University of Ghana  http://ugspace.ug.edu.gh
UNIVERSITY OF GHANA, LEGON 
 
FACULTY OF SOCIAL STUDIES 
 
DEPARTMENT OF ECONOMICS 
 
 
 
EXAMINING THE “IMPACT OF TRADE FACILITATION INDICATORS ON 
BILATERAL EXPORTS” INVOLVING SUB-SAHARAN AFRICA 
 
BY  
 
EDWIN PROVENCAL 
(10051052) 
 
THIS THESIS IS SUBMITTED TO THE DEPARTMENT OF ECONOMICS OF THE 
FACULTY OF SOCIAL STUDIES, UNIVERSITY OF GHANA, LEGON IN 
PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF 
MASTER OF PHILOSOPHY DEGREE IN ECONOMICS  
 
 
 
SUPERVISED BY: DR. FESTUS EBO TURKSON 
DR. YAW ASANTE 
  
University of Ghana  http://ugspace.ug.edu.gh
DECLARATION 
 
I, Edwin Alfred Provencal hereby declare that this thesis is the original 
research undertaken by myself under the guidance of my supervisors; and 
except for references to other people’s works which have been duly cited, 
this thesis has neither in part nor in whole been submitted for another 
degree elsewhere. 
 
 
 
………………………………………
EDWIN ALFRED PROVENCAL 
 
 
……………………………….. 
DATE 
 
 
 
………………………………..    ….……………………….. 
DR. F. EBO TURKSON    DR. YAW ASANTE 
(SUPERVISOR)      (SUPERVISOR) 
 
 
 
………………………………..   ……………………………. 
DATE       DATE 
 
  
2 
University of Ghana  http://ugspace.ug.edu.gh
ABSTRACT 
 
This study examines the impact of trade facilitation measures on bilateral trade involving 
Sub Saharan Africa. It uses country specific data on logistic performance measures for 189 
countries including 47 from SSA over a 9-year period from 2007 - 2015. Leveraging a 
logistics performance augmented gravity model, the findings show trade enhancing effects 
on all the six trade facilitation measures with customs efficiency, trade and transport-related 
infrastructure and competence and quality of logistics services having the greatest impact in 
that order. A unit improvement in the efficiency of customs clearance measure will enhance 
bilateral trade by 19% – 27% whilst that for infrastructure ranges between 16% and 24%. 
Policy recommendations therefore focuses on a two-pronged approach – whilst investment 
in trade-related infrastructure is necessary, the investments required are huge and the benefits 
accrue in the medium to long-term. To achieve quick-wins, it is imperative that a special 
focus be given to initiative that leads to improve the efficiency of customs clearance 
procedures. 
 
  
3 
University of Ghana  http://ugspace.ug.edu.gh
DEDICATION 
 
This thesis is dedicated to my life partner, Roberta Talata Maldima-Provencal whose support 
has brought me this far and my mother, Grace Lamptey whose love, prayers and words of 
wisdom have inspired me all these years and still continues to be my source of inspiration. 
 
  
4 
University of Ghana  http://ugspace.ug.edu.gh
ACKNOWLEDGMENT 
 
First, I would like to express my gratitude to the giver of life, who has blessed me and 
endowed me with wisdom and the unquenchable spirit of enquiry and curiosity, without 
whom this work would have been an impossibility. SHALOM! 
 
I want to express my special appreciation to my supervisors - Dr. Festus Ebo Turkson and Dr. 
Yaw Asante for their enormous support, feedback, contributions and expert guidance in the 
course of this research work.  
 
I owe a huge debt of appreciation to all the lecturers who nurtured and piqued my interest in 
Economics–The Late Dr. Laryea, may his beautiful soul rest in peace, Prof. Quartey, Prof. 
Fosu, Dr Barimah, Dr Senadza, Dr. Gockel, Dr. Osei-Assibey, Dr. Twerefour, Dr. Dankwah, 
Dr. Ebo Turkson and Dr. Yaw Asante.  
 
To my course-mates, I say thank you for giving me the opportunity to engage you in some 
serious intellectual debates that broadened my horizon and contributed to this work. 
 
A special thanks to my mother and late father, the vessels used by God to introduce me into 
this world, to my better half - Mrs. Roberta Talata Maldima-Provencal, who sacrificed 
immensely for two years to enable me to pursue my passion and to my children – Jael-Grace, 
Ella-Jane and Emma-Rose who always brought me so much energy to prod on. 
 
Finally, I want to thank my friend for over 30 years, Lawyer Charles Okyere who encouraged me 
to commence this journey some three years ago. 
  
5 
University of Ghana  http://ugspace.ug.edu.gh
TABLE OF CONTENTS 
Abbreviations ....................................................................................................................... 10 
CHAPTER ONE: ................................................................................................................. 11 
INTRODUCTION ................................................................................................................ 11 
1.1 Background of the study ........................................................................................... 11 
Figure 1.1 - Regional Proportions of LDCs, Source: Economic and Social Council of the 
United Nations...................................................................................................................... 11 
Table 1.1: Types of Exports/Imports from SSA to ROW .................................................... 12 
Figure 1.2: Aid for Trade increases since 2002-05 baseline period (US$ million), Source: 
Aid for Trade, (OECD CRS Database) ................................................................................ 13 
Figure1.3: Regional distribution of Aid for Trade flows (in per cent), Source: Aid for 
Trade, (OECD CRS Database) ............................................................................................. 13 
1.2 Problem Statement .................................................................................................... 14 
Table 1.2: SSA’s Average cost to export/import a 20-footer container ............................... 14 
Figure 1.4 - 2014 SSA Share of World's Exports, WITS 2014 Database ............................ 15 
1.3 Key research question ............................................................................................... 15 
1.4 Objectives of the Study ............................................................................................. 16 
1.4.1 Key Objective: ....................................................................................................... 16 
1.5 Significance of the study ........................................................................................... 16 
1.6 Organization of the study .......................................................................................... 16 
CHAPTER TWO.................................................................................................................. 18 
OVERVIEW AND TRENDS IN GLOBAL TRADE.......................................................... 18 
2.1 Introduction ............................................................................................................... 18 
2.2 Global Trade Trends.................................................................................................. 18 
Figure 2.1: 2015 Global Trade Trends, Exports share of Global GDP, Source WTO 
International Trade Statistics 2015 ....................................................................................... 20 
Table 2.1: Volume of world merchandise exports and gross domestic product .................. 21 
2.3 Export Mix ................................................................................................................ 21 
Figure 2.2:  WTO World merchandise trade volume by major product group, 1950-2014 . 21 
Table 2.2: Average annual change in percentages ............................................................... 22 
Table 2.3: Network of Merchandise Trade by Region 2012-2014 ...................................... 22 
2.4 Trade Costs ................................................................................................................ 23 
Figure 2.3: Declining transport, communication and sea freight costs relative to 1930 ...... 24 
Table 2.4: Transaction Cost and International Trade ........................................................... 25 
2.5 Trade Facilitation ...................................................................................................... 25 
Figure 2.4: Soft and Hard Infrastructure .............................................................................. 26 
2.5.1 Trade Facilitation and Export Performance ........................................................... 28 
Figure 2.5: The four pillars of trade facilitation ................................................................... 28 
6 
University of Ghana  http://ugspace.ug.edu.gh
2.5.2 Trade Facilitation Efforts within SSA ................................................................... 30 
Figure 2.6: Intra-ECOWAS exports, 2014 ........................................................................... 31 
Figure 2.7: Progress on Doing Business indicators, percentage reduction in distance to the 
frontier, 2009-2014............................................................................................................... 32 
Figure 2.8: Intra-African Trade 1999-2011.......................................................................... 37 
Figure 2.9: Intra-regional exports as a proportion of total exports (%) ............................... 38 
Figure 2.10: Sub-Saharan Africa’s trade facilitation performance - OECD indicators ....... 39 
CHAPTER THREE .............................................................................................................. 41 
LITERATURE REVIEW ..................................................................................................... 41 
3.0 Introduction ............................................................................................................... 41 
Figure 3.1: Average SSA Export and Import Costs Trend .................................................. 41 
3.1 Theoretical Review ................................................................................................... 41 
3.1.1 Models of measuring trade facilitation .................................................................. 42 
3.1.2 The iceberg partial equilibrium model .................................................................. 42 
3.1.3 The classical general equilibrium models ............................................................. 44 
3.1.4 The Ricardian model ............................................................................................. 44 
3.1.5 The Heckscher-Ohlin (H-O) model ....................................................................... 45 
3.1.6 The “New Trade Theory” – monopolistic competition ......................................... 46 
3.1.7 The “New Trade” Theory – heterogeneous firms ................................................. 48 
3.1.8 The supply chain models ....................................................................................... 49 
3.2 Despite the trade costs, there are enormous gains from trade ................................... 50 
Figure 3.4: Sources of gains from trade ............................................................................ 50 
3.3 Measuring trade facilitation ...................................................................................... 51 
3.4 Empirical Review ...................................................................................................... 54 
CHAPTER FOUR ................................................................................................................ 59 
METHODOLOGY ............................................................................................................... 59 
4.1 Introduction ............................................................................................................... 59 
4.2 Data Sources .............................................................................................................. 59 
4.3 Methodology ............................................................................................................. 60 
4.3.1 Theoretical underpinnings ..................................................................................... 60 
4.3.2 Estimation Methods..................................................................................................... 63 
4.3.3 The multilateral trade resistance (Remoteness) problem ...................................... 65 
4.4 Empirical Estimation Model ..................................................................................... 67 
4.4.1 The Model.............................................................................................................. 70 
4.4.2 Description of variables and Expected Signs ........................................................ 70 
4.5 A priori-signs............................................................................................................. 71 
4.5.1 Size of Country ...................................................................................................... 71 
7 
University of Ghana  http://ugspace.ug.edu.gh
4.5.3 Distance ................................................................................................................. 72 
4.5.4 Infrastructure (Port Road & ICT) .......................................................................... 72 
4.5.5 Other Country Characteristics ............................................................................... 73 
CHAPTER 5 ......................................................................................................................... 75 
RESULTS AND DISCUSSIONS ........................................................................................ 75 
5.1 Introduction ............................................................................................................... 75 
5.2 Summary Statistics .................................................................................................... 75 
5.3 Diagnostic Tests ........................................................................................................ 80 
5.3.1 Time-Effects .......................................................................................................... 80 
5.3.2 Breusch -Pagan Heterogeneity Test Results .......................................................... 80 
5.3.3 Correlation ............................................................................................................. 81 
5.3.4 Hausman Test for Fixed Effects versus Random Effects ...................................... 82 
5.4 Durbin Wu-Hausman Endogeneity Test ................................................................... 83 
5.5 Diagnostics Results and Model Estimation Strategy ................................................ 84 
5.6 Main Results and Discussions ................................................................................... 85 
CHAPTER 6 ......................................................................................................................... 95 
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ........................................ 95 
6.1 Introduction ............................................................................................................... 95 
6.2 Summary and Conclusions ........................................................................................ 95 
6.3 Policy Recommendation ........................................................................................... 96 
6.4 Limitations of the Study ............................................................................................ 96 
6.5 Further Research Areas ............................................................................................. 97 
7.0 References ................................................................................................................. 98 
Appendix 2: Results for Disaggregated LPI for all trade involving SSA (Full Sample) .. 104 
Appendix 3: Gravity Model Results for Disaggregated LPI for all SSA to ROW trade .. 104 
Appendix 4: Gravity Model Results for Disaggregated LPI for all ROW-SSA trade ...... 106 
Appendix 5: Gravity Model Results for Frequency with which shipments reach consignee 
on time logistics measure – “lpitime” for various groupings/samples ............................... 107 
Appendix 6: Gravity Model Results for the Ability to Track and Trace Consignments 
logistics measure–“lpiatratr” for various groupings/samples............................................. 107 
Appendix 7: Gravity Model Results for Competence and Quality of Logistics Services 
measure–“lpicqols” for various groupings/samples ........................................................... 109 
Appendix 8: Gravity Model Results for Efficiency of Customs Procedures measure–
“Customs Efficiency” for various groupings/samples ....................................................... 110 
Appendix 9: Gravity Model Results for Quality of Trade and Transport Infrastructure 
measure– “Infrastructure” for various groupings/samples ................................................. 111 
Appendix 10: Gravity Model Results for Ease of Arranging competitively priced shipment 
–“ lpieoacps” for various groupings/samples ..................................................................... 112 
8 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 11: Gravity Model Results for the Overall logistics performance indicator – 
“lpioverall” for various groupings/samples........................................................................ 113 
 
