THE IMPACT OF CLIMATE CHANGE AND THE EUROPEAN UNION
GSP-SCHEME ON EAST AFRICA’S HORTICULTURAL TRADE
BY MOSES HERBERT LUBINGA
Submitted in accordance with the requirements for the degree
PHILOSOPHIAE DOCTOR (PhD) in Agricultural Economics
in the
PROMOTER: DR. H. JORDAAN FACULTY OF NATURAL AND AGRICULTURAL SCIENCES
CO-PROMOTER: DR. A.A. OGUNDEJI DEPARTMENT OF AGRICULTURAL ECONOMICS
UNIVERSITY OF THE FREE STATE
NOVEMBER 2014 BLOEMFONTEIN
Declaration
I, Moses Herbert Lubinga, declare that the thesis I hereby submit for the Philosophiae
Doctor (PhD) Degree in Agricultural Economics at the University of the Free State is my
own independent work and that I have not previously submitted it at another university.
Furthermore, I cede copyright of the thesis in favour of the University of the Free State.
-------------------------------------------------------
M. H. LUBINGA
NOVEMBER 2014
i
Dedication
This work is first and foremost dedicated to my wife (Stellah) and the beloved daughter
Sonia. Furthermore, I dedicate it to Engineer C.J. Mutyaba (dad), Ms. H. Nansubuga (Mum-
R.I.P) and my siblings (Dianah, Ibra, Angela and Benjamin).
ii
Acknowledgement
Above all, I am indebted to the Almighty Lord for the gift of life and enabling me accomplish
this task. I thank you Lord!
I am very grateful to the following persons and entities whose support, expertise and advice
has enabled me accomplish this study.
 Profound gratitude goes to my promoters (Dr. H. Jordaan and Dr. A. Ogundeji), who
tirelessly guided me throughout this research. Their succinct advice, constructive
criticisms and encouragement has enabled me reach this far.
 I am indebted to the research directorate at UFS, particularly, Prof. N. Roos, Dr.
Taylor and Mr. W. Nel through whom I was able to source a bursary to partake my
doctoral studies.
 The financial assistance of the National Research Foundation (NRF) towards this
research is hereby acknowledged. Outstandingly, am grateful to Ms. J. Nogabe for the
continuous support rendered by providing apt information regarding scholarship
opportunities.
 Special thanks go to my wife (Stellah), Gertrude, Dr. Kigozi, and the entire family for
being supportive throughout this journey.
 I would also like to register my sincere heartfelt gratitude to all the staff members of
the Department of Agricultural Economics (UFS) for availing an ambient
environment during the course of my studies.
 Friends, I thank you for the amity, prayers and advice.
iii
Table of contents
Declaration..................................................................................................................................i
Dedication ..................................................................................................................................ii
Acknowledgement ................................................................................................................... iii
Table of contents.......................................................................................................................iv
List of tables........................................................................................................................... viii
List of figures............................................................................................................................ix
List of Appendices ....................................................................................................................xi
List of acronyms ......................................................................................................................xii
Abstract ...................................................................................................................................xvi
CHAPTER ONE ......................................................................................................................1
1.1 Background..................................................................................................................1
1.2 Problem statement .......................................................................................................2
1.3 Objectives of the study ................................................................................................5
1.4 Lay out of the study .....................................................................................................7
CHAPTER TWO: LITERATURE REVIEW.......................................................................8
2.1 Introduction..................................................................................................................8
2.2 The concept of competiveness and its measures .........................................................8
2.2.1 The concept and theoretical framework of Comparative Advantage ....................10
2.3 Trade related measures of competitiveness ...............................................................11
2.3.1 The Revealed Comparative Advantage (RCA) and its adjusted indicators...........11
2.3.2 The Revealed Symmetric Comparative Advantage (RSCA) ................................13
2.3.3 The Porter-adapted index of RCA (PRCA) and Dunning index of net
competitive advantage index (DNCA) ..................................................................13
2.3.4 The Net Export Index (NEI) ..................................................................................14
2.3.5 The Grubel-Lloyd index (GLI)..............................................................................14
2.3.6 The export to import price ratio.............................................................................15
2.3.7 The Real exchange rate (RER) ..............................................................................15
iv
2.4 Empirical evidence of competitiveness studies based on RCA methodology
in the agriculture sector ..............................................................................................16
2.4.1 Empirical studies of non-African economies (Rest of the world) .........................16
2.4.2 Empirical studies of some African economies ......................................................17
2.5 Climate change - Agriculture - International trade nexus .........................................19
2.5.1 A review selected empirical effects of temperature and precipitation
as determinants of agricultural productivity..........................................................22
2.5.2 A review of empirical studies relating to climate change effects on
international trade..................................................................................................23
2.6 The European Union's Generalized System of Preferences (EU-GSP) scheme........26
2.6.1 The theoretical framework of preferential treatment effects on
international trade..................................................................................................27
2.6.2 Measures of preferential treatment value: The Preference Margin (PM)..............29
2.6.3 The effect of the EU-GSP scheme on agricultural exports ...................................31
2.7 Predicting trade potential and performance...............................................................34
2.7.1 Trade potential and trade performance measures ..................................................34
2.7.2 Review of empirical studies that predict trade potential and performance ...........35
2.8 Conclusion .................................................................................................................36
CHAPTER THREE: AN OVERVIEW OF EAST AFRICA’S HORTI-
CULTURE SECTOR................................................................................................40
3.1 Introduction................................................................................................................40
3.2 Fruit and vegetable production in East Africa ...........................................................40
3.3 East Africa’s fruits and vegetables trade statistics and the major trade partners.......46
3.4 Fruit and vegetable export trends against temperature and precipitation ..................53
3.5 Conclusion .................................................................................................................59
CHAPTER FOUR:  RESEARCH METHODOLOGY......................................................60
4.1 Introduction................................................................................................................60
4.2 Focus of the study......................................................................................................60
4.3 Determining East Africa's export competitiveness in the fruits and vegetable
sector in the EU market...........................................................................................61
v
4.3.1 Data and data sources ............................................................................................61
4.3.2 Data analysis..........................................................................................................61
4.4 Determining the influence of climate change on East Africa's horticultural
trade flows .................................................................................................................63
4.4.1 A brief overview of the gravity model framework................................................63
4.4.2 Selected trade partners...........................................................................................69
4.4.3 Data description and data sources .........................................................................70
4.4.4 Data management ..................................................................................................72
4.4.5 Computation of anomalies from meteorological data to proxy for climate
change....................................................................................................................76
4.4.6 Specification of the regression model to ascertain the influence of climate
change on East Africa's horticultural trade flows.................................................78
4.4.7 Estimation techniques used ...................................................................................83
4.5 Estimating the effect of the EU-GSP scheme on East Africa’s fruits and
vegetable imports into the EU market.......................................................................84
4.5.1 Data and data sources ............................................................................................85
4.5.2 Data management ..................................................................................................86
4.5.3 The preference margin as a proxy for the effects of the EU-GSP Scheme
on fruit and vegetable imports into the EU market ..............................................86
4.5.4 Specified regression model to capture the effect of the EU-GSP Scheme
on East Africa’s fruits and vegetable trade flows into the EU market .................88
4.5.5 Estimation techniques used ...................................................................................92
4.6 Predicting unilateral Trade Potential and performance .............................................93
4.7 Conclusions................................................................................................................94
CHAPTER FIVE: RESULTS AND DISCUSSIONS .........................................................96
5.1 Introduction................................................................................................................96
5.2 Export competitiveness of East African states in horticultural commodities............96
5.3 The influence of climate change on East Africa's horticultural trade flows............101
5.3.1 Diagnostic test results ..........................................................................................102
vi
5.3.2 The influence of climate change on East Africa's horticultural trade
flows into the EU market......................................................................................104
5.4 The effect of the EU-GSP scheme on East Africa’s fruits and vegetable
imports into the EU market .....................................................................................113
5.4.1 Diagnostic test results ..........................................................................................114
5.4.2 Empirical findings of the effect of the EU-GSP scheme on East Africa’s
fruits and vegetable imports into the EU market.................................................116
5.5  East Africa's unilateral Trade Potential and performance in exporting fruits
and vegetable into the EU market .............................................................................126
5.6 Summary of results and discussions ........................................................................131
CHAPTER SIX: SUMMARY, CONCLUSION AND RECOMMENDATIONS..........135
6.1 Summary and conclusions .......................................................................................135
6.2 Recommendations....................................................................................................137
6.2.1 Recommendations to exporters ...........................................................................137
6.2.2 Policy recommendations......................................................................................137
6.2.3 Recommendations to researchers ........................................................................139
6.2.4 Recommendations for further research................................................................139
REFERENCES.....................................................................................................................140
APPENDICES......................................................................................................................160
vii
List of tables
Table 3.1: Major export markets for fruits and vegetables from East Africa in
December 2011 .....................................................................................................51
Table 4.1: A summary of the expected sign of climate change variables and other
covariates on fruits and vegetable imports into the EU market .............................84
Table 4.2: A summary of the expected sign of the effect of the EU-GSP scheme
and other covariates on fruits and vegetable imports into the EU market ............92
Table 5.1: The mean revealed comparative advantage index at HS-4 Digit level...................97
Table 5.2: Mean revealed comparative advantage index at HS-6 Digit level .........................99
Table 5.3: Over-dispersion test results of horticulture exports by country............................102
Table 5.4: LLC Panel Unit Root test results by country........................................................103
Table 5.5: Empirical effects of climate change on East Africa's horticultural
imports into the EU market...................................................................................105
Table 5.6: East Africa’s selected fruit and vegetable commodities with high
export competitiveness in the EU market ...........................................................113
Table 5.7: VIF test- and over-dispersion test- results for the horticultural
commodities........................................................................................................114
Table 5.8: Panel Unit Root test results by commodity and country ......................................115
Table 5.9: Effect of the EU-GSP scheme on Kenya’s Asparagus and Bean exports ............117
Table 5.10: Effect of the EU-GSP scheme on Tanzania’s Vegetables and Bean
exports...............................................................................................................120
Table 5.11: Effect of the EU-GSP scheme on Uganda’s Banana, Bean and
Pepper exports...................................................................................................122
Table 5.12: Mean Absolute Difference (ADijlt) for East Africa's selected
horticultural commodities at country level with the EU-15 states....................126
viii
List of figures
Figure 3.1: Area harvested under fruits and vegetables in Kenya ...........................................41
Figure 3.2: Area harvested under fruits and vegetables in Tanzania .......................................42
Figure 3.3: Area harvested under fruits and vegetables in Uganda .........................................43
Figure 3.4: Kenya’s trend in fruit and vegetable production ...................................................44
Figure 3.5: Tanzania’s trend in fruit and vegetable production ...............................................45
Figure 3.6: Uganda’s trend in fruit and vegetable production .................................................45
Figure 3.7: Aggregated fruit and vegetable exports from Kenya, Tanzania
and Uganda ..........................................................................................................47
Figure 3.8: Kenya's fruit and vegetable exports to the EU, by value ......................................48
Figure 3.9: Tanzania's fruit and vegetable exports to the EU, by value ..................................49
Figure 3.10: Uganda's fruit and vegetable exports to the EU, by value...................................50
Figure 3.11: Fruit and vegetable net exports from Kenya, Tanzania and Uganda ..................53
Figure 3.12: Trend line of Kenya's fruit and vegetable exports into the EU in
relation to precipitation ........................................................................................54
Figure 3.13: Trend line of Kenya's fruit and vegetable exports into the EU in
relation to temperature ......................................................................................55
Figure 3.14: Uganda's fruit and vegetable exports into the EU in relation to
temperature ........................................................................................................56
Figure 3.15: Uganda's fruit and vegetable exports into the EU in
relation to precipitation ......................................................................................57
Figure 3.16: Tanzania's fruit and vegetable exports into the EU in
relation to precipitation .....................................................................................58
ix
Figure 3.17: Tanzania's fruit and vegetable exports into the EU in
relation to temperature ......................................................................................59
Figure 5.1: The Relative Difference Index for Kenya's beans and asparagus
exports with the EU-15 member states ................................................................129
Figure 5.2: The Relative Difference Index for Tanzania's beans and vegetable
exports with the EU-15 member states ...............................................................130
Figure 5.3: The Relative Difference Index for Uganda's beans, bananas and
pepper exports with the EU-15 member states ...................................................131
x
List of Appendices
Appendix A: The EU-15 member states considered under this study ...................................160
Appendix B: Multi-collinearity test results for the three East African states
(Objective 2: The influence of climate change on East Africa's
horticultural trade flows)................................................................................161
Appendix C: Normality test results for objective two (The influence of climate
change on East Africa's horticultural trade flows).............................................167
Appendix D: Multi-collinearity test results for Kenya's Asparagus- 070920
(Objective three) ..............................................................................................168
Appendix E: Multi-collinearity test results for Kenya's Beans- 070820 (Objective three) ...169
Appendix F: Multi-collinearity test results for Tanzania's Beans- 070820
(Objective three) ..............................................................................................171
Appendix G: Multi-colinearity test results for Tanzania's Vegetables- 070990
(Objective three) ...............................................................................................172
Appendix H: Multi-colinearity test results for Uganda's Beans- 070820
(Objective three) ..............................................................................................173
Appendix I: Multi-colinearity test results for Uganda's Peppers- 070960
(Objective three) ...............................................................................................174
Appendix J: Multi-colinearity test results for Uganda's Bananas- 080300
(Objective three) ................................................................................................175
Appendix K: Kenya's normality test results for Asparagus and Beans .................................177
Appendix L: Tanzania's normality test results for Beans and Vegetables.............................177
Appendix M: Uganda's normality test results for Pepper, Bananas and Beans .....................178
xi
List of acronyms
ACCI : Australian Chamber of Commerce and Industry
ACODE : Advocates Coalition for Development Environment
ACP : African, Caribbean and Pacific
AD : Absolute Difference
AGOA : African Growth and Opportunity Act
AGRODEP ; African Growth and Development Policy Modelling Consortium
ASARECA : Association for Strengthening Agricultural Research in Eastern and
Central Africa
AVE ; Ad valorem Equivalent
BCPR : Bureau for Crisis Prevention and Recovery
BRC : British Retail Consortium
CEECs : Central and Eastern European countries
CEEPA : Centre for Environmental Economics and Policy in Africa
CODED : Eurostat's Concepts and Definitions Database CODED
COLEACP : Europe-Africa-Caribbean-Pacific Liaison Committee
COMESA : Common Market for East and Southern African
COMTRADE : Common Format for Transient Data Exchange
CPI : Consumer Price Index
CPIA : Country Policy and Institutional Assessment
DFID : Department for International Development
DICA : Domestic and Import Competition Adjusted
DGVM : Dynamical Global Vegetation Model
DNCA : Dunning index of Net Competitive Advantage
DSSAT : Decision Support System for Agrotechnology Transfer
EEC : European Economic Community
EA : East Africa
EAC : East African Community
EBA : Everything but Arms
EC : European Commission
EM-DAT : Emergency Disasters Database
xii
EPA : Economic Partnership Agreement
EPOPA : Export Promotion of Organic Products from Africa
EU : European Union
EU-GSP : European Union Generalised System of Preferences
FAO : Food and Agricultural Organisation
FiBL : Research Institute of Organic Agriculture
FPEAK : Fresh Produce Exporters Association of Kenya
FV : Fruits and Vegetables
GDP : Gross Domestic Product
GHG : Green House Gas
GLI : Grubel-Lloyd Index
GLOBALG.A.P: Global Good Agricultural Practices
GNI : Gross National Income
H-O : Heckscher-Ohlin
HS : Harmonised System
HT-test : Harris-Tzavalis test
ICA-PM : Import Competition-Adjusted Preference Margin
ICTSD : International Centre for Trade and Sustainable Development
IFOAM : International Federation of Organic Agriculture Movement
IFPRI : International Food Policy Research Institute
IPPC : Intergovernmental Panel on Climate Change
IRS : Increasing Returns to Scale
ITC : International Trade Centre
LDCs : Less Developed Countries
LLC-test : Levin–Lin–Chu test
MFN : Most Favoured Nation
MRLs : Maximum Residue Levels
NAADS : National Agricultural Advisory Services
NBR : Negative Binomial Regression
NCCRS : National Climate Change Response Strategy
NECOFA : Network for Ecofarming in Africa
NEI : Net Export Index
NOGAMU : National Organic Agriculture Movement of Uganda
NOP : National Organic Program
xiii
OECD : Organisation for Economic Co-operation and Development
OLS : Ordinary Least Squares
PIP : Pesticides Initiative Programme
PM : Preference Margin
PPM : Potential Preference Margin
PRCA : Porter-adapted index of RCA
PRISM : Parameter-Elevation Regressions on Independent Slopes Model
PTA : Preferential Trade Agreement
RC : Revealed Competitiveness
RCA : Revealed Comparative Advantage
RD : Relative difference
RER : Real Exchange Rate
RMA : Relative Import Advantage
RPM : Relative Preference Margin
RSCA : Revealed Symmetric Comparative Advantage
RTA : Relative Trade Advantage
RXA : Relative Export Advantage
SIDA : Swedish International Development Co-operation Agency
SITC : Standard International Trade Classification
SPSS : Statistical Package for Social Scientists
SRES : Special Report on Emissions Scenarios
SSA : Sub-Saharan Africa
SSMI : Special Sensor Microwave Imager
TOL : Tolerance
TRAINS : Trade Analysis and Information System
TRQ : Tariff Rate Quotas
UAE : United Arab Emirates
UIA : Uganda Investment Authority
UK : United Kingdom
UN : United Nations
UNBS : Uganda National Bureau of Standards
UNCTAD : United Nations Conference on Trade and Development
UNDP : United Nations Development Programme
UNFCC : United Nations Framework Convention on Climate Change
xiv
USA : United States of America
USAID : United States Agency for International Development
VIF : Variance Inflation Factor
WBDI : World Bank Development Indicators
WUOGNET : Women of Uganda Network
ZIP : Zero Inflated Poisson
xv
Abstract
With the aim of generating reliable information upon which appropriate decisions can be
based to benefit the various stakeholders, this research at one hand aims at developing a set of
meteorological indices, which are used as proxies to evaluate the impact of climate change on
horticultural trade flows to the European Union (EU) market. On the other hand, the study
examines the role of European Union's Generalised System of Preferences (EU-GSP scheme)
in boosting agricultural imports into the EU. Furthermore, the study assesses the export
competitiveness of various horticultural commodities of East African states within the EU
market, as well as exploring East Africa's trade potential and performance of the selected
commodities within the EU.
Various techniques were used to attain the above objectives. Such techniques include;
Balassa's Revealed Comparative Advantage (RCA) approach, the out-of sample technique,
the relative difference and absolute difference methods. To estimate the various gravity
models specified, a set of the extended Poisson models, viz: Zero Inflated Poisson (ZIP) and
Negative Binomial Regression (NBR) techniques for panel data estimations were employed
so as to deal with the excess zeros and over dispersion problems associated with highly
disaggregated data. Time series data for a period of 23 years (1988-2011) for 15 EU member
states and 3 East African states (Kenya, Tanzania and Uganda) were used for the analysis.
Data was obtained from various sources such as the TRAINS database, World Bank
Development Indicators, African Growth and Development Policy Modeling Consortium
(AGRODEP) database, Food and Agriculture Organisation (FAO) database, and TYN CY
1.11 database provided by the Tyndall Centre for Climate Change Research.
Some of the key empirical findings decomposed at country level reveal that:
 Kenya has export competitiveness in Asparagus, Mushrooms and truffles. Uganda
exhibits competitiveness in exporting pepper, bananas and eggplants while for Tanzania,
vegetables were the most competitive. Therefore, each of these countries should put
much emphasis on producing and exporting commodities over which she has
comparative advantage.
xvi
 Climate change generally has both positive and negative effects on horticultural trade
flows into the EU-Market, depending on the kind of proxy being put into consideration.
Within the EU market, anomalies in precipitation enhance horticultural imports from
East Africa while temperature anomalies tend to hinder trade. Anomalies in temperature
in exporting countries boost horticultural trade flows from Tanzania and Uganda while
the contrary is true for Kenya. Precipitation anomalies in exporting countries favor
horticultural trade flows from Kenya while they curtail trade flows from Tanzania and
Uganda. Thus, results imply that the use of anomalies as proxies for climate change in
agrarian based economies provides a more reliable measure of the effects of climate
change in trade than using the generalized Kyoto Protocol policies.
 The EU-GSP scheme selectively favors importation of certain horticultural commodities
into the EU-market, depending on the country of origin. It promotes importation of
bananas, beans and peppers from Uganda and beans from Tanzania. On the contrary, it
deters asparagus and bean imports from Kenya. Given that the findings concur with
findings of other scholars, it is imperative to argue that the use of preference margin,
based on all policy instruments (tariff rates, MFN, specific duties and Tariff Rate
Quotas) embedded within the EU-GSP scheme provides apt commodity specific
inferences regarding the effect of the EU-GSP scheme on horticultural imports into the
EU-market.
 Kenya and Uganda exhibit existence of un realised trade potential within the EU market.
For Kenya, asparagus has room for further market expansion across all EU-member
states while Uganda's beans and pepper can further be imported many EU member
states like France, Germany, Luxembourg, Portugal and Greece, among others. A
similar scenario applies to beans from Tanzania. This implies there is still have room to
expand East Africa's horticultural trade within the EU-market.
 The three East African states evidently exhibit poor trade performance within the EU-
market in the various commodities. This suggests that there exists some barriers to trade
which limit the proliferation of East Africa's horticultural imports into the EU. Thus, it
is incumbent upon East African states to foster cooperation in horticultural trade with
the EU member states..
xvii
Conclusively, it is commendable that anomalies in temperature and precipitation may be used
as climate change proxies, particularly when evaluating the impact of climate change on
international trade skewed towards agricultural commodities rather than using other based on
Kyoto Protocol policies. It is also recommended that assessment of the influence of non-
reciprocal preferential trade agreement(s) granted to developing countries, based on
preference margins should always take into account all the policy instruments embedded
within the agreement.
xviii
xix
CHAPTER ONE
1.1 Background
Export-driven growth of horticulture has been impressive in a number of countries in Sub-
Saharan Africa (SSA) and the involvement of small-scale growers in the production of fruits
and vegetables, which are exported mainly to the European Union (EU), has contributed to
poverty alleviation and rural development (UNCTAD, 2008). According to Minot and Ngigi
(2004), horticulture has at times been referred to as an “African success story”. In particular,
exports of fresh fruits and vegetables have seen high growth rates and better prices, as
compared with Africa’s traditional agricultural exports (FAO, 2004). In countries such as
Kenya, the subsector has attracted considerable participation of smallholder growers in
production for export. The EU is the key destination market for fruits and vegetables from
East African countries. For instance, the value of Uganda’s horticulture exports to the EU
increased by more than fivefold, from $1.5 million in 1996 to over $8 million in 2006
(UNCTAD, 2008).
The fruit and vegetable exports to the EU mainly go to wholesale markets in the United
Kingdom and to small supermarkets in the Netherlands. In Uganda, the main fruit exports
include off-season fruits (like citrus fruit and pears), major tropical fruits (like bananas,
pineapples, avocados, mangoes and papayas) and other fruits, such as passion fruit.
Furthermore, the major vegetable exports are beans, peas, green chillies (cayenne) and hot
peppers (Scotch Bonnet), among others. The leading Kenyan vegetable exports are French
beans, mixed vegetables, runner beans, okra snow peas and “Asian vegetables”, while the key
fresh fruit exports include avocados, mangoes, passion fruit and pine apples (UNCTAD,
2008).
According to Petriccione et al. (2011), imports into the EU market for fruits and vegetables
are subject to two types of duties, viz, the ad-valorem duties and specific duties. In addition,
the EU largely categorizes a majority of the products as being sensitive which are thus
subjected to a special entry price system. This is aimed at ensuring price stability and to
prevent very cheap products entering the European market. With this approach, each product
is accorded a trigger price such that when the import price surpasses this threshold, a specific
duty is applied. However, when the import price is less than this trigger price, the
commodity is then levied both the specific and the ad-valorem duty. More often than not, the
1
commensurate value of the specific duty is equivalent to the difference between the import
price and the trigger price.
However, in a scenario where the import price is lower than 92 % of the trigger price, the
specific duty is then fixed and equals to the maximum specific duty as specified by the EU.
The EU market also employs a mechanism of altering tariff levels of fruits and vegetables
within a calendar year. This is probably aimed at favouring EU’s production calendar,
thereby protecting the domestic producers within the EU market. In most cases, altering of
tariffs arises during harvesting periods which coincide with the northern hemisphere winter
season. For instance, Uganda Investment Authority (UIA) (2001) notes that the November to
February harvesting period in Uganda coincides with the winter season in Europe and during
this period, the demand for fresh fruits and vegetables is relatively low.
As with many other parts of the world where climate change has become a critical
predicament, Sub-Saharan Africa is not exceptional. Globally, climate change has been
distinguished as one of the major challenges man is facing. Despite the fact that Less
Developed Countries (LDCs) have negligibly contributed to causing climate change, coupled
with their limited capacity to adapt, they have succumbed to its harshest impacts (Dinda,
2011). This phenomenon has led to melting glaciers, more precipitation, more and more
extreme weather events, and drastic changes in seasons.
According to Nelson et al. (2009), the hastening pace of climate change, coupled with global
population and income growth, is a threat to the agricultural sector, hence to food security
globally. Notably, increasing temperatures cause yield loss of desirable crops, while boosting
weed and pest proliferation. The variation in precipitation patterns enhances the likelihood of
short-run crop failures and long-run production declines (Nelson et al., 2009). On the other
hand, climate change can truly provide opportunities to re-design economic activities, for
instance through the formation of non-traditional production technologies and use of
enhanced technological developments.
1.2 Problem statement
International trade is a crucial mechanism for industrialization and sustainable economic
development. The gravity flow model has been used in various studies to evaluate how
various trade policy issues, such as the effects of openness of an economy or protectionist
policies and the merits of proposed regional trade arrangements (such as the Common Market
2
for East and Southern African (COMESA), European Economic Community (EEC), and East
African Community (EAC)), affect trade flows. Notably, the gravity flow model is at the
forefront in enhancing a better understanding of the determinants of a country’s / region’s
trade flows from an empirical point of view. The model broadens the horizons of a country’s
/ region’s trade policies (Deardorff, 1998; Eichengrean and Irwin, 1997; Luca and Vicarelli,
2004).
Despite the fact that a large volume of literature evaluates the role of trade agreements (for
instance, the European Union Generalised System of Preferences (EU-GSP Scheme)) in
enhancing trade, a majority of these studies (Nakakeeto et al., 2011; Teweldemedhin and Van
Schalkwyk, 2010; Korinek and Melatos, 2009; Martìnez-Zarzoso et al., 2009; Caporale et al.
2009; Naude and Saayman, 2005; Péridy, 2005) use a dummy variable to proxy for such
trade policies. On the contrary, scholars (Aielo and Damalia, 2009; Cardamone, 2009; 2007;
2011) argue that this approach does not adequately describe the trade preferences granted,
hence it can be misleading. In detail, the use of dummy variables is inadequate because; (i) it
also captures all other factors that are specific to the country-pair and concomitant to the
preferential trade agreements; (ii) it does not discriminate among different instruments
adopted for non-reciprocal preferential treatment; (iii) it does not recognize the level of trade
preferences and it does not capture the strength of preferential access. Thus, this traditional
approach does not allow for appropriate estimation of the effect of non-reciprocal preferential
treatment on trade flows.
In light of the above setbacks, the literature has drifted towards the use of a continuous
variable, generally referred to as the preference margin. However, the current literature
(Cipollina et al., 2013; Raimondi et al., 2011; Cirera et al., 2011; Cipollina and Salvatici,
2010; 2009; 2008; Philippidis et al., 2011; Emlinger et al., 2008) reveals that this continuous
variable is calculated basing on at least one of the policy instruments, viz, the tariff rate, the
Most Favoured Nation (MFN) rate, specific duties and tariff rate quota embedded within the
non-reciprocal preferential treatment (the EU-GSP scheme). None of the studies uses a
combination of all the policy instruments, yet ignoring any of them jeopardizes the true value
of the preferential margin. Thus, the existing approach (preferential margin) used to proxy
the role of the trade policies, particularly in the EU-GSP scheme, under the gravity model
framework does not allow for appropriate estimation of the effect of the non-reciprocal
preferential treatment granted by the EU.
3
Additionally, significant progress has recently been made in terms of quantifying the effects
of climate change on international trade flows, thus leading to a better understanding of the
associated barriers it imposes on doing business. However, this advancement in academic
research has led to various measures, such as greenhouse gas emissions, environmental
permits, regulations, directives, emissions trading certificates, and tradable renewable energy
certificates, being used to proxy climate change. For instance, the World Bank (2008) used
carbon/energy tax and energy efficiency standards to study the impact of climate change on
the exports of OECD countries. Climate change proxies, such as the carbon tax and
greenhouse gas emissions used in capturing climate change effects among developed
(industrial) countries, are less reliable, especially in the context of developing regions like
East Africa (EA), given that the composition of their exports are skewed towards agriculture
(Hoekman and Nicita, 2011; Bineau and Montalbano, 2011), which is directly influenced by
consequences of weather-related natural factors, such as temperature, rainfall, cloud cover
and humidity, among other climatic factors.
Specifically, Melo and Mathys (2010) mention that measuring greenhouse gas in the
agriculture sector is very difficult, thus complicating the actual quantification of the effects of
climate changes on agricultural trade. According to Bineau and Montalbano (2011), this is
compelling developing countries to substitute machinery of poor energy efficiency with
modern machinery that is energy efficient, so as to catch-up with industrialization. Notably,
given that this transition is unprecedented and requires heavy initial investment costs, the
United Nations (UN) (2009) asserts that this is the major obstacle in curbing climate change
effects. The World Bank (2008) reveals that most of these climate change measures do not
directly target any particular product, but rather focus on the method by which greenhouse
gases may implicitly be related to production.
Therefore, climate-related policies based on those measures may have implications for trade
(Bineau and Montalbano, 2011), especially in agricultural commodities. Better measures
should be based on temperature, precipitation, humidity and other weather-related factors
since these directly affect the agricultural sector. The most plausible way to assess climate
change effects on the architecture of international agricultural trade is to redefine the proxy
measures of climate change, which can be easily and directly linked to agriculture.
Therefore, the current approach employed to model climate change effects on trade does not
appropriately reflect how this phenomenon influences trade in agricultural commodities. So,
evaluation of the influence of climate change on trade in agricultural commodities should be
4
based on variables that directly relate to the agriculture sector, which are temperature and
precipitation.
1.3 Objectives of the study
The overall objective is to develop and illustrate an improved methodology for evaluating the
impact of climate change on international trade in agricultural commodities by using climate
change proxies based on meteorological data; and to provide empirical evidence on the
relationship between the European Union non-reciprocal preferential trade agreement and
agricultural trade flows. Successfully achieving this objective will enhance the making of
informed trade related and climate change adaptation policy decisions. This will enable the
realization of the full trade potential of the East African States.
The overall objective will be met through the following sub-objectives:
Sub-objective 1
To determine the export competitiveness of East Africa’s fruit and vegetable exports within
the European Union market. The identified horticultural commodities for each country will
then be used to demonstrate how climate change and preferential treatment affect
international trade in agricultural commodities.
This objective will be attained by using the index of Revealed Comparative Advantage
(RCA). This index measures the export competitiveness in a given horticultural product by
beneficiaries of the trade agreement relative to other countries of the world. The RCA uses
actual trade flows to ascertain the competitiveness of exporters in fruit and vegetable
products. Attainment of this objective will enable EA states to identify the fruit and vegetable
commodities over which they have export competitiveness. This implies that if such
economies allocate adequate resources to these commodities, more benefits could be realized
instead of thinly spreading limited resources over a wide spectrum of products.
Sub-objective 2
To investigate the effects of a developed set of climate change proxies, based on
meteorological data, on international trade by using panel estimation techniques. As an
alternative to climate change proxies based on Kyoto Protocol policies, such as the carbon tax
and energy efficiency standards, anomalies in temperature and precipitation will be
developed and used as proxies for climate change. This set of climate change variables will
5
then be incorporated into the gravity model and run using the family of Poisson model
estimators.
Sub-objective 3
To determine the effect of the EU-GSP preferential trade agreement on East Africa’s fruit and
vegetable imports into the European Union market.
Unlike other scholars who use dummy variables, preferential treatment will be measured at
HS 6-Digit level as a continuous variable (absolute preference margin), while following
Cardamone (2011). The absolute difference will be measured as the difference between the
trade-weighted applied MFN rate and the Ad Valorem Equivalents (AVEs). The
computation of the preference margin that is employed in this study differs from Cardamone's
in two aspects: (i) the reference tariff, viz, the trade-weighted applied MFN rate, takes into
consideration competition within the EU market, and (ii) the preferential tariff (AVEs)
accounts for all the policy instruments (tariff rates, MFN, specific duties and Tariff Rate
Quotas) embedded within the EU-GSP scheme. The obtained preference margin per selected
horticultural commodity, at a given time, will then be used as the variable within the
augmented gravity model framework to run the family of Poisson model estimators to predict
the effect of the non-reciprocal preferential treatment.
Sub-objective 4
To predict East Africa’s unilateral trade potential and performance.
This study will employ the out of sample approach to predict East Africa’s potential unilateral
trade flows. With this approach, the exact parameters estimated by the gravity flow model
will be used to project the “natural” trade relations between the trading partners, such that the
difference between the actual and predicted trade flows represent the un-exhausted export
potential (Wang and Winters, 1992; Hamilton and Winters, 1992; and Brulhart and Kelly,
1999).
Realization of this objective will enable each East African state to comprehend the level of its
trade with the EU at commodity level. Succinctly, this will enhance the ascertainment of
how much more of the selected fruits and vegetables need to be exported to the EU market so
as to fully benefit from the non-reciprocal preferential trade. Furthermore, accomplishment
6
of this objective will enable the identification in detail of the specific EU member states with
which East African states have room for trade expansion with respect to particular
commodities.
Following Lie et al., (2002) and Amita (2004), trade performance will be analysed using two
indices, that is, the Relative Difference (Rd) and Absolute Difference (Ad). Although Rd can
be a convenient index to describe the relative relation of actual and simulated trade volume, it
does not explain the deviation volumes between them. However, use of Ad enables
computation of the gain or owned trade potential value, hence identifying the future trade
partner of the exporting country (Chen et al., 2007). All in all, the study uses the Absolute
difference index to cross check findings obtained while employing the Relative Difference
index.
1.4 Lay out of the study
The subsequent chapters are organized as follows. In Chapter Two, relevant literature
relating to export competitiveness and preferential treatment (EU-GSP Scheme), as well as
the agriculture-climate change nexus and how it affects trade, are discussed in detail.
Furthermore, a literature review of trade potential and trade performance is also presented in
this chapter. Chapter Three presents an overview of the horticulture sector in Kenya,
Tanzania and Uganda. In Chapter Four, a brief overview about the gravity model, the study
area, data and data management procedures, and the data sources, as well as the estimation
techniques used to achieve the set objectives, are discussed. Detailed results and discussions
of the results of each objective are presented in Chapter Five. Lastly, Chapter Six provides
the conclusions with regard to the objectives and recommendations generated from the results
of the study.
7
CHAPTER TWO:
LITERATURE REVIEW
2.1 Introduction
This chapter presents relevant literature relating to the concept of, and measures for
evaluating, export competitiveness, the impact of the non-reciprocal EU-GSP preferential
trade agreement on international trade, and the predicting of trade potential and trade
performance. The purpose of this review is to ascertain what other scholars have done with
regard to the above-mentioned aspects so as to establish the existing knowledge gap(s). The
identified knowledge gap(s) will then be addressed through this research. At the end of each
main sub-section, limitations and probable means of improving the existing pool of
knowledge are highlighted.
2.2 The concept of competiveness and its measures
Despite the fact that the concept of competitiveness is well known in economics, no
definition based on economic theory exists (Latruffe, 2010; Sharples, 1990; Ahearn et al.,
1990), and previous studies have adopted definitions depending on the context of analysis, as
well as the measurement approach to be used. According to the Eurostat's Concepts and
Definitions Database (CODED), competitiveness refers to “The ability of companies,
industries, regions or supranational regions to generate while being and remaining exposed to
international competition, relatively high factor income and factor employment levels on a
sustainable basis.” On the other hand, the International Institute for Management
Development (IMD) (2009) defines export competitiveness as the country’s ability to create
and maintain a suitable environment that can sustain more value creation for its enterprises
and increased prosperity for its populace. According to the European Commission (EC)
(2009), competitiveness refers to “a sustained rise in the standards of living of a nation or
region and as low a level of involuntary unemployment as possible”.
In the scientific literature, more often than not, the concept of competitiveness is used to
assess a region's or a country's macroeconomic performance by comparing a number of key
economic features that may influence international trade flows. Theoretically, scholars
8
(Meiliene and Snieska, 2010; Saboniene, 2009; Anisimovaite and Marcisauskiene, 2008)
argue that a country’s export competitiveness for a given product depends on the concept of
comparative advantage. That is, a country has increasing competiveness if it exhibits an
increase in exports, a rise in particular exports in the external markets, or an increase in
revenues and variety within the exports. Therefore, for the purpose of this study, export
competitiveness is defined as the proportionate share of a country’s products in the world
markets (Michael et al., 2008).
It is worthwhile to mention that the literature presented in this sub-section does not claim to
be exhaustive in reviewing all possible measures of competitiveness. However, it showcases
a general insight into the most often used measures of analysing competitiveness. According
to Latruffe (2010), competitiveness measures can broadly be categorized into two, viz, trade-
related measures of competitiveness and strategic management measures of competitiveness.
Trade related measures of competitiveness are grounded in neoclassical economics and they
employ real exchange rates, comparative advantage indices, and import or export indices. On
the other hand, strategic management measures are defined as those measures that dwell
much on the firm's structure and strategy. Their relevance was first brought to light by Porter
(1990), when he proposed "the diamond model". According to Kleynhans (2003), the model
provides an insight into the determinants of export competitiveness of firms and it is founded
on demand conditions, factor conditions, and related firms, as well as firm strategy, structure
and rivalry (Porter, 1998).
Within this framework, commonly used measures under this category are further subdivided
into cost measures (Domestic Resource Costs ratio, Social cost-benefit ratio, cost of
production); profitability measures (gross margins, cost to revenue ratio, value added to
sales); and productivity and efficiency measures (total factor productivity, growth of labour
productivity, technical efficiency, allocative efficiency) among others. However, given that
this study focuses more on trade, strategic management measures are not discussed in detail.
Particular emphasis is accorded to trade related measures. Because most of the trade related
measures are based on the concept of comparative advantage, it is prudent that this concept
be introduced first and then followed by the trade related measures.
9
2.2.1 The concept and theoretical framework of Comparative Advantage
One of the most firmly established ideas in economics is that a country's or a region's
competitiveness depends on its comparative advantage. The concept of Comparative
Advantage was first described by David Ricardo in the 1800s (Esterhuizen, 2006) in his book
entitled "On the principles of political economy and taxation" but the concept was later
refined and popularized by Balassa (1965). According to Balassa's (1965) index,
comparative advantage is revealed through the assessment of actual commodity trade patterns
on the assumption that the real exchange of goods and services depicts the relative costs and
the divergences in factors that may not easily be quantified in monetary terms. This index
has been widely used to identify international trade related patterns across borders in an effort
to determine a country’s export competitiveness position.
There are two popular trade related theories that can be used to explain the concept of
comparative advantage, viz, the Ricardian theory and the Heckscher-Ohlin (H-O) theory.
According to the Ricardian theory, it is assumed that comparative advantage is the result of
technological differences across countries, while on the other hand, the H-O theory opines
that comparative advantage is attributable to differences in production costs across countries.
The H-O theory further argues that all countries are technologically indifferent. Therefore, a
country is expected to export goods produced by its reasonably abundant factors of
production and to import goods that are intensive in the rather scarce factors.
As an example in support of the H-O theory, Utkulu and Dilek (2004) assert that many non-
industrialized nations are skewed towards producing primary products rather than
manufacturing products because they have land and labour in abundance but are constrained
by capital, education and technology. However, according to Balance (1988), the major
limitation of this theory is that the concept of comparative advantage is expressed in non-
concrete terms, based on relative prices which hypothetically prevail in a completely closed
economy.
The H-O theory is associated with two problems: (i) it is practically difficult to quantify
comparative advantage, given that all nations to some extent take part in international trade;
(ii) hardly any facts on autarkic prices can be accessed (Balassa, 1989; Utkulu and Dilek,
2004). That is, the prices for specific commodities cannot be observed in ex-post trade
equilibrium, thus consequently rendering use of this theory in estimating comparative
advantage challenging.
10
Acknowledging the above-mentioned drawbacks of the H-O theory, Balassa (1965)
developed and popularized the “Revealed” Comparative Advantage (RCA) index, which is
based on Ricardian theory. Balassa (1965; 1977) noted that the index differs from that based
on Heckscher-Ohlin theory in that it is assumed that a nation’s comparative advantage is
“revealed” in its observed trade patterns, rather than in focusing on factors that determine
comparative advantage.
2.3 Trade related measures of competitiveness
2.3.1 The Revealed Comparative Advantage (RCA) and its adjusted indicators
The index is computed as:
 *
RCA   x
  
ik   xwk 
k  X   *  .................................................... (1) ip   X 
Where the variables xik and Xip denote the value of exports of product k from country i and
total exports (p) from country i, respectively. The variables x*wk and X* represent the value
of world exports of product k and total world exports, respectively. Thus, a country is said to
have a revealed comparative advantage in commodity i, if (xik / Xip) > (x*wk / X*) (Kulapa et
al., 2013; Török and Jámbor, 2013; Athanasoglou et al., 2010; Latruffe, 2010).
At this point, commodity i’s export market share is greater than the country’s total export
market share, hence implying that the country is competitive in exporting commodity i. The
major limitation of this index centres on the fact that its value is asymmetric, viz, for
commodities that register comparative advantage, the RCA value ranges from one to infinity,
while for those commodities regarded as comparatively disadvantage, the index starts from
zero and stops at one (Mirzaei et al., 2006).
The definition of RCA has been revised and modified (Kunimoto, 1977; Bowen, 1983; and
Vollrath, 1987, 1989 and 1991). For instance, Vollrath (1987; 1989; and 1991) introduced
three alternative RCA indices, that is, Relative Trade Advantage (RTA), the logarithm of
relative Export Advantage (lnRXA) and Revealed Competitiveness (RC). These different
modifications of Balassa's index were set to measure RCA at different levels, that is, at global
level, at regional or sub-regional level and others to limit the analysis to trade flows between
only two trading partners (Fertő and Hubbard, 2001; 2002). By definition, the RTA index
11
refers to the difference between Relative Export Advantage (RXA) and the Relative Import
Advantage (RMA). This index (RTA) accounts for both imports and exports. Conspicuously,
it is postulated that RXA is the same as the commonly used Balassa’s index.
Following the work of Fertő and Hubbard (2002), Utkulu and Dilek (2004), and Vollrath
(1991), RTA is computed as:
i  i iRTA RXA  RMA .......................................................................... (2)a a a
where RXA and RMA denote relative export advantage and relative import advantage,
respectively. The disaggregated indices are obtained as:
 i
 X a 
 i 
i
RXA   X n  .......................................................(3) r
 X a 
 rX  n 
 i
 M a 
 i 
i
RMA   M n  ..................................................(4) r
 M a 
 rM  n 
where superscript r refers to the world without country i, while subscripts a and n refer to the
commodity of concern and all traded commodities minus commodity a, respectively. In the
case of the RMA index, a value of less than one implies revealed comparative advantage,
hence a country is said to be competitive in that particular product. It is worthwhile to note
that the lnRXA index and Revealed Competitiveness Index can ably overcome the
asymmetric problem associated with Balassa’s index (Fertő and Hubbard, 2002; and Utkulu
and Dilek, 2004). Positive index values imply that the country has comparative advantage,
thus being competitive in exporting that particular commodity, while negative values denote
competitive disadvantage.
The logarithm of relative Export Advantage (lnRXA) is defined as the natural logarithm of
the commonly used Balassa index. That is:
12
 i 
 X a 
 i 
i
ln RXA  ln  X n  .................................(5) r 
 X a 
 r 
 X n 
The third index, Revealed Competitiveness (RC), refers to the difference between the natural
logarithms of Balassa’s index and the relative import advantage index. Mathematically, RC
is expressed as:
i i i
RC  ln RXA  ln RMA ................................................................ (6)a a a
2.3.2 The Revealed Symmetric Comparative Advantage (RSCA)
The RSCA index, developed by Dalum et al. (1998) and Laursen (1998), is a simple
decreasing monotonic transformation of Balassa's index. According to Nwachuku et al.
