policy Conclusions. 
Forum fo r  Development Studies, Vol.  1: 117-34.
Hylleberg, S. and G.E. Mizon (1989). A Note on the Distribution o f Least Squares 
Estimator o f Random Walk with Drift, Economics Letters, 29, 225-30.
Jaegar, W.K. (1992). The Effects of Economic Policies on African Agriculture, World 
Bank, Washington, D.C.
Jebuni, C., Oduro A., Asante Y., G.K. Tsikata (1992). Diversifying Exports, ODI, 
London, UK.
Johansen, S. (1988). Statistical Analysis of Cointegration Vectors, Journal o f  Economic 
Dynamics and Control, Vol. 12: 231-54.
Johansen, S. and K. Juselius (1990). Maximum Likelihood Estimations and Inference on 
Cointegration -  With Application to the Demand for Money. Oxford Bulletin o f  
Economics and Statistics, Vol. 52: 169-210.
Johnson, D.G. (1948). Mobility as a Field of Research, Southern Economic Journal, Vol. 
40: 152-61.
Johnston, B.F. and P. Kirby (1975). Agriculture and Structural Transformation: 
Economic Strategies in Late-Developing Countries, New York, USA: Oxford 
University Press.
Johnston, J. and J. Dinardo (1997). Econometric Methods, Fourth Edition, McGraw-Hill.
Jones, D.W. and R.V. O’ Neill (1994). Development Policies, Rural Land Use, and 
Tropical Deforestation. Regional Science and Urban Economics, Vol. 24: 757-71.
Jones, D.W. and R.V. O’ Neill (1995). Development Polices, Urban Employment and 
Deforestation: The Role of Infrastructure and Tax Policy in a Two-Sector Model.
91
Journal o f  Regional Science. Vol. 35, No. 1: 135-53.
Kahn, J and J. McDonald (1994). International Debt and Deforestation, In: The Causes of 
Tropical Deforestation: The Economic and Statistical Analysis o f Factors giving 
rise to the Loss o f the Tropical Forests, Katrina Brown and David W. Pearce 
(eds.), University College, London (UCL) Press, 1994.
Judges, G. G., R. C. Hill, W.E. Griffiths, H. Lutkepohl and L. Tsoung-Chao (1988). 
Introduction to the Theory and Practice of Econometrics. John Wiley and Sons, 
New York.
Kannapan, S. (1983). Employment Problems and Urban labour Market in Developing 
Countries. Aim Arbor: University of Michigan, Graduate School o f Business 
Administration. Study Supported by a grant from the U.S. Agency for 
International Development, Employment and Enterprise Division.
Kannapan, S. (1985). Urban Employment and the Labour Markets in Developing 
Nations, Economic Development and Cultural Change. Vol. 33, No. 4 (July): 
699-730.
Kannapan, S. (1988). Urban Labour Markets and Development, The World Bank 
Research Observer, Vol. 3, No. 2 (July): 198-205.
Khan, M.S. and M.D. Knight (1985). "Fund-supported Adjustment Program and 
Economic Growth", Occasional Paper No. 41, International Monetary Fund, 
Washington DC, USA.
Kimsley, M. (1991). An Analysis of the Spatial and Temporal Variation in Tropical 
Deforestation. Ph.D. dissertation. University o f Georgia, Department of 
Geography, Athens.
92
Knerr, B. (1992). "Agricultural Policies and Tropical Forests: Issues in Agricultural 
Development (Sustainability and Cooperation)", IAAE Occasional Papers, No. 6: 
313-321.
Knight, J.B. (1972). Rural-Urban Income Comparisons and Migration in Ghana. Oxford 
Bulletin o f  Economics and Statistics, Vol. 34, No. 2: 199-228.
Krutilla, K., William, F.H. and D.F. Barnes (1995). Urban energy Consumption and Peri- 
Urban Deforestation, Forest Ecology and Management, Vol. 74, No.2: 181-95.
Kwiatkowski, D. Phillips, C.B., Schmidt, P and Y. Shin (1992). Testing the Null 
Hypothesis of Stationarity Against the Alternative o f a Unit Root: How Sure are 
that Economic Time Series Have a Unit Root?, Journal o f  Econometrics, Vol. 54: 
159-178.
Lai, J.B. (1989). India’s Forests: Myths and Reality. Dehradrun: Natraj Publishers.