  
9 
University of Ghana  http://ugspace.ug.edu.gh
Abbreviations 
ASYCUDA Automated System for Customs Data  
EAC  East Africa Community 
EAP East Asia Pacific 
ECCAS Economic Community of Central African States 
ECOWAS Economic Community of West African States 
EU European Union 
FTA Free Trade Agreements 
GVC Global Value Chain 
GTAD Global Trade-Related Technical Assistance Database  
IGAD Inter-governmental Authority on Development 
LPI Logistics performance Indicators 
MENA Middle East and North Africa 
MDG 8 Millennium Development Goal #8 
MTR Multilateral Trade Resistance 
NA North America 
NTB Non-Tariff Barrier 
OSBP One-Stop Border Post 
PTAs Preferential Trade Agreements 
RECs Regional Economic Communities 
RTAs Regional Trade Agreements 
ROW Rest of the World 
SA South Asia 
SADC Southern African Development Community 
SSA Sub Saharan Africa 
TFI Trade Facilitation Indicators 
TMS Transit Management System 
WTO  World Trade Organization 
  
  
10 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER ONE:  
 
INTRODUCTION 
1.1 Background of the study 
Trade facilitation simply means the guidelines and actions that are aimed at decreasing the 
costs in trade and ensuring that the efficacy of each process in the global trade value chain is 
improved. The World Trade Organization (WTO) defines trade facilitation as the 
“simplification of trade procedures”, understood as being the “activities, practices and 
formalities involved in collecting, presenting, communicating and processing data required for 
the movement of goods in international trade”, Hoekman and Kostecki (2001). 
The significance or otherwise of the impact of trade facilitation comes from the thorough 
understanding of trade costs. It is an undeniable fact that lowering trade costs will boost the 
performance of trade. For trade facilitation to achieve a meaningful impact, trade costs should 
be brought downwards which pre-supposes that there must be a wider scope to cut transaction 
costs. We can therefore safely infer that detailed information on the components or sources of 
trade costs is key in shaping which specific trade facilitation indicator(s) will have the highest 
payoff for trade among regional economic communities (RECs) , intra-regional trade, income 
groups and in particular sectors (primary non-oil, oil and manufactures). 
International trade has the capacity to lift nations out of poverty towards economic 
development and prosperity. It is also a truism that the 
world trade order has not favored least-developing 
countries of which a majority can be found in Sub-
Saharan Africa (SSA) (see figure 1.1). The failure of 
countries within SSA to leverage trade as a tool for 
sustainable economic growth and poverty reduction 
Figure 1.1 - Regional Proportions of LDCs, originates from three fundamental features of 
Source: Economic and Social Council of the 
United Nations 
their economies - structure, market size and 
11 
University of Ghana  http://ugspace.ug.edu.gh
direction of trade. In 2014, SSA’s share of international trade was 1 per cent and has been 
marked by a decline over the decades with products’ exports falling by 2.4 percent in 2013 
(WTO, 2013). African exports are focused on a very narrow range of products (Wohlmuth, 
2007; Collier 1998) making African economies susceptible to unfavorable trade shocks. 
Additionally, the continent’s exports are not diversified with a heavy focus on primary goods, 
having relatively slow growth potential, suffering from erratic prices and a reduction in terms 
of trade in the long term. As shown in table 1, in 2014, almost 74% of all exports from SSA 
were either raw materials or intermediate goods equivalent to US$123,377million. However, 
63% of all imports equivalent to $140,016 were either consumer goods (36%) or capital goods 
(27%). This over-dependence on raw materials for exports and over-reliance on imports of 
finished products imposes a huge barrier to economic development in the region. 
Table 1.1: Types of Exports/Imports from SSA to ROW 
SSA Exports (Millions) SSA Imports (Millions) 
 
Raw Materials $77,180 46.30% $26,256 11.86% 
Intermediate Goods $46,197 27.23% $43,674 19.73% 
Consumer Goods $28,853 17.31% $80,709 36.46% 
Capital Goods $13,099 7.86% $59,307 26.79% 
Source: WITS Database (2014) 
Other contributing factors include lack of competitiveness and supply-side restrictions which 
have limited the continent’s share of international trade specifically with regards to 
manufactures and services. 
In view of the above, the WTO identified the need to support developing and least developed 
countries (LDCs) to fully align into the global trading system to enable them exploit and 
optimize the benefits of trade. This is evident in the recent Doha negotiations leading to the 
Doha Development Agenda. It is also captured in the Millennium Development Goals (MDG 
8) focused on building a global partnership for development”. Within the sub-objectives 8.6 and 
8.7 which deal with duty free and preferential tariffs respectively, attempts have been made by 
WTO to boost market access for exports that originate from developing and LDCs. Despite the 
12 
University of Ghana  http://ugspace.ug.edu.gh
general acceptance that the outcome of the Doha Round of negotiations will benefit developing 
countries by improving their chances of gaining a greater share of global trade, it is now widely 
accepted that improved market access opportunities alone may not suffice as many of these 
countries do not have the capacity to take advantage of the opportunities that may present 
themselves. This triggered the “aid for trade” initiative which has seen a general increment in 
support with a huge chunk of resources flowing to Africa and Asia as shown in charts 2 and 3 
below. This WTO-led initiative encourages donors and governments of developing countries 
to identify one of the most important drivers for socio-economic development - trade. 
Additionally, the initiative seeks to marshal resources to resolve the challenges related to trade 
identified by the developing and the LDCs. Alaba (2009) highlights the importance of “Aid-
for-Trade” to enhance transport infrastructure from the European Union to ECOWAS to 
achieve unhampered flow of goods and services across borders. 
 
 
Figure 1.2: Aid for Trade increases 
Figure1.3: Regional distribution of Aid for Trade 
since 2002-05 baseline period (US$ 
flows (in per cent), Source: Aid for Trade, (OECD 
million), Source: Aid for Trade, 
CRS Database) 
(OECD CRS Database) 
 
  
13 
University of Ghana  http://ugspace.ug.edu.gh
1.2 Problem Statement 
There have been numerous attempts by various stakeholders at reducing trade costs to ensure 
LDCs and developing countries attract their fair share of gains from trade. In SSA, regional 
initiatives such as preferential trade agreements (PTAs) and regional trde agreements (RTAs) 
have shown some improvement in intra-regional trade. Turkson (2011) found out that apart 
from the European Union, SSA economies traded with other regions at a significantly higher 
cost than they did with each other. Conversely, within SSA, trade costs among countries within 
the same regional economic blocs are comparatively lower as compared with non-SSA 
countries. Despite these favorable findings, the trade situation is still far from satisfactory 
taking SSA as a region trading with the rest of the world (ROW) and this is evidenced by 
increasing trading costs as shown in table 2 below. The average costs to export and import a 
container have been increasing since 2010 (World Bank Doing Business Report, 2015). 
Table 1.2: SSA’s Average cost to export/import a 20-footer container 
 2010 2012 2014 
Cost to export (US$ per container) $1,915 $1,998 $2,156 
Cost to import (US$ per container) $2,352 $2,573 $2,874 
Source: World Bank Doing Business Report, 2015 
The implications of this is that SSA’s share of total exports in world trade stands at a paltry 1% 
as shown in figure 1.4 below.(Author’s calculation from World Integrated Trade Solution 
(WITS) 2014 Database ) 
14 
University of Ghana  http://ugspace.ug.edu.gh
 
Figure 1.4 - 2014 SSA Share of World's Exports, WITS 2014 Database 
 
1.3 Key research question 
• Are there different trade facilitation levers peculiar to trade involving SSA? And if there 
are, what are they? Should all of them be given the same priority by policy makers? 
 
The motivation for this study seeks to answer these questions and to add to the academic 
discourse on bilateral trade and trade facilitation including SSA by unearthing the key 
determinants of trade relevant to SSA’s trade with the rest of the world. The findings in this 
study may also go a long way to contribute towards informing trade policy within the sub-
Saharan region. 
  
15 
University of Ghana  http://ugspace.ug.edu.gh
1.4 Objectives of the Study 
This study seeks to achieve the singular objective as stated below. 
1.4.1 Key Objective:  
To examine the impact of trade facilitation on bilateral exports involving Sub-Saharan African 
countries and the rest of the world (ROW) over a nine-year period from 2007 -2015.  
1.5 Significance of the study 
Trade and its facilitation are key issues for developing countries and LDCs with over 65% of 
nations that participated in a survey rated trade facilitation on their first three “aid-for-trade” 
focus areas. Other areas included network, transport, cross-border infrastructure, trade 
negotiations, WTO accession, competitiveness, diversification of exports, connectedness to 
global value chains, costs of adjustment and regional integration. When countries and their 
governments seek to improve the socio-economic conditions of their citizens by increasing 
output, a key contributing factor is to significantly increase bilateral exports through 
international trade. Trade facilitation has been touted by the WTO as the largest contributing 
factor to radically increase bilateral exports as reforms targeting trade facilitation are expected 
to achieve a reduction in trade costs (cause) leading to enhanced trade (effect) in Sub-Saharan 
Africa (World Trade Report, 2015, Pg 34).  
The findings from this paper will go a long way to support the trade facilitation efforts from 
WTO and other stakeholders (donors, academics, economists, governments) by assisting 
various countries within SSA target their trade policies, strategies and limited resources for 
maximum impact in their bid to be relevant in the global trading system whilst at the same time 
improving the welfare of their people through economic growth and development. 
1.6 Organization of the study 
There are six chapters in this study. Chapter one includes the background, the problem 
statement, the research questions, objectives, justification and organization of the study. A brief 
16 
University of Ghana  http://ugspace.ug.edu.gh
overview of the study is captured in chapter two. Chapter three focuses on the literature review 
(both theoretical and empirical) associated with this study. The methodology and empirical 
approaches for this study are captured in chapter four with discussions on empirical results of 
the study dealt with in chapter five. Conclusions and recommendations are found in Chapter 6. 
  
17 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER TWO 
 
OVERVIEW AND TRENDS IN GLOBAL TRADE 
 
2.1 Introduction  
This chapter sheds light on the benefits of trade, triggers for increasing global trade, global 
shares of trade, especially SSA’s proportion of global trade and efforts made so far by SSA 
countries in their bid to get connected into the global value chain (GVCs). It also sheds light 
on trade costs and the status of trade facilitation in the sub-region. 
 