(2010), the introduction of this index was aimed at controlling the asymmetry problem
associated with the original Balassa index. Mathematically, it is expressed as:
RSCA  RCA1 RCA1 ......................................................(7)
where RCA is Balassa's index. The index ranges between −1 and +1, and a country is said to
exhibit higher competitiveness in exporting a particular commodity if the values tends
towards +1.
2.3.3 The Porter-adapted index of RCA (PRCA) and Dunning index of net competitive
advantage index (DNCA)
In order to account for production by a firm in foreign countries, Pitts and Lagnevik (1998)
argued that the RCA index should be adjusted and two indices were developed by Porter and
Dunning, henceforth, referred to as Porter-adapted index of RCA (PRCA) and Dunning index
of net competitive advantage index (DNCA). In practice, the PRCA index is founded on the
assumption that national firms that produce abroad retain their country of origin as their home
base. Thus, all production generated abroad by these firms is treated as exports of the
country from which they originate, and hence added to exports. On the contrary, the DNCA
index deducts all production by foreign firms from total exports (Latruffe, 2010).
Symmetrically, these indices are expressed as:
13
X     PRCA  ic Y ic X ik Y ik    .............................................................(8)X  nc Y nc X nk Y nk
where Yic denotes outbound production. This is the value of output c produced by firms of
country i in foreign countries. n denotes all countries other than i.
X Y    DNCA  ic ic M ic Pic  ..............................................................(9)X  ic Y ic Pic
where Pic denotes inbound production, viz, the value of output c produced by foreign firms
operating within country i. X and M denote exports and imports, respectively.
2.3.4 The Net Export Index (NEI)
Banterle and Carraresi (2007) and Latruffe (2010) define the NEI as the ratio of the
difference between a country's or sector's exports and imports to the total value of trade by
that country or sector. Mathematically, it can be expressed as:
X M 
NEI  ij ij   ..........................................................................................(10)X ij M ij
where X represents exports; M symbolizes imports; while subscripts j and i denote a sector or
commodity and the country under consideration, respectively. The index assumes a negative
value of (-1) if the country/sector is a net importer; this implies negative competitiveness,
while a positive value implies increasing competitiveness in exporting that particular good.
2.3.5 The Grubel-Lloyd index (GLI)
The GLI was proposed by Grubel and Lloyd (GL) [1971]. It takes into consideration the fact
that products are often exported and imported during the same period. It is computed as
follows:

 X M ij ijGLI 1ij    ...............................................................................(11) X ij M ij 
where X represents exports; M symbolizes imports; while subscripts j and i denote a sector or
commodity and the country under consideration, respectively. Index values range from 0 to
14
1. A value of 0 implies the country is undertaking inter-industry trade, while a value of 1
depicts intra-industry trade flows. That is, exports are equal to imports (Latruffe, 2010).
2.3.6 The export to import price ratio
According to Bojnec (2003), this is the ratio of the unit value per ton of exported product to
the unit value per ton of imported product. Values greater than one (1) imply that the exports
represent goods of higher quality, as they command a higher price than the imports and vice
versa. Going without saying, the reverse is true.
2.3.7 The Real exchange rate (RER)
The RER index is defined as the ratio of the price index of tradable goods to the price index
of non-tradable goods (Latruffe, 2010). It is expressed as:
T
RER  P
T * ...............................................................................................(12)
P
where PT and PT* denote the price index of tradable goods and the price index of non-tradable
goods, respectively.
As put by Fertő and Hubbard (2002), the major drawback of all indices based on comparative
advantage is that they can be misleading if the underlying comparative advantage is
misrepresented, especially in the presence of government policies and interventions which
tend to distort actual trade flow patterns. Government interventions and policies, such as
export subsidies and import restrictions, may distort trade.
There are a number of RCA indices that can be employed to assess a country’s export
competitiveness (Yilmaz, 2002; Akgüngör et al., 2002; and Lohrmann, 2000). However,
scholars (Bruneckiene and Paltanaviciene, 2012; and Fertő and Hubbard, 2002) assert that,
“There is no common scientific approach regarding the most efficient measure of export
competitiveness, or reliable indicators, able to reflect the country’s export competitiveness
position at the international level.” Secondly, despite the fact that Krugman (1994) disputes
the use of this concept of comparative advantage, especially while measuring competitiveness
at national level, it remains the most common basis for measuring export competitiveness
(Palit and Nawani, 2012; Gilbert, 2010).
15
Thirdly, researchers such as Vollrath (1991) and Fertő and Hubbard (2002) note that in
instances of government intervention, Balassa's index is less susceptible to policy-induced
distortions in trade flow patterns, given that the computation of the index relies only on
export data. Furthermore, it is argued that trade flow distortions are more evident at the
import side than at the exporters' side. Therefore, taking the above considerations into
account and given that this study is based on highly disaggregated data, which Capalbo et al.
(1990) argue should be the basis for measuring competitiveness, it can be posited that
Balassa’s index be accepted as an appropriate measure of export competitiveness.
2.4 Empirical evidence of competitiveness studies based on RCA methodology in the
agriculture sector
The literature on the competitiveness of individual commodities, as well as the agricultural
sector as a whole, is addressed here, first for non-African economies and then for Africa.
Literature focussing on specific agricultural sectors, specifically the fruits and vegetable
sector, is very scanty, especially for the African economies. Most studies either deal with
entire sectors within an economy or focus on sectors, such as manufacturing.
2.4.1 Empirical studies of non-African economies (Rest of the world)
Akgüngör et al. (2002) measured the competitiveness of Turkey's tomato, grape, and citrus
fruit processing industry exports to the EU market. Empirical results showed that Turkey's
competitive power was higher than that of Spain and Portugal in processed grape exports, and
was higher than Greece and Portugal in citrus fruit exports. The results further revealed that
Turkey had a competitive disadvantage in exporting processed tomato products.
Utkulu and Dilek (2004) analysed the competitiveness of Turkey’s agricultural exports within
the EU market, using time series data from 1990 to 2003. The results showed that Turkey
was competitive in its many exports, fruits and vegetables included, within the EU market,
while some sectors registered a comparative disadvantage. It is worthwhile to note that
although the fruits and vegetables sector presented the highest RCA values, the results were
unstable, given that their level of comparative advantage was on a declining trend.
Carraresi and Banterle (2008) investigated the competitiveness of the agric-food and
agricultural sectors in European Union (EU) countries during the 1991- 2006 period, using a
number of RCA indices. Their findings revealed a mixed level of competitiveness across the
16
countries. For instance, Denmark, France, Greece, Ireland, Luxembourg, the Netherlands and
the United Kingdom exhibited a declining trend in export competitiveness, while countries
such as Belgium, Finland and Portugal registered increasing competitiveness in the
agriculture sector. Germany, Italy, Spain and Sweden revealed increasing competitiveness
throughout the entire period.
Palit and Nawani (2012) used Balassa’s Index (RCA) for individual sector groupings to
measure the competitiveness of Indian exports to China for the period 2004–06. Their
findings reveal that India is more competitive in the Chinese market, relative to other
Southeast Asian economies in some product categories such as vegetable products and food
preparations.
With the aim of examining the export competitiveness of the canned tuna export industry in
Thailand between 1996–2006, Kulapa et al. (2013) employed Balassa’s index approach to
estimate RCA indices for both exporters (Ecuador, Spain, the Seychelles, Mauritius,
Indonesia, and the Philippines) in the world market and for contenders in individual export
markets. Despite the fact that empirical results show that Thailand’s comparative advantage
deteriorated markedly in Australia, it still commands high and stable comparative advantage
in all major export markets, such as the United States of America (USA), the Middle East,
Japan and Canada.
Török and Jámbor (2013) analysed the competitiveness of fruit spirits in six Central and
Eastern European countries (CEECs) following the enlargement of the EU market. With the
exception of Hungary and Poland, their findings show that all countries were competitive in
the EU-15 beverages market. The authors argue that despite the weakening drift in
competitiveness since the EU accession, CEEC fruit spirits were equally competitive and had
a comparative advantage in the EU-15 beverages market.
2.4.2 Empirical studies of some African economies
With the exception of Laibuni et al., (2012), Shinyekwa and Othieno (2011), Sebaggala
(2008), and Esterhuizen and Van Rooyen (2000), there has been limited research using the
RCA index in the East African region. For instance, Esterhuizen and Van Rooyen (2000)
investigated the competitiveness of Rwanda's agricultural exports for the period 1990–99.
Their study applied the adjusted Balassa index and the results revealed that Rwanda's
agricultural sector was competitive in exporting beans, coffee, tea and frozen vegetables,
17
among other products. Commodities such as maize, sugar and beer were positioned at a
competitive disadvantage.
Sebaggala (2008) assessed the competitiveness of Uganda’s exports to the rest world over a
period of two years (2000 and 2005) while using aggregated data at SITC Revision 1.
Empirical results showed that Uganda was generally competitive in food and live animal
exports. At sub-sector level, fruits and vegetables exhibited a low level of export
competitiveness.
Shinyekwa and Othieno (2011) evaluated the competitiveness of Uganda’s exports relative to
the East African Community (EAC) member nations. The authors used various indices to
measure Uganda’s revealed comparative advantage (RCA) on all products at Harmonized
System (HS)-4-digit product levels. The findings revealed that Uganda had an increasing
RCA, hence export competitiveness, in leguminous vegetables, shelled or unshelled, fresh or
chilled; frozen vegetables; pineapples, mangoes, avocadoes, guavas over Kenya; in manihoc,
arrowroot salem (yams) over Burundi and Rwanda; and in dried vegetables over all East
African states (Burundi, Kenya, Rwanda and Tanzania).
In order to ascertain the competitiveness of Morocco's fruit and vegetable sector exports to
the European Union over its trading partners, Pappalardo et al (2012) used the revealed
comparative advantage (RCA) approach for a period of 11 years (2000–2010). Empirical
findings showed that Morocco was competitive in the fruits and vegetable sector over its
major EU trade competitors. The most significant types of goods for which Morocco held a
global advantage over the EU included tomatoes, pulses; preserved vegetables; other
vegetables; melons, watermelons and papayas; and citrus.
Laibuni et al. (2012) used the International product specialization index to evaluate the export
competitiveness of Kenyan cut-flowers and fruits and vegetables in the EU-25 market. The
study used SITC-rev.3 disaggregated data and the empirical results indicated that Kenya’s
exports of flowers, fruits and vegetables were very competitive in the EU-25 market. Boansi
(2013) used Balassa's index and its derivative, the Revealed Symmetric Comparative
Advantage index, to assess the competitiveness of Ghana's cocoa exports during the 1960s,
1980s and 2000s. Study results showed that Ghana was more competitive in exporting cocoa
beans than cocoa processed products, especially during the 1960s. Esterhuizen (2006)
assessed the competitiveness of South Africa's agribusiness sector while using Balassa's
18
methodology. The results divulged that the agribusiness sector was marginally competitive,
relative to its competitors.
2.5 Climate change - Agriculture - International trade nexus
Climate scientists seem to have reached a consensus that the Earth’s climate will change at a
unique rate over the 21st century, especially in the form of global warming, with an estimated
temperature increase of 5.8 °C by 2050 (IPPC, 2007). The IPPC (2007) shows that the global
average temperature has increased by approximately 0.76 °C on average over the last 100–
150 years. It is postulated here that African countries which are largely reliant on agriculture
seem to be vulnerable to this phenomenon (Hope, 2009; Muller et al., 2011). With reference
to Sub-Saharan Africa, Traore et al. (2013) note that a temperature rise of about 3.3 °C is
anticipated within this region by the end of the 21st Century. However, it remains unclear
whether rainfall will increase or decrease within Sub-Saharan Africa. The various simulation
models used by the IPPC so far provide divergent results, depending on the scenario under
consideration (Cooper et al., 2008; Berg et al., 2013; Traore, et al., 2013).
According to Derksen and Jegou (2013), the nexus between climate change, agriculture and
trade consists of four categories: (i) when climate change physically distorts trade volumes
and trade patterns; (ii) through the effects of climate change policies on trade; (iii) through
the interactions of trade policies as a means of addressing climate change; and (iv) through
the effects of trade on climate change, especially via aircraft emissions. For the agricultural
sector, for instance, climate change fluctuations negatively alter the productive capacity of
firms during the production phase (Berg et al., 2013; Roudier et al., 2011).
Productivity is hampered through a number of aspects and this culminates in limited
availability of agricultural produce, hence hampering trade both at local and international
level in general. In this regard, the country’s or region’s export competitiveness and trade
patterns also change. Moreover, in cases of extreme weather catastrophes, like floods,
infrastructure necessary for trade is also adversely affected. Thus, in a bid for countries to
adjust and adapt to the alterations imposed by climate change, trade volumes and trade
patterns are also affected.
On the other hand, linkages between agriculture, climate change and trade can be explained
through policies that aim to mitigate the climate change phenomenon. For instance, Derksen
and Jegou (2013) mention that these policies can have both social and economic negative
19
impacts on trade, if not adequately designed and implemented. Such policies include carbon
taxes, national promotion of low-carbon technologies and clean energy, emissions trading
schemes, border carbon adjustments, standards and labelling schemes, the allocation of
emissions allowances free of charge, technical requirements, and the regulation of bunker
fuels.
With reference to the interaction of trade policies as a means of addressing climate change,
trade policies can act as drivers in containing the global problem of climate change. This
school of thought argues that this could be achieved through promotion of adaptation and
mitigation mechanisms. For instance, Derksen and Jegou (2013) remark that the removal of
trade barriers, especially on climate-smart goods, would inevitably promote climate change
adaptation and countries would be in position to curb Green House Gas (GHG) emissions.
Finally, through direct and indirect means, trade has been known to influence climate change,
especially through transport-related emissions. Transport is noted to be one of the major
components of trade, through which a significant level of GHGs are directly emitted
(Derksen and Jegou, 2013).
In agriculture both temperature and precipitation are key climate factors in influencing crop
productivity (Lobell and Field, 2007; Hansen, 2002; Alexandrov and Hoogenboom, 2001),
hence trade flows. For example, climate change may disrupt trade flows through a sudden
disaster, like floods, which may destroy crops and other facilities or through some gradual
changes to an ecosystem which also incapacitate production. According to Brockett et al.
(2005), the significance of temperature alone as a climatic factor across all sectors accounts
for over 90 per cent in influencing productivity, followed by rainfall, among others.
According to Drine (2011), climate change is responsible for low agricultural productivity,
given the fact that uncertainty inhibits innovation and imitation. In addition, it is argued that
uncertainty about agricultural production is bound to increase as severe climate events, such
as droughts and floods, are anticipated to recur more frequently and to cause more
catastrophes. Therefore, given the pervasive risky environmental effects on farming practices
and farm performance, the increasing uncertainty may perhaps dishearten farmers from
adopting new production technology, thus affecting productivity.
Marchiori et al. (2010), World Bank (2010), IPCC (2007), and Deschenes and Greenstone
(2007) argue that climate change has fewer detrimental effects on the manufacturing industry
than on the agricultural sector, which is vulnerable to this phenomenon. The adverse effects
20
are also more likely to be present among the poor economies which are reliant on agrarian
activities.
The direct effects of climate change may be exhibited in form of rural-urban migration
(Marchiori et al., 2010), which culminates in the reallocation of scarce labour from the
agricultural sector in rural areas to the non-agricultural sectors in urban areas. Although the
populace may have a genuine cause to change from one sector to another, Collier et al.(2008)
and Barrios et al. (2006) argue that this reallocation of labour is simply a mechanism of
adapting to climate change. As a result, this is likely to transform into reduced production on
farms, thereby causing deficits in agricultural produce, thus causing a dent in trade flows.
Given that many Sub-Sahara African (SSA) countries rely mainly on small-scale,
subsistence, rain-fed agriculture (i.e. farmers produce mainly for home consumption and only
sell in instances of surpluses), reduced farm production due to rural-urban migration will
inevitably curtail trade in agricultural commodities. The effect of rural-urban migration, thus
the reallocation of scarce labour, is actually greater in communities characterized by non-
functioning rural markets, like Uganda.
In instances where rural markets are functional, households affected by rural-urban migration
would probably be in position to hire labour to substitute for what would have been provided
by out-migrants on the farm. Alternatively, such households would also borrow money for
agro-inputs to boost production. However, labour and credit markets are not functional,
which affects farm production (Barrios et al., 2006), and hence trade flows also decline.
Ogang (2013) argues that accessing loans to finance agriculture related activities is very low
in Uganda, accounting for only 7 per cent of the total private sector credit.
According to Barrios et al. (2006), unpredictable rainfall has an extensive assortment of
commercial repercussions in developing economies, given that it is the main source of water.
For instance, water shortages are associated with detrimental effects such as hunger, and in
extreme cases, death. In Africa per se, variability in rainfall is important because of its
significance in the agricultural sector. In most of SSA, agriculture depends on rainfall to
provide crops with water, as only a small proportion of arable land is irrigated.
The productivity of various crops has been shown to reduce owing to variability in
temperature and precipitation (FAO, 2001; Kumar et al., 2004; Parry et al., 2004; Schlenker
and Lobell, 2010; Tao et al., 2003; 2008; Sivakumar et al., 2005; Xiong et al., 2007). In
21
Uganda for example, annual crops like maize and beans are more susceptible to climate
change than perennial crops, such as tea, coffee and bananas. Maize is generally most
sensitive to drought, while beans tend to be most sensitive to excessive rainfall (UNDP and
BCPR, 2013). The higher vulnerability of annual crops is attributable to the fact that intense
events can wipe out the annual crop, leaving farmers with no harvest, while perennial crops
might often survive, but with lower yields or reduced quality. Thus, these drastic changes in
climatic conditions can impact on the length of a crop’s growing period, and therefore yields,
among other aspects.
Temperature as a climatic factor also presents a number of effects on trade through various
avenues. According to Dell et al. (2008), temperature can also affect agriculture through its
effects on investments or institutions that influence productivity growth. These in the long
run affect a country’s economic activities, where trade is inclusive. For instance, higher
temperatures are known to lead to conflict and political insecurity in poor countries (Dell et
al., 2008; Field, 1992; Jacob et al., 2007; Miguel et al. 2004; Boyanowsky, 1999) and during
such periods of unrest, there is limited agricultural production.
Dello et al., (2008) goes further to show that a 1 °C increase in temperature in developing
economies leads to approximately 2.37 per cent loss in the growth of agricultural output.
This decline then affects the total GDP, which is a key determinant for trade according to the
gravity model theory. Furthermore, it is affirmed that for every increase in 100 mm of annual
precipitation, there is an accompanying 0.24 per cent increase in agricultural output growth in
developing countries, and a 0.14 per cent rise in agricultural output in developed countries.
2.5.1 A review selected empirical effects of temperature and precipitation as
determinants of agricultural productivity
McCandless et al. (2012) used an ecophysiological crop model called the Decision Support
System for Agrotechnology Transfer (DSSAT) to study the impacts of temperature and
precipitation on the yield of maize and bean crops in the Rakai and Kapchorwa districts of
Uganda. Their empirical results project that bean production in Kapchorwa district will
decline by approximately 6 per cent, while maize production may experience an 8 to 10 per
cent decrease by 2050. With regard to Uganda’s major cash crop (coffee), Simonett (1989)
shows that a 2 °C increase in temperature would lead to a significant fall in the production of
Robusta coffee in the country.
22
Berg et al. (2013) employed the agro-Dynamical Global Vegetation Model (DGVM) and two
SRES scenarios to simulate the impact of climate change on the productivity of C4 crops
over Africa and India from 1960 to 2100. In general, the empirical findings divulge that a
discernible yield decrease, ranging from -10 to -20 % is anticipated by the end of the century.
Moreover, the authors also mention that long-term impacts are more than twice those of the
short-term basis.
Traore, et al. (2013) analysed the effect of temperature and rainfall on the productivity of a
number of crops in Southern Mali using a dataset spanning from 1965 to 2005. Their
findings show that there was a declining trend in cotton yields, attributable to the unreliable
precipitation pattern. For instance, a 24 kg/ha yield loss of cotton was registered for every
0.08 °C increase of the maximum temperature during the rainy season.
2.5.2 A review of empirical studies relating to climate change effects on international
trade
For over a decade, the climate change phenomenon has attracted increasing attention at
various levels and a number of reports quantifying the economic effects of climate change in
Africa have been produced, for example by the World Bank and the Centre for
Environmental Economics and Policy in Africa (CEEPA). In the international trade domain
per se, scholars like Folfas et al. (2011), Aichele and Felbermayrz (2010), Kim and Koo
(2010), Kee et al. (2010), and McKibbin et al. (1998) have used one or a combination of
Green House Gas (GHG) emissions, environmental permits, environmental regulations and
permits, emission trading certificates and tradable renewable energy certificates to quantify
the effects of climate change on international trade. Therefore, the literature presented in this
sub-section is grouped according to the quantification measure, or combinations thereof,
used.
2.5.2.1 Literature based on policies that regulate Greenhouse Gas (GHG) emissions
According to the United Nations Framework Convention on Climate Change (UNFCC)
(2009), regulation of GHG emissions is based on two policies, viz, the carbon tax and the
cap-and-trade scheme. These policies were agreed upon under the first international
agreement on GHG emissions, the Kyoto Protocol, which took effect in February 2005.
However, owing to the limited access to comprehensive and comparable information about
countries' specific climate policies and how they relate to the ratification of the Kyoto
23
Protocol (Aichele and Felbermayrz, 2010), some studies simply use a binary variable of 1 if
both members of a country pair commit to the agreement.
For instance, Kim and Koo (2010) evaluated the impact of regulating greenhouse gas
emissions on livestock trade flows among member countries of the Organisation for
Economic Co-operation and Development (OECD). They used dummy variables if a given
country had enacted any GHG emission regulating policy. Their findings indicate that
regulating greenhouse gas emissions has a deterring effect on the flow of livestock from all
countries (regulating and non-regulating) into regulating countries.
Kee et al. (2010) investigated the effects of a carbon tax and energy efficiency standards on
competitiveness in trade of a number of industries among OECD countries. Their findings
show that a carbon tax, imposed by either importing or exporting countries, boosted trade
competitiveness in energy-intensive industries, while energy efficiency standards deter trade
competitiveness, irrespective of which country imposes them.
According to Ma and Keating (2011), the Australian Chamber of Commerce and Industry
(ACCI) argues that imposing a carbon tax would potentially curtail the Australian economy.
ACCI reckons that “The fact is that carbon tax will have a negative impact on all trade-
exposed industries which actually can't pass on the costs associated with a carbon price,
because they're competing internationally either through import or through exporting
competitions.”
The International Centre for Trade and Sustainable Development (ICTSD) (2007) notes that:
… the effects of some of the climate policies such as measures addressing energy
efficiency have resulted in several challenges for developing country exporters in
terms of being able to comply swiftly with changing and increasingly stringent
market access requirements. In the absence of a clear regulatory forum for
addressing these emerging tensions –as they relate both to the trade and climate
policy arenas –there is a fear that countries may increasingly recourse to
unilateral approaches through measures such as antidumping and border measures
in order to solve perceived competitiveness concerns.
24
2.5.2.2 Studies based on environmental permits and embodied carbon content or carbon
dioxide equivalent in traded products
Aichele and Felbermayrz (2010) assessed the influence of Kyoto policies on the bilateral
imports into countries that have ratified to the Kyoto Protocol. To quantify the GHG
emissions, they used the total carbon content embodied in imported goods. Their results
indicate that the policies had had non-negligible effects on the quantity of bilateral import
flows. Folfas et al. (2011) also assessed the impact of GHG emissions on trade flows from
steel and cement industries among developed economies. The findings show that countries
with low GHG emissions had intense export trade in steel products with economies
characterized by a high level of GHG emissions.
McKibbin et al. (1998) estimated the potential effects of the Kyoto Protocol policies
(particularly international permit trading) on international trade under different scenarios.
Their study used the G-cubed multi-region, multi-sector, inter-temporal general equilibrium
model of the world economy. Their findings reveal varying results, depending on the
scenario under consideration. For example, under the assumption that no other region other
than the USA meets its commitment under the Kyoto Protocol, exports of durable goods
would be negatively affected through the appreciation of the exchange rate. Similarly, under
the assumption that all countries within a given region impose the policy, durable export
flows from developing countries would decline as they become very expensive to produce,
unlike for developed economies. Overall, the results reveal that the USA, and to a lesser
extent Australia, would experience a decline in their exports of durable goods as a result of
the policy.
2.5.2.3 Literature based on meteorological data (temperature and precipitation)
Notably, only one study that used meteorological data has been found. Jones and Olken
(2010) examined the effects of temperature and precipitation on the annual growth rate of
exports between developing and developed economies. Their results reveal that, unlike in
developed economies, an increase in temperature negatively affects developing countries'
export flows, while precipitation fluctuations showed no significant deterrent effects in export
growth. A unit rise in temperature would cause a drop in a developing country's export
growth ranging between 2.0 and 5.7 percentage points.
25
2.6 The European Union's Generalized System of Preferences (EU-GSP) scheme
The GSP is an autonomous non-reciprocal trade arrangement through which the European
Union (EU) provides non-reciprocal preferential access to 176 developing countries and
territories into the EU market. The EU-GSP scheme was adopted following the second
United Nations Conference on Trade and Development (UNCTAD) held in 1968, during
which the idea of establishing a generalized, non-reciprocal, non-discriminatory system of
preferences in favour of the developing countries was presented (UNCTAD, 1968). The
initiative aimed at increasing export earnings, promoting industrialization and accelerating
the rates of economic growth among these countries.
The EU-GSP scheme was first introduced in 1971 and since then, it has evolved from time to
time, with the European Commission (EC) making changes in product coverage, tariff
treatment and differentiation among beneficiary countries. For instance, the EC has been
reducing its Most Favoured Nation (MFN) tariffs, thereby narrowing the preference margin
under the GSP Scheme. Between 1981 and1991, the GSP was reviewed annually and this
involved changes in product coverage, quotas, ceilings and their administration, beneficiaries
and depth of tariff cuts for agricultural products. Particularly, the 1981–91 scheme was
extended until early 1995 when another 10 year GSP Scheme was initiated (UNCTAD,
2001).
According to UNCTAD (2008) and European Communities (EC) (2001), the third phase of
GSP came into effect from 1 January 2002 until 31 December 2005. Changes that took
centre stage in the 3rd phase include, among others, the introduction of special incentive
arrangements for the protection of labour rights, special incentive arrangements for the
protection of the environment, special arrangements to combat drug production and
trafficking, and special arrangements for LDCs: the “Everything but Arms” initiative for the
Least Developed Countries (LDCs). On 27 June 2005, the subsequent EU-GSP scheme was
adopted and was enacted on 1 January 2006, to endure until 31 December 2008.
The number of arrangements under this phase was reduced from five to only three, viz, the
general arrangements, special incentive arrangements for sustainable development and good
governance (“GSP Plus”), and the Everything but Arms initiative for LDCs (EU, 2005). As
put by the EU (2008), the structure of this GSP scheme was meant to be extended to cover
the 2009–2011 period. The EU-GSP scheme has three key features, namely: tariff
26
modulation, country/sector graduation, and special incentive arrangements, among other
control structures, such as temporary withdrawal of scheme benefits and rules of origin. For
the purpose of this study, tariff modulation is accorded more attention.
Originally, the 1971 GSP Scheme accorded different tariff treatments to agricultural and non-
agricultural products. Agricultural commodities enjoyed selective preferential treatment until
1995 when all commodities modulated according to product sensitivity. At this time, four
product categories were established, viz, Very sensitive products were subjected to a
preferential tariff 85 % of the MFN rate; Sensitive products were accorded 70 %; while for
Semi-sensitive products, a 35 % preferential tariff was granted. Non-sensitive products were
subjected to duty-free entry into the EU market. However, by adopting EC (2001), the
foundation of the 2001 EU-GSP Scheme, preferential tariffs were then restructured basing on
two product groups, viz, sensitive and non-sensitive products. Thus, all non-sensitive
products were to enjoy duty-free entry into the EU market, except where the MFN tariff had
an agricultural component.
On other hand, with the exception of textile products, all sensitive products with ad valorem
duty were granted a reduction of 3.5 percentage points. Generally, sensitive products with
specific duties were subjected to a 30 per cent reduction, while for sensitive products with
mixed tariffs, the specific duty component was not reduced (UNCTAD, 2001). Interestingly,
the 2009–2011 GSP Scheme retained the basic features of the 2001 scheme with regard to
tariff treatment, though it was adjusted by differentiating the beneficiaries into three
categories. That is, the general GSP beneficiaries; the ‘GSP Plus’ scheme specifically for
vulnerable countries with special development needs; and the Everything But Arms (EBA)
initiative. Commodities from the ‘GSP Plus’ scheme beneficiaries enter the EU market duty-
free, while beneficiaries of the EBA initiative are granted duty-free access to the EU market
without any restrictions.
2.6.1 The theoretical framework of preferential treatment effects on international
trade
In basic terms, the analytical framework of preferential tariffs is presented as a partial
equilibrium model of three country groupings and one traded good. Following the work of
Low et al. (2005), this theoretical framework is based on two general assumptions, (i) That
the preference-receiving country group is not the most competitively advantaged producer of
the traded commodity for which preferential treatment is granted, and (ii) that the initial Most
27
Favoured Nation (MFN) tariff rate is not prohibitive. Let us consider a bloc of developed
economies (European Union (EU)) granting a preference on a given set of imported
agricultural commodities (e.g. Citrus), a set of developing countries benefiting from this
preferential treatment (e.g. Kenya, Uganda and Tanzania (KUT)), and the rest of the world
(RoW) which encounters the Most Favoured Nation (MFN) tariff rate. Initially, let us
suppose that irrespective of any changes in the demand for citrus imports in the EU, the RoW
supplies citrus at a fixed price, while KUT supply more citrus at even higher prices.
Assuming that the RoW has a competitive advantage in producing citrus while the EU has a
competitive disadvantage, in the absence of preferential treatment the EU would obtain citrus
imports from both KUT and RoW at a fixed price. However, the introduction of preferential
treatment alters the relative prices in favour of citrus produced in KUT. This will inevitably
cause the import demand for citrus in the EU to shift from RoW to KUT. In this scenario, the
EU incurs a loss in tariff revenue from citrus; the RoW faces a loss in the volume of citrus
exports; and KUT benefits from the losses incurred by the other two country groups. The
change in sourcing of citrus imports into the EU from RoW, which has a competitive
advantage in producing citrus to KUT, leads to a negative allocative efficiency effect (Low et
al., 2005).
Particularly, exporters in KUT will earn a better price, higher by the margin of preference
between the MFN and the preferential tariff rate, which results in an increase in the supply of
exports from the KUT region. It is, however, argued that the extent to which exports increase
depends more on KUT's export supply elasticity, viz, the export supply response in relation to
the price change. Hence, the higher the elasticity, the greater the trade effects, which results
in larger gains for KUT. For the non-benefiting country group (RoW), preferential treatment
of citrus imports from KUT makes RoW imports into the EU more costly. This inevitably
causes a decline in the demand and production of citrus in RoW.
All in all, preferential treatment results in a shift from competitively advantaged producers of
a given tradable commodity and the government in the preferential treatment granting
country/region to producers in the preferential treatment receiving country/region. Markedly,
preferential treatment may also alter trade from non-beneficiary countries/regions, thus
lessening their welfare. In instances where the above assumptions do not hold, preferential
treatment non-receiving countries may not necessarily lose out because of the preferences.
28
2.6.2 Measures of preferential treatment value: The Preference Margin (PM)
The value of preferential treatment can broadly be categorized into two measures, that is, the
traditional preferential treatment value measures and the adjusted preference value measures.
Generally, traditional measures estimate the value of the preferential treatment accruing to
the beneficiary country in terms of the Preference Margin (PM). The term Preference Margin
(PM) refers to the difference in percentage points between the Most Favoured Nation (MFN)
rate and the preferential tariff rate (Cipollina et al., 2013; Low et al., 2005) and it is
computed at tariff line level.
According to Cipollina et al. (2013) and Low et al. (2005), this approach is limited by the
fact that it does not address the concern of whether the treatment boosts the benefiting
countries' exports; and in the event that the preferential tariff excludes Ad Valorem
Equivalents (AVEs), the actual value of preferential treatment may be under- or over-
estimated. The erroneous estimation of the PM may also be attributed to the fact that the
value of preferential treatment to a given country or region practically depends on other
competing countries within the same market. Nicita (2011) argues that it is a less accurate
measure, given that it does not account for the composition of exports.
The trade-weighted PM is another traditional measure of preferential treatment. It is defined
as the product of the margin of preference per unit of imports and the bilateral value of
imports (Low et al., 2005). This measure takes into account bilateral trade flows between
any two trading partners. The major drawbacks of this measure are: (i) it is assumed that all
countries supplying the same market are subjected to the same MFN rate, yet the rates vary
depending on the trade agreement under consideration and given the fact that these
agreements often overlap across countries. Also, given that a given country's PM depends on
other competing countries in the same market, this would be an inappropriate approach of
computing the value of preferential treatment. (ii) The assumption is utilized that preferential
treatment exists for all exports. In reality, it is noted that utilization rates vary greatly across
countries and sectors (Low et al., 2005).
Adjusted preferential treatment value measures include the competition-adjusted preference
margin and the utilization-adjusted preference margin (Low et al., 2005), as well as others
based on domestic competition within a given market (Carrère and de Melo, 2010; Carrère,
2011). The competition adjusted preference margin is defined as the weighted average tariff
rate applied to the rest of the world minus the preferential rate applied to the preferential
29
treatment receiving country. It takes into account competition from other countries exporting
into the same market while considering the overlapping nature of other unilateral and
bilateral agreements. On the other hand, the utilization-adjusted preference margin takes into
account that the granted preferential treatment may not be fully utilized by the benefiting
country. Thus, the value of the measure of preferential treatment is weighted by the volume
of trade that actually benefits from the preferential treatment.
According to Low et al. (2005), the values of the adjusted preferential treatment measures
may not duly reflect the actual margins. given that “Actual gains from preferences enjoyed
by exporters may be lessened if monopolistic distributors are operating in the importing
market, or if third parties not receiving preferences strategically cut their prices.”
Furthermore, it is noted that while adjusting preferential treatment value measures,
particularly the utilization-adjusted preference margin, to take into consideration other
preferential treatments, one could mistakenly assume that the other preferential treatments are
maximally harnessed, yet in reality they are not. According to Carrère (2011), the major
limitation of these adjusted measures of the preferential treatment value is their lack of
microeconomic foundations.
Other adjusted measures include: the Import Competition-Adjusted (ICA) Preferential
Margin (ICA-PM) (Carrère and de Melo, 2010); the Domestic and Import Competition
Adjusted (DICA) Preferential Margin (Carrère, 2011); the Relative Preferential Margin
(RPM); and the Potential Preferential Margin (PPM) proposed by Nicita (2011). The ICA-
PM measure is adjusted for competition among exporters but it does not consider trade flows
among EU member countries, while the DICA measure was derived under the imperfect
competition framework to take into consideration competition across competitors and within
the EU.
On the other hand, the RPM quantifies the comparative value of preferential treatment on a
country’s observed exports and shows the advantage offered to particular imports from a
certain country in comparison to those exports from other competing countries. With regard
to the PPM approach, this predicts the potential or anticipated value of a given preferential
treatment, depending on the future market tradable commodities (Nicita, 2011). The main
limitation of the RPM measure is that it focuses on a given country’s exports and it does not
spell out the particular instruments through which the preferential treatment benefiting
30
country could gain advantage. On the contrary, the PPM measure is limited by the fact that it
is based only on tariffs (Nicita, 2011) and does not consider other policy instruments that
could be embedded within the preferential trade agreement.
2.6.3 The effect of the EU-GSP scheme on agricultural exports
This sub-section presents the existing literature that evaluates the effect of the EU-GSP
Scheme in enhancing imports from developing countries into the European market. From the
scholarly point of view, this topic has received considerable attention. Some studies
discussed below have focused on the textiles and the manufacturing industry, and others on
the agriculture sector. Both econometric and non-econometric methods have been used on
various datasets, viz, either cross-sectional, time series or panel data.
The literature presented in the following sub-section is limited to studies that focus on
agricultural commodities. Secondly, only those studies that capture the effect of the EU-GSP
scheme as a continuous variable within the gravitational framework are considered. The
literature is grouped by the type of measures used to quantify the value of preferential
treatment.
2.6.3.1 Empirical studies based on the traditional measures of Preferential Margin (PM)
Cardamone (2010) employed a gravity flow model to assess the effect of preferential trade
agreements on monthly exports of fresh fruits to the European Union (EU) during 2001–
2004. The study employed preferential margins (expressed in absolute terms as the
difference between the applied MFN duty minus the preferential tariffs) to capture the effect
of the GSP scheme on fruit and vegetable exports to the EU. From the findings, it is evident
that the Generalized System of Preferences (GSP) only benefits exports of fresh grapes to the
EU.
During the evaluation of the performance of oranges, mandarins, apples and fresh grapes
using highly disaggregated monthly data, Cardamone (2009) calculated the preferential
margin used to capture the effect of preferential treatment as the difference between the
highest tariff applied by the EU and the duty paid by an exporter for a given product. The
results show that the impact of unilateral trade preferences varies depending on the
31
commodity under consideration. In that regard, the author notes that the GSP scheme
effectively increases exports of apples and mandarins to the EU market.
Aiello and Demaria (2009) used the preference margin to evaluate the impact of the EU-GSP
Scheme in enhancing twelve agricultural exports from 169 developing countries to EU
markets over a period of four years (2001–2004). The preference margin was captured in
relative terms, as the ratio between the preference margin and the Most-Favoured Nation
(MFN) tariff. Noticeably, the margin of preference was denoted as the difference between
the MFN tariff and the preferential tariff. Their empirical findings reveal that the EU-GSP
scheme positively impacts on fruit and vegetable exports from developing countries to the
EU market.
2.6.3.2 Literature based on adjusted measures of Preference Margin (PM)
Among other policies, Cipollina et al. (2013) investigated the impact of the EU-GSP scheme
on disaggregated trade flows from developing counties into the EU market. They used
relative preference margins, obtained as the ratio of the reference tariff to the applied tariff
rate subjected to each exporter by the EU. The reference tariff was estimated as duties paid
by all exporting countries. The findings suggest that the EU-GSP scheme plays a role in
boosting the volume of import trade into the EU.
Raimondi et al. (2011) evaluated the effects of the Everything But Arms (EBA) initiative on
rice imports from developing countries. The researchers computed the preference margin as
the percentage difference between the tariff encountered by an MFN exporter and the Tariff
Rate Quota Equivalent faced by the beneficiary country when it exports to the EU. They find
that preferential treatment had a significantly positive impact on rice imports into the EU
from some developing countries.
Cirera et al, (2011) evaluated the impact of the non-reciprocal GSP/EBA scheme on
developing country exports. The researchers used a number of competition-adjusted
measures of the value of preferential treatment. Their findings generally indicate that
preferential treatment has had a relatively small but positive impact in promoting trade flows
from developing economies.
Philippidis et al. (2011) evaluated the impact of EU preferences on European Union imports
based on 20 agro-food sectors during 2 specific years (2001 and 2004). The EU preference
32
variable was measured as a factor of the import tariff rate applied by the importer in terms of
ad valorem equivalents. The results show that increasing import tariff rates deters exports of
fruits and vegetables from developing countries into the European market.
Under the gravity model framework, Cipollina and Salvatici (2010) assessed the impact of
European Union (EU) trade policies on agricultural trade flows from developing countries.
They used an explicit measure for relative preference margins, defined as the ratio of the
maximum applied duty to the applied duty to capture the preferential treatment effects. Their
results reveal that the largest coefficients of the impact of PTAs on trade are registered by
tropical products, most especially the fruits and vegetable sectors.
Cipollina and Salvatici (2008b) examined the impact of the EU-GSP Scheme on
disaggregated agricultural trade flows from 161 developing economies. They used the
relative preference margin measure to proxy for the EU-GSP scheme. Specifically, the
relative preference margin was measured as the proportion of the maximum applied duty
factor subjected by the importer in the EU on a given commodity to the actual duty factor
faced by a specific exporting country. The results reveal that preferential trade schemes have
a significant positive impact on agricultural trade flows, with over nine (9) per cent influence
on the fruit and vegetable sector.
Emlinger et al. (2008) computed the weighted value of preferential margins as a measure of
the level of gains linked to these granted preferences with the aim of evaluating the
advantages accruing to Mediterranean countries resulting from the EU preferential treatment.
The authors considered actual tariffs applied by the EU to its trading partners. This indicator
compares the amount of custom duties paid by countries supplying the EU with the amount
of customs duties these countries would have paid if they did not benefit from tariff
preferences. Emlinger et al. (2008) found that Lebanon and Turkey do not benefit from large
preferential margins for access to the European market, despite the fact that they enjoy tariff
concessions for most products. On the other hand, Egypt, Morocco and Jordan enjoy large
preferential margins from the European Union, given that their exportable products have high
MFN duties on which the EU grants significant tariff reductions. This improves their already
highly favourable access to the European market.
33
2.7 Predicting trade potential and performance
2.7.1 Trade potential and trade performance measures
The term ‘trade potential’ refers to the maximum possible trade that can be achieved
(Armstrong, 2007). It is used to predict the hypothetical level of trade under assumption of
frictionless and free trade under given conditions at a certain time. Within the gravitational
framework, there are two measures for predicting potential trade flows (Gul and Yasin, 2011;
Karagoz and Saray, 2010; Helmers and Pasteels, 2003; Egger, 2000; Nilsson, 2000, Baldwin,
1994). These are:
(i) The within-sample predictions measure, also known as the “Out-of-sample”. This
measure is based on coefficient estimates obtained from the gravity model. Under the gravity
model framework, the measure is executed in two steps, first by estimating the determinants
of trade flows, and secondly, the estimated coefficients of the determinants are used in the
simulations so as to predict the trade volume between any given pair of trading partners.
Thereafter, the predicted trade volumes are compared with the actual trade flows so as to
deduce trade performance. A country’s trade performance can be inferred using either
absolute or relative indicators. The absolute indicator is defined as the absolute difference
between the predicted potential and actual trade flows. Strikingly, positive values suggest
there exists untapped trade that could be harnessed (trade expansion), while negative values
imply that actual trade flows exceed the predicted trade potential. On the other hand, the
relative indicator is defined as the ratio of predicted trade potential to the actual trade flows.
Relative values of greater than one imply that a country under consideration has a good trade
performance with the partners, while the opposite is also true (Gul and Yasin, 2011).
(ii) The relative residual measure. This also known as the “In-Sample approach” (Egger,
2000) and it uses residual values of the estimated gravity model and ranges between −100
and +100. Thus, an approximate value close to zero denotes that the predicted trade potential
is almost equal to the observed/actual trade flows, while a value greater than 30 % implies
that there exists unreleased trade potential. That is, there exists more room to conduct trade,
given that the prevailing conditions are unchanged. On the contrary, if the relative residual
value lies below −30 %, it means that the actual trade flows by far exceed the predicted trade
flows. Worthwhile to note, Egger (2000) argues that the in-sample approach leads to
misleading results, since no systematic variations in residuals can be obtained, even in an
34
econometrically well-specified model. This challenge presents a major drawback of this
approach.
2.7.2 Review of empirical studies that predict trade potential and performance
A plausible volume of studies aiming to predict trade potential and performance have been
carried out in different regions of the globe. However, much of the literature focuses on
aggregated trade flows and few studies focus on particular commodities, particularly in East
Africa. Taking this into consideration, therefore, literature presented in this sub-section is
categorized based on the measure used to predict trade potential and performance,
irrespective of the geographical region and the level of data disaggregation.
2.7.2.1 Literature based on the “In-sample” approach
As a preamble, it is worth mentioning that only one study using the in-sample approach has
been found in the existing literature. Egger (2000) simulated the trade potential between
countries of the Organization for Economic Co-operation and Development (OECD) and the
Central and Eastern European countries (CEEC) for the 1986–97 period. The study aimed at
comparing the influence of various estimators in predicting trade potentials while using the
in-sample methodology. Unfortunately, the author was indeterminate in providing the
information about the trade potential, but concluded by emphasizing that, “The in-sample
approach to the prediction of trade potentials is inappropriate.”
2.7.2.2 Literature based on the “Out-sample” approach
Batra (2006) used the augmented gravity model to estimate India’s trade potential in a two-
step (out-of-sample) approach. In the first step, determinants of India’s trade flows with the
rest of the world were ascertained. In the second step, the estimated coefficients were then
used to predict India’s trade potential as a proportion of predicted trade to actual trade.