Lambardini, C. (1994). Deforestation in Thailand, In: The Causes o f Tropical 
Deforestation: The Economic and Statistical Analysis o f Factors Giving Rise to 
the Loss of the Tropical Forests, Katrina Brown and David W. Pearce (eds.), 
University College, London (UCL) Press, 1994.
Leach, M. and J. Fairhead (2000). In West Africa’s Dynamic Forests, Trees Come and 
Go. Population and Development Review, Vol. 26, No.l, March.
Levi, M. and W. Wadyacki (1973). The Influence of Family and Friends on Geographic 
Labour Mobility: An International Comparison, Review o f  Economics and 
Statistics, Vol. 55, No. 2: 198-203.
Lewis, W.A. (1954). Development with Unlimited supplies o f Labour, The Manchester 
School 22:3-42 (May).
Lewis, W.A. (1966). Development Planning, New York, USA.
93
London Economics (1996). The Impact o f Privatisation on Labour in Africa: The Ghana 
Case Study. A World Bank Research Project, London.
Lopez, R. (1997). Environmental Externalities in Traditional Agriculture and the Impact 
of Trade Liberalisation: The Case of Ghana. Journal o f  Development Economics, 
Vol. 53: 17-39.
Lopez, R. (1998). Agricultural Intensification, Common Property Resources and the 
Farm-Household. Environment and Resource Economics, Vol. 11, No. 3-4: 443- 
58.
Lucas, R.E.B. and O.Stark (1985). Motivations to Remit: Evidence from Botswana, 
Journal o f  Political Economy, Vol. 93, No. 5: 901-18.
MacKinnon, J.G. (1991). "Critical Values for Cointegration Tests" In R.F. Engle and 
C.W.J. Granger (eds.), Long-run Economic Relationships: Readings in 
Cointegration, Oxford, UK: Oxford University Press, Chapter 13.
Maddala, G.S. (1983/ Limited-Dependent and Qualitative Variables in Econometrics, 
Cambridge, UK: Cambridge University Press.
Maddala, G.S. (1992/ An Introduction to Econometrics, 2nd edition, Englewood Cliffs, 
New Jersey 07632, USA: Prentice-Hall, Inc., A Simon & Schuster Company.
Mainardi, S. (1996). An Econometric Analysis o f Factors Affecting Tropical and 
Subtropical Deforestation. University of Natal, Department o f Economics, South 
Africa.
McHone, W.W. and B. Rungeling (1993), Replacement Income, Duration of 
Unemployment and the Migration o f Unemployed Prime Age Males, Applied 
Economics, Vol. 25: 905-9.
94
Mertens, B. and E.F. Lambin (1997). Spatial Modelling o f Deforestation in Southern 
Cameroon. Applied. Geography, Vol. 17, No.2: 1-19.
Molle, W.T.M. and van Mourik, A (1989). A Static Explanatory Model o f International 
Labour Migration to and in Western Europe, In Gordon and Thirwall: 30-52.
Monela, G.C. (1995). Tropical Rainforest Deforestation, Biodiversity Benefits and 
Sustainable Land Use: Analysis of Economic and Ecological Aspects Related to 
the Nguru Mountains, Tanzania. Ph.D. dissertation. Agricultural University of 
Norway, Department o f Forest Sciences, Aas.
Morrison, A.R. (1993). Violence or Economics: What Drives Internal Migration in 
Guatemala? Economic Development and Cultural Change, Vol. 41, No. 4: 816- 
31.
Munasinghe, M. (Ed.) (1996/ Environmental Impacts o f  Macroeconomic and Sectoral 
Policies. The International Society for Ecological Economics (ISEE), The World 
Bank and The United Nations Environmental Programme (UNEP).
Mundlak, Y. and D. Larson (1997). On the Intersectoral Migration o f Agricultural labour, 
Economic Development and Cultural Change, Vol. 45, No. 2 (January).
Myers, N. (1984). The Primary Source: Tropical Forests and our Future, New York: 
Norton.
Myers, N. (1989). Deforestation Rates in Tropical Forests and their Climatic 
Implications. London, UK: Friends o f the Earth.
Myers, N. (1994). Tropical Forests: Rates and Patterns, In: The Causes o f Tropical 
Deforestation: The Economic and Statistical Analysis o f Factors Giving Rise to 
the Loss of the Tropical Forests, Katrina Brown and David W. Pearce (eds.), 
University College, London (UCL) Press, 1994.