2.2 Global Trade Trends 
Due to its welfare implications, exchange of goods and services among nations has been of 
concern to many stakeholders - governments, economists, academics, citizens, non-
governmental organizations and the private sector throughout much of recorded history.  
The 19th century, heralded the industrial revolution which brought in its wake a vast expansion 
in international trade with nations specializing in commodities in which they were efficient in 
producing. This reinforced and accelerated the international division of labour which led to 
inequitable and uneven development and industrialization – benefitting the North, led by 
Britain (Western Europe, North America and later Japan followed suite) and the South 
relegated to a raw material supplying role. 
The wake of the second industrial revolution brought with it vast technological developments 
in communications and transportation – from steamships, railroads airplanes, and telegraphs to 
automobiles. Subsequently the technological revolution in the twenty-first century brought in 
its wake improvement in communication such as fixed telephony, mobile telephony (2G, 3G, 
4G, LTE) and the proliferation of fiber optics leading to the world super-highways (world wide 
web), also known as the internet. These technological advancements progressively lowered the 
costs incurred to move people, capital, goods and technology around the world. For instance, 
18 
University of Ghana  http://ugspace.ug.edu.gh
with respect to sea travel, in 1914, out of the total trading fleet worldwide, 97% were fired by 
coal. “This fell by 27 percentage points to 70% in the 1920s and was less than 50% by the end 
of the 1930s. Only 4% of the world fleet, measured in tonnage, were coal-burning ships in 
1961(Lundgren, 1996). “ 
According to Lundgren (1996), the world maritime trade grew sharply from 500 million tons 
in 1950 to 4,200 million tons in 1992.  He found that this trade volume was necessitated largely 
by the replacement of steam engines by diesel and electric locomotives. Maddison (2008) 
reveals that there was one commercial motor vehicle for every 85 Americans in 1921 and one 
for every 29 in 1938. In 1913, the fleet of passenger cars was about 1.5 million; by 2002, it was 
530 million. Similarly, by 1937, about 5.7% of the world’s railway mileage was in Africa, 
10.2% in Latin America and 10.9 % in Asia (O’Rourke and Findlay, 2007). With respect to air 
travel, air passenger miles also increased substantially from 28 billion in 1950 to 2.6 trillion in 
1998 (Maddison, 2008). 
This phenomenon has significantly eroded what the economic historian Geoffrey Blainey 
termed “the tyranny of distance” (Blainey, 1968) with its associated huge costs burden. 
According to Cairncross (1997), this “death of distance”, has been one of the most significant 
drivers shaping global economic development since the early 1800s. Over time, Anderson and 
Wincoop (2004) later proved this theory to be over-exaggerated even though it still plays a 
significant role in trade costs. 
The falling transportation and communication costs brought together tremendous 
diversification in tradable goods leading to an exponential growth in global trade. Maddison 
(2008) reports that since the 19th Century the world’s population has increased almost six-fold, 
world output 60-fold and world trade- over 140 times. WTO claims that the value of world 
merchandise exports increased from US$ 2.03 trillion in 1980 to US$ 18.26 trillion in 2011, 
equivalent to an average growth rate of 7.3 per cent in current dollar terms. A faster growth in 
19 
University of Ghana  http://ugspace.ug.edu.gh
trade was also recorded in commercial services over the same period from US$ 367 billion in 
1980 to US$ 4.17 trillion in 2011, which translates to a yearly growth of 8.2%. In volume terms, 
accounting for differences in exchange rates and prices, global trade grew more than four-times 
the growth between 1980 and 2011 (WTO, 2015). Figure 2.1 below shows that exports as a 
percentage of world GDP has grown steadily from 5% in 1950 to 48% in 1990, 70% in 2000 
and 149% in 2014. 
Figure 2.1: 2015 Global Trade Trends, Exports share of Global GDP, Source WTO 
International Trade Statistics 2015 
 
In general, growth rates in global exports have exceeded GDP growth rates even though global 
events have made the rate of growth decline in recent times. The rate of growth was 7% for the 
period 1995-2000, declining gradually to 3.3% in 2010-2014. Similarly, annual global GDP 
growth declined from 3.4% within 1995-2000 to 2.5% in 2010-2014 (Table 2.1). The 
narrowing gap between exports and GDP growth over time can be attributed to several factors 
notably Mexico’s monetary crisis in 1995-2001, the Asian financial crisis of 1997, and the 
bursting of the dotcom bubble in 2001. In addition, prices of crude oil and other primary 
products went up due to high Chinese demand for natural resources between 2002 and 2008. 
Despite China’s assent to the WTO in December 2001 which paved the way for its significant 
economic development and as such contributed to increasing world trade from 2002 to 2008, 
it was still not enough to deliver the growth in previous periods. The 2008 financial crisis, 
triggered by the subprime lending crisis in the United States, led to a global recession between 
20 
University of Ghana  http://ugspace.ug.edu.gh
2008 and 2011. The volume of world exports plunged 12% in 2009 while world gross domestic 
product reduced by 2% (WTO, 2015). 
Table 2.1: Volume of world merchandise exports and gross domestic product 
Annual % Exports Annual % GDP 
Period Difference 
Growth Growth 
1995-00 7.0 3.4 3.6 
2000-05 4.9 2.9 2.1 
2005-10 3.4 2.3 1.1 
2010-14 3.3 2.5 0.8 
Source: World Trade Organization Statistics 2015 
 
2.3 Export Mix 
The growth in trade volumes fueled by technological advances in transportation and 
communication led to increased diversification of commodities that could be traded. Over time 
the world merchandise trade mix in terms of volume has shifted tremendously since 1950  (base 
year) with a major shift from agricultural and fuel/mining products to manufactures as depicted 
in figure 2.2  with manufactures 79%, fuels and mining 11% and agricultural products 9% as 
at 2014. Similarly, in terms of annual growth rates in the export mix, we see clearly from the 
table 2.2 the increasing dominance of manufactures over the respective time periods. From 
1950 to 1973, agricultural produce grew at a rate of 4.3% compared with fuel and mining 7.4% 
and manufactures representing the highest growth rate of 9.8%. 
 
Figure 2.2:  WTO World merchandise trade volume by major product group, 1950-2014 
 
21 
University of Ghana  http://ugspace.ug.edu.gh
The importance of manufactures in trade and economic development of nations cannot be 
overstated in the creation of employment and improvement of national welfare therefore 
economies whose fundamental structures have not aligned to this trend will be left behind in 
the global trading system.  
Table 2.2: Average annual change in percentages 
 Agricultural Fuels and mining Manufactures 
products products 
1950-1973 4.3 7.4 9.8 
1973-1990 2.4 0.5 5.5 
1990-2014 3.7 2.9 5.6 
1950-2014 3.6 3.9 7.1 
Source: World Trade Organization Statistics 2015” 
 
Because of the polarized global trade structure where the developed North sourcing their inputs 
from the under-developed or developing South, the winners and losers of global trade became 
obvious over time. Most of the beneficiaries of the global trading system were largely from the 
North with some outliers in Asia, like China1. Sub-Saharan Africa so far has performed sub-
optimally and has benefited least from the gains of global trade. China’s trade alone with the 
rest of the world for the period 2012-2014 is more than double the whole of Africa’s trade as 
shown in Table 2.3 below. 
Table 2.3: Network of Merchandise Trade by Region 2012-2014 
Africa 
 North Europe Africa Asia Excluding China 
America 
South Africa 
2012 16.86% 36.47% 3.27% 29.91% 2.63% 7.44% 
2013 16.73% 36.34% 3.35% 30.13% 2.73% 7.68% 
2014 17.28% 36.72% 3.45% 29.66% 2.86% 7.61% 
Source: World Trade Organization Statistics 2015 
 
                                                          
1 China’s accession to the WTO in December 2001 opened up their economy to the ROW 
22 
University of Ghana  http://ugspace.ug.edu.gh
This poses some interesting questions to managers of economies, economists, donors and 
NGOs with respect to the state of affairs in SSA – What are the components of trade costs and 
can they be identified? Which of these components identified are direct or indirect? Are the 
indirect costs measurable and is data available? What is their impact on international trade? 
Why are these costs different for different countries, regions and income groups? What should 
trade policy focus on to enhance a country’s competitiveness?  
2.4 Trade Costs 
Generally defined, trade costs are the sum total of all costs accrued in transporting a good from 
the location of origin to the final destination. This excludes the direct costs incurred in the 
production of the good. Trade costs include, but are not exclusive to transportation, policy 
barriers, information and search, foreign exchange, contract, enforcement, legal and regulatory 
and local distribution costs. 
Trade costs can be grouped into two - natural and artificial costs. Natural costs are normally 
due to geography - landlocked vs port, distance, country-related, and time which directly 
affects the cost of transporting the good from one location to the other. Artificial costs are those 
associated with tariffs and non-tariff barriers (standards, logistics, facilitation, infrastructure), 
information and search costs, contract enforcement costs, foreign exchange costs, legal and 
regulatory costs, and local costs incurred in wholesaling and retailing. Trade costs are reported 
in terms of their ad-valorem tax equivalent and vary extensively across nations, regions and 
income groups. 
Trade costs are large and immensely shaped by a nation’s economic policy, having huge 
welfare implications (Andersen and Wincoop, 2002). Trade costs are so important that Obstfeld 
and Rogoff (2000) posited that all the six puzzles of international macroeconomics can be 
explained by trade costs. In effect, trade costs definitely do matter! (Andersen and Wincoop, 
2004). 
23 
University of Ghana  http://ugspace.ug.edu.gh
Since 1930, global trade costs have been declining due to technological advances in 
communication, transport and productivity of marine transport as shown in Figure 2.3 leading 
to expansion in world trade. 
Figure 2.3: Declining transport, communication and sea freight costs relative to 1930 
 
Source: Transaction Costs – OECD Economic Outlook (2007) 
 
Table 2.4 below shows the transactions costs from various regions for the year 2014. The 
comparison of the six dimensions of the transaction costs reveal that on exports, (SSA) has the 
largest costs incurred $2,152 and it takes up to 30 days requiring 8 documents to process. This 
compares to $839 in 20 days requiring 6 documents for East Asia Pacific, $1141, in 20 days 
with 6 documents for Middle East and North Africa, $1,343 in 17 days with 6 documents for 
Latin America and Caribbean. Lastly in South East Asia, it costs $1,636 in 29 days requiring 8 
documents to process for South Asia. 
On imports, SSA is at a great disadvantage to the extent that importing a container costs $2,868 
and takes an average of 38 days. In comparison, East Asia Pacific’s cost is $867, taking 22 
days, Middle East and North Africa incurs $1,304 and 24 days, Latin America and Caribbean 
incurs $1,722 and it takes 18 days, South Asia incurs $1,835 taking 31 days.   
24 
University of Ghana  http://ugspace.ug.edu.gh
Table 2.4: Transaction Cost and International Trade 
Cost to Cost to 
Documents Time to Documents Time to 
export import 
Sub-Region to export export to import import 
(US$ per (US$ per 
(number) (days) (number) (days) 
container) container) 
South Asia 8 29 1,636 9 31 1,835 
Middle East & 
6 20 1,141 8 24 1,304 
North Africa 
Latin America 
6 17 1,343 7 18 1,722 
& Caribbean 
Europe & 
7 25 2,138 8 26 2,376 
Central Asia 
East Asia 
6 20 839 7 22 867 
Pacific 
Sub-Saharan 
8 30 2,152 9 38 2,868 
Africa 
 Source: World Bank Doing Business Database, 2014” 
In effect, the cost to export and import one container in SSA is 2.5 times and 3.3 times 
respectively the cost incurred in East Asia Pacific. These costs are significant and overall, SSA 
compares rather poorly with the other regions leading to trade diversion. 
2.5 Trade Facilitation 
It is true that SSA region has the highest trade costs and these costs may arise as a result of 
transportation and/or logistics. The costs can be monetary (levies, custom, transit, 
administrative fees) or related to time spent completing the paperwork, procuring licenses and 
permits, loading, unloading, inspection and transporting goods across borders. These costs may 
be double or triple for landlocked countries. Limao and Venables (1999), reveals that the 
median landlocked country has 30% of the trade volume of the median coastal country. These 
cross-border time and monetary costs can be so high that it may cancel any gains arising out of 
elimination or reduction of tariffs and non-tariff barriers. Perez and Wilson (2012) grouped the 
trade facilitation indicators into two main types – 1) hard infrastructure comprising information 
25 
University of Ghana  http://ugspace.ug.edu.gh
and communications technology (ICT) and physical infrastructure and 2) soft infrastructure 
made up of the business environment and border and transport efficiency as shown in figure 
2.4 below. They concluded that the poor state of either of the two indicators will likely 
contribute to high trade costs leading to low levels of trade in Africa. They also found out that 
most challenges relating to infrastructure mostly explains the relatively lower trade levels in 
Africa.  
Limao and Venables, (1999) also found out that the transport elasticity of trade flows is very 
high – (-2.5) hence when transport costs are halved, the volume of trade increases five-fold. 
Similarly, increasing infrastructure from the 75th to the 50th percentile increases trade by 50%. 
Figure 2.4: Soft and Hard Infrastructure 
 