Empirical results divulged that India had the highest trade potential with countries like China,
the United Kingdom, Italy, France, Pakistan, the Philippines and Cambodia.
Rojid (2006) used panel data over 21 years to estimate unilateral trade flows from 147
exporting countries with the aim of estimating trade potentials of COMESA member
countries within the COMESA region. The estimated coefficients were then employed to
simulate the trade potentials. Empirical results revealed that there was limited trade potential
within the region, owing to the fact that most COMESA member states were actually trading
35
more than they ought to have been. However, the author opines that Angola and Uganda still
had room to expand their trade horizon within the region.
Rahman (2009) investigated Australia’s trade potential by employing the augmented gravity
models and cross-sectional data from 50 countries for 2001 and 2005. He used the estimated
coefficients from the model to predict Australia’s trade potential. The results showed that
Australia had remarkable trade potential with Austria, Argentina, Singapore, the Russian
Federation, New Zealand, Turkey, Portugal, Greece, Chile, the Philippines, Norway, Israel,
Brazil and Bangladesh.
While using the gravity model approach, Karagoz and Saray (2010) employed a sample of 23
APEC countries, with the exception of Laos, Cambodia and Myanmar, to estimate Turkey’s
trade potential over a period of five years. The scholars used the out-of-sample methodology
to determine Turkey’s trade potential. Firstly, the determinants of Turkey’s trade flows were
analysed. The obtained coefficients were then used to predict trade potential by means of the
absolute indicator. According to the findings, Turkey had a high potential of expanding its
trade with Papua New Guinea, Peru, Myanmar, Mexico, Laos, and Brunei.
Gul and Yasin (2011) estimated Pakistan’s trade potential under the gravity model framework
using panel data for 15 years across 42 countries. They also used the out-of-sample
methodology, by first estimating the determinants of Pakistan’s trade flows and then using the
obtained coefficients to predict global trade potential by using the relative indicator. At
regional level Pakistan had very high trade potential with the Asia-Pacific region, the
European Union (EU), the Middle East, Latin America, and North America. At country
level, Japan, Sri Lanka, Bangladesh, Malaysia, the Philippines, New Zealand, Norway,
Sweden, Italy, and Denmark registered the highest trade potential.
2.8 Conclusion
There are various trade related measures used to quantify a country’s export competitiveness.
There is no commonly agreed upon scientific approach regarding the most efficient measure
of export competitiveness, or reliable indicators that are able to reflect the country’s export
competitiveness. However, Balassa's index is a commonly used measure of a country’s
export competitiveness. From the perspective of the East African states, little work has been
done to assess the export competitiveness of their fruits and vegetables. The few studies
36
either focused on a single country within the region, or did not use highly disaggregated data,
resulting in generalized policy recommendations.
The major drawback of such recommendations is that they may not be fruitful in explaining
commodity specific trade flows, yet effective trade-related policies should be based on
specific commodities rather than on a generalized basket of goods. Therefore, there exists a
knowledge gap about the export competitiveness of highly disaggregated commodities under
the fruit and vegetables sector among East African states. The current study intends to use
Balassa’s index (RCA) and highly disaggregated data at HS-6 digit level to assess export
competitiveness of horticultural commodities from East African states.
In international trade, climate change influences trade flows through various mechanisms,
such as through the imposition of climate change policies, like carbon tax and environmental
permits, among others. Climate change as a factor affecting international trade flows has
received very limited attention from scholars. Most of the existing literature on trade uses
Kyoto Protocol policies, like the carbon tax and tradable permits, to proxy for climate
change, especially while assessing the trade flows of manufactured goods, such as cement,
steel and iron. This may lead to misleading generalized recommendations that are of less
relevance to developing economies reliant on agriculture, especially those south of the
Sahara.
The use of Kyoto Protocol policies as climate change proxies is not appropriate for
agricultural-based economies, given that the composition of exports from these economies is
skewed towards agricultural products, and that Green House Gas (GHG) emissions are very
difficult to quantify in the agriculture sector. One study used meteorological data to assess
climate change impacts on international trade, but the study was based on aggregated data.
This implies that general results and recommendations were obtained. However, different
agricultural commodities have specific optimum temperature and precipitation ranges within
which minimal damage is caused. Therefore, the generalized results and recommendations
could also be misleading. Moreover, average meteorological values were used. The major
drawback of using average values is based on the fact that the values do not account for the
likelihood of the higher variance in the data for the more arid countries and such data is
susceptible to potential scale effects.
37
Therefore, climate change proxies based on Kyoto Protocol policies are not appropriate for
agricultural-based economies when assessing the impact of climate change on trade. In
addition, to fill this knowledge gap, the current study intends to develop meteorological
indices, that is, temperature and precipitation anomalies to be used as proxies for climate
change. The indices will be used within the gravity flow model framework, based on
disaggregated horticultural data, to assess the influence of climate change on international
trade flows. Temperature and precipitation are key factors that directly influence the
agriculture sector, while the use of anomalies enables one to overcome the limitations
associated with average meteorological data values.
With regard to the EU-GSP scheme, preference margins can be measured either in absolute
or relative terms. Despite the fact that adjusted preference margin measures are the most
commonly used, they lack microeconomic foundations. Various policy instruments
embedded within the EU-GSP scheme are used to compute the value of preference margin.
These are MFN, tariff rates, tariff rate quotas and specific duties. Different approaches are
used to calculate the margin and this may be associated with the mixed results about the
influence of the EU-GSP scheme on exports from developing countries into the EU market.
None of the studies which have been traced used a combination of all the instruments (MFN,
tariff rates, specific duties and tariff rate quotas) embedded in the EU-GSP scheme to
quantify the value of preference margin, yet ignoring any of these could overestimate the
value of the preference margin. In addition, generalized results and recommendations were
obtained by various scholars, either because they used aggregated data or fruit and vegetable
commodities that are of less relevance to the three East African states.
Therefore, there is inconclusive knowledge regarding the impact of the EU-GSP scheme on
agricultural export flows from developing countries into the EU market. In this study, this
knowledge gap will be filled by using the competition-adjusted PM measure. The measure
will be based on a combination of trade-weighted applied MFN rates, tariff rates, specific
duties and Tariff Rate Quotas within the gravity model framework.
There are two approaches to predicting potential trade flows, viz, the “Out-of-sample” and the
“In-sample”, but literature argues that the latter leads to misleading results. With the “Out-
of-sample” approach, trade potential is computed either as the difference between the
simulated trade potential and the actual trade flows, or as the ratio of the simulated trade
potential to the actual trade flows. Existing literature leads to generalized results and
38
recommendations being given that are based on either data that is aggregated across sectors,
or cross-sectional data in a given economy. Thus, the actual insight into potential markets for
trade expansion based on highly disaggregated commodities is lacking. Analysis based on
cross-sectional data leads to inconsistent estimates, thus implying that the trade potentials
simulated based on such estimates may also be misleading. Therefore, there exists a general
knowledge gap about East African economies’ trade potentials with their trade partners. It is
thus prudent to suggest that the use of recent and highly disaggregated panel data at sector
level, particularly for the fruit and vegetable sub-sector, may provide an insight into sector-
specific results upon which commodity-specific markets for East African economies’ may be
identified so as expand their global trade potential.
39
CHAPTER THREE:
AN OVERVIEW OF EAST AFRICA’S HORTICULTURE SECTOR
3.1 Introduction
This chapter provides a description of the fruits and vegetable sector in Kenya, Tanzania and
Uganda. Specifically, it presents a focus on the acreage covered under fruits and vegetables,
production trends, the major export destinations of fruits and vegetables, net export trends
across the globe, and the export trends in relation to temperature and precipitation.
3.2 Fruit and vegetable production in East Africa
Over the past 2 to 3 decades, it has been argued that the fruit and vegetable sector in Kenya,
Tanzania and Uganda has the potential to draw the rural populace out of poverty. For
instance, Bear and Goldman (2005) put it that “Fruits and vegetables are among the sectors
where Uganda can achieve growth in coming years.” In light of the above, many small-scale
farmers have embarked on the production of these commodities, and in some cases with
technical and financial support from donor agencies such as the United States Agency for
International Development (USAID), the British Department for International Development
(DFID), and the Europe-Africa-Caribbean-Pacific Liaison Committee (COLEACP). Many
development-oriented programmes, such the Pesticides Initiative Programme (PIP), have also
been introduced in the East African region.
Such initiatives have generally boosted the production of fruits and vegetables, to the extent
where they have become recognized as income generating crops, e.g. pepper, bananas,
asparagus, beans, pineapples and French beans, on top of the traditionally grown crops, such
as coffee, tea, cotton and tobacco (in Uganda); coffee, tea, cotton, cashew nuts and tobacco
(in Tanzania); and coffee, tea, cotton, maize, sorghum and millet (in Kenya). Figure 3.1
below shows the trend in the area harvested under fruits and vegetables in Kenya from 1970
to 2012.
40
Figure 3.1: Area harvested under fruits and vegetables in Kenya
Source: FAO database (2013)
Figure 3.1 illustrates that there has been a gradual increase in the area harvested under fruits
and vegetables in Kenya. Fruits generally assume a larger proportion of area harvested than
do vegetables. On average, fruits assume 119,354 hectares (ha) while vegetables account for
111,511 ha. During the late 1970s and the period from the mid-1990s until the early 2000s,
the acreage under fruit production declined below that for vegetables. For Tanzania, Figure
3.2 below also shows that fruits cover a larger acreage than vegetables do. However,
Tanzania's trend in acreage is characterized by drastic fluctuations for both fruits and
vegetables.
For instance, fruits covered more than twofold the acreage (397,475 ha) relative to vegetables
(159,318 ha), but exhibit more sharp fluctuations along the trend, particularly during the
1982–85, 1989–91, 1997–1999 and 2009–12 periods. The lowest acreage in fruit production
was observed in 1990 (40,931 ha) while the highest was 885,182 ha in 2009. Generally,
vegetables assumed the lowest acreage in 1994 (18,638 ha), followed by 19,091 ha in 1985.
The largest acreage (348,694 ha) under vegetable production was observed in 2011, followed
by 316,472 ha in 2009.
41
Figure 3.2: Area harvested under fruits and vegetables in Tanzania
Source: FAO database (2013)
In Uganda, trends presented in Figure 3.3 below show that fruits assume more acreage than
vegetables. Acreage under fruit production gradually increased over the years with hardly any
drastic fluctuations. On average, acreage fruit production was 1.52 million ha, while
vegetables accounted for 51,835 ha only. Between the early 1970s and the early 1980s, the
acreage under vegetable production was very low, ranging between 50 ha in 1978 to 595 ha
in 1984.
Since then, the acreage under vegetables has drastically fluctuated, with the sharpest decline
(99 %) during the 2011–2012 and 2009–2010 periods. During the 2011–2012 period, the
acreage dropped from 195,475 ha to 204 ha and from 186,624 ha to 200,5 ha between 2009
and 2010. Other periods characterized by drastic changes in acreage under vegetable
production are between 1991 and 2001. However, despite the drastic fluctuations in acreage
under vegetable production, there is an increasing trend in general.
42
Figure 3.3: Area harvested under fruits and vegetables in Uganda
Source: FAO database (2013)
Trends presented in Figure 3.4 below show that Kenya has experienced an increase in fruit
and vegetable production over the years, with a sharp rise in 1991 for vegetables, while fruit
production drastically increased in 1979 and 2012. The 1991 fluctuation accounts for a 92.8
per cent increase in vegetable production from 1990 (743,08 tonnes) to 1991 (1,029,450
tonnes), while the changes in fruit production account for 95.3 per cent and 46.4 per cent for
the 1978–79 (from 52,616 tonnes to 1,121,810 tonnes) and 2011–12 (from 2,876,276 tonnes
to 5,364,506 tonnes) periods, respectively. Kenya produces more fruits than vegetables.
Between 1970 and 2012, Kenya produced 1.6 million tonnes of fruits and 0.93 million tonnes
of vegetables.
43
Figure 3.4: Kenya’s trend in fruit and vegetable production
Source: FAO database (2013)
Figure 3.5 below shows Tanzania's trend in the production of fruits and vegetables. In
general, trends show that fruit production outweighs vegetable production. However, both
trends show that Tanzania experienced a gradual increase in the production of fruits and
vegetables over the years. Drastic increases in production were observed after late 1996.
After the mid-1990s, Tanzania’s highest fruit production was estimated at 4.9 million tonnes
in 2006, followed by 4.5 million tonnes in 2011. On the other hand, the lowest volume of
fruit production (1.85 million tonnes) was registered in 2000, followed by 1.91 million tonnes
in 1999. For the entire period between 1970 and 2012, Tanzania produced 1.98 million
tonnes of fruits. Vegetable production was highest in 2012 (2.2 million tonnes) and in 1973
(885 tonnes), while the average volume produced during the 1970–2012 period was 1.11
million tonnes.
44
Figure 3.5: Tanzania’s trend in fruit and vegetable production
Source: FAO database (2013)
Figure 3.6 below illustrates that Uganda mainly produces fruits, as compared to vegetables.
Figure 3.6: Uganda’s trend in fruit and vegetable production
Source: FAO database (2013)
45
Despite the general increasing trend, fruit production experienced fluctuations, particularly
from the late 1970s until the early 1980s. The drastic decline in fruit production occurred
from 92.4 million tonnes in 1978 to 61.1 million tonnes in 1980. Overall, the average fruit
production between 1970 and 2012 is estimated at 88 million tonnes. The trend for vegetables
also shows that production fluctuated more after 1990. Vegetable production drastically fell
from 0.74 million tonnes in 2009 to 986 tonnes in 2010, and thereafter increased by about
100 times to 1.1 million tonnes in 2011. Other periods during which vegetable production
registered drastic fluctuations include 1997–2000 and 2002–2005.
Overall, Uganda produced 0.24 million tonnes of vegetables between 1970 and 2012. Based
on the average values between 1970 and 2012, Uganda is the leading producer of fruits (88
million tonnes), followed by Tanzania (1.98 million tonnes) and then Kenya with 1.59
million tonnes. In the case vegetables, Tanzania, Kenya and Uganda produced 1.11 million
tonnes, 0.93 million tonnes and 0.24 million tonnes, respectively.
3.3 East Africa’s fruits and vegetables trade statistics and the major trade partners
Over the past two decades, the value of global agricultural trade from the East African
economies has been increasing (FAOSTAT database, 2014). The constitution of trade shows
an increasing trend in high-value, non-traditional cash agricultural commodities, such as
fruits and vegetables in world agricultural trade among the traditional cash crops. The
increase in agricultural exports may generally be attributable to the topical climatic
conditions that enhance production throughout the whole year, as well as to the technical and
financial support from donor agencies, among other factors. According to Figure 3.7 below,
Kenya is the number one exporter of fruits and vegetables, followed by Tanzania and then
Uganda. Between 1990 and 2011, Kenya exported fruits and vegetables worth US$ 232.5
million, while the value of Tanzania's and Uganda's exports were valued at US$ 97.1 million
and US$ 13.5 million, respectively.
For Kenya, the increase in the value of fruit and vegetable exports drastically rose after 2002
(US$ 260.7 million) and by 2008 (US$ 453.2 million) the growth rate in exports was 66 per
cent. The sharp rise may be attributable to the fact that Kenya uses advanced technology,
such as green houses, in the production of some horticultural commodities. Tanzania
experienced fluctuations in the value of fruit and vegetable exports over the years.
46
Figure 3.7: Aggregated fruit and vegetable exports from Kenya, Tanzania and
Uganda
Source: FAOSTAT database (2014)
A 57 per cent increase in the value of exports was observed between 1997 (US$ 74.7 million)
and 1999 (US$ 173.4 million), probably due to the full liberalization of the economy during
the mid-1990s. Thereafter, the trends show that the values of exports declined, probably due
to the drought that affected the country from late 1999 until 2000. Uganda's low value of
fruit and vegetable exports in comparison with Kenya and Tanzania may be associated with
the fact that the country is landlocked.
According to Agribusiness East Africa (2013), the European Union (EU) remains the key
export market for fruits and vegetables from all the East African region member countries.
Fresh fruit and vegetable exports are mainly destined to specific ethnic buyers within the EU.
Figure 3.8 below shows the trend in the value of fruit and vegetable exports from Kenya to
the EU. Unlike vegetable exports, fruits exhibit a lower gradual increase. A drastic rise in the
value of vegetable exports was observed between 2002 (US$ 134.8 million) and 2008 (US$
47
289.1 million), representing a 53.4 per cent increase. The highest value of vegetable exports,
worth US$ 289.1 million, was registered in 2008.
Figure 3.8: Kenya's fruit and vegetable exports to the EU, by value
Source: COMTRADE database (2013)
In general, vegetables destined for the EU market generate more foreign exchange than fruit.
The value of Kenya's fruit exports to the EU was also highest in 2008 (US$ 48.4 million),
while the lowest (US$ 15.5 million) was experienced in 1998. By 2009, exports in both fruits
and vegetables declined by 41.9 per cent and 25.5 per cent, respectively.
For Tanzania, Figure 3.9 below shows that EU-bound fruit exports gradually increased over
the years, with a sharp rise between 2006 and 2009. During the 1996–2012 period, Tanzania
exported fruits worth US$ 4.3 million, on average, with the lowest value in 1996 (US$ 0.45
million) and highest value in 2009 (US$ 14.2 million). From 2007, fruit exports brought
more foreign currency into Tanzania than vegetables did. Between 2007 and 2012, EU-
bound fruit exports were worth US$ 11.1 million, on average, while vegetables were worth
48
US$ 7.5 million only. The figure also shows that the value of Tanzania's vegetable exports
fluctuated highly between 1996 and 2012.
Generally, the value of Tanzania's vegetable exports to the EU varied greatly between 1996
and 2012. The highest value of vegetable exports was registered in 2004 (US$ 14.1 million)
while the lowest was US$ 2.2 million in 2000. The declining trend for vegetable exports
during the late 1990s may be associated with the drought that affected the country at the time,
while the increasing trend in fruit exports during the 2000s may be attributable to the
investment ventures by donor agencies, such as DFID, COLEACP and USAID.
Figure 3.9: Tanzania's fruit and vegetable exports to the EU, by value
Source: COMTRADE database (2013)
Figure 3.10 below shows that the value of Uganda's fruit and vegetable exports to the EU has
increased over time. Based on mean values, Uganda received more foreign currency from
vegetable (US$ 4.71 million) than fruit exports ( US$ 2.41 million) between 1996 and 2012.
Vegetable exports were highest in 2011 (US$ 7.32 million) and lowest in 1996 (US$ 1.41
million). Despite the general increasing trend, the value of vegetable exports dropped sharply
from US$ 6.8 million in 2005 to US$ 4.9 million in 2006, probably due to the strict
certification requirements needed to meet the private standards such as the GLOBALG.A.P.
49
On the other hand, the value of fruit exports rose sharply from US$ 0.572 million in 2003 to
US$ 6.32 million in 2008, but by 2010 (US$ 4.4 million), a drastic 30 per cent decline had
been registered.
Figure 3.10: Uganda's fruit and vegetable exports to the EU, by value
Source: COMTRADE database (2013)
The other major destination markets for fruit and vegetable exports from East Africa include
the Common Market for Eastern and Southern Africa (COMESA), East Asia community
countries, and the Middle East. More often than not, owing to the perishability of these
commodities, shipment is done by airlifting the produce to the destination markets. Table 3.1
below shows the main fruit and vegetable exports destinations, disaggregated by the
monetary value of imports by the leading trade partners as at the end of 2011.
50
Table 3.1: Major export markets for fruits and vegetables from East Africa in
December 2011
Exporter Top partners in the Value of trade ('000 Total value
Export market US$) ('000 US$)
market
HS-07 HS-08
United Kingdom 150,384 5,107 155,491
Kenya France 15409 10,166 25,575
Netherlands 32,752 5,559 38,311
EU-27*
Tanzania Netherlands 4,139 7,396 11,535
United Kingdom 661 1,715 2,376
Uganda United Kingdom 3,644 121 3,765
Netherlands 859 1 860
Kenya UAE 5,429 14,124 19,553
Saudi Arabia 203 4,198 4,401
United Arab Emirates 5,119 1,507 6,626
Middle Tanzania Saudi Arabia 62 689 751
East Uganda Oman 115 2 117
Bahrain 13 46 59
Kenya Uganda 651 293 944
Sudan 409 104 513
COMESA Tanzania Kenya 24,469 25,735 50,204
Rwanda 1,179 115 1,294
Uganda Kenya 10,473 654 11,127
Sudan 2,676 29 2,705
EU-27* denotes the 27 members of the European Union. UAE denotes United Arab
Emirates
Source: International Trade Center (ITC) database (2013)
The East African states also import some fruits and vegetable commodities from other
countries. For instance, the International Trade Center (ITC) database (2013) shows that in
2012 Uganda imported fruits and vegetables, amounting to about 73, 63 and 24 thousand US
dollars, from France, the Netherlands and Italy, respectively. Kenya also imported fruits and
vegetables, estimated at 2.4 million US dollars, from France, the Netherlands, Italy and the
United Kingdom, among other European Union (EU) countries. Similarly, an estimated 0.41
million US dollars’ worth of horticultural commodities were imported by Tanzania from the
EU, mainly from Belgium, the Netherlands, the United Kingdom and Italy.
51
In 2012 alone, Kenya also imported fruits and vegetables from the Middle East, amounting to
US$ 4.73 million, mainly from Egypt, Saudi Arabia, Turkey and the United Arab Emirates.
Such commodities mostly comprised citrus fruits and grapes. For Uganda, fruit and
vegetable imports from this group were estimated at about US$ 1 million, and were mainly
citrus fruits, grapes and dried vegetables from Egypt, Turkey and the United Arab Emirates.
In the case of Tanzania, horticultural produce worth about US$ 16.6 million was imported
from the Middle East. Outstandingly, dates, figs, pineapples, mangoes, avocadoes, guavas
and grapes dominated the imported fruits, mainly from the United Arab Emirates and Saudi
Arabia, while onions, garlic and leeks were the key vegetable imports, solely from the United
Arab Emirates (ITC database, 2013).
Within the COMESA trade bloc, to which the three East African states ascribe, there also
exists cross border trade in fruits and vegetables. In 2012 Rwanda and Uganda supplied
Tanzania with vegetables (potatoes in particular) while Kenya supplied Tanzania with fruits
(dates, pineapples, mangoes, avocadoes, guavas, Brazil nuts, cashew nuts and coconuts). In
total, US$ 2.9 million worth of horticultural produce were imported into Tanzania. Uganda
imported fruits and vegetables from within the COMESA region worth US$ 1.9 million.
A large proportion of vegetables (mainly carrots, turnips, salad beetroot, onions and leeks)
were obtained from Kenya, while Egypt supplied citrus fruits. On the other hand, more than
85 % of Kenya’s fruits and vegetable imports were sourced from Uganda in 2011. For
instance, of the total US$ 21.5 million worth of vegetable imports into Kenya, US$ 21.2
million worth of produce was supplied by Uganda, while about US$ 1.2 million worth of
fruits was also sourced from Uganda. All in all, Kenya imported fruits and vegetables worth
US$ 25.2 million in the year 2011(ITC database, 2013).
In general, statistics for the 1997–2013 period show that trade balances for the three East
African states were not in a deficit, that is, countries exported more fruits and vegetables than
what was imported. With the exception of Uganda (See Figure 3.11), Kenya and Tanzania
were net exporters of fruits and vegetables. Kenya's and Tanzania's mean net export value
were US$ 165 million and US$ 132 million, respectively. Uganda has the lowest mean net
export value (US$ 5 million). During 2003, 2004, 2005 and 2007, Uganda was a net importer
of fruits and vegetables and net imports were valued at US$ 1, US$ 7, US$ 9 and US$ 4,
respectively. However, since 2007, the country has become a net exporter of fruits and
52
vegetables. By 2013, Uganda's net exports of fruits and vegetables were valued at US$ 20
million.
Figure 3.11: Fruit and vegetable net exports from Kenya, Tanzania and Uganda
Source: COMTRADE database (2013)
3.4 Fruit and vegetable export trends against temperature and precipitation
Derksen and Jegou (2013) argue that changes in climate may physically distort trade patterns.
Distortions in agricultural exports arise when climatic factors, particularly temperature and
precipitation, affect the production phase (Berg et al., 2013; Roudier et al., 2011). According
to Figure 3.12 below, Kenya's vegetable exports to the EU continued to increase, despite a 22
per cent reduction in precipitation between 1999 and 2000. However, the value of fruit
exports appeared to decline by US$ 2.6 million during the same period.
53
Figure 3.12: Trend line of Kenya's fruit and vegetable exports into the EU in relation to
precipitation
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
On the other hand, Figure 3.13 below shows that a 0.04 per cent increase in temperature from
22.9 degrees Celsius (°C) in 1997 to 23 degrees Celsius in 1998 may have influenced the 49
per cent fall in fruit exports to the EU from US$ 30.3 to US$ 15.5 million. Moreover, the
increase in temperature by 0.1 degree Celsius also seems to be associated with the 13.4 per
cent decline in fruit exports from US$ 19.6 million in 1999 to US$ 16.9 million in 2000.
Increases in temperature, however, seem not to have influenced Kenya's vegetable exports.
54
Figure 3.13: Trend line of Kenya's fruit and vegetable exports into the EU in relation to
temperature
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
In Uganda's case, Figure 3.14 below shows that fluctuations in temperature may have
tendencies for distorting fruit export trends. The general trend shows that when temperatures
are high, fruit exports seem to decline, but as the temperature fall, fruit exports tend to
increase. However, the graph for vegetables exports reveals an increasing trend, irrespective
of the fluctuations in temperature.
During the , a 7.1 per cent decline in the value of vegetable exports was observed between
1999 (US$ 2.9 million) and 2000 (US$ 2.8 million).
55
Figure 3.14: Uganda's fruit and vegetable exports into the EU in relation to temperature
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
In the case precipitation, Figure 3.15 below shows that changes in this climatic factor may be
associated with trends for fruit and vegetable exports. For instance, both fruit and vegetable
exports increased as precipitation also increased, but between 1997 and 1998, fruit exports
declined sharply. This decline may probably be linked with the 1997 floods that occurred in
Uganda. During that period, the value of fruit exports declined by US$ 0.12 million. The
Figure also shows that the value of fruit exports dropped by US$ 4.1 million when
precipitation declined between 1998 and 1999. EM-DAT (2013) argues that it was this
prolonged period of drought that affected the economy.
56
Figure 3.15: Uganda's fruit and vegetable exports into the EU in relation to
precipitation
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
For Tanzania, Figure 3.16 below shows a general declining trend for the value of vegetable
exports, while fruit exports increased until 1998, and dropped sharply thereafter until 2000.
Vegetable exports fell by 54 per cent during the 1999–2000 period when precipitation
dropped by 22 per cent from 1,088 mm to 850 mm. During the same period, the value of
fruit exports also dropped by 70 per cent, from US$ 25,409 to US$ 7,616.
57
Figure 3.16: Tanzania's fruit and vegetable exports into the EU in relation to
precipitation
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
In the case of temperature, Figure 3.17 below shows that changes in temperatures may, to a
little extent, be associated with Tanzania's trend in fruit exports to the EU. For instance, the
increase in temperature between 1996 and 1998 reveals that the value of fruit exports also
increased. When temperatures dropped 0.4 degrees Celsius between 1998 and 1999, the value
of fruit exports also declined. However, when temperature increased thereafter, the value of
fruit exports continuously declined until 2000. The value of vegetable exports dropped
throughout the period between 1996 and 2000, irrespective of the fluctuations in temperature.
58
Figure 3.17: Tanzania's fruit and vegetable exports into the EU in relation to
temperature
Source: COMTRADE database (2013) and TYN CY 1.11 database (2013)
3.5 Conclusion
The main aim of Chapter Three was to present an insight into the fruits and vegetable sector
in Kenya, Tanzania and Uganda, with focus on trends in acreage, production, and exports, as
well as on major trade partners. The three East African countries exhibit increasing trends for
acreage, production and exports of fruits and vegetables. The trends for acreage, production,
and exports of fruits and vegetables vary greatly from one country to another. When climatic
factors such as temperature and precipitation fluctuate, EU-bound fruit and vegetable exports
also appear to vary across the three East African countries. Kenya, Tanzania and Uganda are
net exporters of fruits and vegetables, in that order. In conclusion, it is prudent to mention
that the fruits and vegetable sector plays a contributory role in generating foreign currency for
East African states. Fruits and vegetable exports from all three East African are influenced
differently by changes in temperature and precipitation.
59
CHAPTER FOUR:
RESEARCH METHODOLOGY
4.1 Introduction
This chapter presents the methods, data, data management procedures and the data sources
used to achieve the set objectives. Given that each objective has specific data requirements
and estimation techniques, methodological details are presented in such a way that each
objective has a designated section detailing the data issues and estimation techniques. The
chapter begins with a general overview of the trading parties considered in this study,
followed by procedures used to measure export competitiveness and the influence of climate
change on the horticulture sector. The chapter also describes the approaches used to analyse
the impact of the EU-GSP scheme on East Africa’s fruits and vegetable commodities, as well
as the estimation procedures employed to predict trade potential and performance. The
chapter dwells much on how key variables are transformed from raw data and on how to deal
with zero trade flows within a matrix. The chapter wraps up with a general conclusion.
4.2 Focus of the study
With regard to trade in fruits and vegetable commodities, the study generally focuses on three
East African states (Kenya, Tanzania and Uganda) and their European Union (EU) trade
partners, the largest world importer of fruits and vegetables (Cardamone, 2010; COMTRADE
Database, 2012). The number of EU member states considered under each of the objectives
differs owing to the different data requirements. Details about the EU member states
considered are provided under each objective.
The research focuses on the fruits and vegetable sector, given that for about three decades
now, fruits and vegetable exports have become a strong trade item among the East African
economies. Additionally, despite the fact that the EU grants developing countries non-
reciprocal preferential treatment on their exports into the EU market, fruits and vegetables are
among the most sensitive products and are subjected to a number of tariffs and regulations
which may be seen as barriers to trade.
60
4.3 Determining East Africa's export competitiveness in the fruits and vegetable
sector in the EU market
This sub-section describes the kind of data, its sources and the method of analysis used to
ascertain East Africa’s competitiveness in exporting fruits and vegetable to the European
Union (EU) market. For this particular objective, the EU market entails all the 27 member
countries (EU-27), while East African states refer to Uganda, Kenya and Tanzania. Inclusion
of all the EU member states gives a true picture of export competitiveness of fruit and
vegetable products from the East African states.
4.3.1 Data and data sources
This analysis focuses on chapter 7 and chapter 8 of the Harmonized System (HS) of
nomenclature. These chapters are basically devoted to Vegetables (fresh, chilled or frozen)
and Fruits, respectively. This system of nomenclature was established in 1988 by the United
Nation (UN) so as to ensure uniformity in the recording of cross border trade flows. Data on
the monetary value of exports for the period 1997–2011 on all fruits and vegetable exports at
HS 4-digit and HS 6-digit level tariff lines to the EU-27 market by Uganda, Kenya and
Tanzania were extracted from the COMTRADE database of the United Nations. This
database provides detailed trade flow data for all commodity groupings up to the HS 6-Digit
level of disaggregation for all reporting countries.
4.3.2 Data analysis
Evaluation of East Africa’s export competitiveness was achieved through the use of Balassa’s
Index, i.e. The Revealed Comparative Advantage (RCA). This is the most commonly used
measure of export competiveness globally (Palit and Nawani, 2012; Gilbert, 2010).
Furthermore, Vollrath (1991) and Fertő and Hubbard (2002) show that government
intervention, especially through trade-related policies, does not affect the RCA index, given
that computation of the index relies only on export data. The distortion of trade flows
attributable to government intervention is more evident at the import side than the exporters'
side. Finally, the use of Balassa's index was also supported by the arguments of Capalbo et
al. (1990) who noted that export competitiveness should be based on highly disaggregated
data, upon which this study is founded. The index uses actual trade flows to ascertain the
comparative advantage of exporters in fruit and vegetable products. At country level, the
61
export competitiveness of fruits and vegetable commodities from East Africa into the EU
market was measured using the monetary value of traded commodities disaggregated at two
levels, namely (i) at HS 4-Digit, and (ii) at HS 6-Digit levels. Data analysis was carried out
using an MS-Excel package. Balassa's index (RCA) was calculated at two levels.
Firstly, at HS 4-Digit level, the analysis aimed at reducing the number of tariff lines that
would later be focused on at the highest level of disaggregation. At this level, 27 tariff lines
(commodities) were considered for analysis, fourteen (14) of which are under Chapter 7 for
vegetables, while the other thirteen (13) tariff lines represented Chapter 8 (fruits). After
taking note of each East African country's annual exports and the world annual export for the
entire period (1997–2011), Balassa's Index was then computed following Kulapa et al.
(2013), Török and Jámbor (2013), and Shinyekwa and Othieno (2011).
Mathematically, Balassa’s index is computed as:
   * 
RCA   xik   xwk k  X   *  ................................................................... (13) ip   X 
Where the variables xik and X ip denote the value of exports of product k from country i and
*
total exports (p) from country i, respectively. The variables x wk and X * represent the
value of the world’s exports of product k and total world exports, respectively. The monetary
value of exports is expressed in thousand US dollars. These variables are exclusive of the
exports of country i. The RCA ranges from zero to infinity. As a rule of thumb, a value of
greater than one (1) implies competitiveness in exporting a given horticultural commodity.
However, to ensure comparability over time, the average RCA Index for each commodity at
country level was computed for the most recent four years, viz, from 2008 until 2011.
Because of data limitations for 2011, the average RCA index for Kenya was computed using
the 2007–2010 period.
At HS 6-Digit level (second stage), only the top two commodities that revealed export
competitiveness in the EU-27 market at the first stage of analysis (HS 4-Digit level) were
selected for further analysis. However, non-sensitive commodities , as categorized by the
EU-GSP scheme, were not considered for analysis at this stage because such products are not
highly protected (Bouët et al., 2004) within the EU-27 market, relative to the sensitive ones.
62
Again, to ensure comparability of RCA values over time, averaging for each commodity at
country level as described above in stage one (at HS 4-Digit level) was done. Selected
commodities that were found to have export competitiveness in the EU market were then
considered for further analysis.
4.4 Determining the influence of climate change on East Africa's horticultural trade
flows
This sub-section presents the data and estimation techniques that were employed to develop a
set of climate change indices based on meteorological data and how these indices were
thereafter used to ascertain the influence of climate change on horticultural trade flows into
the EU market. To achieve this, the study builds on the classic gravity flow model that is
briefly described in the following subsection.
4.4.1 A brief overview of the gravity model framework
The gravity model was first conceptualized by Tinbergen (1962) and Pöyhönen (1963) during
the late 20th century. Leamer and Levinsohn (1995), Bergstrand (1985; 1989), Deardorff
(1998), Eichengrean and Irwin (1997) and Luca and Vicarelli (2004) argue that it is the
workhorse of international trade because of its capability to correctly estimate trade flows. It
originates from Newton’s “Law of Universal Gravitation”, which states that the attractive
force (F) between two objects i and j is a positive function of their respective masses (Mi and
Mj) and a negative function of the distance (R) between them. It can be expressed as:
 
   
 M M 
F  i j


 
ij G    …………………………………….............….(14)

 R 
 ij 
where G is a constant proportion. The basic gravity model has, however, been modified with
several additional variables (Linnemann 1966; Bergstrand, 1985; 1989; Oguledo and
Macphee, 1994; Deardorff, 1998). Saltatici (2013) argues that the model was initially
criticized for lacking sound theoretical foundations. However, Linnemann (1966), Anderson
(1979), Bergstrand (1985; 1989), Frankel et al (1995) and Le et al (1996) used various
approaches to validate the theoretical foundation of the model.
63
4.4.1.1 Empirical success in the estimation of the gravity model
Matyas (1997), Cheng (1999), Wall (2000) and Glick and Rose (2001) argued about the
influence of the heterogeneous nature of relationships between trading partners, noting that
heterogeneity between partners leads to biased results. According to Aiello and Demaria
(2009) and Blasi et al. (2007), heterogeneity arises owing to differences in cultural, political,
ethnic, geographical and historical factors, which are often difficult to observe and quantify,
yet those factors play a significant role in explaining trade flows between trading partners.
It was agreed that heterogeneity across countries should be controlled while estimating the
gravity model in order to minimize instances of biased estimates. Aiello and Demaria (2009)
argue that heterogeneity is associated with both observable and non-observable factors.
Furthermore, they note that unobserved heterogeneity is attributable to omitted variables,
which if not taken into account, will result in the estimates becoming inconsistent and
inefficient. When specifying the model, heterogeneity due to observable factors is overcome
by introducing a set of dummy variables, while unobservable heterogeneity models are
controlled by using country-fixed effects.
Therefore, by introducing the country-fixed effects term, the linear analytical form becomes
as illustrated in equation 15:
lnF  lnM  lnM  lnij i j R   ln ij Gj   ij ................................ (15)ij
where  is the specific country-pair effect, while  lnG (hereafter, µ0) denotes the interceptij j
common to all countries. Because of the lack of concrete evidence regarding the actual
causes of heterogeneity in such analysis, Cheng and Wall (2005) argue that each country pair
is represented by a unique dummy variable within the dataset so as to capture the effect(s)
within a given pair of trading partners.
While using panel datasets, the gravity model also provides room to capture linkages of the
different variables over time. According to Blasi et al. (2007), this can be achieved through
the inclusion of a term to capture "time-invariant specific effects".
64
Equation (16) represents this advancement:
ln F  ln M   ln M   R    ij i j ij t  ij ........................(16)0 ij
where ϑt denotes a time invariant effect that occurs in each year and common to all country
pairs.
4.4.1.2 Disaggregated data challenges and econometric issues
Recent analytical trends within the gravity model framework show a drift from the use of
aggregated data towards the use of highly disaggregated data. However, estimating a
linearized gravity model based on disaggregated data becomes more difficult because of the
high proportion of zero trade flows. This presents a problem in executing the log-linear form
of the gravity equation since the logarithm of zero is undefined, coupled with the fact that the
zero trade flows are not randomly distributed (Saltatici, 2013).
To deal with the zero trade flows within the matrix, various estimation methods have been
proposed and explored. For instance, Linders and De Groot (2006) note that an arbitrarily
small positive number (usually 0.5 or 1) is added to all trade flows to ensure that the
logarithm becomes well defined. However, given the fact that it is arbitrary means that it
lacks both theoretical and empirical justification. According to Saltatici (2013), this approach
concurs with the econometric theory if it is based on the assumption that data is censored.
However, the approach is based on restrictive assumptions that do not hold, given that the
censoring of data at zero is not a due to the fact that trade cannot be negative.
Saltatici (2013) argues that zero trade flows are a result of economic decisions based on
possibilities of undertaking profitable bilateral trade between partners. Thus, zero trade flows
do not necessarily show unobservable trade values but may be a result of a decision not to
trade because the venture is not profitable. The use of arbitrarily small positive numbers
leads to inconsistent estimates. Flowerdew and Aitkin (1982) and King (1988) demonstrated
that the arbitrary numbers also distort parameter estimates and can be manipulated to
generate any estimates to suit the modeller's liking.
In some cases, analysts simply drop the country-pairs with zero trade flows from the dataset
so as to create room for estimating the log-linear form of the gravity model using the
65
Ordinary Least Squares (OLS) estimator. Cipollina and Pietrovito (2011) Eichengreen and
Irwin (1998) and Saltatici (2013) argue that exclusion of the zero trade values leads to a
reduced number of observations, as well as loss of important information, which causes a
problem of selection bias. This problem is highly prevalent when the zero trade flows are
non-randomly distributed. To control for this bias, various econometric approaches, such as
the Tobit model, the Heckman two-step estimator and Poisson models, have been suggested.
As an example, Cipollina and Pietrovito (2011) commend the use of Tobit models, especially
when the dependent variable exhibits a significant proportion of zero flows within the sample
and positive flows for the smaller part of the sample. However, according Cipollina and
Salvatici (2007), Tobit models are unfit for this kind of analysis because they are founded on
interim and idealistic assumptions. It is further opined that these assumptions do not
essentially hold, given that censoring at zero may not necessarily be correct because of the
fact that trade cannot be negative. In detail, scholars unequivocally put it that zero trade
flows are as a result of economic decision making established on the possibility of making
profits when bilateral trade has occurred, rather than unobservable trade values.
Helpman et al. (2008) have argued that the “Heckman two-step estimator" best addressed the
problem of zero trade flows, but Cipollina and Pietrovito (2011) discredit the procedure
because it is susceptible to the presence of heteroskedasticity. Moreover, authors mention
that estimates of the log-linear form of the gravity equation obtained by use of this estimator
are biased and inconsistent. In light of the above-mentioned limitations of the Tobit model
and Heckman two-step model procedures, the use of the standard Poisson model and its
extended derivatives has been proposed.
4.4.1.3 The Poisson model and its modifications
The Poisson model and its modifications are derived from the analysis of count data. This
family of estimators can be used on non-negative continuous variables and the estimators are
not susceptible to heteroskedasticity, as well as to zero valued flows (Wooldridge, 2002).
The model’s invulnerability against such major drawbacks associated with highly
disaggregated trade data arises from the fact that it generates actual estimates of trade flows
using the log-linear rather than the log-log function. In this context, a generic gravity model
is specified as:
66
TX   K    ………..........…………..(17)ijt 0 ijt t ij ijt
where TXijt denotes trade flows from county i to j; μ0 denotes the constant; Kijt is a vector of
independent variables; and β is the vector of the corresponding coefficients of the
independent variables. ϑt and μij denote the time invariant and country-pair fixed effects,
respectively, while εijt denotes the idiosyncratic error term.
Martijn et al. (2009) argues that the use of the log-linear rather than the log-log function
prevents the under-prediction of large trade flows and total trade volumes. The predicted
values estimated by this model are almost identical to the actual input data, probably because
the Poisson model is estimated by use of the maximum likelihood model. Irrespective of the
presence of heteroskedasticity, Martijn et al. (2009) and King (1988) posit that estimates
obtained by the Poisson model are consistent and relatively more efficient.
Andersen and Van Wincoop (2003) argue that the Poisson estimator is built on two
assumptions; (i) that the actual trade volume between countries i and j has a conditional mean
(ijt) which is a function of the independent variables, and (ii) that the conditional variance of
the dependent variable equals to its conditional mean (equidispersion). Thus, given that trade
flows from country i to j (TXijt) are assumed to have a non-negative integer value, the
exponential of the independent variables can be taken, such that the conditional mean (ijt)
between country i and j is zero or positive. Mathematically, this is expressed in equation
(18).
 
  TX
exp  
    ijt
   ijt  ijtPr TX   ijt  K , ………………………......................……….......….(18)  !ijt
The conditional mean (ijt) is associated with the exponential function of a vector of
regression variables, Kijt.
Thus,
  exp    K    
ijt 0 ijt  t ij  ijt ........................................ (19)
where μ0 denotes the constant, Kijt is a vector of independent variables and β is the vector of
the corresponding coefficients of the independent variables. ϑt and μij denote the time
67
invariant and country-pair fixed effects, respectively, while εijt denotes the idiosyncratic error
term.
However, the standard Poisson model is susceptible to two problems, that is, over-dispersion
and an excess number of zero trade flows. Over-dispersion refers to a condition where the
conditional variance deviates strongly from the conditional mean. Under the Poisson model
framework, over-dispersion leads to consistent but inefficient estimates (Martijn et al., 2009).
According to Gourieroux et al. (1984) and Cameron and Trivedi (1986), this is usually
exhibited as large spurious z-values and spuriously small p-values. However, modifications
of the standard Poisson model, namely; the Negative Binomial Regression (NBR) and the
Zero Inflated Poisson (ZIP) address these problems.
The Negative Binomial Regression (NBR) model
Despite that fact that the Poisson model is grounded on the equidispersion assumption, more
often than not, this condition is not realized. In order to account for this drawback, scholars
such as Greene (1994) recommend the use of the Negative Binomial Regression (NBR)
model (See generic equation 20 below).
 1     TX ijtPr TX   1  ijt   Pr ijt TX          , ..................................(20)ijt !  1  1   1 TX ijt  ijt   ijt 
where   exp    K        , φ denotes the gamma function andijt 0 ijt t ij ijt
(∞) is the dispersion parameter.