95
Nakajima, C. (1986). Subjective Equilibrium Theory o f the Farm Household. Elsevier, 
Amsterdam
Ngihiep, L. T. (1986). "The Land Rotating Farming in Northern Brazil: Conditions for its 
Continuation and Transition to the Sendatary Cultivation System", Working Paper 
Series 34. International Development Center of Japan, Tokyo.
Nurkse, R. (1953). Problems of Capital Formation in Underdeveloped Areas, New York.
Palo, M. (1994). Population and Deforestation. In: The Causes o f Tropical Deforestation: 
The Economic and Statistical Analysis of Factors Giving Rise to the Loss o f the 
Tropical Forests, Katrina Brown and David W. Pearce (eds.), University College, 
London (UCL) Press, 1994.
Panayotou, T. and S. Sungsuwan (1994). "An Econometric Study o f the Causes of 
Deforestation: The Case of Northeast Thailand". Discussion Paper No. 284, 
Harvard Institute o f International Development. Harvard University, Cambridge, 
Mass.
Panayotou, T. and S. Sungsuwan (1994). An Econometric Analysis o f the Causes of 
Tropical Deforestation: The Case Study o f Northeast Thailand, In: The Causes of 
Tropical Deforestation, The Economic and Statistical Analysis o f Factors Giving 
Rise to the Loss of the Tropical Forests, Katrina Brown and David W. Pearce 
(eds.), University College, London (UCL) Press, 1994.
Parmeiter, C. (1995). Entwurzelung, Globalisierung undue Migration. Ausgewahlite 
Fragestellungen, Journal fu r  Entwicklung politik, Vol. 11, No. 3: 245-60.
Pfaff, A.S.P. (1997). "What Drives Deforestation in the Brazilian Amazon? Evidence 
from Satellite and Socioeconomic Data". Policy Research Working Paper 1772. 
World Bank, Policy Research Department, Washington, D.C. Processed.
96
Phillips, P.C.B. and P. Peron (1988). Testing for a Unit root in Time Series Regression, 
Biometrika, Vol. 75: 335-46.
Pissarides, C.A. and I. McMaster (1990). Regional Migration, Wages and 
Unemployment: Empirical Evidence and Implications for Policy, Oxford 
Economic Papers, Vol. 42: 812-31.
Pissarides, C.A. and Wadsworth, J. (1989). Unemployment and the Inter-Regional 
Mobility o f Labour, The Economic Journal, Vol. 99: 739-55.
Prothero, R.M. and M. Chapman (1985). "Themes on Circulation in the Third World", In 
R.M. Prothero and Chapman (eds.), Circulation in Third World Countries, 
London, UK: Routeledge and Kegan Paul.
Pindyck, R.S. and D.L. Rubinfeld (1991). Econometric Models and Econometric 
Forecasts, McGraw Hill, 3rd edition.
Ravallion, M. (1986). Testing Market Integration. American Journal o f  Agricultural 
Economics, Vol. 68: 102-9.
Rock, M.T. (1996). The Stock, the Plow, Rural Social Structures, and Tropical 
Deforestation in Poor Countries? Ecological Economics, Vol. 18: 113-31.
Roy, A.D. (1951). Some thoughts on the Distribution o f Earnings, Oxford Economic 
Papers, Vol. 3: 135-46.
Ruben, R. Kruseman, G. and H. Hengsdijk (1994). "Farm Household Modelling for 
Estimating the Effectiveness o f Price Instruments on Sustainable Land Uses in the 
Atlantic Zone of Costa Rica". DLV Report 4. Wageningen Agricultural 
University, The Netherlands.
Ruf, F. (1997). In: Ghana Forest Loss, Ted Case Studies, Case Number 193.
Rothenberg, H. (1980). Farming Systems in the Tropics. Third Edition, Oxford, UK: 
Claredon Press.
Sabot, R.H. (1977). The Meaning and Measurement of Urban Surplus Labour. Oxford 
Economic Papers, Vol. 29, No.3: 389-411.
Sahota, G.S. (1968). An Economic Analysis of Internal Migration in Brazil. Journal o f  
Political Economy, Vol. 76, No. 2 (March): 218-245.
Saikkonen, P. (1991). Asymptotically Efficient Estimation o f Cointegrating Regressions, 
Econometric Theory, 7, 1-21.
Sakyi-Dawson, O. (2000). Wenchi Farming Systems Study. Final Report for CARE- 
GHANA, Department of Agricultural Extension, University o f Ghana, Legon, 
May 2000.