Source: Author’s adaptation from World Development Report Vol. 40 
There is overwhelming agreement and evidence that initiatives aimed at facilitating trade could 
reduce these costs whilst bringing significant economic gains from trade to SSA.  
Generally, trade facilitation includes "measures aimed at streamlining trade procedures and 
reducing the cost and uncertainties of international trade transactions" (UNESCAP, 2011, Pg 
26 
University of Ghana  http://ugspace.ug.edu.gh
3). In addition to normal issues with respect to customs, time for transit and logistics, trade 
facilitation may look at broader features related to transport and port infrastructure, business 
practices, information and telecommunications, regulatory environment, corruption and even 
organized crime (Moisé, 2013; Clark et al., 2002).  
According to the Global Trade-Related Technical Assistance Database (GTAD), trade 
facilitation refers to the "simplification and harmonization of international trade procedures", 
or more precisely, to the "activities, practices and formalities involved in collecting, presenting, 
communicating and processing data required for the movement of goods in international trade". 
With respect to the soft infrastructure, trade facilitation brings value through three key areas: 
1. Transparency in customs procedures and regulations –Trade costs are minimized 
when there is clarity in the regulations and procedures relating to customs. These 
customs procedures need to be made available and known to all traders prior to trade 
and applied consistently and fairly over time across all ports of entry and exit. When 
these procedures change, the impact on traders will be minimized if the changes are 
made known to traders on time to allow them to make the required adjustments.  
2. Harmonization and standardization of customs procedures – This is ensuring that 
all processes and procedures related to customs are aligned to international best 
practices. When all traders have to meet a single standard instead of multiple ones, trade 
costs are reduced. Ensuring alignment of procedures and adherence to international 
conventions, international exchange of trade data and memoranda of understanding 
between customs goes a long way to harmonize and standardize procedures. 
3. Simplifying documentation and procedures – This is the reduction of the paperwork 
needed and procedures related to the clearing of goods. Elimination of duplicate 
procedures, automation and operational flexibility are some of the actions needed to 
achieve simplification. 
27 
University of Ghana  http://ugspace.ug.edu.gh
2.5.1 Trade Facilitation and Export Performance 
The concept of trade facilitation entails transparency, simplification, harmonization, and 
standardization (World Bank, 2014). It is widely believed that these four pillars are key to 
ensuring increased flow of goods and services across borders all things being equal. (See figure 
2.5) 
Figure 2.5: The four pillars of trade facilitation  
Pillar 1: Transparency Pillar 2: Simplification
Trade Facilitation Principles
Pillar 3: Harmonization Pillar 4: Standardization
 
Source: National Board of Trade, Sweden (also cited in UNECE, 2016). 
These basic principles of trade facilitation are explained below: 
Transparency  
This is the integral part of trade facilitation as it ensures that there is openness and 
accountability for the actions of governments. It encompasses the full disclosure of information 
such as laws, regulations, budgets, procurement decisions among others to the public for easy 
access and consumption to enable timely decision-making. It is advised that stakeholders and 
the public ought to participate and contribute substantially in the legislative process (World 
Bank, 2014). 
Simplification  
Simplification refers to the process of removing all unnecessary components and duplications 
in the trade formalities, procedures and processes. It should be based on an thorough business 
process analysis of the current “as-is” situation. 
 
Harmonization  
28 
University of Ghana  http://ugspace.ug.edu.gh
Harmonization in trade facilitation simply means the alignment of processes, procedures, 
standards, operation of documents with international conventions and practices. World Bank 
(2014) explains that harmonization can also mean the adoption and implementation of the same 
standards of partner countries which can be because of business decisions or part of regional 
integration processes. 
Standardization  
The formats for practices and procedures in trade facilitation are very crucial. Standardization 
creates the format for procedures, documents and information agreed by all parties involved in 
trade. These are then used to harmonize the methods and practices between the parties. To 
optimize the benefits from these principles, stakeholder engagement is paramount as 
government and the private sector must cooperate to achieve mutual understanding on the 
strategic intent. 
The trade facilitation theory suggests a direct chain of cause and effect relationships – an 
improvement in trade facilitation may lead to reduced trade costs. The reduced cost is expected 
to deliver improved cross-border trade leading to income growth and finally to improved 
human and socio-economic development of nations. 
Trade facilitation can enhance export performance through several channels: 
• Reducing input cost – like reduction in tariffs, trade facilitation will reduce cost of 
imports. This benefits firms in the local market as they reduce their final costs and make 
available more of the imported inputs. This lower cost of inputs will make traditional 
products more competitive in newer markets, thus improving diversification by 
exporting new products to both existing and new markets. 
• Increased Domestic Competition – Simplification and harmonization of customs 
processes and procedures will improve domestic competition benefitting consumers 
directly. Elimination of inefficient firms will provide opportunity to increase economies 
29 
University of Ghana  http://ugspace.ug.edu.gh
of scale for remaining firms. This increased competitiveness may translate into 
improved exports via diversification of export supply, with the effect stronger where 
multilateral agreements exist 
• Enabling smooth integration into GVC – The reduction of trade costs as a result of 
trade facilitation will make the supply and availability of final and intermediate goods 
less prone to delays and more predictable. Domestic firms can, through the reliability 
of supply participate more effectively in the global and regional supply chains – 
whether as an importer or an exporter of intermediate goods. 
2.5.2 Trade Facilitation Efforts within SSA 
In the past decades, there have been tremendous efforts by SSA countries to integrate 
effectively into the global value chain to tap the enormous benefits of globalization and 
international trade. Regional trade facilitation agreements and strategies have been 
implemented by several RECs in SSA to reduce trade cost, improve border procedures and 
boost trade flows. The efforts to improve trade facilitation in SSA are explained below. 
Economic Community of West African States (ECOWAS) 
Since its inception in 1975, ECOWAS has remained focused on ensuring  economic integration 
in Africa despite its challenges. The bloc seeks to promote economic prosperity of member 
countries through effective economic integration, easy access to market, harmonization of 
policies, institutional-building, and increased trade flows. However, till date, several trade 
facilitation challenges prevent it from achieving its desired goals.  
Jebuni (1997) observed that the efforts made by ECOWAS to promote regional integration 
have not been effective, and intra-regional trade was still very low at least before the 1990s.  
This was partly due to the fact that most member countries still had trade walls which was 
against the community protocol of opening markets to each other. There were huge regional 
30 
University of Ghana  http://ugspace.ug.edu.gh
conflicts, weak governance and doubts were raised about the impact of this intra-regional trade 
liberalization on member’s government revenue and balance of payments.  
In the 1990s ECOWAS went through a revamp, restoring the hopes and prospects of its 
member countries. ECOWAS revised its treaty in 1993 redefining measures and strategies for 
economic transformation leading to the ECOWAS Trade Liberalization Scheme (ETLS).   
This move has caused a marginal improvement in intraregional trade in recent time. Per the 
IMF (2015), from 2000 to 2014, the share of intra-ECOWAS exports in the region’s total 
exports rose from 7.7% to 8.3%. Due to lack of diversification of commodities or products, the 
intra-regional trade remains low. Almost all countries rely heavily on a few primary 
commodities, making up for about 75% of their exports. In Nigeria for example, though crude 
oil represents less than 10% of total export, it is the major intra-ECOWAS exports (as shown 
in Figure 2.6) 
Figure 2.6: Intra-ECOWAS exports, 2014 
 
 6000 Value of Exports to ECOWAS Millions $              70 
Percent Exports to ECOWAS (RH scale) 
5000 60 
4000 
50 
 
3000 
 
2000 
 
 10 
1000 
  
 
 
Source: IMF, Direction of Trade Statistics, 2014 
The major hindrance to trade facilitation in ECOWAS countries is non-tariff barriers (NTBs) 
such as trade border post checks, duplication of inspection, corruption and poor infrastructure. 
As shown in figure 2.7, efforts have been made to remove the barriers of trade across borders, 
and this has improved marginally. From 2009 to 2014, the average progress has not been more 
than 20%. In 2013 a supplementary protocol Act/SA.1/07/13 was signed by member states to 
31 
University of Ghana  http://ugspace.ug.edu.gh
create joint border posts so that the number of checkpoints would be reduced to sanitize the 
border processes.  
Figure 2.7: Progress on Doing Business indicators, percentage reduction in distance to 
the frontier, 2009-2014 
Togo 
Sierra Leone 
Senegal 
Niger 
Mali 
Liberia 
Guinea-Bissau 
Guinea 
Ghana 
Gambia, The 
Côte d'Ivoire 
Cabo Verde 
Benin 
Trading across borders 
-10% -5% 0% 5% 10% 15% 20% 25% 30% 
Percentage reduction in distance to the frontier 
 
Source: World Bank, Doing Business. 2014. 
There has been limited progress in the common goal of improving the payment systems and 
increasing intra-regional trade via the adoption of common or single currency. Member 
countries have consistently been unable to meet both the primary and secondary convergence 
criteria2.  
Looking at the limited progress, Amoako-Tuffour et al. (2016) remarked that the lack of 
progress is mainly due to lack of implementation of agreed polices and not lack of policies. 
Notable amongst them are the implementation of the ECOWAS Trade Liberalisation Scheme  
(ETLS), the Protocol on the Free Movement of Goods and Persons and the Right of Resident 
and Establishment. Other problems that are inimical to regional openness are non-compliance 
                                                          