The dispersion parameter reveals the extent to which estimates are dispersed and the larger it
is, the larger the degree of over-dispersion in the data. Unlike the Poisson model, the
variance under the NBR model depends upon both the conditional mean (ijt) and a
dispersion parameter (∞). Accordingly, Martijn et al. (2009) show that under this
specification, unobserved heterogeneity is introduced into the conditional mean. Thus, when
the dispersion parameter (∞) is allowed to assume values other than 1, then over-dispersion is
catered for by explicitly modelling between subject heterogeneity. However, just like the
standard Poisson model, the major drawback of the NBR model lies in the existence of a
larger number of observed zero trade flows, which by far outweigh the quantity of zeros that
68
the model can competently predict (Martijn et al., 2009). In this instance, the modified
version of the standard Poisson model (Zero Inflated Poisson model) was developed. It takes
this drawback into account.
The Zero Inflated Poisson (ZIP) estimator
The ZIP estimator is best used where there is an excessive number of zero trade flows within
the dataset, hence rendering the standard Poisson and NBR estimators inadequate. The Zero
Inflated model accounts for two latent groups within the population; that is, a group with
strictly zero counts and a group with a non-zero probability of having counts other than zero
(Martijn et al., 2009). The estimation process also consists of two parts. The first is the logit
or probit regression which indicates the probability of complete non-existence of trade flows.
Mathematically, it is expressed as specified in equation 21.
PrTX   1 exp ijt ij ij ijt , ...............................................(21)
The second is shown in equation 22 and consists of a Poisson regression of the probability of
each count for the group that has a non-zero probability or interaction intensity other than
zero.
   exp  
TX ijt
Pr  1 ijt ijtTX ijt ij , .................................................(22)K !ijt
Where  exp    K 
ijt 0 ijt       and  ij is the proportion oft ij ijt
observations with a strictly zero count (0 ≤  ij ≤ 1
4.4.2 Selected trade partners
Of the 27 EU member states, only 15 states (see Appendix A) are considered at this level in
order to evaluate the effect of climate change on fruit and vegetable trade flows from East
African states owing to the paucity of comparable meteorological data. Omitted countries
include Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland,
Slovakia and Slovenia, which joined the EU in 2004 while Bulgaria and Romania joined in
2007.
69
4.4.3 Data description and data sources
The empirical analysis is based on panel data collected over a period of 13 years (1988–2000)
basically attributable to time series data limitations. Choice of this time period was
specifically limited by the availability of both comparable climatological data and the trade
flow dataset based on the Harmonized System of nomenclature. The climatological dataset
(TYN CY 1.11 dataset) provided by the Tyndall Centre for Climate Change Research spans
the period 1901 to 2000, while the trade flow dataset from Trade Analysis and Information
System (TRAINS) database only begins from 1988. Therefore, this period is designated for
ascertaining the effect of climate change on East Africa's horticultural trade.
Five databases were used to obtain all the required data. These include the TRAINS database
(2013), World Bank Development Indicators (WBDI) database (2012), the Food and
Agricultural Organization (FAO) database (2013), the African Growth and Development
Policy Modelling Consortium (AGRODEP) database (2013) and the TYN CY 1.11 database
(2013). The United Nations Statistics Division Common database (COMTRADE) provides a
dataset on trade flows. The COMTRADE database (2013) is used to gather data regarding
the imports into the EU on a country basis with reference to the exporting country. Although
products are classified according to different international classifications, under this study net
imports for the EU members at HS 2-Digit level are used.
Climatological data (temperature and precipitation) were obtained from the TYN CY 1.11
database elaborated by Mitchell et al. (2004, 2005). This database comprises nine variables,
viz, daily mean, minimum and maximum temperature (degrees Celsius); daily temperature
range (degrees Celsius); frost day frequency (days); precipitation (millimetres); wet day
frequency (days); vapour pressure (hectaPascals); and cloud cover (percentage).
Meteorological observations were modelled on a 0.5° latitude by 0.5° longitude grid that
covers the world land surface (New et al., 1999, 2000). For transforming the gridded data
into country-level mean values, each grid box was allocated to a single territory and the
weighted mean calculated. Mitchell et al. (2002) candidly note that weights were chosen
according to climatological reasons. In instances where data were insufficient to acquire a
value, it was relaxed towards the 1961–1990 mean (Mitchell et al., 2002). The value
represents the long-run climatological mean (IPCC, 2007). Accordingly, these data
70
transformations result in a dataset with country level climatological information from 1901 to
2000 (Mitchell et al., 2002, 2004, 2005).
The existing literature (Barrios et al., 2010; Deschenes and Greenstone, 2007; and
Kurukulasuriya et al., 2006) identifies other meteorological data sources, such as the
Parameter-Elevation Regressions on Independent Slopes Model (PRISM) dataset which was
developed by the Spatial Climate Analysis Service at Oregon State University for the
National Oceanic and Atmospheric Administration; the Africa Rainfall and Temperature
Evaluation System, created by the Climate Prediction Center of the US National Oceanic and
Atmospheric Administration and the US Department of Defense satellites, where data are
measured by a Special Sensor Microwave Imager (SSMI), to mention but a few. However,
the TYN CY 1.11 database was selected because of two reasons; (i) the dataset has no
missing values, and (ii) unlike other databases, the dataset is availed after aggregation of
monthly averages into annual averages at country level, hence making it more user-friendly.
Data relating to Gross Domestic Product (GDP), per capita GDP, air transport (registered
carrier departures worldwide), the crop production index (2004–2006 = 100), and inflation
were obtained from the World Bank Development Indicators (WBDI) databases. As
measured by the consumer price index, inflation reveals the annual percentage change in the
cost to the average consumer of acquiring a basket of goods and services that may be fixed or
changed at specified intervals (WBDI, 2012).
To proxy for crop production in the importing EU member states, the crop production index
based on the production mean of 2004–2006 was used as the base. Notably, Belgium and
Luxembourg had missing data between 1981 and 1999. To fill the missing gaps, the annual
average of the crop production indexes of neighbouring countries sharing a common border
was used. That is, for Belgium, the annual averages for Spain, Germany, Italy and France
were used, while for Luxembourg, this index was based on data for France and Germany.
Use of this approach was based on the argument that climatic conditions of closely
neighbouring countries may not differ so much as to significantly affect agricultural
production much differently (Cardamone, 2011). The direct use of production data was
curtailed by unavailability of data over long periods.
71
For the exporting East African states, two sets of data sources were used. In the case of
Kenya and Tanzania, data for the total volume of fruits and vegetables produced was used to
proxy for the relevance of this sub-sector in international trade of horticultural produce with
the EU member states. The fruits and vegetable production data were obtained from the Food
and Agricultural Organization (FAO) database. The use of the fruits and vegetable
production data, rather than the general crop production index (as done for the importing
countries), was motivated by the desire to particularly capture whether fruits and vegetable
production influences trade flows in these commodities. However, in the case of Uganda,
production data was obtained from the African Growth and Development Policy Modeling
Consortium (AGRODEP) database.
Despite the fact that many studies have used the cardinal “great circle” formula which
estimates the earth’s shape as a sphere and computes the minimum distance along the surface,
this study uses distances acquired using the geographical distance. This was motivated by the
ease with which the data can be accessed. Distance data is measured as the air distance
between the economic centres of selected trade partners with reference from Nairobi (Kenya),
Dodoma (Tanzania) and Kampala (Uganda). These data were taken from
www.mapcrow.info/distance and www.worldatlas.com.
4.4.4 Data management
The data was managed in MS Excel and the Statistical Package for Social Scientists (SPSS)
computer program, and thereafter transferred to STATA 12 for further empirical analysis.
With the exception of the dependent variable (imports into the EU market) and the dummy
variable, all other variables were transformed into natural logs. Imports of fruits and
vegetables into the EU were used in a semi-log form so as to take into account zero trade
flows. As a rule of thumb, time series were subjected to a number of diagnostic tests as
described below.
4.4.4.1 Multi-collinearity
A number of tests can be used to ascertain if a specified model is susceptible to multi-
collinearity. Commonly used tests include: Variance Inflation Factor (VIF), Tolerance
(TOL), Pearson’s correlation, and Klein's test (Gujarati, 2003). For the purpose of this study,
72
three tests (VIF, TOL and Pearson's correlation) were used to check for multi-collinearity
between model variables.
4.4.4.2 The Tolerance (TOL) test
According to Belsley et al. (1980), the TOL test for any given dependent variable Xm can be
expressed as:
(TOL) 2m = 1 - Rm m = 1, 2, ... p-1, ......................................................................(23)
where R 2m is the R-square, when Xm is regressed on all independent variables, with the
constant term inclusive. If the TOL value is less than or equal to 0.1, this then implies that
there is supporting evidence for the existence of collinearity.
4.4.4.3 Variance Inflation Factor (VIF) test
With regard to the VIF test for the dependent variable Xm, it is computed as the inverse of the
tolerance (TOL)m.
(VIF)m = [1 / (TOLm)]
= [1 / (1 - R 2m m = 1, 2, ... p-1)], ...........................................................(24)
The VIF test quantifies the extent to which the variance of the standardized regression
coefficient bm* is inflated as a result of collinearity (Belsley et al., 1980). The term
standardized regression model refers to a regression model in which both the dependent and
independent variables are standardized into z-scores with mean 0 and standard deviation 1,
and divided by (n - 1)1/2.
In a standardized regression model framework, it is postulated that the normal
equations X'Xb = X'Y turn out to be
rXXb* = rYX, ...........................................................................................(25)
where rXX is the correlation matrix of X, rYX are the correlations of Y with the X, and b* is
the vector of standardized regression coefficients (Belsley et al., 1980). Hence, it can be
proven that (VIF)m is the k
th diagonal element of (r -1XX) . This then implies that
s2{bm*} = (s*)
2(VIF)m = (s*)
2/(1 - R 2m ), ...........................................................(26)
73
where (s*)2 is the error variance of the standardized model (Belsley et al., 1980). As a rule of
thumb, a VIF value greater than 10 implies that there is supporting evidence for the existence
of multi-collinearity which may be problematic while running regressions.
4.4.4.4 Correlation matrix test
The correlation matrix test is a measure used to show the relationship between any two
variables. According to Walker and Madden (2008), multi-collinearity between any two
different variables ranges between −1 (perfect negative correlation) and +1 (perfect positive
correlation), while the relationship between a variable with itself is +1. Correlation values
equalling to zero imply that there is no linear relationship between the two variables. As a
rule of thumb, scholars (Anderson et al., 2008, Walker and Madden, 2008, Griffiths et al.,
1993) note that if the value is not greater than the threshold value of 0.7, then the available
data poses no statistical estimation problems.
4.4.4.5 Stationarity (Unit root) test
Before estimating the models, the stationarity properties of the series were examined to
establish the order of integration of the variables. A univariate analysis of each of the time
series was carried out by testing for the presence of a unit root. When time series are non-
stationary or exhibit a unit root, the conventional econometric procedures may not be
appropriate (Engle and Granger, 1987; Enders, 1995). Notably, Granger and Newbold (1974)
posit that in the presence of non-stationary variables, ordinary least squares regression (OLS)
might result in a spurious regression; hence leading to biased and meaningless results. It was
therefore important to test for stationarity of the series data so as to set up an appropriate
methodology in the development of econometric models.
There is a variety of tests that can be used to test for unit roots of panel datasets. According
to StataCorp (2013), stationarity tests can be grouped into two categories, if one uses the
criterion based on the null hypothesis being validated by a given tests. The first category
comprises tests that have the null hypothesis that all the panels contain a unit root. Such tests
include the Levin–Lin–Chu (2002), Harris–Tzavalis (1999), Breitung (2000; Breitung and
Das 2005), Im–Pesaran–Shin (2003) (hereafter, IPS-test), and Fisher-type (Choi, 2001)
(hereafter, Choi-test). The second category comprises tests that are based on the null
74
hypothesis that all the panels are stationary. For example, Hadri (2000) uses the Lagrange
multiplier (LM) test. With the exception of the IPS-test and Choi-test which permit for
testing unbalanced panels, all of the other tests assume that the dataset has well-balanced
panels.
In that regard, two panel data tests were used to check for unit roots in the series, viz, (i) the
Levin–Lin–Chu (2002) test (hereafter, LLC-test) and (ii) the Harris–Tzavalis test (hereafter,
HT-test). Choice of these tests is based on the fact that the dataset consists of strongly well-
balanced panels. Secondly, it was desirable to explore the consistence of test results, given
the fact that the two tests are grounded in different asymptotic assumptions with regard to the
number of panels in the dataset and the number of time periods in each panel. According to
StataCorp (2013), the LLC-test requires that the ratio of the number of panels to time periods
tends to zero asymptotically, while the HT-test assumes that the number of panels tends to
infinity while the number of time periods is fixed.
Notably, these tests are constructed on the t-statistic which corresponds to the least-squares
(LS) estimator of the autoregressive parameter. As suggested by Dickey and Fuller (1979),
the tests are panel data versions of the unit root test in single time series. A variable is said to
be integrated of order I(1) if it must be differenced once to become stationary I(0). The
integration test is based on the following supporting equation:
 y   t  y    y  …………………………....(27)
t t1 i t1 t
where (yt) is the relevant time series variable, (t) is a linear deterministic trend and (ηt) is an
error term with zero mean and constant variance. The general regressions are based on the
Ordinary Least Squares (OLS) estimator. Thereafter, the estimated error terms from the final
co-integration regressions are tested for unit roots using the tests. The lagged term (yt-1) is
included to make certain that the residuals are white noise.
4.4.4.6 Normality and Over-dispersion tests for disaggregated data
Given that this analysis was grounded on disaggregated (count) data, it was necessary to
establish the nature of the distribution of the data (Normality test). According to Stata FAQ
(2013), normally-distributed data can appropriately be analysed by the Ordinary Least
75
Squares (OLS). However, in the event that data symmetry abrogates the normal distribution
assumption, the OLS estimator would then be an unsuitable estimator. Following Stata FAQ
(2013), a simple histogram was used to show the distribution pattern of the data.
As disaggregated (count) data is associated with over-dispersion (Martijn et al., 2009), it was
imperative to determine the level of dispersion within the dependent variable series. Over-
dispersion refers to a condition where the conditional variance deviates strongly from the
conditional mean. The existence of such a discrepancy leads to consistent but inefficient
estimates. Inefficiency in coefficient estimates is usually exhibited as large spurious z-values
and spuriously small p-values (Gourieroux et al., 1984; Cameron and Trivedi, 1986).
According to Stata FAQ (2013), statistical theory under the Poisson distribution, which is
also associated with disaggregated (count) data, assumes that the mean and variance are the
same. Therefore, a large deviation between the mean and the variance provides adequate
supporting evidence for the existence of over-dispersion within the series. In order to
ascertain if the data was overly dispersed, descriptive statistical analysis was carried out.
4.4.5 Computation of anomalies from meteorological data to proxy for climate change
In order to analyse the influence of climate change on East Africa’s fruits and vegetable
imports into the EU, two measures (temperature and precipitation anomalies) derived from
meteorological data were used to proxy climate change. The use of anomalies from
meteorological data is grounded on the fact that the commonly used Kyoto Protocol policies,
such as the carbon tax, do not ably reflect issues pertaining to developing economies reliant
on agriculture (Hoekman and Nicita, 2011; Bineau and Montalbano, 2011). Thus, this
renders such policies being apt for industrious economies. Secondly, such measures often
ignore the fact that climate change is a complex phenomenon, characterized by
interdependence between climatic- and weather-related natural factors.
From another point of view, Bettin and Nicolli (2012) argue that there exists no consensus
about the best indicators of climate change to be fitted in models, amidst a set of 23 aggregate
environmental indexes which could be used to proxy climate change. Furthermore, it is
postulated that most of these environmental indexes are inappropriate in explaining the trend
of climate change at a global level. Worse still, authors mention that these environmental
indicators are grossly limited by their incapability to cover many countries and cannot
76
distinguish local effects of climate change. Notably, IPCC (2007) shows that climate change
manifests itself through temperature and precipitation fluctuations (among others ways), thus
affirming that environmental indexes and Kyoto Protocol policies, as climate change proxies
may not be limited to being mutually inclusive.
In light of the above-mentioned challenges, the trend in empirical research has moved
towards the use of meteorological data, from which anomalies used to proxy climate change
have been generated. The proxies were computed at country level following the earlier work
of Bettin and Nicolli (2012), Barrios et al. (2010), Marchiori et al. (2010), and Barrios et al.
(2006). However, with due acknowledgement, it is prudent to mention that none of the above
scholarly works focuses on international trade per se.
Anomalies were computed as expressed in the general formula below:
  
T  
T i,t T i  …………………………………………………………............(28)
n,t
 SDi,t 
where Tn,t denotes temperature anomaly in degrees Celsius (°C) of country n at time t.
Country n represents either an exporter I, or importer j, Ti,t denotes temperature of exporting
country i at time t, Ti is the long-run country average temperature value, while SDi,t denotes
the long-run standard deviation of temperature in country i. Notably, the same formula
applies for computing temperature anomalies in an importing country, except that instead of
i, symbol j is used to denote an importer. A similar approach is used to obtain precipitation
anomalies in millimetres (mm/year), denoted as Pren,t. in both the importing and exporting
country
Bettin and Nicolli (2012) mention that the use of meteorological data to proxy for climate
change is advantageous over other measures because (i) it gives a better reflection of the real
effects of climate change, and (ii) such data are available in long time series for a majority of
the economies internationally. Despite the fact that meteorological data could be used in the
form of year by year variation to proxy climate change, Barrios et al. (2010), Nicholson
(1986) and Munoz-Diaz and Rodrigo (2004) argue that the use of anomalies reduces the
potential scale effects, which are associated with the year by year variation approach.
Secondly, unlike the year-by-year variation in meteorological data as a proxy for climate
change, anomalies take into consideration the likelihood of higher variance in the data for the
77
more arid countries. Furthermore, Berg et al. (2013) argue that anomalies are methodically
appropriate when undertaking large-scale estimation of climate change impacts on agro-
ecosystems, given that they eliminate any mean biases from model outputs. The authors
further argue that, "It is arguably the only direct method to get around large-scale climate
model biases for impact assessments."
4.4.6 Specification of the regression model to ascertain the influence of climate change
on East Africa's horticultural trade flows
The specified model generally builds on the gravity flow model framework. Over the years,
the gravity model has been acknowledged to be the utmost authoritative tool in explaining
bilateral trade flows (Anderson, 1979). Mathematically, empirical specification is expressed
as:
X ijt  lnY1  ln     ij 1 it 2 Y2 jt 3ln Dij 4DTit 5DT jt 6Preit  Pre jt7
8ln Agri
 ln Agri  ln Incomijt11lnInfra  lnInflat it 9 jt 10 it 12 it 13
Dlangijt
........(29)
where ln denotes natural logarithms, Xijt represents net imports of horticultural commodities
into the EU market (where j denotes a specific country within the EU) from East African
states (denoted as i if it is either Uganda, Kenya or Tanzania) in year t in ('000) US Dollars.
Y1 and Y2 represent the Gross Domestic Product (GDP) in current terms (US$) of the
exporting and importing country in year t, respectively. The term Dij denotes the
geographical distance between any two trading partners in miles, while anomalies in
temperature in the exporting and importing country at a given time are represented as DTit
and DTjt, respectively. Variables Preit and Prejt denote anomalies in precipitation in the
exporting and importing country, respectively. Variables Agriit and Agrijt represent level of
crop production in the exporting and importing country, respectively. Incomijt, infrait and
inflatit denote Linder’s income similarity index (measured as the squared per capita
differential between any two trading partners in year t), the level of infrastructure and the
level of inflation in the exporting country i in year t, respectively. Dlang is a dummy variable
for a common official language between any two trading partners (= 1 if any two trading
partners share a common official language; and Zero otherwise). εij is the error term.
78
To evaluate the influence of climate change on horticultural trade flows, total net imports of
fruits and vegetables into the EU at HS 2-digital level (chapters 7 and 8) were used. The use
of imports as the dependent variable instead of exports is motivated by the fact that imports
are more reliable since it is easier to check for the incoming flows of goods. Furthermore,
given that this study in part evaluates the impact of the non-reciprocal EU-GSP scheme, it is
prudent to consider only imports into the EU from EAC countries, rather than exports or total
trade flows. The use of combined data for fruits and vegetables rather than dealing with them
separately was aimed at overcoming the problem of non-convergence, which is associated
with disaggregated data owing to the excessive zero trade flows. The high zero trade flows
during this period (1988–2000) can probably be attributed to the fact that the East African
states had not fully liberalized their economies by the early 1990s, and this could have played
a role in deterring trade with EU member states. Hence, the high proportion of zero trade
flows.
Even after the full economic liberalization of East African (EA) states in the mid-1990s,
some EU countries, such as Luxembourg, did not trade at all in fruits and vegetables with any
of the EA states. Portugal also registered no trade at all with Uganda (for Fruits and
vegetables) and Tanzania (for Fruits only). Other EU member countries that registered zero
trade flows in either vegetable commodities (07) or Fruits commodities (08) throughout the
entire period of 1988–2000 with at least one or two EA states include Austria, Belgium,
Denmark, Finland, Greece, Ireland, Italy, Spain and Sweden. Chapter 8 (Fruits) registered a
total of nine (9) EU member states with no trade flows with EA states during the 13 year
period, while data shows that 8 EU states did not trade at all in vegetable products (chapter 7)
with EA states.
Thus, use of combined data would greatly reduce the incidences of zero trade flows, thereby
enhancing model convergence during the analysis. Rather than real values, nominal
monetary values of both fruit and vegetable imports into the EU market were used as the
dependent variable. Trade flows should be measured in nominal terms, given that the use of
price indices such as the GDP deflator or the Consumer Price Index (CPI) to deflate trade
flows does not appropriately capture the unobservable multilateral resistance terms, leading
to misleading results.
79
4.4.6.1 Other covariates used
Gross Domestic Product (Y1it / Y2jt): Gross domestic product of trading partners represents
both their production and consumption capacity, hence largely influencing the trade flow
amongst them. The GDPs are used to proxy for the economic sizes of the countries and it is
expected that an importing country’s GDP (Y2jt) plays a significant role in determining the
trade flow originating from exporting countries. Like the income of the consumer, the
importing country’s GDP determines the demand for the goods originating from exporting
countries. An exporting country’s GDP (Y1it) also helps in ascertaining the productive
capacity of the exporting country, that is, the amount of the goods that could be supplied.
Since it is expected that an exporting country’s GDP influences the trade flow of goods and
services originating from that country, an increase in the GDP of any two or more trading
countries also causes a rise in trade flows. Thus, GDP coefficients are expected to be
positive.
Distance (Dij) is a key variable used to proxy for the associated trade costs between trade
partners. Distance is a trading resistance factor that represents trade barriers such as
transportation costs, delivery time, cultural unfamiliarity and market access barriers. Among
other factors, higher transportation costs reduce the volume of trade and increase information
costs. Countries with short distances between each other are expected to trade more than
those that are far apart because of reduced transaction costs. Distance can also be used as a
proxy for the risks associated with the quality of some of the goods and the cost of the
personal contact between managers and customers.
The use of this approach follows Giorgio (2004) and Keith (2003), and is intended to avoid
the shortcomings associated with the “great circle” formula. In addition, Qadri (2012) and
Christie (2002) mention that alternative variables like "transport time", which could be used
instead of geographical distance, do not significantly improve the model's performance.
Despite the fact that the coefficient of distance is theoretically expected to negatively
influence the flow of trade between countries, some scholars (Andre and Joel, 2012;
Marimoutou et al., 2009; and Brun et al., 2005) show that distance may well positively
influence trade. However, based on the gravity model framework, the distance coefficient is
expected to be negative.
80
Basing on the work of Carrère (2006), Acosta et al. (2005) and Longo and Sekkat (2004), the
model was also augmented to capture the effect of infrastructure on trade flows. Unlike the
commonly used proxies (average of the road density, railway and the number of telephone
lines per capita), the infrastructure variable for the exporting EA states was measured by the
number of aircraft departures, following André and Joel (2012). Use of this proxy was
motivated by two factors: (i) there was limited data on road and railway networks for East
African states. For some countries, data is either unavailable or incomplete (WBDI database,
2012). (ii) The study commodities (fruits and vegetables) are very perishable and more often
than not, they are always airlifted in planes into the EU markets before they go bad. This
data was obtained from the WBDI database (2012). Availability of good infrastructure
reduces trade barriers, thereby minimizing the associated trade costs. Thus, the coefficients
of infrastructure variables are expected to be positive.
The income similarity index (Incomijt) refers to the squared per capita GDP differential. This
explanatory variable was developed by Arnon and Weinblatt (1998). In some cases, it is also
referred to as the Linder’s income similarity. Inclusion of the square of the difference in per
capita income between country i and j follows Philippidis et al. (2011) and Tang (2005). In
the gravity model framework, the index aims at distinguishing between the Linder hypothesis
and Heckscher-Ohlin (H-O) hypothesis effects. The Linder hypothesis states that countries
with similar demand patterns (often measured by a small difference in their per capita income
levels, Tang, 2005) tend to develop similar industries and these countries end up trading more
with each other in similar, but differentiated products. This variable is expected to negatively
affect trade flows between the East African region and the EU market. A negative value
implies that trade between countries with similar incomes would increase as their income
difference decreases. However, this contradicts the traditional Heckscher-Ohlin theories of
trade which argue that countries with differing per capita GDP levels tend to trade more than
those at the same level do. Thus, the coefficient may well be positive.
In order to capture an exporting country’s propensity to trade in fruits and vegetables and to
represent the relevance of the agriculture sector, the model was augmented with a variable
Agriit. The variable denotes the total volume of fruits and vegetables produced in a given EA
state i during period t in tonnes. Fruit and vegetable production data was obtained from the
FAO database (2013). According to Alvarez-Coque and Martì-Selva (2006), the coefficient
for an exporting country’s propensity to trade in fruits and vegetables (Agriit) is expected to
81
be positive, given that increased production in exporting countries may result in increased
supply of tradable commodities.
On the contrary, EU trading partners’ predisposition to import fruits and vegetables was
modelled with the use of the crop production index, denoted as Agrijt, and crop production
data were obtained from the WDI database of 2012. Ceteris paribus, increased crop
production, more so in fruits and vegetables in the EU member states, creates a reliable and
sufficient supply of such commodities. This renders importation of fruits and vegetables into
the EU market undesirable and in lieu to that, a negative coefficient for the importing
country’s crop production index is expected.
For inflation (Inflatit), an exporting country’s annual average inflation rate was also added to
the gravity model. Inflation data was obtained from the WDI database of 2012. According to
the database, inflation was measured by the consumer price index which reflects the annual
percentage change in the cost to the average consumer of acquiring a basket of goods and
services that may be fixed or changed at specified intervals. Generally, inflation is used to
measure a country’s state of macroeconomic stability. Low inflation regimes imply that a
country is in a more stable economic condition, and this is an incentive for global trade.
According to Barro (1991), lower inflation may promote higher economic growth through the
development of capital markets and trade amongst others. On the other hand, high inflation
deters global trade. Thus, the expected sign of the inflation coefficient is indeterminate.
Dlang is a dummy variable which denotes common language (=1 if share a common
language; and = 0 otherwise). This dummy captures the influence of common language on
trade flows among the trading partners. Given that countries sharing a common language
tend to have common historical ties, among other cultural similarities, this presents more
trading opportunities. Thus, the expected sign of the estimated coefficients is positive.
To control for heterogeneity across countries, this study employs a similar approach to that of
Aiello and Demaria (2009). A dummy variable (Dlang) and country-fixed effects were used.
In this context, other than estimating the effects of having a common official language on
trade flows between a country pair, the dummy variable (Dlang) is also used to overcome
heterogeneity due to observable factors. On the other hand, the unobservable heterogeneity is
overcome through the inclusion of country-fixed effects. In this regard, unobserved
82
heterogeneity was accounted for through the decomposition of the error term of equation (30)
as presented below:
 ijt   ijt ………………………….…………………………....(30)ij
where μij denotes time-invariant country-fixed effects and λijt is an idiosyncratic error term.
Accordingly, Aiello and Demaria (2009) posit that the country-fixed effects capture all
unobserved factors that influence trade flows. Thus, the specified model in equation 30
transforms into:
Xijtij lnY1it lnY2jt    1 2 3lnDij 4DTit 5DTjt 6Preit  Pre jt7
 .(31)8ln Agri
 
it 9ln Agrijt 
lnIncomijt11lnInfrait12lnInflatit13Dlang 10 ij ijt
4.4.7 Estimation techniques used
To determine the effects of climate change on East African states’ fruit and vegetable imports
into the EU market, various estimation techniques were used, depending on the properties of
the data at country level. Generally, count data estimation techniques, namely the Negative
Binomial Regression (NBR) and the Zero Inflated Poisson (ZIP) estimators were used. For
Uganda’s case, the NBR estimator was used, while the ZIP was employed on Kenya’s and
Tanzania’s datasets. The use of these count data estimation techniques closely relates to
work done by Santos Silva and Tenreyro (2006), Linders and De Groot (2006), Helpman et
al. (2008), Martin and Pham (2008), Proenca et al. (2008), Siliverstovs and Schumacher
(2009), and Burger et al. (2009). These estimation techniques overcome the challenges of
over-dispersion and excessive zero trade flows, which are highly associated with
disaggregated data (Greene, 1994; Stata FAQ, 2013). Additionally, these estimation
techniques are able to estimate the multiplicative form of the gravity equation, thus giving
more reliable results relative to the log-linear specification estimations of the model based on
the standard methods.
83
A quick overview of the expected signs of the coefficients of all variables is presented in the
table below.
Table 4.1: A summary of the expected sign of climate change variables and other
covariates on fruits and vegetable imports into the EU market
Variable Expected sign
Dependent variable (Xijt) = Net imports of commodity k to a specific country j within the
EU from East African states (denoted as i if it is either Uganda, Kenya or Tanzania) in year t
in (000) US$
Temperature anomalies in the exporting country at a given time in 0C (DTit) +/-
Temperature anomalies in the importing country at a given time in 0C (DTjt) +/-
Precipitation anomalies in the exporting country at a given time in +/-
millimetres (mm) (Preit).
Precipitation anomalies in the importing country at a given time in +/-
millimetres (mm) (Prejt).
Exporter’s Gross Domestic Product (GDP) in current US$ (lnY1it) +/-
Importer’s Gross Domestic Product (GDP) in current US$ (lnY2jt) +/-
Distance in miles between trading partners (lnDij) -
Crop production (tonnes) in the exporting country (lnAgriit) +
Crop production (index) in the importing country(lnAgrijt) -
Linder’s income similarity index (Incomijt) +/-
Level of infrastructure in the exporting country i in year t (i.e. the number of +
registered aircraft carrier departures worldwide) (lninfrait)
Inflation level in the exporting country i in year t (lninflatit) +/-
Dummy variable for common language (=1, if share common official +
language; = 0 otherwise) (Dlangij)
4.5 Estimating the effect of the EU-GSP scheme on East Africa’s fruits and vegetable
imports into the EU market
This sub-section aims at evaluating the impact of non-reciprocal preferential treatment,
particularly the EU-GSP Scheme on fruits and vegetable imports from East African states
(Uganda, Kenya and Tanzania). Achievement of this objective was based on the first
objective, under which an evaluation of the export competitiveness of the fruits and vegetable
84
commodities from East African states into the EU market was conducted. From each East
African state, two commodities were chosen following two basic principles. (i) If the
commodity exhibited an average export competitiveness index (RCA) of greater than one
across all the three countries, and (ii) if the commodity revealed the highest RCA amongst all
commodities exported from a given country. The gravity model was used, as recommended
by Linder (1961), Linnemann (1966) and Anderson and van Wincoop (2003), given that it
takes into account a broad spectrum of factors that inevitably influence trade flows.
4.5.1 Data and data sources
Evaluation of the effect of the EU-GSP Scheme on horticultural trade flows was grounded on
the 15 EU member states (EU-15) which joined before the year 2000. Focusing only on these
EU member states was based on the assumption that these states had established adequate
trade relations during the 1990s when Uganda, Kenya and Tanzania fully liberalized their
economies. Thus, late entrants into the EU were left out (See Appendix A). The research
focused on panel data collected over a period of seven (7) years (2005–2011), with the aim of
tracking the changes in the tariffs, specific duties and Tariff Rate Quotas (TRQs) of the EU-
GSP scheme during this period.
Secondly, the choice of this period was motivated by the fact that it was during this period
that the fruit and vegetable sector registered a significant increase in exports from East
Africa, coupled with drastic changes in the tariffs, specific duties and TRQs under the EU-
GSP scheme. These drastic changes in the tariffs, specific duties and TRQs could probably
be associated with the prevalent preference erosion at the time. Given the aim to evaluate a
specific policy (EU-GSP scheme) applied at product level, disaggregated data at the HS 6-
Digit level was used. These trade flow data were obtained from two databases, that is, the
ITC-Market Access Map (MAcMap) database and the TRAINS database. These provide data
on imports into the EU on a country basis with reference to the exporting country. Although
products are classified according to different international classifications, under this study net
imports by EU member states at HS 6 were used.
Data on tariff rates, specific duties and TRQs were also obtained from UNCTAD’s Trade
Analysis and Information System (TRAINS) and COMTRADE databases, a World Bank
initiative which provides information on highly disaggregated products. These databases
85
provide ad valorem equivalents (AVE) transformed from specific and complex duties by
using an estimate of unit values based on EU import statistics from the COMEXT database.
The databases use UNCTAD method 1 (U1) to convert all tariffs, duties and quotas into
AVEs. Data relating to Gross Domestic product (GDP), crop production index (2004–2006 =
100), and inflation were obtained from World Bank Development Indicators (WBDI)
databases.
Despite the fact that many studies have used the cardinal “great circle” formula which
estimates the earth’s shape as a sphere and computes the minimum distance along the surface,
this study uses distances acquired using the geographical distance. The choice to use
geographical distances is motivated by the ease with which the data can easily be accessed.
Distance is measured as the air distance between the economic centres of selected trade
partners with reference from Nairobi (Kenya), Dodoma (Tanzania) and Kampala (Uganda).
4.5.2 Data management
The data was managed in MS Excel and the Statistical Package for Social Scientists (SPSS)
computer program, and thereafter transferred to STATA 12 for further analysis. With the
exception of the dependent variable (imports into the EU market) and the dummy variables,
all the other variables were transformed into natural logs. Given that the natural log of zero is
undefined, imports of fruits and vegetables into the EU were used in a semi-log form to take
into account zero trade flows. Panel data was the subjected to a number of diagnostic tests,
as described in detail (See section 4.4.3).
4.5.3 The preference margin as a proxy for the effects of the EU-GSP Scheme on fruit
and vegetable imports into the EU market
More often than not, variables that capture unilateral preferential treatment, such as the EU-
GSP Scheme, are measured using dummy variables. However, this is problematic given that
dummy variables simultaneously capture a range of other country-pair-specific effects. In
addition, dummy variables do not differentiate between the different preferential trade policy
instruments (preferential tariff margins, preferential quotas, reduced ‘entry prices’), nor do
they tell the difference in the level of trade preferences. This therefore implies that dummy
variables assume that the level of preference margin across products under a given
86
preferential treatment (like the EU-GSP Scheme) is the same. However, in reality,
preferential margins may greatly differ, depending on the commodity under consideration,
and may also differ across the various preferential treatments (MAcMap, COMTRADE and
TRAINS databases).
Thus, in order to analyse the effects of the EU-GSP Scheme on East Africa's fruits and
vegetable imports into the EU, a count variable (preference margin) was used to capture such
effects. For each East African state and for each selected commodity, a preference margin
was computed. A competition-adjusted Preferential Margin measure proposed by Low et al.
(2005) was used. Following Cardamone (2010), the preference margin was computed as the
absolute difference between the trade weighted applied Most Favoured Nation (MFN) rate
and the Ad Valorem Equivalents as expressed below:
Preference Margin = Trade-weighted applied MFN rate – AVE ……………..(32)
where AVE denotes the Ad Valorem Equivalent. An ad valorem equivalent tariff refers to a
tariff presented as a proportion of the value of goods cleared through customs. It is the
equivalent of a corresponding non-ad valorem tariff measure based on unit quantities, such as
weight, number or volume.
The measure of preference margin was based on a combination of trade-weighted applied
MFN rates, tariff rates, specific duties and Tariff Rate Quotas, which differs from the
approach taken by other scholars (Cipollina et al., 2013; Raimondi et al., 2012; Cipollina and
Salvatici, 2009). The trade-weighted applied MFN rate takes into account the global
competitors at tariff line level and the weights are based on reference group imports. In this
context, East Africa’s reference group is the set of all countries categorized as the Least
Developed Countries (LDCs). These countries are granted similar preferential treatment
within the EU market; hence they are the major competitors with Kenya, Tanzania and
Uganda at that level. According to Fugazza and Nicita (2010), the use of a combination of
these trade policy instruments enhances the disclosure of other additional advantages or
disadvantages that come along with preferential treatment.
87
4.5.3.1 Computation of the Ad Valorem Equivalents (AVEs)
Agriculture is known to be one of the most protected sectors, globally (Bouët et al., 2004).
Agricultural protection is extended to sensitive commodities in the form of various
instruments, such as ad-valorem tariffs, specific duties or a combination of the two,
antidumping duties and tariff rate quotas (TRQs). These instruments cannot be directly
compared or summed, thus implying that they cannot readily be used in large-scale modelling
exercises. For this reason they are converted into Ad Valorem Equivalents (AVEs) using
various methods.
For the purpose of this study, AVEs were obtained using UNCTAD's three-step method for
estimating unit values of the commodities, viz, (i) the use of tariff line import statistics of the
destination country available in TRAINS; (ii) but if (i) is not available, then HS 6-digit
import statistics of the market country from COMTRADE are opted for. In the event that
both (i) and (ii) are not available, then the HS 6-digit import statistics of all OECD countries
are employed. Thus, when the unit value has been estimated, it is then used for all types of
rates (MFN and preferential rates).
This conversion does not, however, take into account mixed tariffs which involve the use of
either a maximum or a minimum operator. The choice of this approach over the others relies
on the following facts: (i) the revenue method is cumbersome, disregards the question of
quality differences, and is clearly unfit in the presence of many preferential agreements; and
(ii) the price wedge estimation method is hardly tractable when using highly disaggregated
data (Bouët et al., 2004; WTO, 2003; Gibson et al., 2001).
4.5.4 Specified regression model to capture the effect of the EU-GSP Scheme on East
Africa’s fruits and vegetable trade flows into the EU market
An augmented gravity model presented in the equation below was used, with total monetary
value of commodity l from the i th East African state to j th EU member state in year t in
thousand US Dollars (Mijlt) as the dependent variable.
88
M ijlt ij  1lnY1it  2lnY 2 jt  3ln Dij  4ln PMijlt  5ln Inflit   lnCOSTEXP6 it
(33)
  lnGOVit   ln FDIit   lnCOSTBIZit   Dlang7 8 9 10 ij  ijlt
lnY1it and lnY2jt denote the natural logarithm of current Gross Domestic Product (GDP) of
each i th EA state and j th EU member state, respectively in year t in US Dollars; lnDij is
distance between the economic centres (Nairobi, Dodoma and Kampala) and their jth trading
partner’s commercial centre in miles. The variable lnPMijlt represents the preference margin
granted under the EU-GSP scheme (excluding the Drugs Regime and the Everything But
Arms (EBA initiative). Lninflatit is the mean annual inflation rate of each EA state in year t;
lnCOSTEXPit denotes the cost to exporting a 20-foot container in US dollars, while lnGOVit
captures role of the public sector and government institutions in fostering trade. The variable
lnFDIit represents net inflows of foreign direct investment of each EA state in year t in
current US Dollars. lnCOSTBIZit refers to the cost of establishing a business. Dlangij is a
dummy variable that values 1 when countries i and j share the same official language, and 0
otherwise. εijlt denotes the associated error term.
Gross Domestic Product (Y1it / Y2jt): The gross domestic product of trading partners
represents both the productive and consumption capacity, hence largely influencing the trade
flow amongst them. Thus, GDP coefficients are expected to be positive. Similarly, the
distance variable (Dij) is used to proxy the associated trade costs between trade partners.
Theoretically, the coefficient on this variable is expected to negatively influence trade flows.
However, scholars (André and Joel, 2012; Marimoutou et al., 2009; and Brun et al., 2005)
show that distance may not necessarily deter trade, especially in instances where the
importing country's economy is very large relative to the exporter's and depending on the
type of commodities being transacted. Based on the gravity model architecture, the
coefficient of distance is expected to be negative.
The preference margin variable (lnPMijlt) captures the effect of the EU-GSP scheme at
specific commodity level. Depending on the type of commodity and country of origin, the
expected sign of the coefficient on the variable is indeterminate, given that in some instances
a higher preference margin may favour other reference group countries (LDCs), thus these
outcompete the EA states in the EU market. On the other hand, higher preference margins
89
may boost imports of commodities into the EU, given that Uganda, for example, produces
organic agricultural products which are in high demand in developed economies.
To capture the effect of macro-economic stability on trade, the model was augmented with
the exporting country's mean annual inflation rate (Inflait). This data was obtained from the
WBDI database (2012). Low inflation regimes imply that a country is in a more stable
economic condition, and this is an incentive for global trade. According to Barro (1991),
lower inflation may promote higher economic growth through the development of capital
markets, trade among others. On the other hand, high inflation deters global trade. Thus, the
expected sign of the inflation coefficient may be positive or negative.
lnCOSTEXPit denotes the cost to exporting a 20-foot container in US dollars. The cost
entails all the associated fees (documents fees, customs clearance and technical control
administrative fees, customs broker fees, terminal handling charges and inland transport)
required to accomplish the procedures to export. Notably, the variable does not include tariff
or trade costs (WBDI database, 2012). Increasing export costs are a barrier to trade, thus a
negative sign is expected for this variable.
The variable lnGOVit represents the role of the public sector and government institutions in
fostering trade. The average un-weighted Country Policy and Institutional Assessment
(CPIA) index for transparency, accountability, and corruption was used. The index ranges
between one (1) and six (6) (1 = low, to 6 = high) and data was obtained from the WBDI
database (2012). The index evaluates the extent to which public sector administrators can be
held answerable for the use of public resources and for the results of their actions by the
electorate and by the legislature and judiciary, and the extent to which public employees
within the executive are required to account for administrative decisions, use of resources,
and results obtained. The estimated coefficient for this variable is expected to be positive.
The net inflow of foreign direct investment (lnFDIit) measures the influence of FDI on
international trade flows. Data on FDI was obtained from the WBDI database (2012). On
the one hand, FDI may be seen as a means through which to boost trade, given that it may
serve the purpose of enhancing the efficient use of the factor endowments in a given country,
thereby boosting production and trade. On the other hand, FDI in a country may deter trade
90
flows from a country after a certain time lag, probably due to focusing on producing for the
domestic market. Thus, the expected sign on the coefficient of this variable is indeterminate.
The cost of establishing a business (lnCOSTBIZit), expressed as the proportionate share of
per capita Gross National Income (GNI), was obtained from the WBDI database (2012). The
existence of many bureaucratic procedures and the associated costs while opening up a
business tend to discourage investment in business ventures. Thus, the expected sign of the
estimated coefficient is negative. Dlangij is a dummy variable which denotes common
language ( = 1 if sharing a common language; and = 0 otherwise). This dummy captures the
influence of common language on trade flows among the trading partners. Given that
countries sharing a common language tend to have common historical ties, among other
cultural similarities, this represents more trading opportunities. Thus, the expected sign of
the estimated coefficients is positive.
To control the heterogeneity across countries, country pair time-invariant effects and a
dummy variable were used. Other than estimating the effects of having a common official
language on trade flows between a country pair, the dummy variable (Dlang) is also used to
overcome heterogeneity due to observable factors. The unobservable heterogeneity is
overcome through the inclusion of country-fixed effects. In this regard, unobserved
heterogeneity was accounted for through the decomposition of the error term of equation
(34), as expressed below:
 ijlt   ijlt ……………………….………………………….......… (34)ij
where μij denotes time-invariant country pair effects and λijlt is an idiosyncratic error term.
Accordingly, Aiello and Demaria (2009) posit that the country pair time-invariant effects
capture all unobserved factors that influence trade flows. Thus, the specified model in
equation 34 becomes:
Mijltij1lnY1it2lnY2jt3ln Dij4ln PMijlt5lnInflit lnCOSTEXP6 it
....(35)
 lnGOVit lnFDIit lnCOSTBIZit Dlang 7 8 9 10 ij ijijlt
An overview of the expected signs of the coefficients of all variables is presented in the table
below.