Shane, D.R. (1986). Hoofprints in the Forests: Cattle Ranching and Deforestation o f  
Latin America’s Tropical Forests. Philadelphia: Institute for the Study of Human 
Issues.
Shields, G.M. and M.P. Shields (1993). A Theoretical and Empirical Analysis of Family 
Migration and Household Production: U.S. 1980-1985, Southern Economic 
Journal, Vol. 59, No. 4: 768-82.
Silvapulle, P. and S. Jayasuriya (1994). Testing for Philippines Rice Market Integration: 
A Multiple Cointegration Approach, Journal o f  Agricultural Economics, Vol. 54: 
369-80.
Singh, K.D. (1993). "The 1990 Tropical Forest Resources", In: Unaslylva, An FAO 
International Journal o f  Forestry and Forest Industries, Vol. 44, No. 174: 10-19.
Singh, M. (1995). Uneven Development in Agriculture and Labour Migration: A Case o f  
Bihar and Punjab, Shimla Institute.
Sjaastad, L.A. (1962). The Cost and Returns of Human Migration. Journal o f  Political 
Economy, Vol. 70, No. 5 (October): 80-94.
98
Southgate, D. (1990). The Causes of Land Degradation along Spontaneously Expanding 
Agricultural Frontiers in the Third World. Land Economics, Vol. 66, No. 1: 93- 
101.
Stark, O. (1991). The Migration o f  Labour. Oxford, UK: Basil Blackwell.
Stewart, M.B. and K.F. Wallis (1981). Introductory Econometrics, 2nd Edition, Oxford, 
UK: Basil Blackwell.
Stock, J.H. and M.W. Watson (1993). A Sample Estimator o f Cointegrating Vectors in 
Higher Order Integrated Systems, Econometrica, 61: 738-820.
Sunderlin, W.D. and Resosudarmo, I.A.P. (1999). The Effect o f Population and 
Migration on Forest Cover in Indonesia. Journal o f  Environment and 
Development, Vol. 8, No. 2, (June): 152-169.
Tailor, J.E. (1987). Undocumented Mexico-U.S. Migration and the Returns to 
Households in Rural Mexico. American Journal o f  Agricultural Economics 9: 
626-38.
Tambi, N.E. (1999). Cointegration and Error-Correction Modelling o f Agricultural 
Export Supply in Cameroon, Agricultural Economics 20:57-67.
Ted Cases Studies (1997). Ghana Forest Loss, Case Number 193.
Thomas, J.J. (1992). Informal Sector Activity, LSE Handbooks in Economics, Ann Arbor: 
University o f Michigan Press.
Todaro, M.P. (1969). A Model of Labour Migration and Urban Unemployment in Less 
Developed Countries, American Economic Review, Vol. 59, No. 1: 138-48.
Todaro, M.P. (1985). Economic Development in the Third World. 3d. Ed. New York: 
Longman.
The Ecologist. (1990). Increasing Aridity in Ghana, The Ecologist, Vol. 20, No. 2: 50.
99
Treyz, G.I. (1993). The Dynamics of U.S. Internal Migration, Review o f  Economics and 
Statistics, Vol. 84, No.2: 17-22.
Vincent, J.R., R.M. Ali and Associates (1997). "Forests" In: Environment and 
Development in a Resource-Rich Economy: Malaysia Under the New Economic 
Policy, Harvard Institute for International Development, USA.
Winkelmann, D. (1966). A Case Study of the Exodus of Labour from Agriculture: 
Minnesota Journal o f  Farm Economics, Vol. 48, No. 1: 211-25, (February).
Withers G. and D. Pope (1985). Immigration and Unemployment, Economic Record, 
Vol. 61, No. 173:554-63.
World Bank (1989). World Development Reports, Oxford University Press, New York 
and Oxford.
World Bank (1990). World Development Reports, Oxford University Press, New York 
and Oxford.
WRI (1985). Tropical Forests: A Call for Action. Washington, D.C., World Resources 
Institute.
WRI (1986). World Resources 1986, World Resources Institute, New York: Basic Books.
WRI (1989). World Resources Report, 1988-1989, World Resources Institute.
Yap, L. (1977). The Attraction o f Cities: A Review of the Migration Literature, Journal 
o f  Development Economics, Vol. 4, No. 3: 239-64.
Zulu, J.B. and M. Nsouli (1985). "Adjustment Programs in Africa": The Recent 
Experience, Occasional Paper No.34, International Monetary Fund, Washington, 
DC, USA, 1985.
100