2 “The four primary criteria are (1) single-digit inflation; (2) fiscal deficit of no more than 3% and 4% of GDP with 
and without grants, respectively; (3) limiting central bank financing of deficit of no more than 10% of previous 
year’s tax revenue; and (4) maintaining gross external reserves of not less than three months of import cover. The 
six secondary criteria are (5) prohibition of new domestic default payments; (6) tax revenue must exceed 20% of 
GDP; (7) wage bill to tax revenue ratio must not exceed 35% of GDP; (8) public investment to tax revenue ratio 
must not be less than 20%; (9) maintain stable exchange rates; and (10) maintain positive real interest rates 
(Amoako-Tuffour, et al., 2016) ” 
32 
University of Ghana  http://ugspace.ug.edu.gh
of member countries with the provisions of the Community Levy Scheme and the fragile 
political and security situation in some member states. 
South African Development Community (SADC) 
The trade facilitation efforts by SADC started with the development of SADC protocol on 
Trade in 1996. This protocol encompassed the simplification, harmonization and 
modernization of various custom procedures and transit. With respect to the customs 
procedures, for example, the protocol demands that SADC member countries adopt common 
nomenclature to ensure simplicity and modernize their custom procedures by employing a 
system of goods valuation aligned to the WTO systems. They were also to cooperate in training 
each other’s staff, communicate information, investigate and address customs offences (SADC, 
2012) 
A range of trade facilitation strategies preceded these protocols. SADC developed the 
Coordinated Border Management (CBM) guidelines and systems to reduce the disruption of 
movement of people and goods across borders. To ensure the success of this protocol, SADC 
has established many regional single windows. Also, initiatives such as one-stop border post 
(OSBP) systems has been implemented to minimize the time and cost that comes with clearing 
goods through border crossings. In addition, to harmonize customs administrations and IT 
systems within SADC, there has been the introduction of automation and custom reform and 
modernization (CRM) programs.  These initiatives have helped connect most member 
countries over the years.  
Other measures include the transit management system (TMS) adopted in 2009 and meant to 
harmonize and standardize procedures for transit of goods within SADC. Although there had 
been models, as of 2014, the TMS had not been fully implemented. Attempts had been made 
to also develop the necessary trans-boundary infrastructure to boost trade within SADC. 
Finally, the SADC Integrated Regional Electronic Settlement System (SIRESS) was created to 
33 
University of Ghana  http://ugspace.ug.edu.gh
cut out intermediaries and simplify cross-border regional payment transactions that boosts 
intra-SADC trade. 
Recently, the SADC implemented a one-stop border system at Chirundu, between the Zambia 
and Zimbabwe border which has resulted in many benefits. First, smuggling activities at border 
posts has reduced drastically with the introduction of Chirundu One-Stop Border Post (OSBP) 
(Muqayi and Manyeruke, 2015). The challenge of border delays was one of the key drivers 
motivating smuggling. However, the Chirundu OSBP has helped reduce significantly the 
border delays thereby minimized smuggling activities. The OSBP has further improved border 
security and reduced border jumping activities.  
In addition, this measure is said to have reduced clearance times for cargo and associated costs 
at specific border crossings (SADC, 2015). The introduction of the OSBP has led to a 
substantial reduction in the time taken to clear customs by streamlining procedures for crossing 
the border thereby significantly reducing processing times (SADC, 2015; Zimbabwe Ministry 
of Industry and Commerce, 2011). This has caused a tremendous decline in the time taken for 
freight crossing the border from days to hours. Some studies show that the waiting times for 
commercial traffic have reduced significantly from an average of four days to just few hours 
(Kassee, 2014). 
It is also noteworthy that the rate of clearing goods and movement of people has improved post 
OSBP. Nkwemu (2011) estimates that the process of daily export declaration at Chirundu 
increased from 128 in 2003 to 380 in 2009, an increase of 670% over seven years. On port 
congestion, Nkwemu (2011) reveals that on a daily basis, an average of 470 buses was cleared 
at the border in 2011 an increase from 100 buses in 2004, equivalent to 370%. Consequently, 
total revenue received by the Zimbabwean government has been impacted positively. Revenue 
losses that were previously experienced because of unnecessary delays at the border have 
reduced drastically. 
34 
University of Ghana  http://ugspace.ug.edu.gh
East African Community (EAC) 
The earlier EAC was birthed in 1967 but dissolved in 1977 due to ideological differences and 
mistrust among member countries – Uganda, Kenya and Tanzania. Later in 1999, a new treaty 
was signed and entered in force in 2000 by the original members.  
Currently, Burundi, Rwanda, Kenya, Tanzania and Uganda form the EAC, originally started 
as a customs union in 2005. In 2010, they moved to a common market. Trade in this part of 
SSA is very difficult due to poor infrastructure, overwhelming lengthy procedures and the 
existence of many trade barriers therefore, trade facilitation agreements have been suggested 
as a means of increasing trade in this region according to the Kenya Private Sector Alliance 
and the East African June 18 2016 publication titled “Intra-EAC trade falls to $5.63 billion.  
The member countries of the EAC have made frantic efforts to ensure easy trade facilitation 
among them which has led to the agreement and the signing of the Framework for the 
Attainment of the EAC Single Customs Territory (SCT) in 2013. The objective of the SCT is 
to ensure effective and efficient trade processes in EAC via trade facilitation. The framework 
is to ensure that the cost of doing business in the region is reduced as well as improving intra-
regional trade by reducing internal border controls.   
However, the existence of NTBs impedes the free movement of goods and people in the region. 
Although some successes have been made in reducing NTBs, the presence of old and new 
barriers has limited the trade facilitation among member countries (Amoako-Tuffour, 2016). 
To solve this, information about existing NTBs are released frequently to member countries 
through its publication. The EAC then puts a time-bound programme for member states to 
eliminate the NTBs before a given deadline. For example, four new barriers and 18 old and 
unresolved barriers were reported in March 2015 (Amoako-Tuffour, 2016).  
The member countries of the EAC have also embarked on many initiatives to improve their 
customs procedures. The adoption of electronic processing for customs procedures has 
35 
University of Ghana  http://ugspace.ug.edu.gh
improved trade facilitation in the region. Kenya has adopted a system called Simba, Tanzania 
uses the Tanzania Customs Integrated System, and Rwanda and Uganda made a shift from 
ASYCUDA++ to ASYCUDA World, though Burundi still uses the ASYCUDA++ (Chimilila 
et al., 2014).  In addition, the implementation of the electronic single window has allowed for 
a greater integration of data from many sources in the region.  
Just like SADC, EAC passed an OSBP Bill in 2013 to reduce delays at the border, smuggling 
activities, improve processing of declaration time and hence improve revenue collection and 
management. Though the OSBP is still a work in progress, there are many similar initiatives 
taking place in the region. The impact of OSBPs in the region is being assessed now, although 
research by Tyson (2015) indicates that at Busia border between Kenya and Uganda, trade 
facilitation improved the economic livelihoods of traders (both formal and informal) through 
the expanded market opportunities brought about by the improved crossing procedures 
Economic Community of Central African States (ECCAS) 
ECCAS aims to promote regional economic cooperation among its member states in Central 
Africa3. Created in 1983, the community started formal operations in 1985. It became active 
only in 1999 when the Brazzaville treaty recognizing it as a REC within the African Economic 
Community was signed. 
Among the various RECs in Africa, progress in the functioning of ECCAS has lagged 
considerably behind for a host of reasons, including the series of conflicts in the Great Lakes 
region; weak financial commitments of member states to support the common cause of the 
community; overlapping membership of member states in other regional blocs, which also 
makes it difficult for member states to fully honour their financial commitments; weak 
commitments to regional policy partly because of overlapping membership in other regional 
                                                          
3 “The 10 member countries of ECCAS are Angola, Burundi, Cameroon, Central African Republic, Chad, 
Congo, DRC, Gabon, Equatorial Guinea and São Tomé & Príncipe. ” 
36 
University of Ghana  http://ugspace.ug.edu.gh
groupings; and a fairly heterogeneous group by way of language and culture (ADB’s Regional 
Integration Assistance Strategy Paper (RIASP) For Central Africa 2005-2009, Pg 1). For 
example, although all members of the Economic and Monetary Community of Central Africa 
region are members of ECCAS, there are members in ECCAS, such as Burundi, that have 
joined the East African Customs Union. 
The ECCAS region has not been able to do much with respect to the level of integration. In 
2004, ECCAS set up a free trade area with the hope of transforming it into a customs union by 
2008 and this has so far been unsuccessful. Implementation of the free trade area in the ECCAS 
region is yet to gain wider cooperation from all member states as bottlenecks still persist in 
some countries. 
Concluding Remarks  
SSA has chalked some successes from the implementation of various trade facilitation 
initiatives on the continent as evidenced in growth of intra-African trade from $49 billion in 
1995 to $130 billion in 2011 as shown in figure 2.8. 
Figure 2.8: Intra-African Trade 1999-2011 
 
Source: UNCTADstat Database, 2013 
 
37 
University of Ghana  http://ugspace.ug.edu.gh
While good progress has generally been made in reducing tariffs on intra-African trade and 
more generally on Africa’s trade with the rest of the world, the persistent presence of NTBs, 
together with inadequacies in soft infrastructure such as inefficient customs procedures and 
trade logistics services as well as poor-quality hard infrastructure such as roads, railways and 
ports continue to act as impediments to trade on the continent. Generally, as figure 2.9 depicts, 
despite all these conscientious efforts by RECs in SSA to facilitate trade among each other, the 
impact on global trade is still small relative to the other regions, suggesting that there is more 
to be done. 
Figure 2.9: Intra-regional exports as a proportion of total exports (%) 
 
Source: Adapted from UNCTAD 2013 
 
A cursory look at SSA performance of trade facilitation using OECD’s trade facilitation 
indicators suggests that the region lags the average performance of groups surveyed by OECD. 
With various trade facilitation indicators used, SSA differs the most in the areas of involvement 
38 
University of Ghana  http://ugspace.ug.edu.gh
of trade community, advance rulings, harmonizing and streamlining of documents, and to a 
lesser extent, appeal procedures, fees and charges as shown in figure 2.10 below. 
Improvements in these areas can potentially produce significant gains. Trade facilitation plays 
an important role in stimulating economic transformation such as raising exports, stimulating 
export diversification, reallocating resources to more productive activities, improving access 
to cheaper and better-quality imported inputs and enabling participation in global value chains. 
Figure 2.10: Sub-Saharan Africa’s trade facilitation performance - OECD indicators 
 
Source: OECD, 2014 
 
Many African regions have begun to formulate regional approaches to trade facilitation. There 
are important examples where particular approaches have worked well. “For example, some 
trade facilitation initiatives have generated positive effects directly, such as the reduction in the 
time taken for freight to cross the Zambia–Zimbabwe border since the introduction of the 
OSBP at Chirundu. Similarly a reduction in border crossing times along the Trans-Kalahari, 
Northern and Maputo Development Corridors (the latter has also benefited from improved road 
39 
University of Ghana  http://ugspace.ug.edu.gh
and rail infrastructure) and the expansion in market opportunities that has improved the 
livelihoods of small traders (both formal and informal) at the Busia OSBP. Other trade 
facilitation initiatives may generate positive effects indirectly, for instance by improving value 
chain and network connectivity in Africa (Shepherd, 2015).”  
40 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER THREE 
LITERATURE REVIEW 
3.0 Introduction 
Some key questions arise such as “after all the efforts and focus on ensuring LDCs and 
developing countries benefit from the gains of world trade - investments, aid for trade amongst 
others, why has the situation not improved for SSA as a region even though there has been 
improvement for different countries within SSA (author)?” In fact, on the average, the export 
and import costs have been worsening for the region as evidenced in the figure 3.1 below.  
Figure 3.1: Average SSA Export and Import Costs Trend 
 $3,000
$2,573 $2,874 
$2,352 
 $2,000 $2,156 
$1,915 $1,998 
 $1,000
 $-
2010 2012 2014
Cost to export (US$ per container) Cost to import (US$ per container)
Source: Doing Business Report, 2015 
 
This chapter will provide both the theoretical and empirical review, evidence of trade 
facilitation and its related problems in SSA. 
3.1 Theoretical Review 
In this section, the conceptual framework of trade facilitation is provided. It offers an 
understanding of the economic effects of trade facilitation using various concepts, models and 
indicators to measure and improve trade procedures which eventually reduces trade cost. We 
shall also look at the transmission mechanism for trade facilitation initiatives to maximize the 
economic gains of trade.  
 
41 
University of Ghana  http://ugspace.ug.edu.gh
3.1.1 Models of measuring trade facilitation  
The measurement and impact of trade facilitation is made possible by utilizing models from 
the classical to the new models of global value chains of international trade.  
The classical trade model was one of the earlier models that emerged to explain the 
dissimilarities of countries pivoted around variation in productivity (Ricardo, 1817) or factor 
endowments (Heckscher and Ohlin, 1991). Although these models do not explicitly analyze 
trade costs, recent models incorporate trade cost to explain differences in trade between 
countries (Krugman, 1980). A more recent model explains the impact of differences in firm 
productivity and its ability to overcome the fixed trade cost of entering the export market 
(Melitz, 2003). Finally, models have been developed to focus on fragmented productions and 
value chains (Yi, 2010). These models are explained below.  
 