91
Table 4.2: A summary of the expected sign of the effect of the EU-GSP scheme and
other covariates on fruits and vegetable imports into the EU market
Variable Expected sign
Dependent variable (Mijlt)  = Total monetary value of commodity l from the i th East
African state to j th EU member state in year t in '000 US Dollars (Mijlt)
Preference margin of a specific commodity, expressed as a percentage of +/-
the product value (lnPMijlt)
Exporter’s Gross Domestic Product (GDP) in current US$ (lnY1it) +/-
Importer’s Gross Domestic Product (GDP) in current US$ (lnY2jt) +/-
Distance in miles between trading partners (lnDij) -
Exporting country's mean annual inflation rate (lninflatit) +/-
Cost to exporting a 20-foot container in US$ per container (lnCOSTEXPit) -
The role of the public sector and government institutions, expressed as an +
index (from 1=low to 6= high) (lnGOVit)
Net inflow of foreign direct investment in current US$ (lnFDIit) +/-
The cost of establishing a business, expressed as percentage of Gross -
National Income (GNI) per capita (lnCOSTBIZit)
Dummy variable for common language (=1, if share common language; = +
0 otherwise) (Dlangij)
4.5.5 Estimation techniques used
To ascertain the effects of the EU-GSP scheme on East Africa’s horticultural exports into the
EU market, the Zero Inflated Poisson (ZIP) and Negative Binomial Regression (NBR)
estimators were used. The use of the NBR estimator follows Green (1994) and Stata FAQ
(2013) who argue that it is in most instances an apt estimator of datasets characterized by
over-dispersion. On the other hand, the use of ZIP model relates to work done by Santos
Silva and Tenreyro (2006), Linders and de Groot (2006), Helpman et al. (2008), Martin and
Pham (2008), Proenca et al. (2008), Siliverstovs and Schumacher (2009), and Burger et al.
(2009) who urge that it can deal with excessive zero values and over-dispersion. Highly
disaggregated data, upon which this analysis is grounded, is very susceptible to over-
dispersion of the data and too many zero trade flows (Martijn et al., 2009). Additionally, the
estimation techniques are able to estimate the multiplicative form of the gravity equation,
92
thus giving more reliable results relative to the log-linear specification estimations of the
model based on the standard methods.
4.6 Predicting unilateral Trade Potential and performance
The out of sample approach was used to calculate each East African state's potential in
exporting a given commodity to the EU market. This approach uses the estimated variable
coefficients to predict the trade potential. The specified model below was used to predict the
trade.
M ijlt ij  1lnY1it  2lnY 2 jt  3ln Dij  4ln PMijlt  5ln Inflit   lnCOSTEXP6 it
....(36)
  lnGOVit   ln FDIit   lnCOSTBIZ7 8 9 it   Dlangij  10 ij ijlt
where αij denotes the constant and β1- β10 represent coefficients of the variables already
defined in equation (34). According to researchers (Wang and Winters, 1992; Hamilton and
Winters, 1992; and Brulhart and Kelly, 1999) the difference between the actual and predicted
trade flows represent the export potential. Thus, a negative value implies that there exists un-
exhausted export potential, hence supporting evidence that there is room for trade expansion.
On the other hand, a positive value implies that there is hardly any room for expansion of
trade. The same data spanning a seven-year period (2005–2011) that were employed to
evaluate the impact of the EU-GSP Scheme were also used in this scenario.
The analysis followed Lubinga (2009), Lie et al., (2002) and Amita (2004). Trade
performance was evaluated using two indices, the Relative difference (Rd) and Absolute
difference (Ad). The Relative difference (Rd) index was computed as expressed in equation
(37). The mean predicted trade value together with the mean actual trade value were used as:
 ijltijlt
Rdijlt    *100..........................................................................(37) ijltijlt
where Rdijlt denotes relative difference of each East African state's trade flows with trade
partner j. ψijlt denotes mean actual trade and  ijlt is the mean predicted trade. The relative
difference index varies between −1 and 1. Relying on the existing status quo, the Rd index is
an indicator of the status of trade performance between trade partners and it gives an insight
93
into the future direction of trade (Chen et al., 2007). Positive values imply that there exists
good trade performance, an indication of cooperation between the trading parties.
The Absolute difference (Adijlt) Index was also used to analyse trade performance.
ADijlt ijltijlt ........................................................................................(38)
where ψijlt denotes mean actual trade,  ijlt is the mean predicted trade, and Adijlt is the
absolute difference between a given East African state and its EU trading partner j.
The Absolute difference index can also be used to analyse the good or bad trade performance
between trade partners on top of analysing the future direction of trade of the exporting
country. Notably, Rd is an opportune index in determining trade performance but its major
drawback lies in its failure to explain the relative nexus between actual and predicted trade
volumes, given that it does not explain the divergence in volumes between them (Lubinga,
2009). Thus, if 0 < Rd < 1, it is hard to articulate the un-exhausted trade. Also, when -1<Rd
<0, hardly will one know the trade potential of East African economies and their trade
partners. However, by using Ad it is possible to calculate the gain or owned trade potential
value to identify the future trade partner of the exporting country (Chen et al., 2007). This
study thus employs the Absolute difference index to validate the results obtained by the
Relative difference index.
4.7 Conclusions
Chapter 4 has aimed at elaborating the procedures that were used to achieve the objectives of
the study. The Revealed Comparative Advantage (RCA), also known as Balassa’s index, was
used to determine the export competitiveness of various fruit and vegetable products from
Kenya, Tanzania and Uganda into the EU market. Generated sets of temperature and
precipitation anomalies were used as proxies for climate change to determine the influence of
climate change on East Africa’s horticultural trade flows within the gravity model
framework. Depending on the dataset properties of a given country, two estimation
techniques, viz, the Zero Inflated Poisson (ZIP) estimator and Negative Binomial Regression
(NBR) estimator were used to take into account over-dispersion and excess zeros. The use of
meteorological data to proxy for climate change over other measures reveals the real effects
of climate change and the data can be obtained over long time series. In addition, the use of
94
anomalies reduces the potential scale effects that are evident if annual variations are used,
and anomalies take into account the likelihood of higher variance in the data for arid areas.
Finally, the use of anomalies is so far the only direct method of dealing with the large-scale
model encountered while undertaking impact assessments.
In order to determine the effect of the EU-GSP Scheme on the various East African fruit and
vegetable commodities imported into the EU, the preference margin was used within the
gravity model framework. The preference margin is a count variable rather the commonly
used dummy approach. Preferential margin was computed as the difference between the
average trade-weighted MFN rate and the Ad Valorem Equivalents granted under the EU-
GSP scheme. Again, depending on the dataset properties of a given commodity from a given
East African state, two estimation techniques (Zero Inflated Poisson (ZIP) and Negative
binomial Regression (NBR)) were used to take into account over-dispersion and excess zeros.
Over-dispersion and excess zeros are key data problems encountered when dealing with
highly disaggregated data.
The “out-of-sample” approach was used to simulate trade potential and trade performance.
Two indicators were used, that is, the Relative difference index and the Absolute difference
index. Both indicators were used concurrently to validate results obtained from the other
index.
95
CHAPTER FIVE:
RESULTS AND DISCUSSIONS
5.1 Introduction
This chapter presents the results and discussion of the export competitiveness of East African
states in horticultural commodities, the influence of climate change on EA's selected
horticultural exports flows, and the effect of the EU-GSP scheme on East Africa’s fruits and
vegetable imports into the EU market, as well as the projected unilateral Trade Potential and
performance of the selected commodities from EA states. The sub-sections begin with results
relating to export competitiveness, followed by the influence of climate change on the
horticulture sector and then the findings concerning the impact of the EU-GSP scheme on
East Africa’s fruits and vegetable commodities. Results and discussion of the predicted trade
potential and performance are presented last.
5.2 Export competitiveness of East African states in horticultural commodities
The Revealed Comparative Advantage (RCA) index was used to establish the export
competitiveness of fruit and vegetable exports from the East African region into the European
Union (EU). At country level, competitiveness was calculated using the monetary value of
traded commodities, disaggregated at two levels, namely (i) at HS 4-Digit, and (ii) at HS 6-
Digit levels of data disaggregation.
The results are presented in Table 5.1. All RCA values greater than one (1) depict that a
given commodity exhibits export competitiveness within the EU market. Out of the 27 tariff
lines considered at HS 4-Digit level for each country (i.e. 81 observations), only 16 tariff
lines (with asterisks in table 5.1) were observed to have export competitiveness within the EU
market for at least one of the EA states. Furthermore, 30 observations exhibited no
competitiveness within the EU across the three EA states – potatoes (0701), tomatoes (0702),
Lettuce and chicory (0705), Cucumbers and gherkins (0707), Vegetables (0711), grapes
(0806), Apples, pears and quinces (0808), Stone fruit (0809), Fruits (0812), Fruit, dried, n.e.s
(0813) and Peel of citrus fruit or melons (0814).
96
Table 5.1: The mean revealed comparative advantage index at HS-4 Digit level
No CODE Product description RCA index (Mean 2008- 2011)
(HS-4) Kenya (Av. Tanzania Uganda
2007-10)
1 0701 Potatoes, fresh or chilled 0.01 0.00 0.46
2 0702 Tomatoes, fresh or chilled 0.23 0.00 0.34
3 0703 Onions, shallots, garlic, leeks, etc. 23.08* 0.11 3.39*
4 0704 Cabbage, cauliflower, kohlrabi & kale, 0.01
fresh, chilled 45.49* 0.12
5 0705 Lettuce and chicory, fresh or chilled 0.83 0.00 0.04
6 0706 Carrots, turnips, beetroot, etc. fresh or
chilled 1.62* 0.01 3.08*
7 0707 Cucumbers and gherkins, fresh or chilled 0.04 0.00 0.15
8 0708 Leguminous vegetables, shelled or
unshelled, fresh 2,262.37* 24.86* 32.12*
9 0709 Other vegetables, fresh or chilled 364.76* 0.55 3,978.5*
10 0710 Vegetables (uncooked, steamed, boiled) 42.95* 0.02 3.68*
11 0711 Vegetables provisionally preserved, not
ready to eat 0.00 0.00 0.73
12 0712 Vegetables, dried, not further prepared 3.78* 0.00 2.43*
13 0713 Vegetables, leguminous dried, shelled 1.99* 43.41* 20.57*
14 0714 Manioc, arrowroot, salep, etc. 0.05 0.00 259.63*
15 0802 Nuts except coconut, brazil & cashew 23.49* 0.08 8.24*
16 0803 Bananas, including plantains 38.03* 0.12 2,265.9*
17 0804 Dates, figs, pineapple, avocado, guava,
fresh or dried 288.67* 1.13* 810.98*
18 0805 Citrus fruit, fresh or dried 0.81 0.00 6.52*
19 0806 Grapes, fresh or dried 0.16 0.02 0.63
20 0807 Melons, watermelons and papaws 2.65* 0.00 70.81*
21 0808 Apples, pears and quinces, fresh 0.43 0.00 0.17
22 0809 Stone fruit, fresh (apricot, cherry, etc.) 0.10 0.52 0.30
23 0810 Fruits n.e.s, fresh 46.90* 5.37* 120.25*
24 0811 Fruits and nuts, uncooked boiled or
steamed, frozen 1.41* 0.00 156.23*
25 0812 Fruits, nuts provisionally preserved, not
ready to eat 0.23 0.00 0.03
26 0813 Fruit, dried, n.e.s, dried fruit and nut
mixtures 0.06 0.14 14.10*
27 0814 Peel of citrus fruit or melons 0.00 0.67 0.00
Mean RCA index across all commodities per country 116.67** 2.85** 287.38**
* Denotes export competiveness at country level.
** Mean RCA Index across all commodities was computed as the average of four years
(2008–2011). Owing to limited data for 2011, Kenya's mean RCA index was based on the
2007–2010 period.
Source: Author’s own calculations
97
Generally, Uganda is the most competitive in fruits and vegetables within the EU market,
followed by Kenya and Tanzania. According to the mean RCA index across all commodities
at country level, Uganda (287.38) surpasses Kenya (116.67) by approximately 2.5 times,
while Tanzania lies at the bottom, with an estimated index of 2.89. Among the competitive
horticultural commodities, Uganda ranked highest (3,979) in exporting “Other vegetables,
fresh or chilled (0709)”, followed by Kenya (2,262) in the exports of “Leguminous
vegetables (0708)”, while Tanzania had “Vegetables, leguminous dried, shelled (0713)” as its
best exportable horticultural commodity.
Notably, at the HS-4 digit level the high RCA index associated with Uganda’s and Kenya’s
products may be attributable to a tariff line aggregation effect. On the other hand, Uganda’s
general outstanding export competitiveness may be explained by the convenient climatic and
soil factors which favour production of these crops. However, the poor export
competitiveness of exotic commodities such as apples, pears and quinces; stone fruit, fresh
(apricot, cherry, plum, peach, etc.); and preserved fruits and vegetables may be on one hand
attributed to the tropical climatic conditions that do not favour production of these
commodities, and the limited technological advancements in adding value to such products.
Subsequently, analysis at HS-6 digit level focused on only sensitive tariff lines as categorized
under the EU-GSP scheme, and only the top two tariff lines which presented the highest RCA
values per country (at HS-4 digit level) were considered. At country level, the selected HS-4
digit level tariff lines for further assessment in the case of Kenya included “Leguminous
vegetables (0708)” and “Other vegetables, fresh or chilled (0709)”. For Tanzania,
“Leguminous vegetables (0708)” and “Vegetables, leguminous dried, shelled (0713)” were
chosen, while Uganda registered “Bananas, including plantains, fresh or dried (0803)”, as
well as “Other vegetables, fresh or chilled (0709)”. Further disaggregation of the selected
tariff lines led to a total of twenty-one (21) horticultural commodities (see Table 5.2 below)
that were evaluated at HS-6 digit level.
98
Table 5.2: Mean revealed comparative advantage index at HS-6 Digit level
CODE CODE Product description RCA index (Mean 2008- 2011)
(HS-4) (HS-6) Kenya Tanzania Uganda
070810 Peas (Pisum sativum) 17.88** 0.56 0.00
0708 070820 Beans (Vigna spp., Phaseolus 3.70** 2.23** 1.23**
spp.)
070890 Other leguminous vegetables 0.55 0.01 0.26
(excl. of 070810, 070810)
070920 Asparagus, fresh/chilled 8,504.32** 0.00 0.00
070930 Aubergines (eggplants) 409.99** 0.00 47.08**
070940 Celery other than celeriac 0.12 0.00 0.00
070951 Mushrooms of the genus 0.09 0.00 0.00
Agaricus
070959 Mushrooms & truffles (excl. of 750.34** 0.00 0.73
0709
070951), fresh/chilled
070960 Fruits of the genus Capsicum or 164.28** 0.00 27,668.87**
of the genus Pimen
070970 Spinach, New Zealand spinach 509.19** 0.00 0.00
and garden spinach
070990 Vegetables, n.e.s. in 07.01- 0.15 24.60** 1.67**
07.09, fresh/chilled
071310 Peas dried, shelled 0.02 0.00 0.00
071320 Chickpeas, dried, shelled 0.00 0.06 0.00
071331 Urd, mung, black/green gram 1.33** 1.06** 0.00
beans
071332 Beans, small red (Adzuki) dried 0.00 0.00 0.00
071333 Kidney beans & white pea 1.73** 13.51** 0.00
beans
0713 071339 Beans dried, shelled 0.02 0.03 0.00
071340 Lentils dried, shelled 0.09 0.00 1.18**
071350 Broad beans &horse beans dried 0.00 0.00 0.00
071390 Leguminous vegetables dried 0.00 0.00 0.08
0803 080300 Bananas, including plantains 31.09** 2.40** 25.98**
Mean RCA index across all commodities 494.99*** 2.12*** 1,321.29***
** Denotes export competiveness at country level.
*** Mean RCA Index across all commodities was computed as the average of four years
(2008–2011). Owing to limited data for 2011, Kenya's mean RCA index was based on the
2007–2010 period.
Source: Author’s own calculations
99
At HS-6 digit level of disaggregation, empirical results still show that Uganda out-performs
the other East African states. Uganda’s mean export competitiveness across all commodities
is estimated at 1,321; followed by Kenya (495) and then Tanzania (2). However, the results
show that Kenya exhibits export competitiveness in more commodities (10) than Uganda and
Tanzania, which are only competitive in six (6) and five (5) commodities, respectively.
Uganda’s high mean RCA index (1,321) across all commodities is attributable to the “Fruits
of the genus Capsicum or of the genus Pimen, fresh/chilled (070960)”.
According to Kleih (2007), fruits of the genus Capsicum or of the genus Pimen, most
especially scotch bonnet, dominate Uganda’s horticultural exports and Uganda is one of the
top suppliers of hot pepper to Europe. The performance of this commodity may also be
attributed to the fact that it is less bulky to airlift and has a high monetary value. This greatly
reduces the associated transaction costs. In addition, Uganda is endowed with an abundance
of natural resources, such as fertile soils, a bi-modal rainfall pattern, and the tropical climatic
conditions which are very conducive to a wide spectrum of horticultural crops. All these
factors enhance organic production of the horticultural commodities which are in high
demand in the developed economies, hence the competitiveness. This argument is supported
by the fact that Uganda has the most developed sector of certified organic producers (IFOAM
and FiBL, 2006). Further, ACODE (2006) postulates that Uganda has the lowest agro-
chemical usage in Africa.
For Kenya, “Asparagus, fresh/chilled (070920)” was noted to be the most competitive
horticultural commodity within the EU market, followed by “Mushrooms & truffles,
fresh/chilled (070959)” and “Spinach, New Zealand spinach and garden spinach, fresh/chilled
(070959)”, in that order. Conversely, some commodities, like Adzuki beans (071332),
exhibited a competitive disadvantage. Kenya’s strong export competitiveness in the 10
horticultural commodities may be associated with the country’s close proximity to the sea,
thus greatly reducing the burden of transport costs. In addition, Kenya’s success story may
be attributable to the efforts of both the government and the private sector in developing the
industry.
For instance, private institutions, such as the Fresh Produce Exporters Association of Kenya
(FPEAK), have been in the lead to ensure that most farmers within the horticulture sector are
100
duly certified under trade enhancing international standards like GLOBALG.A.P. To this
effect, Kenya is rated to have the highest number of certified farmers of horticultural
commodities within the EA region (Government of Kenya, 2012). This precedent inevitably
serves as the impetus from which Kenya’s exportable horticultural products have become
very competitive within the EU market.
In the case of Tanzania, “Vegetables, n.e.s, fresh/chilled (070990)” were the most
competitive within the EU market, with an estimated RCA index of about 24.5. Other
commodities that exhibited export competitiveness include: Kidney beans & white pea beans
dried shelled, whether or not skinned or split (071333); Bananas, including plantains, fresh or
dried (080300); Beans (Vigna spp., Phaseolus spp.) (070820); and Urd, mung, black/green
gram beans dried shelled, whether/not skinned/split (071331), in that order.
The research findings imply that the respective countries would be better off if they invested
more resources in the production and exports of horticultural commodities over which they
have export competitiveness. Uganda's study findings relate with the results of Kleih et al.
(2007) who reported that pepper (particularly, Scotch Bonnet), the East African Highland
banana and apple banana are among the chief fruit and vegetable exports into the EU market.
Results concerning Kenya concur with those of Nyangweso and Odhiambo (2004) who note
that asparagus, among other vegetables, is one of the major horticultural commodities in the
country. In the case of Tanzania, study findings are in tandem with results of MMA (2008)
who note that vegetables, especially baby vegetables, destined for the EU market present a
strong pillar for the horticulture sector in the country.
For the subsequent analysis, a few selected commodities from each country were considered.
Detailed discussions are in the following sub-sections.
5.3 The influence of climate change on East Africa's horticultural trade flows
The relationship between climate change and imports of horticultural commodities into the
EU market was examined using the gravity flow model framework. However, as a rule of
thumb while using time series data, a number of diagnostic tests were undertaken to ascertain
whether the variables were appropriate for the subsequent econometric analysis.
101
5.3.1 Diagnostic test results
Multi-collinearity test results presented in Appendix B reveal that serial correlation among
the variables was not a serious problem for the various datasets used to assess the nexus
between climate change and trade flows of horticultural commodities from EA states. That
is, in the case of the Tolerance (TOL) test, values were more than 0.1, while for the mean
Variance Inflation Factor (VIF) test all values were less than the threshold value of 10.
Results of the correlation matrix test also divulge that there was no correlation problem, since
correlation values were not greater than the threshold value of 0.7.
Over-dispersion test results presented in Table 5.3 show that there exists adequate supporting
evidence for the existence of over-dispersion within the data series since the conditional
variance deviates by far from the conditional mean. Similarly, normality test results
(Appendix C) reveal that the disaggregated data series does not comply with the normal
distribution assumption. This implies that Ordinary Least Squares (OLS) could not be used
for this scenario.
Table 5.3: Over-dispersion test results of horticulture exports by country
Kenya (N=195) Tanzania (N=195) Uganda (N=195)
Variable Mean ('000 Variance Mean ('000 Variance Mean ('000 Variance
US$) US$) US$)
Horticulture
imports into EU 4,164.16 1.59e+08 323.094 1,041,719 84.15 138,017
from EA (Xijt)
Source: Author’s own calculations
The unit root test results grounded on the Levin–Lin–Chu test (LLC-test) presented in Table
5.4 below signify that the variables are integrated of various orders. With the exception of
the data series for horticultural exports (Xijt) to the EU market, Linder’s income similarity
index (Incomijt) and the level of infrastructure in the exporting country (lninfrait), Kenya's
time series were inherently stationary. Consequently, data series for horticultural exports
(Xijt) were found to be integrated of order 2, while Linder’s income similarity index
(Incomijt) and the level of infrastructure in the exporting country (lninfrait) were integrated
of order 4.
102
Table 5.4: LLC Panel Unit Root test results by country
Kenya Tanzania Uganda
Variable LLC-test LLC-test LLC-test
Levels Levels I(1) Levels I(1)
Horticultural imports into jth -0.72 1.75 7.52
state within EU from EA state i
('000 US$) (Xijt)
Exporter's temperature -13.7*** -2.84*** -2.70***
anomalies (DT 0it) ( C)
Importer's temperature -3.59*** -3.59*** -3.59***
anomalies (DTjt) (
0C)
Exporter's precipitation -15.49*** -0.88 -3.19*** 0.36 -7.06***
anomalies (DPreit) (mm)
Importer's precipitation -3.47*** -3.47*** -3.47***
anomalies (DPrejt) (mm)
Exporter’s GDP in current US$ -1.75** 2.21 -4.76***
(lnY1it)
Importer's GDP in current US$ -8.92*** -8.92*** -8.92***
(lnY2jt)
Exporter's Crop production -4.03*** -3.00*** 2.23
(lnAgriit) (tonnes)
Importer's Crop production -3.24*** -3.24*** -3.24***
index (lnAgrijt)
Linder’s income similarity index 6.08 - 6.07
(Incomijt)
Exporter's infrastructure level 14.01 - -4.76***
(lninfrait) (aircraft departures)
Exporter’s inflation level -5.11*** 10.52 -5.50***
(lninflatit)
***, **, * denote significance at 1%, 5% and 10% level respectively. I(1) denotes integration
of first order.
Similarly, Tanzania's data series for horticultural exports (Xijt) and Gross Domestic Product
(lnY1it) became stationary after the seventh lag, hence are said to be integrated of order seven
(7). On the other hand, the series for Tanzania's inflation and precipitation anomalies were
integrated of order four (4) and two (2), respectively. In the case of Uganda, data series for
horticultural exports (Xijt) were integrated of order six (6) while Linder's income similarity
index (Incomijt) became stationary after the fourth lag (Integrated of order 4). Uganda's data
series for precipitation anomalies and crop production (lnAgriit) were integrated of order one
(1) and two (2), respectively. Otherwise, Uganda's other data series were also inherently
103
stationary. Given the evidence that the series were stationary, appropriate econometric
estimations were made to establish the nexus between climate change and horticultural trade
flows into the EU market.
5.3.2 The influence of climate change on East Africa's horticultural trade flows into
the EU market
The influence of climate change on trade flows was examined using anomalies in temperature
and precipitation in both the exporting and importing countries within the gravity flow model
framework. The study employed count data estimation techniques to overcome the
challenges of over-dispersion and excessive zero trade flows, which are usually encountered
with disaggregated data. In the case of Kenya and Tanzania, the Zero Inflated Poisson (ZIP)
estimator was used, while for Uganda the Negative Binomial Regression (NBR) was
employed. The choice of an appropriate estimator was based on the properties exhibited by
each country’s dataset. For instance, datasets comprised of a high proportion of zero trade
flows were estimated at 31.9 % and 75.5 % for Kenya and Tanzania, respectively.
Moreover, the statistically significant Vuong test results 2.85 (p<0.01) for Kenya and 1.51
(p<0.10) for Tanzania imply that the ZIP was a more apt estimator for this analysis, rather
than using the standard Poisson estimator. In the case of Uganda, the highly significant
Likelihood-ratio test results 323.29 (p<0.01) imply that the Negative Binomial Regression
(NBR) estimator was the most suitable estimator, as compared to the Binomial regression.
Regression results at country level are presented in Table 5.58 below.
104
Table 5.5: Empirical effects of climate change on East Africa's horticultural imports into the EU market
Dependent variable (Xijt) = Net imports of horticultural commodities to a specific country j within the EU from EA states (denoted as i if
it is either Uganda, Kenya or Tanzania) in year t in (000) US$
Variable Kenya Tanzania Uganda
Coefficient p-value Coefficient p-value Coefficient p-value
Constant -11.51*** 193.48** 1335.34*
(2.871) 0.000 (78.54) 0.014 (759.452) 0.079
Precipitation anomalies in the importing country at a given 0.01*** 0.07* 2.46***
time (DPrejt) (mm) (0.003) 0.000 (0.041) 0.075 (0.698) 0.000
Temperature anomalies in the importing country at a given -0.03*** -0.76*** -2.13**
time (DTjt) (
0C) (0.003) 0.000 (0.065) 0.000 (1.002) 0.033
Temperature anomalies in the exporting country at a given -0.18*** 0.34*** -3.48*
time (DT ) (0it C) (0.013) 0.000 (0.087) 0.000 (1.929) 0.071
Precipitation anomalies in the exporting country at a given 0.05*** -1.28*** -16.39*
time (DPreit) (mm) (0.004) 0.000 (0.162) 0.000 (8.786) 0.062
Exporter’s Gross Domestic Product (GDP) in current US$ 0.42*** -18.27*** -36.63*
(lnY1it) (0.025) 0.000 (2.484) 0.000 (21.274) 0.085
Importer’s Gross Domestic Product (GDP) in current US$ 0.73*** 0.38* 1.80***
(lnY2jt) (0.025) 0.000 (0.229) 0.098 (0.672) 0.007
Distance between trading partners (lnDij) (miles) -2.05*** -8.73** -14.06**
(0.248) 0.000 (3.956) 0.027 (6.509) 0.031
Crop production in the exporting country (lnAgriit) 0.30*** 16.92*** -36.98
(tonnes) (0.041) 0.000 (0.638) 0.000 (22.942) 0.107
Crop production index in the importing country(lnAgrijt) 0.48*** -5.22*** 0.06
(0.055) 0.000 (1.184) 0.000 (1.195) 0.959
Linder’s income similarity index (Incomijt) 1.61*** 1.66 2.96
(0.099) 0.000 (1.489) 0.265 (3.148) 0.347
Level of infrastructure in the exporting country i in year t 15.05*
(lninfrait) (aircraft departures) - - - - (8.359) 0.072
Inflation level in the exporting country i in year t -0.06*** -13.45*** 1.99**
(lninflatit) (0.006) 0.000 (0.972) 0.000 (0.834) 0.017
Dummy variable for common language (Dlangij) 4.34*** 2.23*** 1.12
105
(=1, if share official language; =0, otherwise) (0.070) 0.000 (0.769) 0.004 (1.412) 0.426
Vuong test (Z-value) 2.85*** 0.002 1.51* 0.066 - -
Likelihood-ratio test of alpha=0 323.39*** 0.000
Fixed effects Yes Yes Yes
Number of observations (N) 135 90 105
Nonzero observations (N1) 92 22
Zero observations (N0) 43 68
Log likelihood -4376.76 -138.68 -101.551
Count data estimation technique used ZIP ZIP NBR
***, **, * denote significance at 1%, 5% and 10% level respectively.
106
These show that precipitation anomalies in the EU market have a significant positive influence
on horticultural trade flows from all three East African states, namely Uganda, Tanzania and
Kenya, in that order. Notably, a unit percentage change in precipitation anomalies within the EU
leads to US$ 2 500 (p<0.01) increase in Uganda's value of exports. In the case of Tanzania and
Kenya, a similar unit change translates into an estimated US$ 70 (p<0.10) and US$ 10 (p<0.01)
increase in the value of horticultural trade flows, respectively. In general terms, this positive
trend may be associated with the unique and advantageous endowment of a tropical climate that
favours production of exportable tropical fruits and vegetables through the year, relative to other
countries which also export to the EU market. In addition, East African states specialize in high
value, annual tropical horticultural commodities, such as asparagus, pepper and other
specialty/exotic vegetables. These horticultural commodities have a short growing period and
can be delivered to the EU market at any time of the year. These results are in concurrence with
Nelson et al. (2010) who note that climate change variability will enhance trade flows of certain
agricultural commodities from regions that are less affected to areas adversely limited in the
production of given commodities.
Temperature anomalies in the importing country (DTjt) were found to have a significant negative
influence on fruits and vegetable imports from EA states into the EU market. Results show that
a unit change in temperature anomalies in the destination markets leads to a loss of US$ 2,130
(p<0.05), US$ 760 (p<0.01) and US$ 30 (p<0.01) in the value of fruits and vegetable imports
from Uganda, Tanzania and Kenya, respectively. Among other factors, the results may be
explained by the European Union's tendency to protect its domestic producers by subjecting
horticultural imports to a series of tariffs, as well as by limiting the importation of specific
horticultural commodities in favour of the EU production calendar (Petriccione et al., 2011).
Such requirements are stipulated under the EU-GSP preferential treatment granted to all
developing countries. Within the production calendar, the EU furthermore alters tariff levels of
some products, particularly fruits and vegetables, and this usually occurs during harvesting
periods which coincide with the northern hemisphere winter season (UIA, 2001; Lubinga et al.
2013).
Given the high perishability of most unprocessed horticultural commodities, coupled with
changes in EU market demand based on their production calendar, East African states tend to be
discouraged from trading to their full potential within the EU market. As an example, UIA
(2001) shows that much of Uganda's harvesting occurs between November and February, which
coincides with the winter season in Europe when there is a high demand for fresh fruits and
107
vegetables. According to the EU-GSP scheme, however, some horticultural commodities cannot
be exported to the EU market. That is, the EU only allows imports of, for example cucumber
(fresh or chilled), between 16 May and 31 December. Other commodities with limited access to
the EU market in specific months include Mandarins (080520), Plums (080940) and fresh
Peaches (080930) (UNCTAD, 2008).
The negative influence of temperature anomalies on East Africa's horticultural imports in the EU
market may also be attributable to the relatively longer summer periods that were generally
experienced in Europe in the late 2000s. Odongo (2007) asserts that during this period
horticultural sales declined. The relatively small impact of temperature on Kenya's imports may
be attributed to the fact that it exports a wide spectrum of fruits and vegetables of high demand
and limited supply within the EU market, for example asparagus. In addition, given that Kenya's
horticulture industry is highly developed given that it is reliant on irrigation and green houses.
Thus, the harvesting periods for these crops can easily be controlled to avoid coinciding with the
production cycle of the EU producers. These technological advancements render Kenyan traders
capable of supplying the EU market only during periods of high demand for the fruits and
vegetables, hence minimizing the negative effects of fluctuations in temperature in the
destination markets. For example, Kenya is the second-leading exporter of horticultural
commodities after South Africa in Sub-Saharan Africa (English et al,, 2004).
With regard to temperature (DTit) and precipitation (DPreit) anomalies in the exporting
countries, a mixture of both negative and positive results was obtained. However, all
coefficients were statistically significant at various levels. A unit percentage change in
temperature anomalies in Tanzania was noted to enhance an increase in value of horticultural
exports to the EU by US$ 340 (p<0.01), while a similar change led to a loss worth US$ 3,480
(p<0.10) and US$ 180 (p<0.01) in Uganda's and Kenya's exports, respectively.
The positive trend for Tanzania may be associated with the agronomic attributes of cashew nuts,
which assume a large proportion the country's horticulture exports. Cashew nut production is
strongly influenced by weather conditions and can withstand water-constrained conditions for a
long period, unlike most annual fruits and vegetables. For instance, ARC (2013) opines that
high temperatures stimulate early flowering. In essence, high temperatures during the
production phase serve as an incentive to include this major crop within Tanzania's horticultural
exports. In addition, it is a perennial crop, with a well-developed root system and renowned for
108
high tolerance to drought conditions. Thus, such agronomic attributes of cashew nuts explain
the positive trends in trade, despite fluctuations in Tanzania's temperature pattern over the years.
On the other hand, the decline in Uganda's and Kenya's exports may be attributed to the fact that
these countries specialize more in annual horticultural crops, which according to researchers
(UNDP and BCPR, 2013; McCandless et al., 2012) are very sensitive to higher temperatures.
The vulnerability of annual horticultural crops is associated with the fact that high temperatures
can quickly destroy the crop, unlike for perennial crops which might survive, but with lower
yields or reduced quality (Lubinga et al., 2013). According to GRID (2002), a 2 °C increase in
annual temperature results in a reduction in the arable land suitable for a wide spectrum of crops
within the East African region.
Furthermore, research shows that a unit change in Kenya's precipitation anomalies would lead to
an increase in the monetary value of horticultural exports by US$ 50 (p<0.01). This positive
observation may be attributable to the fact that Kenya has agriculture- and environment-related
policies, strategies, and programmes which focus on adaptation to climate change (Odera et al.,
2013). Kenya drafted a "National Irrigation Policy" in a bid to promote advanced irrigation
technology. Accordingly, recent data from the World Bank Development Indicators (WBDI)
database (2012) shows that only Kenya, of the three EA states, has embraced irrigation
technology at commercial levels. Within a period of three years (2001–2003), Kenya's
proportion of irrigated land increased by close to threefold, from 0.04 % in 2001 to 0.11 % of
agricultural land. In addition, it is worthwhile to mention that Kenya has a National Climate
Change Response Strategy (NCCRS) with interventions designed for the various sectors of the
economy.
In an effort to buffer crop production failures, probably due to climate change related factors,
Kenyan farmers use more fertilizers than their Tanzanian and Ugandan counterparts. For
instance, in 2002 and 2003, Kenya used on average 30.2 kilograms per hectare of fertilizers,
compared with 3.8 and 1.5 kilograms per hectare of fertilizers in the case of Tanzania and
Uganda, respectively (WBDI database, 2012). The use of such technological advancements acts
as a buffer against the devastating effects of climate change, which would greatly lead to low
productivity and reduced trade flows.
On the contrary, fluctuations in precipitation were observed to significantly influence declines in
the value of Uganda's and Tanzania's trade flows by an estimated US$ 16,390 (p<0.10) and US$
109
1,280 (p<0.01), respectively. The large coefficient associated with Uganda's trade flows may be
explained by Timmers' (2012) argument that rainy seasons have become increasingly erratic
over the years, most especially within the Lake Victoria basin, and this curtails production. As
an example, he argues that the increasingly erratic precipitation has led to an estimated 2.3 m
decline in the water level within the lake over a six-year period (2000–2005). It is known that
Uganda's subsistence agriculture sector depends largely on natural rainfall, hence lower rainfall.
Furthermore, the negative effects of precipitation anomalies for Tanzania and Uganda may be
attributed to the extreme climatic catastrophes, such as floods and droughts, that were
experienced during the period (1988–2000) considered for this analysis. According to Timmers
(2012), floods, for instance, cause damage to agricultural infrastructure such as roads, bridges,
storage facilities and processing lines; aggravate land mass movements; and a number of people
lose their lives. All these problems associated with floods operate to increase trade barriers,
given that they impede easy and quick access to goods and services, thus increasing the cost of
doing business.
As an example, between 1988 and 2000, Tanzania experienced three severe floods. EM-DAT
(2013) notes that the April 1989 floods affected slightly more than 140,000 people, while the
1990 and 1993 floods caused an estimated US$ 0.28 million and US$ 3.51 million worth of
economic damage, respectively. In the case of Uganda, it is also noted that in 1997 the country
experienced floods which affected about 0.15 million people, coupled with massive
infrastructure damages estimated at US$ 1 million. Furthermore, Uganda witnessed a prolonged
drought in 1998 which led to an estimated cost of US$ 1.6 million, affecting 0.13 million people.
Other episodes of severe drought in Uganda were observed in 1987 and 1999, during which 0.6
million and 0.7 million people were affected (EM-DAT, 2013). Similarly, Tanzania also
suffered long intense spells of drought in 1984, 1991 and 1996, during which 1.9 million, 0.8
million and 3 million people were adversely affected.
Extreme climate events like floods and droughts disrupt international trade, largely at the
production stage, through their effect in reducing productivity of agricultural farms.
Furthermore, the low agricultural productivity attributable to erratic climatic change may inhibit
innovation and imitation, given that farmers are uncertain of what the near future has to offer.
Coupled with the pervasive risky environmental effects on farming practices and farm
performance, the increasing uncertainty probably discouraged farmers from upgrading
production technology, thus further affecting productivity.
110
Notably, the negative effects of precipitation anomalies on Uganda's and Tanzania's trade may
also be attributable to the migration of people from the rural areas to urban centres in a bid to
adapt to harsh climatic changes (Marchiori et al., 2010). Accordingly, this migration could have
culminated in the reallocation of the scarce labour force from the agricultural sector in rural
areas to the non-agricultural sectors in urban areas, hence disrupting the production cycle, as
well as agricultural trade in the long-run. For instance, within the 1996-2000 period during
which severe drought was experienced (particularly 1997–1999), Uganda registered over
0.45million emigrants in net terms (WDI, 2012). A similar scenario was observed in Tanzania
during the early 1980s drought, when over 22,000 people emigrated from the country. This is an
enormously plausible amount and quality of labour lost from the agricultural sector.
Furthermore, the deterrent effects of precipitation anomalies in the case of Uganda and
Tanzania's trade may be associated with the growth characteristics of a majority of their
horticultural crops, particularly, the vegetables. These are annual crops with short growth
periods during which they need much water. Thus, severe variations in precipitation patterns are
capable of causing 100 % crop failure (Bashasha et al., 2013; Kilembe et al., 2013) and long-run
production declines, which indeed curtail trade flows. Moreover, in Tanzania and Uganda large
proportions of the populations are still surviving below the poverty line, which Kilembe et al.
(2013) and Bashasha et al. (2013) note as being an enormous stumbling block in adapting to
climate change within the horticulture sector. Under such high levels of poverty, farmers cannot
ably have access to basic agro-inputs and credit through which they could create a buffer against
the hastening climate change. Nelson et al. (2010) succinctly reckon that only smallholder
farmers with fairly high disposable incomes can dare to experiment with new technologies and
management systems, which might be expensive to invest in initially but enhance farmers'
resilience to climate change effects through higher productivity. This guarantees sustainable
trade flows in horticultural commodities.
Uganda's and Tanzania's negative findings in relation to precipitation anomalies are in
accordance with those of other scholars (McCandless et al., 2012; Marchiori et al., 2010;
Schlenker and Lobell, 2010; Nelson et al., 2009; Tao et al., 2008; Xiong et al., 2007; Sivakumar
et al., 2005; Kumar et al., 2004; Parry et al., 2004; FAO, 2001; Timmers, 2012; MWE, 2007)
who argue that climatic factors that perturb agricultural productivity inevitably limit the
economy's available tradable produce. Ludi et al. (2007) argue that climate change affects the
111
volume of agricultural commodities produced within an economy, as well as the price of
tradable agricultural produce from which foreign exchange is generated.
Generally, findings concerning the other variables used in the country models conform to the
theoretical expectations of the gravity flow model. For instance, distance was found to have a
significant negative influence on trade flows across all the three East African states. Uganda's
large drop in trade by US$ 14,060 (p<0.05) may be attributed to the fact that it is a landlocked
country, unlike Tanzania and Kenya, which registered a decline in their trade by US$ 8,730
(p<0.05) and US$ 2,050 (p<0.01), respectively. An increase in an importer’s Gross Domestic
Product (GDP) (lnY2jt) was also found to significantly enhance horticultural trade. Results
reveal that a unit increase in an importer's GDP leads to about US$ 1,800 (p<0.01), US$ 730
(p<0.01) and US$ 380 (p<0.10) rises in the monetary worth of horticultural exports from
Uganda, Kenya and Tanzania, respectively.
With the exception of Kenya, an increase in an exporter's (GDP) (lnY1it) was found to
significantly influence a drop in the value of horticultural exports from Uganda (US$ 36,630,
p<0.10) and Tanzania (US$ 18,270, p<0.01). The declining trends may be attributable to the
fact that these states have not duly recognized these horticultural commodities as foreign
exchange earners (UNBS, 2013). Thus, an increase in GDP would probably foster promotion of
the traditional cash crops like Coffee, cotton, tea and tobacco, rather than the fruit and vegetable
sub-sector. Inevitably, an increase in Kenya's GDP was noted to enhance the exportation of
horticultural exports by approximately 0.42 times. This observation may be associated with the
fact that horticulture is the leading industry within the agricultural sector in Kenya, accounting
for approximately 36 % of the Agricultural GDP (Government of Kenya, 2012).
Other variables, such as Linder’s income similarity index (Incomijt), Level of infrastructure in
the exporting country i in year t (lninfrait) and Dummy variable for common language (Dlangij),
were also found to enhance exports of fruits and vegetables from EA states. However, for
Uganda, the positive coefficient (1.12, p>0.10) for common language (Dlangij) was an
insignificant coefficient. Unlike for Tanzania, a unit percentage change in crop production in the
importing country (lnAgrijt) was observed to promote Kenya's and Uganda's horticultural trade
flows into the EU by US$ 480 (p<0.01) and US$ 60 (p>0.10), respectively. However, the result
relating to Uganda was insignificant. This observation may be associated with the fact that
Kenya has comparative advantage in the production of many high-value tropical fruits and
vegetables which cannot easily be produced within the temperate EU.
112
5.4 The effect of the EU-GSP scheme on East Africa’s fruits and vegetable imports into
the EU market
Assessment of the impact of the EU-GSP scheme was carried out based on the results of
objective one, under which an evaluation of the export competitiveness of the fruits and
vegetable commodities from East African states in the EU market was conducted. The focus
was on a few selected commodities (see Table 5.6 below) that exhibited export competitiveness
within the EU market. From each EA State, two commodities were chosen following two basic
procedures: (i) if the commodity exhibited an average export competitiveness index (RCA) of
greater than one across all the three EA states, and (ii) if the commodity revealed the highest
RCA amongst all commodities exported from a given country. In the case of Uganda, three
commodities were considered, given that fruits of the genus Capsicum (070960) exhibited high
RCA value (>27,600), in addition to the top two commodities that had been selected. Thus,
seven (7) commodities in total were subsequently considered for evaluating the effect of the EU-
GSP Scheme on horticultural commodities from the East African region.
Table 5.6: East Africa’s selected fruit and vegetable commodities with high export
competitiveness in the EU market
Country HS 6- Commodity description Mean RCA Mean
Digit (2007-10) preference
code margin
Kenya 070920 Asparagus, fresh/chilled 8,504.32 9.35
070820 Beans (Vigna spp., Phaseolus spp.) 3.70 10.86
Tanzania 070990 Vegetables, n.e.s. in 07.01-07.09, 24.60 13.69
fresh/chilled
070820 Beans (Vigna spp., Phaseolus spp.) 2.23 10.86
Uganda 070960 Fruits of the genus Capsicum/ Pimen 27,668.87 3.56
080300 Bananas, including plantains, fresh/dried. 25.98 23.15
070820 Beans (Vigna spp., Phaseolus spp.) 1.23 10.86
Source: Author’s calculations
Descriptive statistics of the selected commodities depict that bananas (080300) enjoyed the
largest preferential margin, estimated at 23, followed by vegetables (13.7), beans (10.9),
asparagus (9.6), and lastly, fruits of the genus Capsicum or of the genus Pimen (3.6). Prior to
the econometric estimation of the effect of the EU-GSP scheme on the selected commodities, a
number of diagnostic tests were conducted.
113
5.4.1 Diagnostic test results
According to the multi-collinearity test results presented in Appendices D to J, no serial
correlation was found among the variables. Viz, in all commodities and across all the three EA
states, the Tolerance (TOL) test values were in accord with the expected value (more than 0.1).
On the other hand, the mean Variance Inflation Factor (VIF) test results shown in Table 5.7
below were also less than the threshold value of 10. Furthermore, the correlation matrix test
results (see Appendices D – J ) depict that there exists no serial correlation problem across all
commodities, since the correlation values were not greater than the threshold value of 0.7.