3.1.2 The iceberg partial equilibrium model 
The “iceberg” model was developed by Paul Samuelson to analyze the effect of trade cost 
though the original design was to model transport cost. As depicted in figure 3.3 below, when 
there exists inefficiency in trade procedures, the cost of trade increases considerably. This 
creates a wedge between the price paid by consumers and the price received by producers of a 
good creating a deadweight loss to the society. “Iceberg” was used to capture this loss just like 
the iceberg mass that is melted away as it moves through the ocean (World Bank, 2014). The 
iceberg model shows the value of shipped goods are directly proportional to the trade cost.  
  
42 
University of Ghana  http://ugspace.ug.edu.gh
Figure 3.3: The Iceberg Partial Equilibrium Model 
 
Source: Adapted from World Trade Report, 2015  
 
For this purpose, if we assume that the good is a foreign good, thus imported, then domestic 
demand is given by the marginal benefit curve D while foreign supply is given by the marginal 
cost curve, S. In the initial market equilibrium, trade costs are high, denoted by line AC (δ0). 
These trade costs may be due to both tariffs and non-tariff barriers and because of this 
distortion, domestic consumers pay a higher price of P 0d  while the total quantity imported is 
equal to Q0, far lower than the Q*, quantity at equilibrium. There is therefore and increase in 
both consumer and producer surplus due to trade costs. Welfare is reduced for both consumers 
and suppliers. 
If we assume that that trade procedures are improved by both countries so that trade cost edges 
towards zero, then equilibrium quantity of imports will rise to Q* with domestic (foreign) price 
falling (rising) to P*. At this point B, the price wedge disappears and the utility of both domestic 
consumer and foreign producer is maximized indicated by the trapezoidal areas P 0d ABP* and 
P 0s CBP* respectively. Therefore, trade facilitation becomes a “win-win” outcome which 
43 
University of Ghana  http://ugspace.ug.edu.gh
improves the terms of trade in both countries. However, the gains from trade facilitation will 
be reduced if there are inefficient trade procedures that creates deadweight losses, Dee et al 
(2006). The simple analysis above did not take into consideration the cost of trade facilitation 
reform which, when included equally lowers the gains from trade facilitation.  
3.1.3 The classical general equilibrium models 
Our analysis above was based on a single market (partial equilibrium) without taking into 
consideration the spill-over effect of trade facilitation in other markets. We therefore extend it 
to a general equilibrium setting where more insight is given. 
In the classical models, the gains from trade arise because of differences in productivity or 
endowment of countries. The first reason is attributed to Ricardo (1817) and the latter to 
Heckscher (1949) and Ohlin (1934). These models explain the importance of absolute and 
comparative advantages associated with specialization and international trade. They explain 
that countries should specialize in goods in which they have a comparative advantage relative 
to other countries or produce goods which use more intensively their factors of production. The 
good produced is then exchanged with other trade partners to increase the overall production 
and welfare of society. The classical models provide a reason for inter-industry trade but not 
intra-industry trade. The two basic versions of the classical model are explained below.  
3.1.4 The Ricardian model 
The pattern and motive for trade in this model is based on differences in labour productivity in 
countries. This suggests that when two countries are not engaged in international trade the 
relative price of one good expressed in terms of the other good differs between the countries. 
There is therefore an opportunity for welfare-improving trade between the two countries as the 
world price will lie between the autarky price of the two countries which is determined by the 
relative size of the country and consumption preference (Markusen et al., 1995). 
44 
University of Ghana  http://ugspace.ug.edu.gh
However, as explained above, trade cost caused by inefficient procedures will reduce the 
benefit of international trade. The trade cost will create a wedge between relative price faced 
by the two countries. Even if the countries specialize, consumption and trade will be low, 
reducing welfare. Worst case, if trade cost is very high such that the prices faced by at least 
one country is less favourable compared to the autarky price, there will be no trade. The 
likelihood of this happening depends on the relative sizes of the countries involved. According 
to World Trade Report, (2015) authored by the World Trade Organization, “if one country is 
much larger, then the frictionless world price is already close to the autarky price and hence 
trade ceases for even the smallest of transaction costs.” 
3.1.5 The Heckscher-Ohlin (H-O) model 
The assumption of the H-O model differs from the Ricardian model as the former assumes 
same production functions  in both countries. The model further assumes that there are two 
factors of production in each country - capital and labour, and that the endowments of these 
factors differ in both countries. There are also two industries producing two different varieties 
of a good; for example, an automobile industry and cocoa industry. The automobile sector is 
capital intensive whereas the cocoa sector utilizes labour more intensively.  
Pre-trade, the price ratios of the two countries will vary due to the difference in factor 
endowments. A labour-abundant country will experience a higher price of cocoa relative to 
automobiles and the reverse will be true  for a capital-abundant country. If the countries open 
up to trade, with zero trade cost, production and export of the commodity that uses more 
intensively the abundant factor will increase. However, in the H-O model, complete 
specialization is not likely. The countries will trade at an international price lying between the 
autarky prices of both countries. Another important outcome of free trade is a convergence of 
factor prices in the two countries (factor price equalization). In the real world, trade cost is not 
45 
University of Ghana  http://ugspace.ug.edu.gh
zero and it will reduce the gains from trade compared to a frictionless world. Also, factor price 
equalization becomes impossible.  
Although the two classical models differ in terms of tenets, the impact of trade cost is similar 
and works through the same mechanism. The existence of poor trade procedures creates a 
wedge between the price ratios of the two trading economies. These price ratios edge closer to 
the initial autarky price when there is inefficient trade procedures, thus reducing the gains from 
specialization and trade. This lowers the consumption bundles in both countries and economic 
welfare drops as a result.  
Trade facilitation therefore plays a critical role in improving the economic welfare. The H-O 
model has shown that reducing the trade cost through improved trade facilitation will increase 
real income of the abundant factor of production leading to greater specialization in the 
commodity sector that uses intensively the abundant factor and consequently maximize the 
gains of international trade. 
3.1.6 The “New Trade Theory” – monopolistic competition 
Krugman (1979;1980) developed the New Trade Theory to explain trade within similar 
industries. This is important because a huge chunk of global trade is intra-industry in nature 
(see World Trade Report, 2015) whilst a small fraction is inter-industry in nature. Krugman 
(1979) made a few crucial assumptions such as consumers love variety, market made up of 
firms selling different varieties of goods and increasing returns to scale.  
The theory posits that for small countries, the adverse impact of trade cost can be 
disproportionate (Krugman, 1980). Characteristically, small developing countries have small 
manufacturing sectors but large agriculture sectors operating under constant returns to scale. 
On the other hand, big developed economies have large manufacturing sectors typified by 
increasing returns to scale and a small agriculture sector. The features of these two economics 
imply that trade cost will lead to less trade characterized by what is called the “home market 
46 
University of Ghana  http://ugspace.ug.edu.gh
effect” where a disproportionate reallocation of manufacturing goes to the big developed 
country, leaving the small developing economies to concentrate in the agricultural sector.  
Krugman (1980) explained this result using the obvious tension between the consumer’s 
preference for variety in the consumption bundle and the concept of increasing returns to scale. 
Consensus on the theory states that due to the love for variety, when countries open up to trade 
with zero trade cost, consumers from big developed economies will certainly buy both 
domestic and foreign manufactured goods. Holding all factors constant, the preference for 
variety will lead to more trade. Manufacturing companies in big developed countries will 
benefit from scale economies when they experience increasing returns. All factors being equal, 
consumers in this developed economy will prefer to buy lower-cost domestic varieties than 
foreign varieties which comes at a higher cost. 
The concept of trade facilitation therefore becomes useful here. It is noted that poor trade 
procedures will lead to higher cost of trade which will r increase cost of imports by consumers 
of big developed countries. Thus, consumers in these countries will prefer their domestic 
variety causing a distortion in trade between the big and small economies. Because of the shift 
in demand to the domestic manufactured goods, the already powerful big developed country 
will expand even more, further shrinking the similar industry in the smaller economy. This 
theory proposes that to diversify the economies of small developing countries, they must first 
be concerned about reducing trade costs, which will in effect lower the incentives for 
manufacturing to concentrate in the biggest markets. 
  
47 
University of Ghana  http://ugspace.ug.edu.gh
3.1.7 The “New Trade” Theory – heterogeneous firms 
The New Trade theory offers a different perspective of the classical trade theories by shifting 
the focus from countries to firms or heterogenous firms (Melitz, 2003). These models were 
because of many empirical studies that showed outstanding differences among firms with 
respect to productivity, size and their involvement in trade (Bernard et al., 2007).  
The studies reveal that the number of exporting firms is very small and that a larger proportion 
rather trade in the domestic market. Backing this revelation is the difference in firm 
productivity. A lot more productive firms have a competitive edge over low productivity firms 
in the domestic market making the most productive ones more likely to enter the export market 
to compete. The model provides two productivity thresholds – 1) the minimum productivity 
level required for a firm to remain in operation and 2) the level needed for a firm to participate 
in the export market.  
Obviously, a decline in trade cost will cause an increase in a county’s exports. However, the 
literature on this model shows that there is the need to make a distinction between variable cost 
and fixed cost as they have different consequences for export (Chaney, 2006). Fixed cost is the 
cost that is incurred autonomously of export whilst the variable cost is the cost incurred on each 
unit of export. Trade facilitation will reduce both the variable and fixed cost, thus enabling 
exporters to capture a greater portion of the export market. A fall in the variable cost affects 
both intensive (average firm exports) and extensive margins (number of firms) of trade whilst 
a reduction of fixed cost only influences the extensive margin of trade (see World Trade Report, 
2015).  
The analysis implies that trade will expand along both extensive and intensive margins if trade 
facilitation causes a fall in both variable and fixed costs. Firms that are already engaged in trade 
will have the propensity to increase the volume of their exports. Also, firms that left the foreign 
48 
University of Ghana  http://ugspace.ug.edu.gh
market due to their inability to compete, will find it possible to re-enter and start exporting once 
again. However, these new entrants may be smaller and less productive than the existing firms. 
3.1.8 The supply chain models 
While the classical trade theory argues that final goods are produced exclusively in a single 
country, the supply chain models identify that the components that make up these complex 
final goods are produced in many different countries and hence trade costs are magnified (Yi, 
2010). Because semi-finished goods traverse the national borders multiple times during the 
various stages of the value chain, trade cost is amplified. At any stage, trade cost is incurred 
and must be paid out of the share of value added in the cost of production. 
The presence of this “magnification effect” suggests that inefficient trade procedures which 
increases trade cost have a huge adverse effect on the global value chain-related trade than 
those involving only final goods. This therefore implies that the scope for supply chain trade 
will be inversely related to the trade cost. An increase in trade cost, reduces the scope of supply 
chain trade. In a polar case, if the trade cost is too high, the supply chain trade will collapse 
and only final goods will be traded. This implies that trade facilitation is imperative to the 
continuity and viability of the global value chains, giving room for countries to specialize in 
those stages of production where they have comparative advantage. Any attempt to reduce 
trade cost such as TFA, will amplify the trade among countries.  
Baldwin and Venables (2013) examine a more complex production arrangement in the global 
value chain by making a distinction between “snakes” (where the production process is in a 
sequential order with each operation adding a unique value) and “spiders” (where at every 
assembly stage, there is a combination of intermediate inputs).  
These differences in structure only suggests that trade facilitation will have different and more 
complex consequences on the global value chain depending on the given structure. Firms are 
therefore confronted with trade-offs between reducing their production costs by setting up 
49 
University of Ghana  http://ugspace.ug.edu.gh
production plants in different countries or to limit the trade cost by keeping production in one 
country.  
3.2 Despite the trade costs, there are enormous gains from trade  
In as much as trade costs exists, there are tremendous benefits accrued to countries engaged in 
trade. From the academic literature, the gains from trade arise from  two main sources - static 
and dynamic gains (see Figure 3.4). Dynamic gains – gains to the growth rate of living 
standards or real income. This is important but empirically very complex and difficult to 
ascertain. Gains from trade are normally from improvements in standard of living measured by 
aggregate consumption driven by real income and prices which normally ignores the 
distributional effects from trade. The consensus is that there are more winners than losers. 
Figure 3.4: Sources of gains from trade 
 
Source: Adapted from Professor Ralph Ossa’s trade dialogues lecture, 2016 
 
Dynamic gains involve three key sources first of which is international knowledge spillovers 
which occurs when nations have access to knowledge and goods embodying superior 
technology which may impact domestically when they open up to trade. Next are scale effects 
for example from R&D, which is a huge fixed cost activity. Due to access to a huge market 
50 
University of Ghana  http://ugspace.ug.edu.gh
because of opening up to trade, these fixed costs may lead to scale effects which then justifies 
the needed investment. However, opening up also brings the risk of competition. The net effect 
is dependent on which effect dominates the other - scale vs competition. Lastly, dynamic gains 
arise via general equilibrium effects emanating from changes in relative goods or relative factor 
prices which makes one country focus more on one activity than the other. The impact of the 
dynamic effects empirically is not consistent, for example China and South Korea have 
benefited from these gains but same cannot be said for other countries (Professor Ralph Ossa’s 
trade dialogues lecture, 2016). 
Static gains from trade arise when the level of real income increase one-time. They mainly 
come from specialization where division of labor increases productivity with various countries 
specializing in certain activities. A country therefore exports their goods and use the proceeds 
to import the other goods needed domestically. These benefits derive from comparative 
advantage because national endowments are different. If the fixed costs are high, it makes sense 
for one country to produce a lot more of the goods that leverage their comparative advantage 
in terms of relative endowment benefiting from increasing returns. 
The other group is the “New” gains from trade. Krugman (1980) and Melitz (2003) argues that 
opening up to trade enables access to new foreign varieties hence individuals benefit from the 
consumption of different varieties.  Similarly, free trade introduces competition which leads to 
the reallocation of resources to more productive firms which then leads to productivity gains 
with the weaker companies being driven out of the market. 
3.3 Measuring trade facilitation  
The existing literature captures numerous trade facilitation indicators (TFIs). Orliac (2012) 
finds more than 12 of these indicators signifying the complex nature and the crucial role played 
by trade facilitation. This study will look at the frequently used indicators in the economics 
literature such as the Doing Business “(DB) indicators mainly those related to trading across 
51 
University of Ghana  http://ugspace.ug.edu.gh
borders, the Logistics Performance Index (LPI) by the World Bank; the OECD Trade 
Facilitation Indicators (TFIs); and the World Economic Forum’s Enabling Trade Index (ETI). ” 
It is important for policy makers to distinguish between indicators that measure policy inputs 
and those that track the outcomes of policy. Since they are complementary, these indicators aid 
in tracking the outcomes of trade facilitation as well as identifying appropriate policy measures 
or drivers that will impact a desired outcome. It is worth noting that these indicators are not 
perfect but give relatively better insights into trade facilitation. The “DB indicators” mostly 
measure policy outcomes, the OECD TFIs focuses on inputs to policy and the LPIs and ETIs 
are a combination of the two. The LPI is chosen out of the lot because it combines both inputs 
to policy and its outcomes as well as its simplicity. 
  
52 
University of Ghana  http://ugspace.ug.edu.gh
Table 3.1: List of indicators and indexes 
“ 
 Indicators Index 
1. Dealing with construction and permits 
2. Starting business 
3. Getting electricity 
4. Getting credit There are two major indices 
Doing 5. Registering property  
Business 6. Protecting Minority investors • Distance to Frontier 
(DB) 7. Trading across borders  
8. Paying taxes • Ease of Doing Business. 
9. Enforcing contracts 
10. Selling to the government 
11. Resolving insolvency 
1. Infrastructure  
The Principal component analysis is 
2. Customs  
Logistics used to construct the aggregate LPI 
3. Quality of logistics services; 
Performance and the weighted average of the six 
4. Tracking and tracing 
Index (LPI) indicators in calculated as the Overall 5. Ease of arranging shipments 
Logistics Performance Index 
6. Timeliness. 
1. Availability of information.  
2. Document formalities 
3. Automation formalities  
4. Procedure formalities 
5. Internal border agency cooperation 
6. External border agency cooperation 
OECD Trade 7. Trade community involvement The 16 indicators are made up of 97 
Facilitation 8. Advance ruling  variables which have been normalized 
Indicators 9. Appeal procedures using “multiple binary” scoring 
(TFIs) 10. Charges and fees system (Moïsé et al. (2011) 
11. Consularization  
12. Impartiality in governance  
13. Transit fees and charges 
14. Transit guarantees 
15. Transit formalities 
16. Transit agreements and cooperation 
1. Foreign market access 
2. Domestic market access 
3. Efficiency and transparency of border 
Fifty-six indicators classified into 
administration; 
Enabling seven pillars. 
4. Availability and quality of transport 
Trading  
services; 
Index (ETI) ETI is computed as the unweighted 5. Availability and quality of transport 
average of the various indicators. 
infrastructure 
6. Operating environment. 
7. Availability and use of ICTs; 
” 
Source: World Trade Report (2015)  
53 
University of Ghana  http://ugspace.ug.edu.gh
3.4 Empirical Review 
The assessment of trade facilitation on trade flows and exports has sparked interest among 
economists and policy makers making the studies on this topic quite diverse and interesting. 
This section offers a review of existing studies with emphasis on developing countries, 
especially SSA.  
First, many papers that focused on the analysis of trade facilitation’s impact on export 
performance agree that trade facilitation increases trade volumes in developing countries. To 
find out the benefit of trade facilitation in developing countries, Zaki (2011) used the dynamic 
CGE model (Mirage) and data on 19 regions, 21 sectors over 2004-2008. The findings of the 
study reveal that there are many gains in trade facilitation for developing countries especially 
SSA as compared to developed countries. In SSA trade costs, if reduced by half cause an 
improvement in the terms of trade by 2.3%. Also, trade facilitation causes an increase in export 
by 22.2%. The study further shows that due to an expansion in the manufacturing sector, 
employment increases by 2.7% and the net welfare gain can increase by 4.7%.  
Similarly, Mevel and Karingi (2012) adopted the dynamic CGE model (Mirage) and found a 
complementarity between trade facilitation and trade creation in the Continental Free Trade 
Area. The results showed that the time spent on goods at African ports is reduced by 50%, and 
the efficiency of customs procedures doubles. With the creation of the free trade area, the share 
of intra-trade in African countries will increase by more than twice between 2012 and 2022. 
Also, revenue and income loss as a result of removal of tariff barriers will be offset by the gains 
from trade facilitation 
By using a pooled, cross-country, annual time series data for the period 1995-2004 for twenty 
African countries, Akinkugbe (2009) also finds that trade facilitation could have a significant 
impact on goods exported from Africa. His results showed that improvements in infrastructure, 
institutions and usage of technology would be trade enhancing; whereas regulatory barriers and 
54 
University of Ghana  http://ugspace.ug.edu.gh
the perception of corruption may divert trade. Iwanow, et al. (2009), using panel data analysis 
for 124 developed and developing countries find that trade facilitation could yield higher 
returns on manufacturing goods and export in Africa than the rest of the world. They explain 
that a 10% increase in trade facilitation will cause African export to rise by about 17%.  
Mann et al. (2005) also made use of panel data analysis. They combined data from 75 countries 
of which 3 were African countries spanning 2001-2002 and concluded that trade facilitation 
had a trade enhancing effect. The three African countries were found to have a small export 
gain relative to import gains. They explained that this could be due to Africa’s limited access 
to the OECD market and lack of integration to the global value chain therefore trade facilitation 
therefore becomes important.  
There is a general consensus that quality infrastructure development in the region will 
remarkably reduce the cost of trade and hence increase trade volumes. A study by Limao and 
Venables (2001) which sought to find the impact of infrastructure on trade facilitation and 
export adopted the gravity framework. They found that 40% of transport cost was from 
infrastructure deficit in African countries leading to the conclusion that improvements in 
infrastructure will therefore increase trade across countries. Behar and Venables (2011) 
estimated that a country fixing its infrastructure from the median to the 75 percentile will 
increase trade volumes significantly -by about 68%. However, a country whose infrastructure 
falls below the median will reduce trade by 28%. 
Portugal-Perez & Wilson (2012) investigated the impact of infrastructure quality on exports 
from developing countries. They used panel data of 100 countries covering the period 2004-
2007. The analysis adopted the HMR (Heckman, Helpman, Melitz and Rubinstein) 
international trade model incorporating firm heterogeneity. They included hard and soft 
infrastructure of an exporter as part of trade costs.  Their outcome indicated that trade 
facilitation increased the export performance of developing countries. Infrastructure 
55 
University of Ghana  http://ugspace.ug.edu.gh
improvement and efficient trade procedures should put Africa on a higher trajectory of 
economic development by increasing trade flows and export. For instance, they found that for 
Chad, if investment were channeled to the improvement of infrastructure quality halfway to 
the level of South-Africa, their expansion in export would equal a 24% reduction in tariffs in 
importing countries. 
Again, Portugal-Perez and Wilson (2011) categorized infrastructure into “soft” and “hard” and 
investigated the impact they had on trade volumes and export performance. Using four new 
indicators covering more than 100 countries between the periods 2004–2007, this study found 
out that trade facilitation improved export performance in developing countries. This is 
predominantly valid with investment in physical infrastructure. Contrary to most findings 
however, the effect of physical infrastructure on export performance appeared to become more 
relevant as countries become richer. Their indicators also proved a complementarity between 
the soft and hard infrastructure.  
Using an augmented gravity model and data from 121 countries made up of 20 developed and 
101 developing countries, Turkson (2011) found out that affordability and ease of shipping had 
the greatest impact whereas timeliness had the least impact on bilateral exports with domestic 
logistics costs being insignificant. On primary commodities, logistics cost for destination was 
relevant. For low-income countries, timeliness and efficiency of customs were equally 
important. 
The cost of trade has also become a topical issue as it is a major obstacle to developing 
economies’ participation in the global trade system. With regards to the importance of 
eliminating the huge cost of trade in developing countries, Seck (2014) analyzed the extent to 
which various components of the trade cost in Africa may have contributed to shape trade.  
Simple averages and factor analysis were conducted to aggregate the various trade facilitation 
indicators into four main groups: regulatory environment and ICT, physical infrastructure, 
56 
University of Ghana  http://ugspace.ug.edu.gh
border efficiency and the LPI. These indicators were then used in the gravity framework 
adopted by Seck (2014) after accounting for the several theoretical and empirical issues in 
relation to non-agricultural and agricultural commodity trade among countries. The result 
provided evidence that trade facilitation would yield varying gains from trade based on a 
country’s trade cost landscape, commodity type, and trade partners.  These results provide a 
strong basis for targeted trade facilitation reforms that should put Africa in a better position in 
the international market.  
It has also been found that one of the key contribution to low trade cost and high trade flows in 
most developing countries is Aid-for-Trade (AfT). According to Alaba et al (2006) AfT plays 
an important role in achieving unhindered trade flows from EU to ECOWAS. A study by 
Massimiliano et al. (2011) used panel data covering 100 developing countries over 2002-2007 
to investigate the impact of AfT on trade cost. The results of their analysis show that AfT 
improves trade facilitation and reduces the cost and time to trade. According to them a 100% 
increase in AfT cuts the cost of imports by 5% and the cost of exports by 4.7%.  
Helble, Mann & Wilson (2009)., (2009) after using the gravity model and considering 167 
exporters as well as 172 importers from 1990-2005 also find that trade flows are significantly 
related to AfT for trade facilitation. The statistical evidence points that a 1% increase in AfT 
facilitation, generates global trade growth of about $818 million. 
Other papers have analysed the impact of various trade measures of trade facilitation to ensure 
high trade flows. Decreux, and Fontagne (2006) find that after 2020, a successful trade 
facilitation agenda would be equal to multiplying ODA to SSA countries by two. Trade 
facilitation generates about 7.2% increase in world trade, and welfare gains of about 1% of 
world GDP (approximately US$330 billion). Their results also show that the lion’s share of the 
gains of trade facilitation will be taking by the EU (1/3) while SSA will gain just 6%, an 
equivalent to almost US$20 billion. Also, in SSA, the real returns to unskilled labour would 
57 
University of Ghana  http://ugspace.ug.edu.gh
rise by 9.2%. Similarly, IFPRI (2010) finds that measures to improve trade facilitation would 
cut trade costs by half. The inclusion of trade facilitation in regional FTAs would expand 
exports and subsequently cause national income to rise as compared to a regional FTA with no 
trade facilitation. 
58 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER FOUR 
METHODOLOGY 
4.1 Introduction 
This chapter presents the methodological techniques and data sources used to empirically 
analyze the impact of trade facilitation on measures on bilateral exports in SSA. It explains the 
theoretical framework supporting the research, the model to be estimated, variables description, 
estimation technique and other problem-solving tests. It also gives the underlying assumptions 
and reasons for the choice of the gravity model.   
4.2 Data Sources 
The study uses mainly secondary data from the various sources shown below: 
• Export data from IMF Direction of Trade Statistics (DOTS) 
• Gravity data from Centre d'Études Prospectives et d'Informations Internationales 
(CEPII)4 
• Logistics Performance Indicators (LPI) data from the World Bank Doing Business 
Database 
• Trading Across Borders data from World Bank Doing Business database 
 