Table 5.7: VIF test- and over-dispersion test- results for the horticultural commodities
Country HS 6- Digit code Mean VIF Mean ('000 US$) Variance
value (n=105)
Kenya 070820 2.06 9,488.14 4.68e+08
070920 1.96 22.10 8,017.33
Tanzania 070820 2.17 210.14 353,270.4
070990 2.74 13.88 8,393.40
Uganda 070820 3.97 2.67 64.85
070960 2.11 257.82 422,124
080300 7.33 200.11 373,606
Source: Author’s calculations
The over-dispersion test results presented in Table 5.7 above indicate that the import data series
for the three EA states were highly over-dispersed, given that the conditional variance of each
series deviated by far from the conditional mean. Moreover, the normality test results presented
in Appendices K to M also show that the highly disaggregated data series were not normally
distributed. Inevitably, the existence of over-dispersion, coupled with distribution asymmetry
problems, imply that ordinary econometric estimation procedures cannot be used to obtain
reliable results.
The unit root test results presented in Table 5.8 are based on both the Levin–Lin–Chu test (LLC-
test) and the Harris-Tzavalis test (HT-test). All commodity series were found to be integrated of
order one while using the LLC-test, except for Beans (070820) from Uganda, Asparagus
(070920) from Kenya, and Vegetables (070990) from Tanzania. However, in instances where
the stationarity could not be established, even after first order difference with the LLC-test, the
HT-test was used. Interestingly, all data series upon which the HT-test was used were also
found to be significantly stationary at all levels. Thus far, since the test statistics in both the
114
LLC- and HT-tests are significant at all levels (p<0.01), it is prudent to reject the null hypothesis
of a unit root in the series in favour of the alternative hypothesis that all the series are stationary.
Table 5.8: Panel Unit Root test results by commodity and country
Kenya Tanzania Uganda
Variable
Levels Levels I(1) Levels I(1)
070820 Beans (Mijlt) 0.22***-11.20*** 2.95 -6.41***
('000 US$) (ht)
070920 Asparagus (Mijlt) 0.25*** - - - -
('000 US$) (ht)
070990 Vegetables (Mijlt) 0.24***- - -
('000 US$) (ht)
070960 Peppers (Mijlt) - - - -2.99***
('000 US$)
080300 Bananas (Mijlt) - - - -5.0743***
('000 US$)
Other covariates Levels Levels Levels
Exporter’s GDP (lnY1it) (US$) -14.65*** -13.38*** -20.91***
Importer’s GDP (lnY2jt) (US$) -17.73*** -17.73*** -17.73***
Exporter's mean annual
-7.05*** - -6.25***
inflation rate (lninflatit)
Cost of establishing a business
(lnCOSTBIZit) (% of GNI per -6.84*** - -
capita)
Cost to exporting a 20-foot
container (lnCOSTEXPit) (% - 12.45 -84.83***
of GNI per capita)
Public sector and government
0.46***
institutions role (lnGOVit) (an - -1.19(ht)
index from 1=low to 6= high)
Net inflow of foreign direct
- - -12.05***
investment (lnFDIit) (US$)
***, **, * denote significance at 1%, 5% and 10% level respectively. (ht) denotes that the
Harris-Tzavalis unit-root test was used, otherwise other results are based on the LLC-test.
Source: Author’s own calculations.
115
5.4.2 Empirical findings of the effect of the EU-GSP scheme on East Africa’s fruits and
vegetable imports into the EU market
Analytical results regarding the effect of the EU-GSP scheme on East Africa's horticultural
imports into the EU are presented at country level. Owing to the varying data properties, the
estimation techniques used are also highlighted.
5.4.2.1 Kenya
Econometric estimation results presented in Table 5.9 below were estimated using the Zero
Inflated Poisson (ZIP) technique, given that commodity datasets had excessive zeros, that is,
71 % of the Asparagus export dataset were zeros, while 27 % of the beans export dataset were
also zeros. Moreover, Kenya's export data were also highly over-dispersed. In both cases, the
statistically significant Vuong test results show that the ZIP estimator is preferable to the
standard Poisson estimator.
With regard to the key variable of interest, the preference margin, findings show that the EU-
GSP scheme hampers imports of Kenya's asparagus and beans into the EU market. At a one
per cent level of significance, a unit rise in the preference margin granted under the EU-GSP
scheme leads to a decline of US$ 2,460 (p<0.01) and US$ 280 (p<0.01) in Kenya's asparagus
and bean exports to the EU market. This may be attributed to stiff competition from other
exporters, such as Colombia, Ecuador, Ethiopia, Morocco, Israel and Egypt, partly as a result of
preferential treatment.
116
Table 5.9: Effect of the EU-GSP scheme on Kenya’s Asparagus and Bean exports
Dependent variable (Mijlt)  = Total value of commodity l from Kenya i to jth EU member
state in year t in ‘000 US Dollars
Variable Asparagus (070920) Beans (070820)
Coefficient p-value Coefficient p-value
Constant 65.71 280.51***
(116.36) 0.572 (2.416) 0.000
Preference margin of a specific
commodity, expressed as a share of -2.46*** -0.28***
the product value (lnPMijlt) (0.186) 0.000 (0.019) 0.000
Exporter’s Gross Domestic Product 6.15*** -0.07***
(GDP) in current US$ (lnY1it) (0.312) 0.000 (0.010) 0.000
Importer’s Gross Domestic Product 1.56*** 1.85***
(GDP) in current US$ (lnY2jt) (0.267) 0.000 (0.005) 0.000
Distance in miles between trading -27.71* -39.95***
partners (lnDij) (15.194) 0.068 (0.305) 0.000
Exporting country's mean annual -0.17** -0.05***
inflation rate (lninflatit) (0.068) 0.015 (0.003) 0.000
The cost of establishing a business,
expressed as percentage of GNI per -5.93*** -0.36***
capita (lnCOSTBIZit) (0.646) 0.000 (0.035) 0.000
Dummy variable for common
language (=1, if share common 6.73*** 15.51***
language; = 0 otherwise) (Dlangij) (1.863) 0.000 (0.078) 0.000
Vuong test (Z-value) 4.64*** 0.000 3.25*** 0.001
Fixed effects Yes Yes
Number of observations (N) 105 105
Nonzero observations (N1) 30 77
Zero observations (N0) 75 28
Log likelihood -207.283 -22938.15
Count data estimation technique used ZIP ZIP
***, **, * denote significance at 1 %, 5 % and 10 % levels, respectively.
The deterrent effect of the EU-GSP scheme on Kenya's asparagus and bean exports to the EU
may also be the result of stringent standard requirements, to which horticultural commodities
entering Europe are subject (e.g. GLOBALG.A.P and the British Retail Consortium (BRC)
standard). Compliance with these continually changing standards comes with investment costs,
which discriminates against smallholder farmers. Despite the fact that Kenya is regarded as a
success story in the implementation of the private and voluntary GlobalG.A.P standard,
researchers (Kuwornu and Mustapha, 2013; Asfaw et al.,2010; Aloui and Kenny, 2005; Augier
et al., 2005) argue that sustaining compliance with the standard poses both technical and
financial constraints for smallholder farmers and exporters.
117
Under Option 2, where smallholder farmers are certified as a group, Kuwornu and Mustapha
(2013) show that each group member contributes over 36,000 KSh (US$ 1 =  86.12 KSh), the
equivalent of about 30 per cent of an individual smallholder farmer's annual income realized
from crop enterprises. As a pre-condition to implement the standard, it is also mandatory for
farmers to invest in infrastructure and equipment, such as stores, waste disposal pits and product
handling facilities (grading shed and cooler).
Furthermore, the negative finding may result from Kenya's graduation from a Least Developed
Country (LDC) to a lower-middle income country in December 2007. This change in status
comes with lower benefits under the Lomé Convention between African Caribbean Pacific
(ACP) countries (of which Kenya is a member) and the EU. The United States Agency for
International Development (USAID) (2007) has argued this would have reduced its
competitiveness within the EU market.
The inconsistency and unpredictability of small-scale exporters of the horticultural commodities
may also be associated with the negative effects of the EU-GSP scheme on Kenya's
commodities. Dolan et al. (2000) mention that the regular and consistent supply of specialty
products, such as asparagus, is paramount within the EU market. However, probably owing to
the dwindling production volumes by Kenyan farmers, such market conditions could not be
realized, hence other competitors in the asparagus market (e.g. Lesotho) have increased their
market shares. Dolan et al. (2000) show that there are many licensed exporters in Kenya, but
only a handful are consistently in operation. Most of the exporters only take advantage of
favourable short-term market conditions, especially during the peak season (October–April).
This jeopardizes the benefits that might be realized through the EU-GSP scheme.
In addition, the non-beneficial nexus between the EU-GSP scheme and Kenya's commodities
(asparagus and beans) may be attributed to the existence of other trade enhancing policies that
probably present barriers to trade than the GSP scheme. Such policies include the Economic
Partnership Agreements (EPAs) between the EU and African, Caribbean and Pacific (ACP)
countries, as well as the African Growth and Opportunity Act (AGOA) initiative granted by the
USA. According to FreshPlaza (2014), the AGOA initiative offers more duty-free benefits than
what is granted under the GSP scheme and as a result, Kenya's horticultural exports to the USA
have significantly increased and totalled more than US$ 38 million in 2011. Notably, under the
AGOA initiative, the USA recently granted Kenya for fresh green beans, runner beans, baby
118
carrots, baby corn and shelled beans. This implies that Kenya is experiencing a drift into other
markets because of the existence of more attractive policies.
All in all, the results imply that the EU-GSP scheme is of less importance in boosting Kenya's
asparagus and bean exports into the EU market. The results for Kenya closely relate with results
of other researchers (Philippidis et al,, 2011; Asfaw et al., 2010). Asfaw et al. (2010) argue that
smallholder farmers who cannot cope with the frequently changing and stringent EU market
standards are bound to search for alternative markets. Philippidis et al. (2011) also note that
increasing import tariff rates deter exports of fruits and vegetables from developing countries to
the European market.
Estimated coefficients of all the other covariates are in concurrence with the theoretical
expectations of the gravity model framework. In general terms, a one per cent increase in the
country's inflation rate (lninflatit), the cost of doing business (lnCOSTBIZit) and distance (lnDij)
is associated with varying levels of decline in trade flows, depending on the commodity, viz,
asparagus or beans. Such results mean that high inflation rates, distance and the high cost of
doing business in Kenya are bottlenecks to trading in these commodities.
On the other hand, a unit positive change in the importing country's GDP (lnY2jt) was noted to
cause an estimated US$ 1,560 (p<0.01) and US$ 1,850 (p<0.01) rise in the monetary value of
asparagus and bean trade flows, respectively. This implies that an increase in GDPs of EU-
member states leads to higher demand for these commodities from Kenya. A unit change in
Kenya's GDP (lnY1jt) was found to have mixed but significant effects on trade flows. For
instance, a one per cent (1 %) rise in Kenya's GDP was noted to cause an increase in asparagus
trade flows by US$ 6,200 (p<0.01), while a similar change in the country's economy would lead
to a fall in bean exports by approximately US$ 70 (p<0.01) in monetary terms. This suggests
that Kenya may not have adequate capacity or competitive advantage in the production of beans,
as compared with the production of asparagus.
The results for the dummy variable for common language (Dlangij) also show that sharing
similar cultural ties presents more trading opportunities. Sharing a common language was found
to boost trade in beans and asparagus by US$15,500 (p<0.01) and US$ 6,700, respectively,
relative to trade partners that do not have a common official language. Notably, much of
Kenya's asparagus goes to the Netherlands (where Dutch is the official language), unlike beans
119
which are taken to the United Kingdom, a former colonial master and where English is the
official language.
5.4.2.2 Tanzania
The assessment of the effect of the EU-GSP scheme on Tanzania's bean and vegetable imports to
the EU-market was also based on the ZIP estimation technique for the same reasons (excess
zeros trade flows and over-dispersion of the dependent variable) as in the case of Kenya.
Specifically, the findings in Table 5.10 below show that Tanzania's bean export data contained
more than 70 zero trade flows (approximately 68 % of the data), with vegetable exports
registering 87.6 % zeros. The statistically significant Vuong test results (6.51, p<0.01 for beans;
3.11, p<0.01 for vegetables) used to compare the appropriateness of the ZIP model relative to the
standard Poisson model also show that the ZIP estimator was an apt technique for this analysis.
Table 5.10: Effect of the EU-GSP scheme on Tanzania’s Vegetables and Bean exports
Dependent variable (Mijlt)  = Total value of commodity l from Tanzania i to jth EU
member state in year t in ‘000 US Dollars
Variable Beans (070820) Vegetables (070990)
Coefficient p-value Coefficient p-value
Constant 65.77*** 141.73***
(4.590) 0.000 (46.579) 0.002
Preference margin of a specific
commodity, expressed as a share of 1.01*** -0.09
the product value (lnPMijlt) (0.086) 0.000 (0.275) 0.739
Exporter’s Gross Domestic Product 0.52*** -7.29***
(GDP) in current US$ (lnY1it) (0.078) 0.000 (0.609) 0.000
Importer’s Gross Domestic Product -0.23*** -2.43***
(GDP) in current US$ (lnY2jt) (0 .019) 0.000 (0.434) 0.000
Distance in miles between trading -7.92*** 29.63***
partners (lnDij) (0.491) 0.000 (5.267) 0.000
Cost to exporting a 20-foot container -0.79*** -22.69***
in US$ per container (lnCOSTEXPit) (0.126) 0.000 (2.172) 0.000
The role of the public sector and
government institutions, expressed as 2.76*** 8.05***
an index (lnGOVit) (0.132) 0.000 (0.823) 0.000
Dummy variable for common 3.01*** 5.53***
language (Dlangij) (0.041) 0.000 (0.910) 0.000
Vuong test (Z-value) 6.51*** 0.000 3.11*** 0.001
Fixed effects Yes Yes
Number of observations (N) 105 105
Nonzero observations (N1) 34 13
Zero observations (N0) 71 92
Log likelihood -4604.642 -86.123
Count data estimation technique used ZIP ZIP
***, **, * denote significance at 1%, 5% and 10% level respectively.
120
Estimated coefficients of all other covariates considered for the bean commodity were found to
be in tandem with the theoretical expectations of the gravity model framework. With the
exception of distance, all variables also exhibit significant effects on Tanzania's vegetable
exports to the EU and the variables were found to have the theoretically expected signs. The
positive results (US$ 29630, p<0.01) associated with the distance variable (lnDij) imply that
distance does not necessarily impede exports of Tanzania's vegetables to the EU. André and
Joel (2012) and Marimoutou et al. (2009) opine that the deterring effect of distance ceases to be
an issue if the trading partner's economic size (GDP) is by far larger than that of the exporting
country. According to data provided by the WBDI database (2012), the economy of the United
Kingdom (Tanzania's key destination market for vegetables in the EU) is more than 130 times
larger than Tanzania's economy. Furthermore, Tanzania capitalizes more on high-value
vegetables (USAID, 2007), such as snow peas, sugar snap peas, French green beans, and baby
vegetables (carrots, maize, leeks, zucchini, pattypan squashes, and eggplants, among others),
which are in high demand in the EU.
5.4.2.3 Uganda
The analysis of the effect of the EU-GSP scheme on Uganda's banana, beans and pepper exports
to the EU market was also grounded on the ZIP estimator, given that export data series of each
of the commodities exhibited excess zeros, coupled with over-dispersion. The empirical results
in Table 5.11 below reveal that, of the 105 observations for each of Uganda's banana, beans and
pepper export datasets, there were 47, 86 and 30 zero trade flow values, respectively. 44.8, 81.9
and 28.6 percent for banana, beans and pepper export data. The statistically significant Vuong
test results (2.38, p<0.01 for bananas; 9.65, p<0.01 for beans; 3.99, p<0.01 for pepper) also
imply that the ZIP estimator was the most appropriate model for this analysis, relative to the
standard Poisson model.
The preference margin (lnPMijlt), a count variable used to proxy the role of the EU-GSP scheme,
was found to have a statistically significant positive influence on Uganda's banana, bean and
pepper exports to the EU market. A one per cent increase in the preferential margin granted
under the EU-GSP scheme was observed to lead to an increase in the monetary value of
Uganda's banana, bean and pepper imports into the EU market by US$ 770 (p<0.01), US$ 3050
(p<0.05) and US$ 280 (p<0.01), respectively. This observation implies that the scheme has had
a contributory role in boosting Uganda's exports to the EU.
121
Table 5.11: Effect of the EU-GSP scheme on Uganda’s Banana, Bean and Pepper exports
Dependent variable (Mijlt)  = Total value of commodity l from Uganda i to jth EU member state in year t in ‘000 US Dollars
Variable Bananas (080300) Beans (070820) Pepper (070960)
Coefficient p-value Coefficient p-value Coefficient p-value
Constant -7.77 1442.33*** -81.74***
(16.798) 0.644 (211.209) 0.000 (2.116) 0.000
Preference margin of a specific commodity, expressed as a share 0.77*** 3.05** 0.28***
of the product value (lnPMijlt) (0.084) 0.000 (1.211) 0.012 (0.082) 0.001
Exporter’s Gross Domestic Product (GDP) in current US$ 3.076*** -0.02 1.89***
(lnY1it) (0.203) 0.000 (1.059) 0.986 (0.060) 0.000
Importer’s Gross Domestic Product (GDP) in current US$ 1.63*** 6.64*** 0.99***
(lnY2jt) (0.042) 0.000 (1.035) 0.000 (0.010) 0.000
Distance in miles between trading partners (lnDij) -17.67*** -196.07*** 1.99***
(1.857) 0.000 (27.662) 0.000 (0.194) 0.000
Exporting country's mean annual inflation rate (lninflatit) -0.43*** 0.15 -0.08***
(0.026) 0.000 (0.129) 0.256 (0.017) 0.000
Cost to exporting a 20-foot container in US$ per container -0.75*** -0.09
(lnCOSTEXPit) (0.070) 0.000 (0.365) 0.811 - -
The role of the public sector and government institutions, 1.57*** -5.05*** 1.80***
expressed as an index (from 1=low to 6= high) (lnGOVit) (0.182) 0.000 (1.707) 0.003 (0.182) 0.000
Net inflow of foreign direct investment in current US$ (lnFDIit) 2.08***
(0.086) 0.000 - - - -
The cost of establishing a business, expressed as percentage of -0.77***
GNI per capita (lnCOSTBIZit) - - - - (0.178) 0.000
Dummy variable for common language (=1, if share common 4.46*** -3.14*** 1.98***
language; = 0 otherwise) (Dlangij) (0.110) 0.000 (1.017) 0.002 (0.022) 0.000
Vuong test (Z-value) 2.38*** 0.009 9.65*** 0.000 3.99*** 0.000
Fixed effects Yes Yes Yes
Number of observations (N) 105 105 105
Nonzero observations (N1) 58 19 75
Zero observations (N0) 47 86 30
Log likelihood -803.596 -97.100 -8671.796
Count data estimation technique used ZIP ZIP ZIP
***, **, * denote significance at 1%, 5% and 10% level respectively.
122
In general, the positive effect of the scheme on Uganda's commodities may be attributed to the
fact that Uganda's agricultural commodities are produced organically. Generally, ACODE
(2006) argues that Uganda has the lowest agro-chemical usage on the African continent.
Furthermore, the International Federation of Organic Agriculture Movement (IFOAM) and
Research Institute of Organic Agriculture (FiBL) (2006) also note that Uganda has the most
established sector of certified organic smallholder producers. By the end of 2010, the country
registered the highest number of certified organic farm households, that is more than 188,600,
while many more uncertified smallholder farmers also manage their farms in compliance with
internationally commended organic standards and guidelines (FiBL and IFOAM, 2013; 2005).
Namuwoza and Tushemerirwe (2011) also contend that Uganda has the largest cultivated
organic area, estimated at more than 0.22 million hectares.
Such organically produced agricultural commodities are in high demand in developed
economies. For instance, according to FiBL and IFOAM (2013), Germany and France had the
second- and third-largest organic market globally, after the United States in 2011. Germany's
organic market size was estimated at US$ 9.2 billion, while France accounted for US$ 5.2 billion
and the United Kingdom, US$ 2.6 billion. In terms of per capita consumption, 2013 statistics
show that Switzerland (US$ 250.4), Denmark (US$ 225.7) and Luxemburg (US$ 187.3) are in
the lead.
The positive influence of the EU-GSP scheme may further be linked to the successful
negotiations between the EU and the USA regarding the mutual recognition of their organic
standards and control systems (FiBL and IFOAM, 2013). Organic products, certified by a
control body recognized for operations in the exporting country, can be sold in both regions
without further inspection or certification. This has relieved some of the producers from
implementing more than one organic standard, thus reducing the certification costs and boosting
trade flows.
The generally positive trend of Uganda's horticultural exports under the EU-GSP scheme may be
attributable to the fact that more than 90 % of these commodities are sold in niche markets. That
is, sold in wholesale markets and through the food service sector which are less stringent with a
number of standard requirements. According to Kleih et al. (2007), the strict demand for
standards and other requirements for trade is largely a construct of the supermarkets, which
Ugandan exporters are able to circumvent.
123
Furthermore, this may be associated with the external support accorded to the sector. For
instance, by 2004, the Export Promotion of Organic Products from Africa (EPOPA) programme
financed by the Swedish International Development Co-operation Agency (SIDA) trained and
linked many smallholder farmers/exporters (19,000) to international markets where they are able
to sell their produce (Forss and Lundström, 2004). According to FAO (2005), there are many
other local and foreign associations and Non-Government Organizations (NGOs) promoting
organic agriculture by building the capacity of smallholder farmers. Mentioned institutions
include National Organic Agriculture Movement of Uganda (NOGAMU), Association for
Strengthening Agricultural Research in Eastern and Central Africa (ASARECA), Network for
Ecofarming in Africa (NECOFA), and Women of Uganda Network (WOUGNET).
Given that most exported fruits and vegetables come from certified organic producers
(Namuwoza and Tushemerirwe, 2011), these commodities tend to meet the minimum EU market
safety standards with regard to Maximum Residue Levels (MRLs). This implies that the
interception of Uganda's horticultural consignments within the EU may not, in most cases, be
grounded in a failure to comply with safety standards. Furthermore, organically produced
products are known to have a smaller carbon footprint, which commands a ready market for the
produce within the EU.
These findings may also be attributed to Uganda's comparative advantage in producing
horticulture commodities, such as bananas, beans, and peppers. The country's bi-modal rainfall
pattern, fertile soils, cheap labour and ambient climatic conditions favour the production and
export of horticultural produce to the EU market (Dolan et al., 2000). Although the preference
margin granted under the GSP scheme may not be so large, most exporters' success is based on
trading in high-value commodities and in relatively large quantities to take advantage of
economies of scale. The EU-GSP scheme is therefore a vital policy instrument in promoting
Uganda's banana, beans and pepper exports to the EU market. (Also see Cipollina et al. 2013,
Cirera et al. 2011, Aiello and Demaria, 2009, and Cipollina and Salvatici, 2009).
The coefficients of all the other covariates used for modelling the effect of the EU-GSP scheme
on Uganda's banana imports into the EU market were found to be statistically significant and
exhibit the expected signs according to the gravity flow model framework. With the exception
of Uganda's mean annual inflation rate (lninflatit), the role of the public sector and government
institutions (lnGOVit), and the dummy variable for common language (Dlangij), the other
variables were found to accord with theoretical expectations. Contrary to theory, fluctuations in
124
the mean annual inflation rate do not influence bean exports to the EU market. This observation
may be associated with high demand and the high price paid for organically produced foodstuffs.
Thus, although high inflation could deter bean exports, the high returns realized upon exporting
beans to the EU are probably responsible for offsetting the negative influence of high inflation.
This, therefore, ends up having no influence on Uganda's bean exports to the EU market.
According to the analytical results (-3.14, p<0.01), having a common official language (Dlangij)
significantly deters imports of Uganda's beans into the EU market by US$ 3140, as compared
with other exporting countries within the market that do not share a common official language.
This peculiar observation is attributable to the fact that, among Uganda's bean export
destinations within the EU (the United Kingdom, Belgium, the Netherlands and Denmark, in that
order), it is only the United Kingdom with whom Uganda shares a common official language
(English). The other countries which use English are Ireland and Luxembourg, but existing data
indicates that none of those countries ever imported beans from Uganda during the period
(2005–2011) considered under this particular analysis.
Contrary to the expected results, Uganda's public sector and government institutions (lnGOVit)
were found not to play a contributory role in enhancing bean imports into the EU market. At
5 % significance level, results (-5.05, p<0.05) indicate that a unit change in the Country Policy
and Institutional Assessment (CPIA) index for transparency, accountability, and corruption leads
to a decline worth US$ 5,050 in the importation of Uganda's beans into the EU. This finding
may be explained by the fact that most government interventions in boosting horticultural trade
have been accorded to prioritized enterprises, such as flowers, bananas, citrus, pineapples and
passion fruits. However, no single policy or government intervention has been found promoting
either bean production or value addition, in particular.
In the case of Uganda's pepper imports into the EU, all covariates were found to be statistically
significant at all levels, although the coefficient on the variable for distance between trading
partners (lnDij) had a positive sign (1.99, p<0.01), instead of a negative sign. This result may be
explained by the argument postulated by scholars (Marimoutou et al., 2009; André and Joel,
2012) that the influence of distance ceases to be an issue if the trading partner's economic size
(GDP) is very large in comparison to that of the exporting country. Indeed, on average, Europe's
GDP is 75 times more than that of Uganda, with Germany (highest) and Luxembourg (lowest)
being about 250 and 4 times larger than Uganda, respectively (WDI, 2012). Furthermore, the
positive sign may also be explained by the view of Kuwornu and Mustapha (2013) that the
125
variety of the crop produced has a great influence on smallholder farmers' accessibility into the
export market. In this regard, Uganda is known for producing highly-favoured pepper (Scotch
bonnet). This variety of pepper is characterized by an aromatic flavour and high pungency,
which are great attributes for peppers. According to Abdulla et al. (2008), the high pungency
and aromatic flavour of the Scotch Bonnet are desirable in the pharmaceutical, as well the foods
and beverages, industries.
5.5 East Africa's unilateral Trade Potential and performance in exporting fruits and
vegetable into the EU market
The assessment of East Africa's trade potential and performance was carried out at country level
for the different horticultural commodities considered under the third objective. Notably, the
third objective examined the impact of the EU-GSP scheme on selected horticultural commodity
imports from East African states into the EU market. The results are presented in Table 5.12
below and in the subsequent figures at country level.
Table 5.12: Mean Absolute Difference (ADijlt) for East Africa's selected horticultural
commodities at country level with the EU-15 states
EU-15 states Kenya ('000 US$) Tanzania ('000 Uganda ('000 US$)
US$)
Beans Asparagus Beans Vegetables Beans Bananas Pepper
Austria 127.5 -467.4 -5.7 0.9 15.9 7.8 1.1
Belgium 8,043.7 -466.9 423.6 1.4 15.6 514.9 425.7
Denmark 3.4 -474.6 -5.2 -3.1 6.3 28.0 1.9
Finland 66.8 -469.6 10.2 -4.5 16.2 1.3 -0.6
France 17,981 -459.5 64.4 2.9 -14.9 5.6 169.4
Germany 9172.0 -455.6 -2.2 3.8 -19.3 114.4 178.5
Greece -7.9 -458.5 -7.9 6.9 -72.5 -7.0 -2.4
Ireland 363.9 -473.2 -7.7 -11.6 25.5 2.4 15.2
Italy 121.6 -466.2 -6.3 7.3 -53.5 -7.3 0.3
Luxembourg 857.1 -469.6 -9.0 -11.8 9.8 -2.4 -3.5
Netherlands 20,331 -463.2 550.2 -0.1 -0.3 29.7 731.3
Portugal 6.4 -468.2 -5.6 -2.8 -4.2 -0.5 -3.1
Spain 11.4 -469.4 -5.4 2.7 -24.7 6.2 8.8
Sweden 8.4 -470.8 -4.9 -3.1 21.7 30 -1.2
UK 85,164 -184.2 2,071 199.5 16.8 2,233.2 2,284.6
Mean EU-15 9,483.4 -447.8 204.0 12.5 -4.1 197.1 253.7
Source: Author's own calculation
126
With the exception of asparagus (US$ -0.45 million) from Kenya and beans (US$ -4,100) from
Uganda, the results based on the mean Absolute Difference (AD) measure generally show that
there exists no un-exhausted trade between EA states and the EU market for these selected
horticultural commodities. In the case of Kenya, the results mean that the current exports of
asparagus to the EU market have not reached their full potential by approximately US$ 0.45
million, while for Uganda bean exports fall short by US$ 4,100. Kenya's asparagus and
Uganda's beans have a high trade potential within the EU market and there exists room for
further trade expansion. This may be attributed to the fact that asparagus is a speciality
vegetable in the EU, while Uganda's beans are renowned for being organically produced.
On the other hand, the results show that the other commodities exceeded their trade potential
with the EU. Kenya's beans registered the highest level of trade flows (US$ 9.5 million) that
surpassed the optimum level, followed by Uganda's pepper (US$ 0.25 million), while vegetables
from Tanzania ranked last (US$ 12,500). This implies that there is hardly any potential for
further trade expansion in these commodities with the EU. The observation for Kenya's beans
may be attributed to the fact that Kenya trades with virtually all EU member states, unlike
Uganda and Tanzania. In the case of Uganda, the upper bound levels of trade in pepper with the
EU may be associated with the commodity's attributes (pungency and aroma), hence making it
very desirable in a number of industries. For Tanzania, the results may be associated with the
fact that it consistently trades with very few EU member states, particularly Belgium and the
United Kingdom (COMTRADE data, 2013).
At country level, the findings indicate that Kenya has un-exhausted trade potential in bean
exports to Greece (US$ 7,900). This suggests that there is supportive evidence for Kenya's trade
expansion with Greece for bean imports. For Tanzania, room for trade expansion in bean
exports exists with Luxembourg, Greece, Ireland, Italy, Austria, Portugal, Spain, Denmark,
Sweden and Germany. Similarly, provision for more trade in vegetables from Tanzania still
exists with Luxembourg (US$ 118,000), Ireland (US$ 116,000), Finland (US$ 4,500), Portugal
(US$ 2,800), and the Netherlands (US$ 100), as well as an estimated trade worth of US$ 3,100
for Denmark and Sweden. This also implies that Tanzania has room for trade expansion with
the above-mentioned EU member states in bean and vegetable commodities.
Uganda has un-exhausted trade potential with Greece and Portugal for all three commodities.
Thus, trade in these commodities has the capacity to grow further. With the exception of pepper
imports into Greece in 2006 and 2008 (COMTRADE data, 2013), these results may be
127
attributable to the fact that Uganda registered no exports to any of these countries for the period
2005–2011. For beans, other EU states with un-exploited trade potential include Italy (US$
53,500), Spain (US$ 24,700), Germany (US$ 19,300), France (US$ 14,900) and the Netherlands
(US$ 300). In the case of bananas, Italy and Luxembourg presented un-exhausted trade potential
at an estimated value of US$ 7,300 and US$ 2,400, respectively. Other than Greece and
Portugal, Luxembourg (US$ 3,500), Sweden (US$ 1,200) and Finland (US$ 600) also present
Uganda with un-exploited trade potential in peppers. Thus, these states provide a basis for more
trade expansion in pepper.
In light of the Relative Difference (RD) Index measure, the results provided in Figure 5.1 below
show that Kenya has a very poor trade performance in asparagus with the EU market, given that
the estimated index lies below zero across all the EU 15 member states. Conversely, Kenya
generally exhibits good trade performance (38 %) in its bean imports into the EU. The results
imply that Kenya has not established adequate trade cooperation with the EU in general in
asparagus trade. On the other hand, the results also mean that Kenya has a good trade
relationship in beans with the EU.
At EU state level, Kenya has very poor trade performance with Portugal (100 %), Denmark
(48 %) and Sweden (11 %) in bean imports. This poor performance may be associated with
language barriers, among other factors. With the exception of Spain (3 %), Kenya has very good
trade relationships with all the other EU-15. The low, but positive, trade performance with
Spain may be linked to divergence in cultural ties with Spain.
128
Figure 5.1: The Relative Difference Index for Kenya's beans and asparagus exports with
the EU-15 member states
Source: Author's own calculation
For Tanzania, there is a general poor trade performance with the EU-15 market for both
vegetables (83 %) and beans (45 %). This may be because there are few countries with which
Tanzania trades within the European Union. According to Figure 5.2 below, Tanzania has a
good trade performance in beans with Belgium (96 %), the Netherlands (97 %), the UK (99 %)
and Finland (14 %), while the UK is the only EU-15 member state with which Tanzania has a
good trade performance for both commodities. This may be associated with the fact that
Tanzania was once a British colony, thus there exists long-term trade relations between the two
countries. This observation is supported by the Absolute Difference measure, which shows that
Tanzania has room for trade expansion in beans and vegetables with a number of EU-15 states.
129
Figure 5.2: The Relative Difference Index for Tanzania's beans and vegetable exports with
the EU-15 member states
Source: Author's own calculation
Uganda's trade performance, as measured by the Relative Difference Index, shows a poor trade
performance in all three commodities with the EU in general. Beans have the weakest trade
performance (74 %), followed by pepper (11 %) and then bananas at 6 %. At the commodity
level, Uganda exhibited a strong trade performance in all three horticultural commodities with
Belgium and the UK, only. This may be attributable to the long-term colonial ties with Britain
and similarity in language. Although English is not one of the three official languages in
Belgium (Wikipedia, 2014), it is widely spoken country wide as a second native language by
Belgians. Figure 5.3 below also shows that Uganda has a poor trade performance with Finland,
Greece, Italy, Portugal and Spain, in all three commodities. The results thus imply that Uganda
has more room to trade with these EU states in all three commodities. Findings based on the
Relative Difference (RD) index identify with the results based the Absolute Difference (AD)
measure, which suggests that Uganda has a wide base for trade expansion in these horticultural
commodities, most especially beans.
130
Figure 5.3: The Relative Difference Index for Uganda's beans, bananas and pepper exports
with the EU-15 member states
Source: Author's own calculation
5.6 Summary of results and discussions
Purposively, this chapter aimed to analytically determine the export competitiveness of various
fruit and vegetable products from three East African states in the EU market, to develop a set of
climate change proxies and then use the proxies to ascertain the influence of climate change on
East Africa’s horticultural trade flows, and to examine the impact of the EU-GSP scheme on the
various fruits and vegetable commodities imported into the EU from the East African region.
Furthermore, this chapter aimed at predicting the trade potential and performance of East
Africa's selected horticultural commodities imported into the EU market.
The mean export competitiveness results show that Uganda (1,321) by far out-competes Kenya
(495) and Tanzania (2) within the EU-market. Detailed analysis shows that Kenya exhibits
export competitiveness in more commodities (10) as compared with Uganda and Tanzania,
which are only competitive in six (6) and five (5) commodities, respectively. For Kenya,
“Asparagus (070920)” is the most competitive commodity, followed by “Mushrooms & truffles
(070959)” among others. Kenya’s good export competitiveness in the 10 horticultural
commodities may be attributable to the breath-taking efforts undertaken by both the government
and the private sector in developing the horticulture industry. For Tanzania, “Vegetables
(070990)” were the most competitive, while in the case of Uganda, fruits of the genus Capsicum
or of the genus Pimen, (070960), bananas and eggplants exhibited the highest levels of export
competitiveness within the EU market. Thus, East African countries should capitalize on
131
exporting horticultural commodities over which they have export competitiveness, viz,
Asparagus, Mushrooms and truffles (Kenya); Vegetables for Tanzania; while Uganda should
dwell more on peppers, bananas and eggplants, among other horticultural commodities.
The statistically significant climate change results depict both positive and negative influences
on East Africa's horticultural trade flows into the EU-market, depending on the climate change
proxy being put into perspective. While precipitation anomalies in the importing countries (EU)
are noted to enhance horticultural trade flows from all the three East African states, temperature
anomalies seem to negatively influence trade. On the other hand, temperature anomalies in
exporting countries seem to boost trade in horticultural commodities from Tanzania and Uganda,
but they may limit trade flows from Kenya. Similarly, precipitation anomalies in exporting
countries favour horticultural trade flows from Kenya, while they deter trade flows from
Tanzania and Uganda. Thus, it is prudent to conclude that climate change has both trade-
enhancing and trade-deterring effects, depending on the country and the proxy being put into
consideration. Therefore, owing to the trans-boundary nature of the climate change
phenomenon, East African states should collaborate in designing, coordinating and
implementing pro-growth and pro-poor development policies and investment strategies that will
enhance the sustainability of the horticulture sector, as well as adapt to climate change. Such
ventures may include breeding improved horticultural cultivars that can tolerate extreme climatic
conditions, and investment in physical infrastructure, such as irrigation dams.
To determine the effect of the EU-GSP scheme, the preference margins of each of the seven
horticultural commodities that exhibited export competitiveness within the EU-market were used
for analysis within the gravity flow model framework. For Kenya, results show that the scheme
reduces the value of trade flows in asparagus and beans by an estimated US$ 2,460 (p<0.01) and
US$ 280 (p<0.01), respectively. This implies that the EU-GSP scheme is of less importance in
boosting the importation of Kenya's asparagus and bean commodities into the EU-market. For
Tanzania, findings at one per cent level of significance indicate that the scheme significantly
enhances importation of beans into the EU, thus implying that it is a relevant policy in this
scenario. However, despite the fact that the EU-GSP scheme has negative effects on Tanzania's
vegetable imports into the EU, it is inconclusive since the result was insignificant at all levels.
For Uganda, the statistically significant results indicate that the scheme promotes the importation
of bananas, beans and pepper into the EU market. It is thus insightful to conclude that the EU-
GSP scheme is a very vital policy instrument in promoting importation of Uganda's horticultural
produce. Hence, Kenya should explore other international markets, such as the Middle East and
132
the USA, for its asparagus and bean exports. However, Tanzania and Uganda should continue to
aggressively seize the market opportunity granted by the EU, particularly for bananas and
peppers (Uganda), as well as beans (Uganda and Tanzania).
In light of East Africa's trade potential and performance with the EU market, seven commodities
were analysed. Generally, Kenya and Uganda exhibit un-exhausted trade potential in asparagus
and beans worth US$ 0.44 million and US$ 4,100, respectively. This implies that there is room
for further trade expansion within the EU market for these particular commodities. At
commodity level, there exists a very large possibility for trade expansion for Kenya's asparagus
across all the EU-15 member states that were considered in this study. For beans, results show
that Kenya has un-exhausted trade worth about US$ 8,000 with only Greece, among other EU-
member states. This suggests that Kenya can still by far expand her trade in beans with Greece
than many other EU states. With the exception of Belgium, Finland, France, the Netherlands
and the United Kingdom, all the other EU member states have room for expansion with
Tanzania's trade in beans. In the case of Uganda, key destination markets with further
possibility of trade expansion for beans include France, Germany, Greece, Italy, Spain, Portugal
and the Netherlands. Furthermore, results for Uganda's pepper show that Luxembourg, Portugal,
Greece, Sweden and Finland still have un-tapped trade opportunities that could be exploited by
Uganda.
In terms of trade performance, it is plausible to conclude generally that Kenya has a very poor
trade performance in asparagus within the EU market, while its beans perform fairly well.
Similarly, with the exception of the UK, Tanzania has a very poor trade performance within the
EU market for both beans and vegetables. Likewise, Uganda registered a poor trade
performance in all the three commodities (bananas, beans and pepper) imported into the EU.
The poor trade performance for the EA states within the EU market implies that EA states have
not yet established strong trade relationships within the market.
All in all, empirical findings reveal that by successfully using a heterogeneous set of climate
change proxies (anomalies in temperature and precipitation in both the importing and exporting
countries) to investigate the effect of climate change on international trade, I have been able
affirm that climate change possesses both negative and positive impacts on international trade in
horticultural commodities. The positive and negative effects depend on the type of commodity,
the origin of the commodity and the type of proxy used to quantify climate change.
Furthermore, the results indicate that, by using a preference margin based on the trade weighted
133
applied Most Favoured Nation (MFN) rate and the Ad Valorem Equivalents (AVEs), I have ably
ascertained that the EU-GSP scheme selectively promotes importation of horticultural
commodities into the EU-market, depending on the country of origin of the commodity.
134
CHAPTER SIX:
SUMMARY, CONCLUSION AND RECOMMENDATIONS
In this chapter, a brief summary, conclusions and recommendations are presented with reference
to each objective. In particular cases, conclusions and recommendations are specific to a
particular country.
6.1 Summary and conclusions
As one of the objectives, the study aimed at investigating the effects of a developed set of
meteorological data variables as climate change proxies on international trade by using panel
estimation techniques. The objective was achieved by using generated anomalies in temperature
and precipitation, both in the exporting and importing countries. The use of anomalies in
temperature and precipitation as proxies for climate change, particularly for trade flows skewed
towards agricultural commodities, was motivated by the fact that these two factors are direct
inputs in the agricultural sector. Thus, their fluctuations will directly be reflected in the volumes
of agricultural commodities traded.
Findings based on the developed set of climate change proxies, viz, temperature and
precipitation anomalies, imply that climate change has both positive and negative influences on
horticultural trade flows to the EU-Market. The influence, however, depends on the climate
change proxy being put into consideration. Within the EU market, anomalies in precipitation
enhance horticultural imports from East Africa, while temperature anomalies seem to hinder
trade. Anomalies in temperature in exporting countries boost horticultural trade flows from
Tanzania and Uganda, while the contrary is true for Kenya. Precipitation anomalies in exporting
countries favour horticultural trade flows from Kenya, while they curtail trade flows from
Tanzania and Uganda. These empirical results correlate with findings of other scholars. Thus, it
is prudent to conclude that the proposed approach of assessing the effects of climate change
based on meteorological anomalies in agrarian-based economies may be a more reliable measure
than the use of proxies based on Kyoto Protocol policies.
Furthermore, the study endeavoured to determine the effect of the EU-GSP preferential trade
agreement on East Africa’s fruits and vegetable exports to the European Union market. This
objective was successfully achieved using the preference margin variable based on the trade
weighted applied Most Favoured Nation (MFN) rate and the Ad Valorem Equivalents (AVEs).
135
The trade weighted preference margin measure takes into account all the policy instruments
embedded within the EU-GSP scheme and other competitors within the EU market.
Empirical results suggest that the EU-GSP scheme selectively favours exports of certain
horticultural commodities to the EU-market, depending on the country of origin. Particularly,
the scheme promotes importation of bananas, beans and peppers from Uganda and beans from
Tanzania. On the contrary, this policy instrument does not enhance asparagus and bean imports
from Kenya. In conclusion, the use of preference margin, based on all the policy instruments
embedded within the EU-GSP scheme, provides appropriate commodity-specific inferences
regarding the effect of the EU-GSP scheme on horticultural imports into the EU market.
The study also aimed at determining the export competitiveness of East Africa's fruit and
vegetable exports to the European Union market. The objective was successfully attained by
using Balassa's index. At country level, results show that Kenya has export competitiveness in
more commodities (10) than Uganda (6) and Tanzania (5). For Kenya, “Asparagus (070920)”
and "Mushrooms & truffles (070959)" are the most competitive commodities, while for Uganda,
fruits of the genus Capsicum or of the genus Pimen, (070960), bananas and eggplants registered
the highest levels of export competitiveness within the EU market. In the case of Tanzania,
“Vegetables (070990)” were the most competitive. Conclusively, each of these countries
exhibits comparative advantage in exporting those commodities over which it has export
competitiveness. Thus, East African countries should capitalize on exporting those horticultural
commodities over which they have export competitiveness, viz, Asparagus, Mushrooms and
truffles (Kenya); Vegetables for Tanzania; while Uganda should dwell more on peppers, bananas
and eggplants, among other horticultural commodities.
Lastly, the study aimed at predicting East Africa’s unilateral trade potential and performance.
Kenya and Uganda exhibit the existence of un-realized trade potential, thus implying that these
countries still have room to expand their horticultural trade with the EU market. For Kenya,
asparagus is a key commodity for further market expansion across all EU member states, while
to expand Uganda's market for beans and pepper, trade partnerships should be considered with
countries such as France, Germany, Luxembourg, Portugal and Greece. Other than Belgium,
Finland, France, the Netherlands and the United Kingdom, all the other EU member states have
room for trade expansion for beans from Tanzania. With regard to trade performance, results
indicate that all the three East African states have poor trade performance with the EU market in
the various commodities. This suggests that there exists an array of trade barriers curtailing East
136
Africa's horticultural imports into the EU. Conclusively, there is need for the East African states
to foster trade cooperation in horticultural commodities with the EU member states.
6.2 Recommendations
The following recommendations are based on the empirical results of this study.
Recommendations are categorized into four (4) types, viz, recommendations to exporters; policy
recommendations; recommendations to researchers, and recommendations for further research.
6.2.1 Recommendations to exporters
 Based on empirical results from the analysis for export competitiveness, exporters should
trade more in the following top five horticultural commodities at country level. For
Kenya, the key commodities are: Asparagus (070920), Mushrooms & truffles (070959),
Spinach and New Zealand spinach (070959), Aubergines (070930) and Peppers
(070960). For Tanzania, Vegetables (070990), Kidney & white pea beans (071333),
Bananas (080300), Beans (070820) and Urd, mung, gram beans (071331) are the most
important. For Uganda, peppers (070960), Aubergines (070930), Bananas (080300),
Vegetables (070990) and Beans (070820) are the major horticultural commodities that
should be focused on.
 In light of results for trade potential and performance, it is commended that exporters,
with support from government institutions responsible for promoting trade, should
strengthen trade cooperation with the various EU member states, particularly with
Greece, Portugal, Sweden Luxembourg and Italy, among others. This will greatly
enhance the capacity of East African states to exploit the untapped trade potential within
the EU market.
 Based on trade potential and performance results, it is prudent to recommend that East
African states should consider exploring and strengthening trade linkages with
alternative markets such as the USA and the Middle East. This will probably reduce
greatly the non-tariff barriers to trade associated with the frequently changing strict
"voluntary" standard (Global G.A.P) within the EU market.
6.2.2 Policy recommendations
 In light of the developed set of climate change proxies based on temperature and
precipitation, it is recommended that representatives of agriculture-based economies,
137
such as Kenya, Tanzania and Uganda, should lobby and advocate for putting into
consideration other measures for quantifying the effects of climate change, rather than
relying on measures based on Kyoto Protocol policies. The lobbying and advocating for
such pertinent considerations should be brought forward at the international climate
change negotiation forums, at which agriculture-based economies have representatives.
Measures based on Kyoto Protocol policies are more appropriate and applicable for
industrious countries. Putting into practice such new measures for quantifying climate
change for agriculture-based economies will greatly protect such countries from being
compelled to adopt the unaffordable or costly industrial technologies, which are noted to
be one of the factors hindering the curbing of climate change effects. The use of
anomalies will most likely stimulate development of technological innovations, like
breeding crop varieties that are tolerant to extreme climatic conditions and which will
match the changing climate within the agricultural sector.
 Based on the negative findings relating to climate change effects on trade, it is
recommended that East African states should design and implement good overall
development policies and programmes. Owing to the trans-boundary nature of the
unprecedented climate change phenomenon, the East African states should collaborate to
design, coordinate and implement pro-growth and pro-poor development policies that
enhance the sustainability of the horticulture sector, as well as the adaptation to climate
change.
 Kenya, Tanzania and Uganda should undertake more investments aimed at boosting
productivity of the horticultural sector. Such investments may include:
- Agricultural research and development. Through science- and technology-based
innovations, like the breeding improved horticultural cultivars (biotechnology) that
can withstand extreme climatic conditions and improved farm management
practices, the productivity of this sector will become less vulnerable to climatic
fluctuations. This in the long-run translates into high and sustainable production of
horticultural commodities, thus implying the availability of horticultural
commodities that can be traded globally.
- East African states should also invest in physical infrastructure, such as irrigation
dams so as to enhance efficiency in water use.
138
6.2.3 Recommendations to researchers
 Within the gravity flow model framework, it is recommended that anomalies in
temperature and precipitation should be used to proxy for climate change when
evaluating the impact of climate change on trade flows skewed towards agricultural
commodities. That is, if a country's or a region's exports comprise mostly agricultural
commodities, then anomalies in temperature and precipitation should be used. The other
climate change proxies based on Kyoto Protocol policies should be used when dealing
with manufactured goods from industry-based economies.
 Evaluation of the influence of non-reciprocal preferential trade agreement(s) granted to
developing countries, based on preference margins, should always take into account the
various instruments embedded within the EU-GSP scheme. The omission of any of the
instruments may lead to over-estimation of the preference margin.
6.2.4 Recommendations for further research
 Because of the lack of comparable meteorological datasets, this study was at one hand
based on historical meteorological data (1988–2000), which may not provide an adequate
perspective of the three East African states' horticultural trade flows to the EU market.
Therefore, further research based on temperature and precipitation anomalies is desirable
to affirm the effects of climate change.
 The evaluation of the impact of the EU-GSP scheme, while employing the preference
margin variable, focused on selected horticultural commodities. However, given that the
EU-GSP takes into consideration other commodities, it is recommended that the
preference margin approach used in this study should be employed on the other
commodities as well.
139
REFERENCES
Abdulla, M., Mobley, R. and Carter, L. 2008. Exploring New Opportunities For Scotch Bonnet
Hot Pepper. Extension Fact Sheet No. 1 Fall 2008. Florida A&M University, Florida. Accessed
at: http://gadsden.ifas.ufl.edu/pdfs/SCOTCH%20BONNET%20FLIER.pdf.
ACODE 2006. The Status of Organic Agriculture Production in Uganda. Unpublished report.
Advocates Coalition for Development Environment (ACODE), Kampala.
Acosta, R, G., Calfat, G. and Flôres R.G. Jr. 2005. Trade and Infrastructure: Evidences from the
Andean Community. Ensaios Econômicos No. 580, EPFGV, Brazil.
Ahearn, M., Culver, D. and Schoney, R. 1990. Usefulness and limitations of COP estimates for
evaluating international competitiveness: A comparison of Canadian and U.S. wheat. American
Journal of Agricultural Economics, 72(5): 1283-1291.
Aichele, R.. and Felbermayrz, G. 2010. Kyoto and the carbon content of trade.
Aiello, F., and Demaria, F. 2009. Do trade preferential agreements enhance the exports of
developing countries? Evidence from the EU GSP. MPRA Paper No. 20093. Accessed at:
http://mpra.ub.uni-muenchen.de/20093.
Akgüngör, S., Barbaros, F. and Kumral, N. (2002). Competitiveness of the Turkish Fruit and
Vegetable Processing Industry in the EU Market. Russian and East European Finance and
Trade, 38(3): 34-53.
Alexandrov, V.A. and Hoogenboom, G. 2001. Climate variation and crop production in Georgia,
USA during the twentieth century. Climate Research, 17(1):33-43.
Aloui, O. and Kenny, L. 2005. The Cost of Compliance with SPS standards for Moroccan
Exports: A case Study. Agriculture and rural development discussion paper. Washington, DC:
World Bank. Accessed at:
http://documents.worldbank.org/curated/en/2005/01/10375330/cost-compliance-sps-standards-
moroccan-exports-case-study.
Alvarez-Coque, J.M., Martí Selva, M.L. 2006. A Gravity approach to assess the effects of
associations agreements on Euro-Mediterranean Trade of Fruits and Vegetables, MPRA Paper
No. 4124.
Amita, B. 2004. India’s global trade potential: The gravity model approach. New Dheli, India.
Indian Council for Research on International Economic Relations.
Anderson, J. E. 1979. A theoretical foundation for the gravity equation. American Economic
Review, 69: 106-116.
Anderson, J. E. and VanWincoop, E. 2003. Gravity with gravitas: a solution to the border
puzzle. American Economic Review, 93: 170–192.
Anderson, M.J., Gorley, R.N. and Clarke, K.R. 2008. PERMANOVA+ for PRIMER: Guide to
Software and Statistical Methods. PRIMER-E, Plymouth, UK.
140
Andre, J. C. and Joel, H. E. 2012. Determinants of South Africa's exports of leather products.
Agrekon, 51(2): 38-52.
Anisimovaite, K. and Marcisauskiene J. 2008. Factors affecting the Lithuanian textile and
clothing sector competitiveness analysis. Economics and Management: Current Issues and
Perspectives, 3(12): 14-24.
ARC. 2013. Cultivating cashew nuts. ARC-Institute for Tropical and Subtropical Crops,
Department of Agriculture. Accessed at : http://www.daff.gov.za/docs/Infopaks/cashew.htm
Armington, P.S. 1969 A Theory of demand for products distinguished by place of  production.
IMF Staff Papers, 16: 159-178.
Armstrong, S. 2007. Measuring trade and trade potential: A survey. Asia Pacific economic
papers No. 368, 2007. Australia–Japan Research Centre. Canberra
Arnon, A. and Weinblatt, J. 1998. Linder’s hypothesis revisited: Income similarity effects for
low income countries. Applied Economic Letters, 5: 607- 611.
Asfaw, S. Mithöfer, D. and Waibel, H. 2010. What Impact Are EU Supermarket Standards
Having on Developing Countries' Export of High-Value Horticultural Products? Evidence From
Kenya. Journal of International Food & Agribusiness Marketing, 22(3): 252-276.
Athanasoglou, P. P., Backinezos, C. and Georgiou, E. A. 2010. Export performance,
competitiveness and commodity composition. Working paper No. 114. Bank of Greece.
Augier, P., Gasiorek, M., and Lai-Tong, C. 2005. The impact of rules of origin on trade flows.
Economic Policy, 20: 567–623.
Baier, S. L., and Bergstrand, J. H. 2007. Do free trade agreements actually increase members'
international trade? Journal of International Economics, 71(1): 72 - 95.
Balance, R. H. 1988. Trade performance as an indicator of comparative advantage. In: D.
Greenway, ed. 1988. Economic Development and International Trade. London. Macmillan.
Balassa, B. 1989. Comparative Advantage, Trade Policy and Economic Development, Harvester
Wheatsheaf, New York.
Balassa, B. 1977. Revealed Comparative Advantage Revisited: An Analysis of Relative Export
Shares of the Industrial Countries, 1953-1971. The Manchester School of Economic & Social
Studies, 45(4): 327-344.
Balassa, B. 1965. Trade Liberalisation and ‘Revealed’ Comparative Advantage. The Manchester
School, 33: 99-123.
Baldwin, R.E. 1994. Towards an Integrated Europe. CEPR. Graduate Institute of International
Studies, Ave de la Pais, 1202 Geneva.
Banterle, A. and Carraresi, L. 2007. Competitive performance analysis and European Union
trade: The case of the prepared swine meat sector. Food Economics - Acta Agricult Scand C,
4:159-172.
141
Barrios, S., Bertinelli, L., and Strobl, E. 2010. Trends in Rainfall and Economic Growth in
Africa: A Neglected Cause of the African Growth Tragedy. Review of Economics and Statistics,
92(2): 350-366.
Barrios, S., Bertinelli, L., and Strobl, E. 2006. Climatic change and rural-urban migration: The
case of sub-Saharan Africa. Journal of Urban Economics, 60: 357-371.
Barro, R.J. 1991. Economic Growth in a Cross Section of Countries. Quarterly Journal of
Economics, 106(2): 407 – 433.
Bashaasha, B., Thomas, T.S., Waithaka, M. and Kyotalimye, M.. 2013. Uganda In: M.
Waithaka, G. C. Nelson, T. S. Thomas, and M. Kyotalimye, eds. 2013. East African Agriculture
and Climate Change: A Comprehensive Analysis. Washington, DC: International Food Policy
Research Institute. Ch. 12.
Batra, A. 2006. India’s global trade potential: The gravity model approach. Global Economic
Review: Perspectives on East Asian Economies and Industries, 35(3): 327-361.
Bear, M. A. and Goldman, R. H. 2005. Enhancing local sourcing of fresh fruit and vegetables in
Uganda’s domestic market. DFID.
Belsley, D.A., Kuh, E. and Welsch, R.E. 1980. Regression Diagnostics. Identifying Influential
Data and Sources of Collinearity. New York: John Wiley & Sons.
Berg, A., De Noblet-Ducoudré, N., Sultan, B., Lengaigne, M. and Guimberteau, M. 2013.
Projections of climate change impacts on potential C4 crop productivity over tropical regions.
Agricultural and Forest Meteorology, 170: 89–102.
Bergstrand, J. H. 1989. Trade generalized gravity equation, monopolistic competition, and the
factor-proportions theory in international trade. Review of Economics and Statistics, 71:143 –
153.
Bergstrand, J. H. 1985. The Gravity Equation in International Trade: Some Microeconomic
Foundations and Empirical Evidence. Review of Economics and Statistics, 67: 474 –481.
Bettin, G., and Nicolli, F. 2012. Does climate change foster emigration from less developed
countries? Evidence from bilateral data. Università Degli Studi Di Ferrara.
Bineau, Y. and Montalbano, P. 2011. Selected developmental aspects of trade and trade policies.
Final Report, Project No. 2011/265311.
Blasi, G.D., Seccia, A. Carlucci, D. Santeramo, F.G. 2007. Exports analysis of Italian high
quality wine by using gravity model approach. Contributed Paper prepared for presentation at
the 105th EAAE Seminar "International Marketing and International Trade of Quality Food
Products", Bologna, Italy, March 8-10, 2007
Boansi, D. 2013. Export performance and macro-linkages: A look at the competitiveness and
determinants of cocoa exports, production and prices for Ghana. MPRA Paper No. 48345.
Available online at:  http://mpra.ub.uni-muenchen.de/48345/
Bojnec, S. 2003. Three concepts of competitiveness measures for livestock production in Central
and Eastern Europe. Agriculturae Conspectus Scientificus, 68(3): 209-220.
142
Boyanowsky, E. 1999. Violence and Aggression in the Heat of Passion and in Cold Blood: The
Ecs-TC Syndrome. International Journal of Law and Psychiatry, 22(3): 257-271.
Bouët, A., Decreux, Y., Fontagné, L., Jean, S., and Laborde, D. 2004. A consistent, ad-valorem
equivalent measure of applied protection across the world: The MAcMap-HS6 database.
Working Paper No. 2004-22. CEPII.
Bowen, H.P. 1983. On the theoretical interpretation of Trade intensity and revealed comparative
advantage. Weltwirtschaftliches Archiv, 119: 464-472.
Breitung, J. 2000. The local power of some unit root tests for panel data, In: Baltagi, 2000. ed.
Advances in Econometrics, Volume 15: Nonstationary Panels, Panel Cointegration, and
Dynamic Panels. Amsterdam: JAY Press.
Breitung, J. and S. Das. 2005. Panel unit root tests under cross-sectional dependence. Statistica
Neerlandica, 59: 414–433.
Brockett, P.L., Wang, M. and Yang, C. 2005. Weather derivatives and weather risk
management. Risk Management and Insurance Review, 8(1): 127-140.
Brulhart, M. and Kelly, M. J. 1999. Ireland's trading potential with central and eastern
European countries: A gravity study. Economic and Social Review, 30(2): 159 – 174.
Brun, J., Carrère, C., Guillaumont, P. and De Melo, J. 2005. Has distance died? Evidence from a
panel gravity model. The World Bank Economic Review, 19(1): 99 - 120.
Bruneckiene, J. and  Paltanaviciene, D. 2012. Measurement of Export Competitiveness of the
Baltic States by Composite Index. Inzinerine Ekonomika-Engineering Economics, 23(1), 50-62.
Buch, C. M., Kleinert, J. and Toubal, F. 2003. The Distance puzzle: On the interpretation of the
distance coefficient in the Gravity equations, Kiel Working Paper No. 1159.
Burger, M., Van Oort, F., and Linders, G.J. 2009, On the specification of the gravity model of
trade: Zeros, excess zeros and zero-inflated estimation. Spatial Economic Analysis, 4(2): 167-
190.
Cameron, A. C. and Trivedi, P. K. 1986. Econometric models based on count data: Comparisons
and applications of some estimators and tests. Journal of Applied Econometrics, 1: 29 - 53.
Capalbo, S., Ball, E. and Denny, M. 1990. International comparisons of agricultural
productivity: Development and usefulness. American Journal of Agricultural Economics, 72(5):
1292-1297.
Caporale, G. M, Rault, C., Sova, R., and Sova, A. 2009. On the bilateral trade effects of free
trade agreements between the EU-15 and the CEEC-4 countries. Rev. of World Economy
145:189-206.
Carraresi, L. and Banterle, A. 2008. Measuring Competitiveness in the EU Market: A
Comparison Between Food Industry and Agriculture. Paper presented at the 12th EAAE
Congress, Gent, Belgium, 27-30 August.
143
Cardamone, P. 2011. The effect of preferential trade agreements on monthly fruit exports to the
European Union. European Review of Agricultural Economics, pp. 1–34.
doi:10.1093/erae/jbq052.
Cardamone, P. 2010. The effect of preferential trade agreements on monthly fruit exports to the
European Union. European Review of Agricultural Economics, 38 (4): 553–586.
doi:10.1093/erae/jbq052.
Cardamone, P. 2009. Preferential trade agreements granted by the European Union: an
application of the gravity model using monthly data. Pue&Piec Working Paper n. 09/6.
Cardamone, P. 2007. A survey of the assessments of the effectiveness of preferential trade
agreements using gravity models. TradeAg Working Paper n. 07/09.
Carrère, C. 2011. A new measure of tariff preference margins adjusted for import and domestic
competition. Ferdi, Working paper P19.
Carrère, C. 2006. Revisiting the effects of regional trade agreements on trade flows with proper
specification of the gravity model. European Economic Review, 50(2), 223- 247.
Carrère C. and de Melo, J. 2010. The Doha Round and Market Access for LDCs: Scenarios for
the EU and US Markets. Journal of World Trade, 44(1): 251-290.
Chen, X., Yang, Z. and Lui, X. 2007. Empirical Analysis of Xinjiang’s Bilateral Trade: Gravity
Model Approach. Chinese Geographical Science, 18(1): 9 -16.
Cheng, I. 1999. The Political Economy of Economic Integration.  Unpublished Ph.D. thesis.
University of London. London.
Cheng, I. and Wall, H. J. 2005. Controlling for heterogeneity in gravity models of trade and
integration. Federal reserve Bank of St. Louis Review, 87(1): 49-64.
Choi, I. 2001. Unit root tests for panel data. Journal of International Money and Finance 20:
249–272
Christie, E. 2002. Potential Trade in Southeast Europe: A Gravity Model Approach. Working
Paper No. 21. The Vienna Institute for International Economic Studies-WIIW.
Cipollina, M., Laborde, D. and and Salvatici, L. 2013. Do Preferential Trade Policies (Actually)
Increase Exports? An analysis of EU trade policies. Selected Paper prepared for presentation at
the Agricultural & Applied Economics Association’s 2013 AAEA & CAES Joint Annual
Meeting, Washington, DC.
Cipollina, M. and Pietrovito, F. 2011. Trade impact of EU preferential policies: a meta-analysis
of the literature, In: Luca De Benedictis e Luca Salvatici (eds), 2011. The Trade Impact of
European Union Preferential policies: An analysis through gravity models. Berlin/Heidelberg:
Springer.
Cipollina, M. and Salvatici, L. 2010. The trade impact of European Union agricultural
preferences. Journal of Economic Policy Reform, 13: 87–106.
144
Cipollina, M. and Salvatici, L. 2008a. Measuring protection: mission impossible? Journal of
Economic Surveys, 22: 577–616.
Cipollina, M and Salvatici, L. 2008b. EU and developing countries: what is the impact of
agricultural preferences? Paper presented at the 10th European Trade Study Group conference,
Warsaw.
Cipollina, M and Salvatici, L. 2007. EU and developing countries: an analysis of preferential
margins on agricultural trade flows. Accessed online at:
http://www.fondazionemasi.it.isiportal.com/UploadDocs/187_Cipollina_Salvatici1.pdf
Cirera, X., Foliano, F. and Gasiorek, M. 2011. The impact of GSP Preferences on Developing
Countries' Exports in the European Union: Bilateral Gravity Modelling at the Product Level.
Working Paper Series No. 27-2011. University of Sussex. Brighton
Collier, P., Hoeffler, A. and Soderbom, M. 2008. Post-Conflict Risks. Journal of Peace
Research, 45(4):461– 478.
COMTRADE database, 2013. Market Access map. Improving transparency in international
trade and market Access. Accessed at:
http://www.macmap.org/AdvancedSearch/RawData/TradeResult.aspx?prevpage=TradeByCount
ry.aspx&productlength=6&nomenclature=HS&source=5
Cooper, P.J.M., Dimes, J., Rao, K.P.C., Shapiro, B., Shiferaw, B., Twomlow, S. 2008. Coping
better with current climatic variability in the rain-fed farming systems of sub-Saharan Africa: an
essential first step in adapting to future climate change? Agric. Ecosyst. Environ. 126: 24–35.
Dalum, B., Laursen, K. and Villumsen, G. 1998. Structural change in OECD export
specialization patterns: De-specialization and stickiness. International Review of Applied
Economics, 12: 447-467.
Day, R., Tambi, E. and Odularu, G. 2012. An analysis of compliance with selected sanitary and
phytosanitary measures in Eastrern and Southern Africa. Forum for Agricultural Research in
Africa (FARA), Accra, Ghana.
Deardorff, A. V. 1998. Determinants of bilateral trade: Does gravity work in a neoclassical
world? In J.A. Frankel, ed. 1998. The Regionalization of the World Economy, pp. 7–28. Chicago:
University of Chicago Press.
Dell, M. Jones, B. F., and Olken, B. A. 2008. Climate change and economic growth: Evidence
from the last half century. NBER working paper series, Working Paper 14132. Available at:
http://www.nber.org/papers/w14132.
Deressa, T., and Hassan, R. M. 2009. Economic Impact of Climate Change on Crop Production
in Ethiopia: Evidence from Cross-Section Measures. Journal of African Economics 18 (4): 529–
554.
Derksen, S. and Jegou, I. 2013. The Trade and Climate change linkages. ICTSD Global Platform
on Climate Change, Trade and Sustainable Energy.
145
Deschenes, O. and Greenstone, M. 2007. The economic impacts of climate change: evidence
from agricultural output and random fluctuations. The American Economic Review, 97(1):354–
355.
Dickey, D. A. and Fuller, W. (1979). Distribution of the estimators for autoregressive time series
with a unit root. Journal of American Statistical Association, 74: 427 - 431.
Dinda, S. 2011. Climate Change and Development: Trade Opportunities of Climate Smart
Goods and Technologies in Asia. Accessed at: http://mpra.ub.uni-muenchen.de/34883/
Dolan, C., Humphrey, J. and Harris-Pascal, C. 2000. Horticulture commodity chains: The impact
of the UK market on the African fresh vegetable industry. IDS working paper 96.
Drine, I. 2011. Climate Variability and Agricultural Productivity in MENA region. Working
Paper No. 2011/96.
Egger, P. 2000. A note on the proper econometric specification of the gravity equation.
Economics Letters, 66: 25-31.
EM-DAT. 2013. The OFDA/CRED International Disaster Database. www.em-dat.net -
Université Catholique de Louvain - Brussels - Belgium".
Emlinger, C., Chevasuss Lozza, E. and Jacquet, F. 2008. Tariffs and other trade costs: assessing
obstacles to Mediterranean countries’ access to EU-15 fruit and vegetable markets. European
Review of Agricultural Economics, 35: 409–438.
Enders, W. 1995. Applied econometric time series. New York: Wiley.
Engle, R. F. and Granger, C. W. J. 1987. Co-integration and error-correction: Representation,
estimation and testing. Econometrica, 55: 251—276.
English, P., Jaffee, S. and Okello, J. 2004. Exporting Out of Africa- Kenya’s Horticulture
Success Story. A case study from Reducing Poverty, Sustaining Growth—What Works, What
Doesn’t, and Why A Global Exchange for Scaling Up Success. Scaling Up Poverty Reduction:
A Global Learning Process and Conference. Shanghai, May 25–27, 2004. Accessed at:
http://info.worldbank.org/etools/docs/reducingpoverty/case/120/fullcase/Kenya%20Horticulture
%20Full%20Case.pdf
Eichengrean, B., and Irwin, D. 1998. The role of history in bilateral trade flows, In: ed Frankel J.
1998, The Regionalisation of the World Economy. Chicago, University of Chicago Press, 33 –
57.
Esterhuizen, D. 2006. An evaluation of the competitiveness of the South African agribusiness
sector. Diss. University of Pretoria.
Esterhuizen, D. and Van Rooyen, C. J. 2000. Investment Opportunities in Agribusiness -The
Rwanda Case study. In: Van Rooyen, C. J., Doyer, O. T., Bostyn, F., and Haese, L.D. 2000.
Readings in Agribusiness. A source book for Agribusiness training. University of Pretoria,
Pretoria.
146
European  Communities,. 2009. European competitiveness report. Commission staff working
document, SEC (2009)1657 final, 216p. Available at: http://bookshop.europa.eu/en/european-
competitiveness-report-2009- pbNBAK09001/?CatalogCategory.
European Communities (EC). 2001. Council Regulation (EC) No 2501/ 2001 of 10 December
2001. Official Journal, L 346.
EU.  2008. Council Regulation (EC) No 732/ 2008 of 22 July 2008. Official Journal 2008 L 211.
EU. 2005. Council Regulation (EC) No. 980/2005 of 27 June 2005. Official Journal, 2005 L
169.
Evenett, S.J. and Keller, W. 1998. On the Theories Explaining the Success of the Gravity
Equation. NBER Working Paper, No. 6529, Cambridge, MA
FAO. 2001. Climate Change: Implications for Food Safety. Rome. FAO
FAO. 2005. Uganda Country Profiles for Organic Agriculture. Accessed at
[http://www.fao.org/organicag/display/work/display_2.asp%3Fcountry%3DUGA%26lang%3De
n%26disp%3Dsummaries].
FAO database, 2013. FAOSTAT. Accessed at: http://faostat3.fao.org/faostat-
gateway/go/to/home/E.
Fertő, I. and Hubbard, L.J. 2002. Revealed Comparative Advantage and Competitiveness in
Hungarian Agri-food sectors. Discussion Papers, MT–DP.2002/8. Hungarian Academy of
Sciences, Budapest.
Fertő, I. and Hubbard, L.J.  2001. Competitiveness and Comparative Advantage in Hungarian
Agriculture. Discussion Papers, MT–DP.2001/2. Hungarian Academy of Sciences, Budapest.
FiBL and IFOAM, 2013. The World of Organic Agriculture Statistics and Emerging
Trends2013. Frick and Bonn. IFOAM and FiBL.
Field, S. 1992. The Effect of Temperature on Crime. The British Journal of Criminology 32:340-
351.
Flowerdew, R. and Aitkin, M. 1982. A method of fitting the gravity model based on the Poisson
distribution. Journal of Regional Science, 22: 191-202.
Folfas, P., Nyga-Łukaszewska, H., and Słok-Wódkowska, M. 2011. International trade in steel
and cement industry: gravity model, carbon leakage and border tax adjustments.
Forss, K. and Lundström, M. 2004. An evaluation of the program “Export Promotion of Organic
Products from Africa”, Phase II. Final report  Strängnäs 2004.10.15. SIDA. Accessed at:
http://www.grolink.se/epopa/Publications/EPOPA-Phase-2-Evaluation-04.pdf.
Frankel, J. A., Stein, E. and Wei, S. 1995. Trading Blocs and the Americas: The Natural, the
Unnatural, and the Supernatural. Journal of Development Economics, 47(1): 61–95.
147
FreshPlaza. 2014. Kenya: Export rise driven by fruit and vegetables. Accessed at:
http://www.freshplaza.com/article/94677/Kenya-Export-rise-driven-by-fruit-and-
vegetables#SlideFrame_1.
Fugazza, M. and Nicita, A. 2010. The value of preferential market access. UNCTAD: Geneva.
Gibson, P., Wainio, J., Whitley, D.M. and  Bohman,  M. 2001. Profiles of tariffs in Global
Agricultural markets. Agricultural Economic Report 796. USDA.
Gilbert, J. 2010. Understanding and Using Common Indices of International Trade. ESCAP
Training Session, Ulaanbaatar. Department of Economics and Finance, Utah State University.
Giorgio, R. 2004. Empirical Evidence on the North-South Trade Flows: An Augmented Gravity
Model. MPRA Paper No. 136. Available at http://mpra.ub.uni-muenchen.de.
Glick, R. and Rose, A. K. 2001. Does a currency union affect trade? The time series evidence.
NBER Working Paper No. 8396. National Bureau of Economic Research.
Gourieroux, C., Montfort, A. and Trognon, A. 1984. Pseudo-maximum likelihood methods:
Applications to Poisson models. Econometrica, 52: 701 - 720.
Government of Kenya. 2012. National Horticulture Policy. Agricultural Sector Coordination
Unit (ASCU), Nairobi, Kenya.
Accessed at: http://www.kilimo.go.ke/kilimo_docs/pdf/National_Horticulture_Policy.pdf
Granger, C. W. J. and Newbold, P. 1974. Spurious regressions in econometrics. Journal of
Econometrics, 2: 111-120.
Greene, W. H. 1994. Accounting for Excess Zeros and Sample Selection in Poisson and
Negative Binomial Models. Working Paper 94-10, Stern School of Business, New York
University.
GRID 2002. Vital Climate Graphics Africa. Global Resource Information Database, Arendal.
Accessed at: http://www.grida.no/graphicslib/collection/vital-climate-graphics
Griffiths, H.I., Martin, D.S., Shine, A.J. and Evans, J.G. 1993. The ostracod fauna (Crustacea,
Ostracoda) of the profundal benthos of Loch Ness. Hydrobiologia, 254: 111-117.
Grubel, H. G. and Lloyd, P. J. 1971. The Empirical Measurement of Intra-industry Trade. The
Economic Record, 47: 494-517.
Gujarati, D.N. 2003. Basic Econometrics, International Edition. 4th ed.. McGraw-Hill Higher
Education.
Gul, N. and Yasin H. M. 2011. The Trade Potential of Pakistan: An Application of the Gravity
Model. The Lahore Journal of Economics, 16 (1): 23-62.
Hadri, K. 2000. Testing for stationarity in heterogeneous panel data. Econometrics Journal, 3:
148–161.
Hamilton, C.B., and Winters, A. L. 1992. Opening up International Trade with Eastern Europe.
Economic Policy, 14: 77 – 116.
148
Hansen, J.W. 2002. Realizing the potential benefits of climate prediction to agriculture: Issues,
approaches, challenges. Agric. Syst., 74: 309–330.
Harris, R. D. F. and Tzavalis. E. 1999. Inference for unit roots in dynamic panels where the time
dimension is fixed. Journal of Econometrics, 91: 201–226.
Helmers, C. and Pasteals,  J. M. 2003. A gravity model for the calculation of trade potentials for
developing countries and economies in transition. Geneva, Switzerland: International Trade
Centre.
Helpman, E., Melitz, M. and Rubinstein, Y. 2008. Estimating trade flows: trading partners and
trading volumes. Quarterly Journal of Economics, 123: 441- 487.
Hoekman, B., and Nicita, A. 2011. Trade Policy, Trade Costs, and Developing Country Trade.
World Development, 39(12): 2069-2079.
Hope, K. R. 2009. Climate change and poverty in Africa. International Journal of Sustainable
Development & World Ecology, 16(6): 451–461.
ICTSD, 2007. Dialogue on International Trade, Climate Change and Global Competitiveness,
Jointly organized by ICTSD, UNEP and The World Bank. 4 October 2007, WMO Geneva,
Switzerland, 12:30 – 14:30. Meeting Report.
IFOAM, 2005. The World of Organic Agriculture. Statistics and Emerging Trends, 2005. Frick
and Bonn. IFOAM and FiBL
IFOAM and FiBL. 2006. The World of Organic Agriculture. Statistics and Emerging Trends
2006. Frick and Bonn. IFOAM and FiBL
Im, K. S., Pesaran, M. H. and Shin, Y. 2003. Testing for unit roots in heterogeneous panels.
Journal of Econometrics, 115: 53–74.
International Institute for Management Development (IMD) (2009). World Competitiveness
Yearbook 2009 Lausanne. IMD.
IPCC, 2007. Climate Change 2007: Impacts, Adaptation and Vulnerability. Cambridge
University Press.
International Trade Center (ITC) database, 2013. Trade Map. Trade statistics for international
business development. Accessed at: http://legacy.intracen.org/marketanalysis/TradeMap.aspx
ITC, 2003. TradeSim. A Gravity Model for the calculation of Trade Potentials for Developing
Countries and economies in transition, Technical report, UNCTAD – WTO.
Jacob, B., Lars, L., and Enrico, M. 2007. The Dynamics of Criminal Behavior: Evidence from
Weather Shocks. Journal of Human Resources.
Jones, B. F. and Olken, B. A. 2010. Climate Shocks and Exports. American Economic Review
100(2): 454-459.
149
Karagoz, K., and Saray, M. O.  2010. Trade Potential of Turkey with Asia-Pacific Countries:
Evidence from Panel Gravity Model. International Economic Studies, 36(1): 19-26.
Kee, H. L., Ma, H., and Mani, M. 2010. The Effects of Domestic Climate Change Measures on
International Competitiveness. The World Bank, Development Research Group, Trade and
Integration Team & Environment Department.
Keith, H. 2003. Gravity for beginners. Faculty of Commerce, University of British Columbia,
Canada. At available at www.economics.ca/keith/gravity.pdf
Kilembe, C., Thomas, T.S., Waithaka, M., Kyotalimye, M. and Tumbo, S. 2013. Tanzania. In:
M. Waithaka, G. C. Nelson, T. S. Thomas, and M. Kyotalimye, eds. 2013. East African
Agriculture and Climate Change: A Comprehensive Analysis. Washington, DC: International
Food Policy Research Institute. Ch. 11.
Kim, H.S., and Koo, W.W. 2010. Impacts of regulating greenhouse gas emissions on livestock
trade flows. Selected paper prepared for presentation at the Agricultural & Applied Economics
Association 2010 AAEA,CAES, & WAEA Joint Annual Meeting, Denver, Colorado, July 25-27,
2010.
King, G. 1988. Statistical models for political science event counts: Bias in conventional
procedures and evidence for the exponential Poisson regression model. American Journal of
Political Science, 32: 838 - 863.
Kleih, U., Ssango, F., Kyazze, F., Graffham, F. and MacGregor, J. 2007. Impact of EurepGAP
on small-scale fruit and vegetable growers in Uganda. Fresh Insights, Number 10. Accessed at:
http://www.agrifoodstandards.net/en/filemanager/active?fid=103.
Kleynhans, E.P.J. 2003. The Competitive Platform for Industrial Development in South Africa.
Unpublished PhD Thesis. Potchefstroom University for Christian Higher Education, School of
Economics, Risk Management and International Trade.
Korinek, J. and Melatos, M. 2009. Trade Impacts of Selected Regional Trade Agreements in
Agriculture. OECD Trade Policy Working Papers, No. 87, doi: 10.1787/225010121752
Krugman, P. 1994. Competitiveness: A Dangerous Obsession. Council on Foreign Relations,
Inc., Foreign Affairs. Available online at: http://www.foreignaffairs.com/articles/49684/paul-
krugman/competitiveness-a-dangerous-obsession.
Kulapa, S. K., Dawson, P.J. and Lingard J. 2013. The export competitiveness of the tuna
industry in Thailand. British Food Journal, 115(3): 328–341.
Kumar, K.K., Kumar, K.R., Ashrit, R.G., Deshpande, N.R., and Hansen, J.W. 2004. Climate
impacts on Indian agriculture. International Journal of Climatology, 24:1375–93.
Kunimoto, K. 1977. Typology of Trade Intensity Indices. Hitotsubashi Journal of Economics,
11: 15 – 32.
Kurukulasuriya, P., Mendelsohn. R., Hassan, R., Benhin, J., Deressa, T., Diop, Eid, M.H.M.,
Fosu, K. Y., Gbetibouo, G., Jain, S., Mahamadou, A., Mano, R., Kabubo-Mariara, J., El-
Marsafawy, S., Molua, E., Ouda, S., Ouedraogo, M., Se´ne, I., Maddison, D., Seo, S. N., and
150
Dinar, A. 2006. Will African Agriculture Survive Climate Change? The World Bank Economic
Review, 20(3): 367–388.
Kuwornu, J.K.M. and  Mustapha, S. 2013. Global GAP Standard Compliance and Smallholder
Pineapple Farmers’ Access to Export Markets: Implications for Incomes. Journal of Economics
and Behavioral Studies, 5(2): 69-81.
Laibuni, N., Waiyaki, N., Ndirangu, L., and Omiti, J. 2012. Kenyan cut-flower and foliage
exports: A cross country analysis. Journal of Development and Agricultural Economics, 4(2):
37-44.
Latruffe, L. 2010. Competitiveness, Productivity and Efficiency in the Agricultural and Agri-
Food Sectors. OECD Food, Agriculture and Fisheries Papers, No. 30. OECD Publishing.
Laursen, K. 1998. Revealed comparative advantage and the alternatives as measures of
international specialization. DRUID Working Paper, No. 98-30, Danish Research Unit for
Industrial Dynamics (DRUID).
Le, Q., Nguyen, D. and Bandara, J.S. 1996. Vietnam’s Foreign Trade in the Context of ASEAN
and Asia-Pacific Region: A Gravity Approach. ASEAN Economic Bulletin, 13(2): 185-199.
Levin, A., C.-F. Lin, and Chu. C.-S. J. 2002. Unit root tests in panel data: Asymptotic and finite-
sample properties. Journal of Econometrics 108: 1–24.
Lie, L., Liu Z. and Li C. 2002. Spatial correlation analysis of crop yield in the middle and West
of Jilin Province. Chinese Geographical Science, 12(2): 182 – 185.
Linder, S. B. 1961. An Essay on Trade and Transformation. New York: Wiley and Sons.
Linders, G-J. M and De Groot, H.L.F. 2006. Estimation of the gravity equation in the presence
of zero flows. Tinbergen Institute Discussion Paper, TI 2006-072/3.
Linnemann, H. 1966. An Econometric Study of International Trade Flows. Amsterdam: North-
Holland Publishing Company.
Lobell, D., and Field, C. 2007. Global scale climate–crop yield relationships and the impacts of
recent warming. Environ. Res. Lett. 2, 014002.
Lohrmann, A.M. 2000. Development Effects of the Customs Union between Turkey and the
European Union. Catching-Up – Or the Heckscher-Ohlin Trap. Russian and East European
Finance and Trade, 36(4): 26-44.
Long, R. and Sekkat, K. 2004. Economic obstacles to expanding intra-African trade. World
Development, 32(8): 1309 - 1321.
Low, P., Piermartini, R. and Richtering, J. 2005. Multilateral Solutions to the Erosion of Non-
Reciprocal Preferences in NAMA. WTO. Geneva.
Lubinga, M.H., Kiiza, B.A. and Mugisha, J. 2011. Uganda's bilateral trade flows explained
using the gravity flow model. Saarbrücken: VDM Verlag DR Mller AG and CO KG.
151
Lubinga, M.H., Ogundeji, A. and Jordaan, H. 2013. The influence of climate change on
Uganda’s international trade flows: Evidence from weather anomalies. Selected paper prepared
for presentation at the Biennial Conference of the Economics Society of South Africa (ESSA),
University of the Free State, Bloemfontein, South Africa, 25 – 27 September, 2013. Accessed at:
www.essa2013.org.za/fullpaper/essa2013_2714.pdf
Luca, D. B., and Vicarelli, C. 2004. Trade potential in gravity Panel Data models. The B. E.
Journal of Economic Analysis and Policy, 5(1) Article 20. Available at http://
www.bepree.com/bejeap
Ludi, E., Stevens, C., Peskett, L. and Cabral, L. 2007. Climate Change and Agriculture:
Agricultural trade, markets and investments. Overseas Development Institute.
Ma, V, and Keating, M. 2011. Carbon tax to have negative impact on all Australia's trade-
exposed industries: ACCI Director of Economics. Xinhua, 23 March, Available online at:
http://news.xinhuanet.com/english2010/indepth/2011-03/23/c_13794169.htm.
Marimoutou, V., Peguin, D. and Peguin-Fessoille, A. 2009. The ‘distance-varying’ gravity
model in international economics: Is the distance an obstacle to trade? Economics Bulletin,
2(2):1157-1173.
Match Maker Associates (MMA). 2008. Baby vegetables for EU market Sub sector quick scan
Tanzania. Study commissioned by SME competitiveness Facility (SCF).
Mátyás, L. 1997. Proper Econometric Specification of the Gravity Model. The World Economy,
20(3): 363-369.
Marchiori, L., Maystadt, J. F., and Schumacher, I. 2010. Another inconvenient truth: climate
change and migration in sub-Saharan Africa, paper presented at the Trans-nationality of
Migrants (TOM) Marie Curies Research Training Network 'International Migration:
Transnational links, Effects and Policies' hosted by Venice International University, 16-18
September 2010.
Martin, W. and Pham, C. S. 2008. Estimating the Gravity equation when zero trade flows are
frequent.  MPRA Paper 9453. University Library of Munich, Germany.
Martijn, B., Frank, V. O. and Linders, G.J. 2009. On the Specification of the Gravity Model of
Trade: Zeros, Excess Zeros and Zero-inflated Estimation. Spatial Economic Analysis, 4(2): 167-
190.
Martínez-Zarzoso, I., Nowak-Lehmann, D. F., Horsewood, N. 2009. Are regional trading
agreements beneficial? Static and dynamic panel gravity models. North American Journal of
Economics and Finance, 20: 46-65.
McCandless, M., Mbosa, M., Bagyenda, R., Nakangu, B., Nyombi, K., Kutegeka, S., Dion, J.
and Dekens, J. 2012. Impacts of Climate Change on Maize and Beans in Rakai and Kapchorwa
Districts, Uganda. Unpublished document. International Institute for Sustainable Development.
McKibbin, W. J., Shackleton, R. and Wilcoxen, P. J. 1998. The potential effects of international
Carbon permit trading under the Kyoto Protocol. Asia Pacific School of Economics and
Management Working Papers No. 98-1. The Australian National University.
152
Meiliene, E., and Snieska, V. 2010. Lithuanian industrial competitiveness factors export policy
provisions. Public Policy and Administration, 31: 119-132.
Melo, J., and Mathys, N. A. 2010. Trade and Climate Change: The Challenges Ahead. CEPR
Discussion Paper No. DP8032. Accessed at: SSRN: http://ssrn.com/abstract=1711045
Michael E. P., Christian K. and Mercedes, D. 2008. The Microeconomic Foundations of
Prosperity: Findings from the Business Competitiveness Index, The Global Competitiveness
Report 2007-2008. Available online at:
http://siteresources.worldbank.org/EXTEXPCOMNET/Resources/2463593-
1213989126859/03_Porter_GCI_ch1&2.pdf.
Miguel, E., Shanker, S., and Sergenti, E. 2004. Economic Shocks and Civil Conflict: An
Instrumental Variables Approach. Journal of Political Economy, 112(4): 725-753.
Minot, N. and Ngigi, M. 2004. Are horticultural exports a replicable success story? Evidence
from Kenya and Cote D’Ivoire. EPTD Discussion paper No. 120 and MTID Discussion Paper
No. 73. Washington, DC: IFPRI.
Ministry of Water and Environment (MWE). 2007. National Adaptation Pr ogramme of Action
on Climate Change in Uganda. (NAPA). Accessed at:
http://www.preventionweb.net/english/policies/v.php?id=8578&cid=180
Mitchell T. D. and Jones P. 2005. An improved method of constructing a database of monthly
climate observations and associated high resolution grids. International Journal of Climatology,
25: 693 - 712.
Mitchell, T.D., Carter T.R., Jones, P., Hulme M. and New, M. 2004. A comprehensive set of
high resolution grids of monthly climate for Europe and the globe: The observed record (1901–
2000) and 16 scenarios (2001–2100). Tyndall Working Paper 55.
Mitchell, T.D., Hulme, M. and New, M. 2002.. Climate data for political areas. Area, 34: 109–
112.
Mirzaei, S., Yazdani, S., and Mostafavi, S. M. 2006. The dynamics of Iran’s chicken meat
export partens in the Middle East Region. International Journal of Poultry Science, 5(1): 93-95.
Muhammad, A. and Andrews, D. 2008. Determining tourist arrivals in Uganda: The impact of
distance, trade and origin: Specific factors. African Journal of Accounting, Economics, Finance
and Banking Research, 2(2): 51-62.
Muller. C., Cramer, W., Hare, W. L., and Lotze-Campen, H. 2011. Climate change risks for
African agriculture. Proc. Natl. Acad. Sci. USA, 108 (11): 4313–4315.
Munoz-Diaz, D. and Rodrigo, F. S. 2004. Spatio-Temporal Patterns of Seasonal Rainfall in
Spain (1912–2000) Using Cluster and Principal Component Analysis: Comparison, Annales
Geophysicae, 22(5), 1435–1448.
Nakakeeto, G., Norton, G. W.  and Taylor, D. B. 2011. The Impact of Technical Measures on
Agricultural Trade: A Case of Uganda, Senegal, and Mali. ― Improving Food Security through
Agricultural Trade.  Virginia Polytechnic Institute and State University, USA.
153
Namuwoza, C. and Tushemerirwe, H. 2011. Uganda: Country report. Organic Certified
agricultural land and farmers. Accessed at:
http://www.organic-world.net/fileadmin/documents/yearbook/2011/namuwoza-tushmerirwe-
2011-uganda.pdf
Naude, W. A. and Saayman, A. 2005. Determinants of tourist arrivals in Africa: A panel data
regression analysis. Tourism Economics 11(3): 365-391.
Nelson, G.C., Rosegrant, M.W., Palazzo, A., Gray, I., Ingersoll, C., Robertson, R., Tokgoz, S.,
Zhu, T., Sulser, T.B., Ringler, C., Msangi, S. and You, L. 2010. Food Security, Farming, and
Climate Change to 2050: Scenarios, results, policy options. Washington, DC: International Food
Policy Research Institute (IFPRI).
Nelson, G. C., Rosegrant, M. W., Koo, J., Robertson, R., Sulser, T., Zhu, T., Ringler, C.,
Msangi, S., Palazzo, A., Batka, M., Magalhaes, M., Valmonte-Santos, R., Ewing, M. and Lee D.
2009. Climate Change, Agriculture, and Food Security: Impacts and Costs of Adaptation, Food
Policy Report. Washington, DC: IFPRI.
New, M., Hulme M., and Jones P. 2000. Representing twentieth-century space-time climate
variability. Part I: development of a 1901-96 monthly grids of terrestrial surface climate. Journal
of Climate, 13: 2217-38.
New, M., Hulme M., and Jones P. 1999. Representing twentieth-century space-time climate
variability. Part I: development of a 1961-1990 mean  monthly terrestrial climatology. Journal of
Climate, 12: 829-56.
Nicholson, S.E. 1986. The nature of rainfall variability in Africa south of the equator. Journal of
Climatology, 6(5): 515-530.
Nicita, A. 2011. Measuring the relative strength of preferential market access. Policy issues in
International trade and commodities study series No. 47. United Nations. New York and
Geneva,
Nilsson, L. 2002. Trading relations: Is the roadmap from Lomé to Cotonou correct? Applied
Economics, 34: 439-452.
Nilsson, L. 2000. Trade Integration and the EU Economic Membership Criteria. European
Journal of Political Economy, 85: 615–623.
Nilsson, L., and Matsson, N. 2009. Truths and myths about the openness of EU trade policy and
the use of EU trade preferences. Working Paper, DG Trade European Commission.
Nwachuku, I.N., Agwu, N., Nwaru, J. and Imonikhe, G. 2010. Competitiveness and
Determinants of cocoa export from Nigeria. Report and Opinon, 2(7):51- 54.
Nyangweso, P.M. and Odhiambo, M.O. 2004. Exporting Kenya’s Horticultural Products:
Challenges and Opportunities in The 21st Century. A paper presented at the Inaugural
Symposium of the African Association of Agricultural Economists, 6 to 8 December 2004.
Grand Regency Hotel, Nairobi, Kenya
Odera, M.M., Thomas, T.S., Waithaka, M. and Kyotalimye, M. 2013. Kenya. In: M. Waithaka,
G. C. Nelson, T. S. Thomas, and M. Kyotalimye, eds. 2013. East African Agriculture and
154
Climate Change: A Comprehensive Analysis. Washington, DC: International Food Policy
Research Institute. Ch. 7.
Odongo, M. 2007. The Horticulture Sector in Kenya. Accessed at: http://www.cuts-
citee.org/tdp/pdf/Article-The_Horticulture_Sector_in_Kenya.pdf.
Ogang, C. 2013. Challenges of agricultural financing: The case of Uganda. World Farmers
Organisation (WFO): Roma. Accessed at: http://www.wfo-oma.com/women-in-agriculture/case-
studies/challenges-of-agricultural-financing-the-case-of-uganda.html
Oguledo, V. I., MacPhee C. R. 1994. Gravity models: a reformulation and an application to
discriminatory trade arrangements. Applied Economics, 26:107-120.
Palit, A. and Nawani, S. 2012. India-China Trade: Explaining the Imbalance. ISAS Working
Paper No. 95. http://dx.doi.org/10.2139/ssrn.1986543.
Pappalardo, G., Bucca, M. and Zarbà, A. S. 2012. Mediterranean Agricultural Product Trade
between the EU and Morocco in the Perspective of the New Bilateral Agreement: Main Causes
of Imbalance and Possible Effects. International Agricultural Policy, 3: 29-43.
Parry, M.L., Rosenzweig, C., Iglesias, A., Livermore, M., and Fischer, G. 2004. Effects of
climate change on global food production under SRES emissions and socioeconomic scenarios.
Global Environmental Change, 14: 53–67.
Péridy, N. 2005. The trade effects of the Euro-Mediterranean partnership: what are the lessons
for ASEAN countries? Journal of Asian Economics, 16:125-139.
Petriccione G., Crescenzo D. A., Perito, M. A., Solazzo, R., Cioffi, A., and Garcia-Alvarez-
Coque, J. M. 2011. The EU fruit and vegetable sector: Overview and post 2013 CAP
perspective.
Accessed at: http://ageconsearch.umn.edu/bitstream/131062/2/Dell'Aquila_Petriccione.pdf.
Philippidis, G., Resano-Ezcaray, H., and Sanjuán-López, A. I. 2011. Capturing zero trade values
in gravity equations of trade: A disaggregated sectoral analysis. Paper prepared for presentation
at the EAAE 2011 Congress: Change and Uncertainty, Challenges for Agriculture, Food and
Natural Resources. August 30 to September 2, 2011. ETH Zurich, Zurich, Switzerland
Pitts, E. and Lagnevik, M. 1998. What determines food industry competitiveness? In: Traill, B.,
Pitts, E. eds. 1998, Competitiveness in the Food Industry, Blackie Academic and Professional,
London, Chapter 1.
Porter, M.E. 1998. The Competitive Advantage of Nations: With a New Introduction.
Houndmills, Basingstoke: Palgrave.
Porter, M. 1990. The Competitive Advantage of Nations. The Free Press, New York.
Poyhonen, P. 1963. A Tentative Model for the Volume of Tradebetween Countries.
Weltwirtschaftliches Archiv, 90(1): 93-100.
Proenca, I., Fontoura, M. P. and Martìnez-Galàn, E. 2008. Trade in the enlarged European
Union: a new approach on trade potential. Portuguese Economics Journal. 7: 205-224.
155
Quadri, H. M. 2012. Analysis of trade flows among ECO member countries and potential for
free trade area. Melbourne. Victoria University.
Rahman, M. M., 2009. Australia’s Global Trade Potential: Evidence from the Gravity Model
Analysis. University of Southern Queensland.
Raimondi, V., Scoppola, M., and Olper, A. 2011. Preference erosion and the developing
countries exports to the EU: A dynamic panel gravity approach. Selected Paper prepared for
presentation at the Agricultural & Applied Economics Association’s 2011 AAEA & NAREA
Joint Annual Meeting, Pittsburgh, Pennsylvania, July 24-26, 2011.
Rojid, S. 2006. COMESA Trade Potential: A Gravity Approach, Applied Economics Letters, 13:
947-951.
Roudier, P., Sultan, S., Quirion, P., and Berg, A. 2011. The impact of future climate change on
West African crop yields: What does the recent literature say? Global Environ. Change, 21(3):
1073–1083.
Salvatici, L. 2013. The Gravity Model in International Trade. Version 2. AGRODEP Technical
Note TN-04. April 2013. Washington, DC: International Food Policy Research Institute (IFPRI).
Santos Silva, J. M. C. and Tenreyro, S. 2006. The log of gravity. The Review of Economics and
Statistics, 88: 641-58.
Saboniene, A. 2009. Lithuanian Export Competitiveness: Comparison with other Baltic States.
Engineering Economics, 2: 49-57.
Schlenker W. and Lobell, D. 2010. Robust negative impacts of climate change of African
agriculture. Environment Research Letters, 5(1):1-8.
Schlenker W., and Lobell, D. 2010. Robust negative impacts of climate change of African
agriculture. Environment Research Letters, 5.
Sebaggala, R. 2008. Revealed Comparative Advantage and Competitiveness of Uganda's exports
with the Rest of the World. Ndejje University and Uganda Christian University.
Sharples, J. 1990. Cost of production and productivity in analyzing trade and competitiveness.
American Journal of Agricultural Economics, 72(5): 1278-1282.
Shinyekwa, I. and Othieno, L. 2011. Uganda’s Revealed Comparative Advantage: The Evidence
with the EAC and China. EPRC, Makerere University, Kampala
Siliverstovs, B. and Schumacher, D. 2009. Estimating gravity equations: To log or not to log?
Empirical Economics, 36: 645-669.
Simonett, O. 1989. Impact of Temperature Rise on Robusta Coffee in Uganda. Case Studies on
Climatic Change. Geneva, Switzerland: GRID-Geneva. Retrieved 28 May 2013 from
http://maps.grida.no/go/graphic/impact-of-temperature-rise-on-robusta-coffee-in-uganda
Sivakumar M.V.K., Das, H.P., and Brunini, O. 2005. Impacts of present and future climate
variability and change on agriculture and forestry in the arid and semi-arid tropics. Climatic
Change, 70: 31–72.
156
StataCorp, 2013. Panel-data unit-root tests. Accessed online at:
http://www.stata.com/stata11/xtur.html
Tang, D. 2005. Effects of the Regional Trading Arrangements on Trade: Evidence from the
NAFTA, ANZCER and ASEAN Countries, 1989-2000. The Journal of International Trade and
Economic Development, 14(2): 242 - 265.
Tao, F., Hayashi, Y., Zhang, Z., Sakamoto, T., and Yokozawa, M. 2008. Global warming, rice
production and water use in China: developing a probabilistic assessment. Agricultural and
Forest Meteorology, 148(1): 94–110.
Tao, F., Yokozawa, M., Hayashi, Y. and Lin, E. 2003. Future climate change, the agricultural
water cycle, and agricultural production in China. Agriculture, Ecosystems & Environment, 95:
203–15.
Teweldemedhin, M. Y.  and Van Schalkwyk, H. D. 2010. Regional trade agreements and its
impact on trade flows for South African agricultural products. Journal of Development and
Agricultural Economics, 2(5): 157-165.
Timmers, B. 2012. Impacts of climate change and variability on fish value chains in Uganda.
Project Report 2012-18. The WorldFish Center, Penang, Malaysia.
Tinbergen, J. 1962. Shaping the world economy: Suggestions for an International Economic
Policy. The Twentieth Century Fund.  New York
Török Á. and Jámbor A. 2013. Competitiveness and Geographical Indications: The case of fruit
spirits in Central and Eastern European countries. Studies in Agricultural Economics 115: 25-32.
Traore, B., Corbeel, M., van Wijk, M.T., Rufino, M.C., and Giller, K. E. 2013. Effects of climate
variability and climate change on crop production in southern Mali. Europ. J. Agronomy, 49:
115– 125.
TYN CY 1.11 database, 2013. Climate Data. Tyndall Centre  for Climate Change Research.
Accessed at: http://www.cru.uea.ac.uk/~timm/data/index.html
UIA, (2001). Fruits and Vegetables sector profile.
United Nations. 2009. World Economic and Social Survey 2009: Promoting Development,
Saving the Planet. New York: United Nations.
UNCTAD. 2008. Private-sector standards and national schemes for Good Agricultural
Practices: Implications for exports of fresh fruit and vegetables from Sub-Saharan Africa.
Experiences of Ghana, Kenya and Uganda. New York and Geneva: United Nations.
UNCTAD. 2001. Improving market access for Least developed countries.
UNCTAD/DITC/TNCD/4. 2 May 2001.
UNCTAD. 1999. Quantifying the benefits obtained by developing countries from the
generalized system of preferences.
Accessed at: http://unctad.org/en/docs/poitcdtsbm52.en.pdf
157
UNCTAD. 1968. Proceedings of the United Nations Conference on Trade and Development,
Second Session, vol. 1 and 3 and Add.l and 2], Report and Annexes (United Nations publication,
Sales No. E.68.II.D.14), p. 38.
UNDP and BCPR. 2013. Climate Risk Management for Sustainable Crop Production in
Uganda: Rakai and Kapchorwa Districts. New York, NY: UNDP BCPR.
UNFCCC. 2009. Kyoto Protocol: Status of Ratification. United Nations Framework Convention
on Climate Change. Available at
http://unfccc.int/files/kyoto_protocol/status_of_ratification/application/pdf/kp_ratification_2009
1203.pdf. Accessed on 29 August 2013.
USAID. 2007. Export markets for high-value vegetables from Tanzania. An AMAP BDS K&P
task order study. Accessed at:
http://www.microlinks.org/sites/microlinks/files/resource/files/export_markets_for_high_value_
vegetables_from_Tanzania.pdf.
Utkulu, U. and Dilek, S. 2004. Revealed Comparative Advantage and Competitiveness:
Evidence for Turkey vis-à-vis the EU/15. A paper prepared for the European Trade Group 6th
Annual Conference, ETSG 2004, Nottingham.
Vollrath, T. L. 1987. Revealed competitive advantage for wheat. Economic Research Service
Staff Report No. AGES861030. United States Department of Agriculture, Washington, D.C.
Vollrath, T. L. 1989. Competitiveness and protection in World Agriculture. Agricultural
Information Bulletin No. 567, Economic Research Service. United States Department of
Agriculture, Washington, D.C.
Vollrath, T. L. 1991. A theoretical evaluation of alternative Trade intensity measures of revealed
comparative advantage. Weltwirtschaftliches Archiv, 127: 265 – 280.
Walker, S., Chen, L., and Madden, R. 2008. Deaths due to suicide: The effects of certification
and coding practices in Australia. Australian and New Zealand Journal of Public Health, 32(2),
126-130.
Wall, H.J. 2000. Gravity Model Specification and the Effect of the Canada-U.S. Border.
Working Paper No. 2000-024A , Federal Reserve Bank of St. Louis.
Wang, Z., and Winters, A.L. 1992. The Trading Potential of Eastern Europe, Journal of
Economic Integration, 7: 113-136.
WBDI, 2012. World Development Indicators. World Bank.
Wikipedia. 2014. Languages of Belgium. Accessed at
http://en.wikipedia.org/wiki/Languages_of_Belgium.
Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data. Boston, MIT
Press.
World Bank, 2010. World Development Report 2010: Development and Climate Change.
Washington, DC: World Bank.
158
World Bank, 2008. International Trade and Climate Change: Economic, Legal, and Institutional
Perspectives. The International Bank for Reconstruction and Development. Washington, DC:
World Bank.
Xiong, W., Matthews, R., Holman, I., Lin, E., and Xu, Y. 2007. Modelling China’s potential
maize production at regional scale under climate change. Climatic Change, 85: 433–51.
Yılmaz, B. 2002. Turkey’s Competitiveness in the European Union. A Comparison of Greece,
Portugal, Spain and the EU/12/15. Russian and East European Finance and Trade, 38(3): 54-72.
159
APPENDICES
Appendix A: The EU-15 member states considered under this study
No. Country Year joined EU
1 Belgium 1957
2 France 1957
3 Germany 1957
4 Italy 1957
5 Luxembourg 1957
6 Netherlands 1957
7 Denmark 1973
8 Ireland 1973
9 United Kingdom 1973
10 Greece 1981
11 Portugal 1986
12 Spain 1986
13 Austria 1995
14 Finland 1995
15 Sweden 1995
160
Appendix B: Multi-collinearity test results for the three East African states
(Objective 2: The influence of climate change on East Africa's horticultural trade flows)
1. KENYA
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
DTjt |     27.60    0.036231
Incomijt |     19.01    0.052600
Dlangij |     12.45    0.080296
lnY2jt |     11.89    0.084104
lnAgrijt |      8.68    0.115233
m9 |      8.18    0.122300
m8 |      5.07    0.197310
m5 |      4.62    0.216267
lnDij |      4.53    0.220892
DTit |      4.28    0.233534
m6 |      4.04    0.247460
lnY1it|      3.92    0.255029
lninflatit |      3.48    0.287631
m4 |      3.11    0.321922
lnAgriit |      2.79    0.359017
m13 |      1.99    0.501604
m3 |      1.98    0.505480
m11 |      1.80    0.555235
m1 |      1.69    0.591275
Prejt |      1.33    0.753568
Preit |      1.31    0.762127
-------------+----------------------
Mean VIF |      6.37
NB: Variables m1-m13 within the annexure denote importer fixed effects
161
ii) Pearson's correlation matrix
|   X            lnY1it lnY2jt lnDij DTit Preit DTjt
-------------+---------------------------------------------------------------
X |   1.0000
lnY1it |   0.2588   1.0000
lnY2jt |   0.3992   0.0826   1.0000
lnDij |   0.1360 -0.0000 -0.0717   1.0000
DTit |   0.0664   0.3271   0.0404 -0.0000   1.0000
Preit | -0.1034 -0.3381 -0.0760   0.0000 -0.3461   1.0000
DTjt |   0.0448   0.0141   0.1398 -0.6416 -0.0312   0.0199   1.0000
Prejt | 0.1568   0.1726 -0.0275   0.3564   0.0627 -0.0955 -0.2110
lnAgriit |   0.1954   0.4803   0.1123   0.0000   0.0823 -0.5084   0.0034
lnAgrijt | -0.0797 -0.0146 -0.5532   0.0371   0.0134   0.0349 -0.1041
Dlangij |   0.2769   0.0000 -0.3475   0.2931 -0.0000   0.0000   0.0802
lninflatit | -0.1676 -0.5891 -0.0600 -0.0000 -0.5125   0.2270 -0.0126
Incomijt |   0.1455   0.0899 -0.0711   0.4382 -0.0031 -0.0552 -0.6512
m1 | -0.0853 -0.0000 -0.0621 -0.1779 -0.0000 -0.0000 -0.2121
m3 | -0.0820 -0.0000 -0.1080   0.1578 -0.0000 -0.0000 -0.2189
m4 | -0.0861 -0.0000 -0.1582   0.2327 -0.0000 -0.0000 -0.4477
m5 |   0.1803 -0.0000   0.3434 -0.0961 -0.0000 -0.0000   0.0997
m6 |   0.1030 -0.0000   0.4241   0.0564 -0.0000 -0.0000 -0.1389
m8 | -0.0764 -0.0000 -0.2958 0.2327 -0.0000 -0.0000   0.1636
m9 | -0.0834 -0.0000   0.3078 -0.3779 -0.0000 -0.0000   0.4469
m11 |   0.0437 -0.0000   0.0621   0.1442 -0.0000 -0.0000 -0.0790
m13 | -0.0870 -0.0000   0.1565 -0.0391 -0.0000 -0.0000   0.1730
|   Prejt lnAgriit lnAgrijt Dlangij lninflatit Incomijt m1
-------------+---------------------------------------------------------------
Prejt |   1.0000
lnAgriit |   0.2130   1.0000
lnAgrijt | -0.0398 -0.0166   1.0000
Dlangij |   0.1786 -0.0000   0.5144   1.0000
lninflatit | -0.1323 -0.1415   0.0020 -0.0000   1.0000
Incomijt |   0.1877   0.0653   0.4730   0.3023 -0.0662   1.0000
m1 | -0.0394   0.0000 -0.0722 -0.1336   0.0000   0.0819   1.0000
m3 |   0.0407 -0.0000 -0.0643 -0.1336 -0.0000   0.2669 -0.0714
m4 |   0.0835   0.0000 -0.0701 -0.1336 -0.0000   0.0488 -0.0714
m5 | -0.0261   0.0000 -0.0666 -0.1336 -0.0000 0.0500 -0.0714
m6 | -0.0414   0.0000 -0.0714 -0.1336 -0.0000   0.0996 -0.0714
m8 |   0.1682   0.0000 -0.0702   0.5345 -0.0000 -0.1352 -0.0714
m9 | -0.1425   0.0000 -0.0688 -0.1336 -0.0000 -0.0357 -0.0714
m11 |   0.0478   0.0000 -0.0710 -0.1336 -0.0000   0.0684 -0.0714
m13 | -0.0921   0.0000 -0.0752 -0.1336 -0.0000 -0.3357 -0.0714
162
|       m3       m4       m5       m6       m8       m9      m11
-------------+---------------------------------------------------------------
m3 |   1.0000
m4 | -0.0714   1.0000
m5 | -0.0714 -0.0714   1.0000
m6 | -0.0714 -0.0714 -0.0714   1.0000
m8 | -0.0714 -0.0714 -0.0714 -0.0714   1.0000
m9 | -0.0714 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
m11 | -0.0714 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
m13 | -0.0714 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714
|      m13
-------------+---------
m13 |   1.0000
2. TANZANIA
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lninflatit |     50.19    0.019923
lnY1it |     49.66 0.020135
Incomijt |      6.47    0.154591
lnAgriit |      6.42    0.155884
Preit |      5.95    0.168102
lnAgrijt |      4.05    0.247175
lnDij |      3.36    0.297366
m12 |      3.34    0.298967
DTjt |      2.97    0.336622
DTit |      2.44    0.410151
Dlangij |      2.42    0.414011
lnY2jt |      2.31    0.433775
m13 |      1.96    0.510892
m4 |      1.45    0.687730
m6 |      1.41    0.707788
Prejt |      1.31    0.763305
m11 |      1.18    0.850502
-------------+----------------------
Mean VIF |      8.64
163
ii) Pearson's correlation matrix
|   X            lnY1it lnY2jt lnDij DTit DTjt Preit
-------------+---------------------------------------------------------------
X |   1.0000
lnY1it |   0.0100   1.0000
lnY2jt |   0.3230 -0.0229   1.0000
lnDij |   0.0627   0.0000 -0.0472 1.0000
DTit | -0.0359 -0.4137   0.0615 -0.0000   1.0000
DTjt |   0.0586 -0.2212   0.1144   0.0894 -0.1205   1.0000
Preit |   0.0105   0.4720   0.0084   0.0000   0.1220 -0.0651   1.0000
Prejt |   0.0067 -0.0494 -0.0275   0.3594   0.1791   0.2186 -0.0336
lnAgriit |   0.1377 -0.0502   0.0552 -0.0000   0.3011   0.3034   0.1958
lnAgrijt | -0.0698 -0.0418 -0.5509   0.0216 -0.0000 -0.0446 -0.0011
Dlangij | -0.1326   0.0000 -0.3475   0.3299   0.0000 -0.0068   0.0000
m4 | -0.0719   0.0000 -0.1582   0.2014 -0.0000 -0.1040   0.0000
m6 |   0.6885   0.0000   0.4241   0.0373 -0.0000   0.0139   0.0000
m11 |   0.2318   0.0000   0.0621   0.1240 -0.0000 -0.0134   0.0000
m12 | -0.0608   0.0000 -0.1999   0.0375 -0.0000   0.1013   0.0000
m13 | -0.0720   0.0000   0.1565 -0.0625 -0.0000   0.1360   0.0000
lninflatit |   0.0195 -0.8141 -0.0174 -0.0000 -0.0270   0.1943 -0.4400
Incomijt |   0.0994   0.0124 -0.0724   0.4239   0.0231 -0.0553   0.0305
|    Prejt lnAgriit lnAgrijt Dlangij m4       m6      m11
-------------+---------------------------------------------------------------
Prejt |   1.0000
lnAgriit |   0.2573   1.0000
lnAgrijt | -0.0319 -0.0069   1.0000
Dlangij |   0.1814 -0.0000   0.5188   1.0000
m4 |   0.0850 -0.0000 -0.0741 -0.1336   1.0000
m6 | -0.0398 -0.0000 -0.0783 -0.1336 -0.0714   1.0000
m11 |   0.0493 -0.0000 -0.0768 -0.1336 -0.0714 -0.0714   1.0000
m12 | -0.0038 -0.0000 -0.0614 -0.1336 -0.0714 -0.0714 -0.0714
m13 | -0.0904 -0.0000 -0.0910 -0.1336 -0.0714 -0.0714 -0.0714
lninflatit |   0.0636 -0.0626   0.0258   0.0000 -0.0000 -0.0000 -0.0000
Incomijt |   0.1845   0.0861   0.4771   0.3028   0.0484   0.0994   0.0682
|      m12      m13 lninflatit Incomijt
-------------+------------------------------------
m12 |   1.0000
m13 | -0.0714   1.0000
lninflatit | -0.0000 -0.0000   1.0000
Incomijt | -0.5298 -0.3358 -0.0501   1.0000
164
3. UGANDA
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lninfrait |     58.19    0.017186
lnAgriit |     42.17    0.023716
lnY1it |     20.25    0.049392
Preit |    16.97    0.005077
DTit |      6.34    0.157767
lninflatit |      3.68    0.271697
lnAgrijt |      3.09    0.324018
Incomijt | 2.64    0.379013
DTjt |      2.43    0.412317
lnY2jt |      2.29    0.436104
lnDij |      1.70    0.587806
Dlangij |      1.63    0.615223
m6 |      1.34 0.746740
m3 |      1.29    0.776777
Prejt |      1.28    0.778791
-------------+----------------------
Mean VIF |     11.01
ii) Pearson's correlation matrix
| X            lnY1it lnY2jt lnDij lninflatit lnAgriit lnAgrijt
-------------+---------------------------------------------------------------
X |   1.0000
lnY1it |   0.0367   1.0000
lnY2jt |   0.1986   0.0283   1.0000
lnDij |   0.0821   0.0000 -0.0524   1.0000
lninflatit |   0.0282 -0.0628 -0.0738   0.0000   1.0000
lnAgriit |   0.1935   0.1711   0.0335   0.0000 -0.3435   1.0000
lnAgrijt | -0.0264 -0.0016 -0.5502   0.0126 -0.0006 -0.0258   1.0000
DTit | -0.0155   0.4857   0.0609 -0.0000 -0.0451   0.0230 -0.0119
DTjt |   0.0625   0.1610   0.1144   0.0647   0.1074   0.3884 -0.0430
Preit | -0.0958 -0.1540 -0.0916   0.0000   0.6394 -0.5908   0.0241
Prejt |   0.2219   0.2028 -0.0275   0.3684 -0.1680   0.2765 -0.0302
lninfrait |   0.0037   0.0140 -0.0396 -0.0000   0.0989 -0.0830   0.0467
Dlangij |   0.2718 -0.0000 -0.3475   0.2836   0.0000   0.0000   0.5192
m3 | -0.0379 -0.0000 -0.1080   0.1372   0.0000   0.0000 -0.0528
m6 | -0.0038 -0.0000   0.4241   0.0267   0.0000   0.0000 -0.0793
Incomijt |   0.0808   0.1014 -0.0734   0.4614 -0.0227   0.0794   0.4785
|      DTit DTjt Preit Prejt lninfrait Dlangij m3
-------------+---------------------------------------------------------------
DTit |   1.0000
DTjt |   0.0283   1.0000
Preit | -0.3924 -0.1608   1.0000
Prejt |   0.1255   0.2186 -0.1292   1.0000
lninfrait | -0.4029   0.0646   0.5162 -0.0532   1.0000
Dlangij | -0.0000 -0.0068   0.0000   0.1814   0.0000   1.0000
165
m3 |   0.0000 -0.0435 -0.0000 0.0422   0.0000 -0.1336   1.0000
m6 |   0.0000   0.0139 -0.0000 -0.0398   0.0000 -0.1336 -0.0714
Incomijt |   0.0520 -0.0568 -0.0775   0.1837 -0.0051   0.3032   0.2675
| m6   Incomijt
-------------+------------------
m6 |   1.0000
Incomijt |   0.0993   1.0000
166
Appendix C: Normality test results for objective two (The influence of climate change
on East Africa's horticultural trade flows)
i)  KENYA
0 20000 40000 60000 80000
Exports '000USD (CH. 07 & 08)
ii) TANZANIA
0 2000 4000 6000
Exports of Tanzania ('000USD) (Ch. 07 & 08)
iii)  UGANDA
0 500 1000 1500 2000 2500
Exports of Uganda ('000USD) (ch. 07 & 08)
167
Frequency Frequency Frequency
0 50 100 150 200 0 50 100 150 200 0 50 100 150 200
Appendix D: Multi-collinearity test results for Kenya's Asparagus- 070920 (Objective
three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
------------+----------------------
lnCOSTBIZit |      4.33    0.230803
lnY1it |      3.70    0.270335
lninflatit |      3.43    0.291686
lnPMijlt |      2.26    0.442550
Dlangij |      1.76    0.566968
m8 | 1.43    0.701157
lnY1jt |      1.38    0.724584
lnDij |      1.32    0.760379
m6 |      1.27    0.788595
m3 |      1.19    0.843215
m11 |      1.14    0.874947
m2 |      1.13    0.886589
m1 |      1.10    0.910023
-------------+----------------------
Mean VIF |      1.96
ii) Pearson's correlation matrix
|        Mijkt lnY1it lnY2jt lnDij lnPMijlt lninflatit lnCOSTBIZit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it|   0.1254   1.0000
lnY2jt |   0.2921   0.0735   1.0000
lnDij |   0.1723 -0.0000 -0.0823   1.0000
lnPMijlt |   0.0527   0.7338   0.0263   0.0097   1.0000
lninflatit |   0.0511 -0.0959   0.0206 -0.0000   0.0469   1.0000
lnCOSTBIZit | -0.1185 -0.4780 -0.0595 -0.0000 -0.4232 -0.6858   1.0000
Dlangij |   0.4404   0.0000 -0.2699   0.2931 -0.0192   0.0000   0.0000
m1 | -0.0659 -0.0000 -0.1087 -0.1779 -0.0102   0.0000   0.0000
m2 | -0.0455 -0.0000 -0.0480   0.1043 -0.0102   0.0000   0.0000
m3 | -0.0647 -0.0000 -0.1410   0.1578 -0.0102   0.0000   0.0000
m6 | -0.0663 -0.0000   0.4101   0.0564 -0.0102   0.0000   0.0000
m8 | -0.0538 -0.0000 -0.2099   0.2327 -0.0102   0.0000   0.0000
m11 | -0.0303 -0.0000   0.0731   0.1442 -0.0102   0.0000   0.0000
| Dlangij m1       m2       m3       m6       m8      m11
-------------+---------------------------------------------------------------
Dlangij |   1.0000
m1 | -0.1336   1.0000
m2 | -0.1336 -0.0714   1.0000
m3 | -0.1336 -0.0714 -0.0714   1.0000
m6 | -0.1336 -0.0714 -0.0714 -0.0714   1.0000
m8 |   0.5345 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
168
m11 | -0.1336 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
Appendix E: Multi-collinearity test results for Kenya's Beans- 070820 (Objective three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnCOSTBIZit|      4.34    0.230623
lnY1it |      4.19    0.238787
lninflatit |      3.45    0.289870
lnPMijlt |      2.70    0.369715
Dlangij |      1.81    0.552289
lnY2jt |      1.71    0.583140
lnDij |      1.58    0.634555
m4 |      1.44    0.692105
m6 |      1.40 0.712746
m3 |      1.33    0.754378
m5 |      1.29    0.776327
m11 |      1.26    0.791668
m2 |      1.24    0.806038
m1 |      1.14    0.873757
-------------+----------------------
Mean VIF |      2.06
169
ii)  Pearson's correlation matrix
|        Mijkt lnY1it lnY2jt lnDij lnPMijlt lninflatit lnCOSTBIZit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it |   0.0370   1.0000
lnY2jt |   0.4665   0.0735   1.0000
lnDij |   0.2274 -0.0000 -0.0823   1.0000
lnPMijlt | -0.0415 -0.7833 -0.0602 -0.0000   1.0000
lninflatit |   0.0248 -0.0959   0.0206 -0.0000 -0.0500   1.0000
lnCOSTBIZit | -0.0412 -0.4780 -0.0595 -0.0000   0.4517 -0.6858   1.0000
Dlangij |   0.4488   0.0000 -0.2699 0.2931   0.0000   0.0000   0.0000
m1 | -0.1162 -0.0000 -0.1087 -0.1779 -0.0000   0.0000   0.0000
m2 | -0.0178 -0.0000 -0.0480   0.1043 -0.0000   0.0000   0.0000
m3 | -0.1178 -0.0000 -0.1410   0.1578 -0.0000   0.0000   0.0000
m4 | -0.1169 -0.0000 -0.2012   0.2327 -0.0000   0.0000   0.0000
m5 |   0.1056 -0.0000   0.3517 -0.0961 -0.0000   0.0000 -0.0000
m6 | -0.0038 -0.0000   0.4101   0.0564 -0.0000   0.0000   0.0000
m11 |   0.1348 -0.0000   0.0731   0.1442 -0.0000   0.0000   0.0000
|    Dlangij m1       m2       m3       m4       m5       m6
-------------+---------------------------------------------------------------
Dlangij |   1.0000
m1 | -0.1336   1.0000
m2 | -0.1336 -0.0714   1.0000
m3 | -0.1336 -0.0714 -0.0714   1.0000
m4 | -0.1336 -0.0714 -0.0714 -0.0714   1.0000
m5 | -0.1336 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
m6 | -0.1336 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714   1.0000
m11 | -0.1336 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714 -0.0714
|      m11
---------+---------
m11 |   1.0000
170
Appendix F: Multi-collinearity test results for Tanzania's Beans- 070820 (Objective
three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnY1it |      4.82    0.207394
lnGOVit |      4.07    0.245464
lnCOSTEXPit |      2.15    0.466068
lnPMijlt |      1.74    0.573827
Dlangij |      1.25    0.801374
lnDij |      1.16    0.860909
lnY2jt |      1.09    0.918319
m11 |      1.05    0.949535
-------------+----------------------
Mean VIF |      2.17
ii)  Pearson's correlation matrix
|        Mijkt X    lnY1it lnY2jt lnDij lnPMijlt lnCOSTEXPit lnGOVit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it | -0.0682   1.0000
lnY2jt |   0.3202   0.0653   1.0000
lnDij |   0.2041   0.0000 -0.0460   1.0000
lnPMijlt |   0.0758 -0.5705 -0.0602 -0.0000   1.0000
lnCOSTEXPit |   0.0223 -0.6174 -0.0155 -0.0000   0.1037   1.0000
lnGOVit |   0.0668 -0.8445 -0.0399   0.0000   0.4036   0.6805   1.0000
Dlangij |   0.4081 -0.0000 -0.2699   0.3299   0.0000   0.0000 -0.0000
m11 |   0.1565   0.0000   0.0731   0.1240 -0.0000   0.0000   0.0000
|    Dlangij m11
-------------+------------------
Dlangij |   1.0000
m11 | -0.1336   1.0000
171
Appendix G: Multi-colinearity test results for Tanzania's Vegetables- 070990 (Objective
three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnY1it |      4.82    0.207394
lnGOVit |      4.07    0.245464
lnCOSTEXPit |      2.15    0.466068
lnPMijlt |      1.74    0.573827
Dlangij |      1.25    0.801374
lnDij |      1.16    0.860909
lnY2jt |      1.09    0.918319
m11 |      1.05    0.949535
-------------+----------------------
Mean VIF |      2.17
ii) Pearson's correlation matrix
|        Mijkt lnY1it lnY2jt lnDij lnPMijlt lnCOSTEXPit lnGOVit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it | -0.1422   1.0000
lnY2jt |   0.1868   0.0653   1.0000
lnDij |   0.1008   0.0000 -0.0460   1.0000
lnPMijlt | -0.1070 -0.2250   0.0185 -0.0000   1.0000
lnCOSTEXPit |   0.0126 -0.6174 -0.0155 -0.0000   0.7801   1.0000
lnGOVit |   0.1064 -0.8445 -0.0399   0.0000   0.3651   0.6805   1.0000
Dlangij |   0.2961 -0.0000 -0.2699   0.3299   0.0000   0.0000 -0.0000
m11 | -0.0367   0.0000   0.0731   0.1240   0.0000   0.0000   0.0000
|    Dlangij m11
-------------+------------------
Dlangij |   1.0000
m11 | -0.1336   1.0000
172
Appendix H: Multi-colinearity test results for Uganda's Beans- 070820 (Objective three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnY1it |     16.21    0.061683
lnCOSTEXPit |      6.32    0.158350
lnGOVit |      5.64    0.177156
lnPMijlt |      1.98    0.505827
Dlangij |      1.19    0.841795
lnDij |      1.11    0.898619
lnY2jt |      1.11    0.899153
lninflatit |      1.07    0.931004
m1 |      1.07    0.937790
-------------+----------------------
Mean VIF |      3.97
ii)  Pearson's correlation matrix
|     Mijkt lnY1it lnY2jt lnDij lnPMijlt lninflatit lnCOSTEXPit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it |   0.0894   1.0000
lnY2jt |   0.1985   0.0644   1.0000
lnDij |   0.1399 -0.0000 -0.0617   1.0000
lnPMijlt | -0.0392 -0.6306 -0.0602   0.0000   1.0000
lninflatit |   0.0874   0.2201   0.0290   0.0000 -0.0442   1.0000
lnCOSTEXPit |   0.0832   0.8593   0.0749   0.0000 -0.6430   0.2091 1.0000
lnGOVit | -0.0822 -0.8253 -0.0284 -0.0000   0.3536 -0.1904 -0.5286
Dlangij |   0.2450   0.0000 -0.2699   0.2836   0.0000   0.0000 -0.0000
m1 | -0.0889   0.0000 -0.1087 -0.1811 -0.0000   0.0000   0.0000
|    lnGOVit Dlangij m1
-------------+---------------------------
lnGOVit |   1.0000
Dlangij |   0.0000   1.0000
m1 |   0.0000 -0.1336   1.0000
173
Appendix I: Multi-colinearity test results for Uganda's Peppers- 070960 (Objective three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnGOVit |      6.06    0.165018
lnY1it |      3.45    0.289870
lnCOSTBIZit |      3.34    0.299083
lnPMijlt |      2.14    0.466580
lninflatit |      2.00    0.499633
Dlangij |      1.32    0.756695
lnY2jt |      1.29    0.777170
lnDij |      1.20    0.834398
m5 |      1.18    0.845410
m3 |      1.13 0.886178
m11 |      1.09    0.920564
m1 |      1.09    0.921616
-------------+----------------------
Mean VIF |      2.11
ii)  Pearson's correlation matrix
|        Mijkt lnY1it lnY2jt lnDij lnPMijlt lninflatit lnCOSTBIZit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it | 0.1294   1.0000
lnY2jt |   0.3735   0.0644   1.0000
lnDij |   0.1854 -0.0000 -0.0617   1.0000
lnPMijlt |   0.0594   0.4117   0.0287 -0.0000   1.0000
lninflatit |   0.0331   0.2201 0.0290   0.0000   0.6307   1.0000
lnCOSTBIZit | -0.0593 -0.5571 -0.0286   0.0000 -0.5822 -0.4580   1.0000
lnGOVit | -0.0819 -0.8253 -0.0284 -0.0000 -0.4714 -0.1904   0.7550
Dlangij |   0.3968   0.0000 -0.2699   0.2836   0.0000   0.0000 -0.0000
m3 | -0.1054   0.0000 -0.1410   0.1372   0.0000   0.0000   0.0000
m5 | -0.0345   0.0000   0.3517   0.0569   0.0000   0.0000 -0.0000
m11 |   0.1984   0.0000   0.0731   0.1090   0.0000   0.0000   0.0000
m1 | -0.1063   0.0000 -0.1087 -0.1811   0.0000   0.0000 -0.0000
|    lnGOVit Dlangij m3       m5      m11       m1
-------------+------------------------------------------------------
lnGOVit |   1.0000
Dlangij |   0.0000   1.0000
m3 |   0.0000 -0.1336   1.0000
m5 |   0.0000 -0.1336 -0.0714   1.0000
m11 |   0.0000 -0.1336 -0.0714 -0.0714   1.0000
174
m1 |   0.0000 -0.1336 -0.0714 -0.0714 -0.0714   1.0000
Appendix J: Multi-colinearity test results for Uganda's Bananas- 080300 (Objective
three)
i) VIF and TOL test results
Variable |       VIF       1/VIF
-------------+----------------------
lnY1it |     51.22    0.019523
lnCOSTEXPit |     21.90    0.045666
lnPMijlt |     13.35    0.074911
lnFDIit |      8.02    0.124642
lnGOVit |      6.14    0.162844
lninflatit |      2.96    0.338068
Dlangij |      1.79    0.557569
lnDij |      1.65    0.605607
m14 |      1.52    0.659808
lnY2jt |      1.37    0.732321
m5 |      1.31    0.762590
m3 |      1.30    0.766558
m11 |      1.24    0.807625
m2 |      1.22    0.822081
m13 |      1.17    0.857442
m1 |      1.14    0.875108
-------------+----------------------
Mean VIF |      7.33
ii) Pearson's correlation matrix
|        Mijkt lnY1it lnY2jt lnDij lnPMijlt lninflatit lnCOSTEXPit
-------------+---------------------------------------------------------------
Mijkt |   1.0000
lnY1it |   0.0812   1.0000
lnY2jt |   0.3285   0.0644   1.0000
lnDij |   0.1648 -0.0000 -0.0617   1.0000
lnPMijlt | -0.0778 -0.8409 -0.0565 -0.0000   1.0000
lninflatit |   0.0293   0.2201   0.0290   0.0000 -0.0309   1.0000
lnCOSTEXPit |   0.1034   0.8593   0.0749   0.0000 -0.6538   0.2091   1.0000
lnGOVit | -0.0359 -0.8253 -0.0284 -0.0000   0.6863 -0.1904 -0.5286
lnFDIit |   0.1060   0.5408   0.0638   0.0000 -0.5986   0.4356   0.7038
Dlangij |   0.4508   0.0000 -0.2699   0.2836 -0.0000   0.0000 -0.0000
m1 | -0.0835   0.0000 -0.1087 -0.1811 -0.0000   0.0000   0.0000
m2 |   0.1410   0.0000 -0.0480   0.0681 -0.0000   0.0000   0.0000
175
m3 | -0.0742   0.0000 -0.1410   0.1372 -0.0000   0.0000   0.0000
m5 | -0.0844   0.0000   0.3517   0.0569 -0.0000   0.0000   0.0000
m11 | -0.0734   0.0000   0.0731   0.1090 -0.0000   0.0000   0.0000
m13 | -0.0879   0.0000   0.2109 -0.0968 -0.0000   0.0000   0.0000
m14 | -0.0733 -0.0000 -0.0538   0.3376 -0.0000   0.0000   0.0000
|    lnGOVit lnFDIit Dlangij m1       m2       m3       m5
-------------+---------------------------------------------------------------
lnGOVit |   1.0000
lnFDIit | -0.2890   1.0000
Dlangij |   0.0000   0.0000   1.0000
m1 |   0.0000   0.0000 -0.1336   1.0000
m2 |   0.0000   0.0000 -0.1336 -0.0714   1.0000
m3 |   0.0000   0.0000 -0.1336 -0.0714 -0.0714   1.0000
m5 |   0.0000   0.0000 -0.1336 -0.0714 -0.0714 -0.0714   1.0000
m11 |   0.0000   0.0000 -0.1336 -0.0714 -0.0714 -0.0714 -0.0714
m13 |   0.0000   0.0000 -0.1336 -0.0714 -0.0714 -0.0714 -0.0714
m14 |   0.0000   0.0000 -0.1336 -0.0714 -0.0714 -0.0714 -0.0714
|      m11      m13 m14
-------------+---------------------------
m11 |   1.0000
m13 | -0.0714   1.0000
m14 | -0.0714 -0.0714   1.0000
176
Appendix K: Kenya's normality test results for Asparagus and Beans
Asparagus Beans
0 200 400 600 0 20000 40000 60000 80000 100000
070920 - Asparagus exports ('000USD) 070820 - Bean exports ('000USD)
Appendix L: Tanzania's normality test results for Beans and Vegetables
Beans Vegetables
0 1000 2000 3000 4000 0 200 400 600 800
070820 - Bean exports ('000USD) 070990 - Vegetable exports ('000USD)
177
Frequency Frequency
0 20 40 60 80 0 20 40 60 80 100
Frequency
0 20 40 60 80 100 Frequency
0 20 40 60 80
Appendix M: Uganda's normality test results for Pepper, Bananas and Beans
Pepper Bananas
0 1000 2000 3000 0 1000 2000 3000
070960 - Pepper exports ('000USD) 080300 - Banana exports ('000USD)
Beans
0 10 20 30 40 50
070820 - Bean exports ('000USD)
178
Frequency
0 20 40 60 80
Frequency
0 20 40 60 80 100
Frequency
0 20 40 60 80 100