Data for 189 exporting countries and 194 importing countries spanning a 9-year period between 
2007 and 2015 were collected and used for this work with the number of observations totaling 
278,541. This included 64,507 observations on 47 Sub-Saharan African countries exporting to 
194 countries and 67,358 observations representing imports from 189 countries to 47 sub-
Saharan African countries.   
                                                          
4 4 CEPII is a French research center in international economics which produces studies, research, databases 
and analyses on the world economy and its evolution. It was founded in 1978 and is part of the network 
coordinated by the Economic Policy Planning for the Prime Minister. 
59 
University of Ghana  http://ugspace.ug.edu.gh
4.3 Methodology 
As in most trade related literature, our work takes its source from the workhorse of 
international trade - the Jan Tinbergen’s gravity model. We shall attempt to augment it with 
the logistics performance indicators to find answers to our research questions.  
4.3.1 Theoretical underpinnings 
Jan Tinbergen (1962), in his inspiring academic work demonstrated that the quantum of 
bilateral trade flows between any two nations could be estimated by what is known as the 
“gravity equation”, a law analogous to Newton’s theory of gravity. Similar to the planet world 
where planets are attracted in direct proportion to their proximity and sizes, nations also trade 
in proportion to the distances between them and their relative sizes (GDPs). This “gravity 
equation” was initially thought of as a mere stable illustration of an empirically stable 
relationship between the size of economies, their proximity and the quantum of trade between 
them. 
At that time, the key models of international trade were the Ricardian model, which relied 
heavily on the technology differences between nations to explain the patterns of trade, and the 
Heckscher-Ohlin (H-O) model which depended on the dissimilarities in factor endowments 
among nations. The assumption was that the Ricardian and H-O models were unable to provide 
the foundation for Tinbergen’s gravity model. For example, in the H-O model, country size has 
almost no impact on the structure of trade flows. 
The immense stability of the gravity equation and its supremacy in explaining bilateral trade 
flows triggered the search for its theoretical underpinnings. While empirical analysis preceded 
theory, it was widely accepted that most, if not all of trade models will require gravity for it to 
work. The first person who attempted to offer a theoretical basis of the gravity model was 
Anderson (1979). His context was based on the Armington assumption - where goods were 
differentiated by country of origin, and where consumer’s preference was defined over the 
60 
University of Ghana  http://ugspace.ug.edu.gh
spectrum of all the differentiated products. This assumption would mean some of every good 
from every country will be consumed by a nation at any price. All nations trade in all kinds of 
goods and in equilibrium, the national income for any particular nation is the sum total of both 
domestic demand and consumption of foreign goods by each country. This explains why large 
developed economies export and import more. To account for trade costs, it is assumed that 
only a portion of the good shipped arrives at the destination are accounted for because  
transportation related costs will reduce trade flows at least if imports are measured at the CIF 
value. 
Subsequently, Bergstrand (1985 and 1989) showed that a gravity model is a direct result of a 
model of trade based on monopolistic competition developed by Krugman (1980) where due 
to consumers’ love for variety, similar nations trade in differentiated goods. Monopolistic 
competition models overcome differentiation by production location assumption (Armington 
model) as firm location is determined endogenously and nations specialize in the production 
of different types of goods. A gravity model can also be deduced from the typical factor-
proportions explanation of trade (Deardorff, 1998). Kortum and Eaton (2002) obtained an 
equation similar to the gravity model from the Ricardian model, Helpman & Melitz . (2008) 
and Chaney (2008), controlling for firm heterogeneity, derived the gravity model as a 
theoretical model of international trade in differentiated goods. 
The Gravity model can be explained generally in its multiplicative form as: 
𝑋𝑖𝑗  =  G 𝑆𝑖 𝑀𝑗  Ф𝑖𝑗 ε𝑖𝑗          [1] 
where 
• Xij represents the dollar value of exports from country i to country j,  
• Mj represents all factors that are import specific and make up the entire importer’s 
demand (for instance GDP of the importing country) 
61 
University of Ghana  http://ugspace.ug.edu.gh
• Si indicate all factors that are export specific (for example the GDP of the exporting 
country) which make up the entire amount exporters are willing to supply.  
• G represents a variable that is independent of country i or country j  
• Фij represents the ease of exporter i to enter market j (all factors that pose resistance to 
trade hence generates trade costs, such as distance) 
• εij denotes the error term assumed to be statistically independent of all the explanatory 
variables [E(εij/ Mj, Si, Фij)=1] 
 
Anderson and van Wincoop (2003) argue and prove that controlling for comparative trade costs 
is extremely important for a well-specified gravity model. They correct the error made in 
gravity model estimations by studies that used Si and Mj in equation [1] without controlling for 
multilateral resistance terms (MRT), a fixed effect. Their theoretical outcomes show that 
relative trade cost determines bilateral trade - the propensity of nation j to import from nation 
i is based on nation j’s trade cost toward i. This is also dependent on the general “resistance” 
to import and average “resistance” confronted by the exporting country. According to 
Anderson and Van Wincoop (2003), this is not necessarily the absolute cost between the two 
countries. However, the motive behind the inclusion of these multilateral resistance (MRT) 
cost is that, all other things remaining the same, two economies surrounded by other big trading 
partners will trade much less between themselves compared to if they were surrounded by 
oceans or mountains or vast stretches of desert. This implies that the absence of MRT in the 
specification of the gravity model causes the model to suffer from omitted variable bias.  
Anderson and van Wincoop (2003) therefore show that in a world consisting of N countries 
consuming a variety of differentiated goods by the country of origin, a well-specified 
theoretically funded gravity equation takes the form: 
1−𝜎
𝑌
𝑋 = 𝑖
 𝑌𝑗 𝑇[ 𝑖𝑗𝑖𝑗 𝑤 ]            [2] 𝑌 𝑃𝑖𝑃𝑗
62 
University of Ghana  http://ugspace.ug.edu.gh
where: 
i and j represents importing or exporting countries (from country 1 to N), Yi and Yj represents 
countries i and j’s GDP respectively, Yw is the world’s GDP, Tij is the cost incurred in country 
j in importing goods from country i, σ is the elasticity of substitution, Pi and Pj are multilateral 
resistance terms (MRTs) which are not directly observable, and they capture exporter and 
importer ease of market access. This is expected to be low if a nation is far apart from the world 
market. Remoteness and proximity are determined by factors such as physical distance from 
international market, tariff barriers and other trade costs. For example, neighbours such as 
Netherlands and Belgium bordered by “big countries”, Germany and France respectively, will 
trade less between themselves than if they were surrounded by oceans (New Zealand and 
Australia) or by vast stretches of mountains and/or deserts (such as the Uzbekistan and 
Kazakhstan). McCallum (1995) estimated trade flows between the USA and Canada using the 
traditional gravity model [1] two Ф variables (distance and a dummy variable = 1 if the two 
regions are located in the same country and dummy variable = 0 if not). He found out that trade 
between two provinces in Canada were 22 times more than trade between a province and a 
state suggesting that there were huge trade costs across the Canada-US border.  
4.3.2 Estimation Methods 
The standard procedure for estimating the gravity equation is to transform gravity equations 
[1] and [2] to their log-linear forms so that estimation can be easily done using various 
estimation techniques. Transforming [1] and [2] into a stochastic logarithmic form gives: 
ln 𝑋𝑖𝑗 =  ln 𝐺 +  ln 𝑆𝑖 + ln 𝑀𝑗 +  ln Ф𝑖𝑗  +  𝜀𝑖𝑗      [3] 
and in the case of Andersen and Wincoop’s model, which is our focus for this work, including 
the MTRs transforms equation 2 into: 
ln 𝑋𝑖𝑗 = 𝛽
∗
0 + 𝛽
𝐾
1 ln 𝑌𝑖 + 𝛽2 ln 𝑌𝑗 + ∑𝑘=1 𝛿𝑘 ln 𝑍
𝑘
𝑖𝑗 + (1 − 𝜎) ln 𝑃𝑖 + (1 − 𝜎) ln 𝑃𝑗 +  𝜀𝑖𝑗 [4] 
where: 
63 
University of Ghana  http://ugspace.ug.edu.gh
𝛽∗0 = 𝑙𝑛β0 is a constant, β1 and β2 are elasticities of trade with respect to GDPs of countries i 
and j respectively, Pi and Pj are the MTRs, capturing the resistance of country i and country j 
to trade with all regions and εij is the error term. From [2], trade costs, Tij is usually captured 
through a couple of K variables, 𝑍𝑘𝑖𝑗, m=1...K. Generally, empirical studies use bilateral 
distance (dij) as a proxy for trade costs. Additionally, dummies for common borders 
(contiguity), landlocked countries and islands are used as well. These are often used to support 
the hypotheses that transportation costs rise with distance - higher for landlocked countries and 
islands but are lower for contiguous countries. For information costs, adjacency, common 
language, or additional cultural characteristics like colonial history are preferred. There is a 
high probability that search costs will be lower for trade between countries who are familiar 
with each other’s competitiveness, delivery and business practices. Contiguous firms’ 
countries, common language countries or other countries with similar cultural characteristics 
are likely to know and understand each other’s business environment better than firms 
operating in different contexts. In this case, most firms will more than often look for partners 
(suppliers or customers) in countries with similar business practices. Largely, tariff barriers are 
included as dummies for the existence of RTAs. However, because of lack of data over time, 
most studies prefer information on bilateral tariffs.  
Trade costs, Tij can therefore be expressed as: 
 
𝐾 𝑘
T𝑖𝑗  =   𝑑
𝛿1
𝑖𝑗 ∗ (𝑒
∑𝑘=1 𝛿𝑘 ln 𝑍𝑖𝑗)                   [5a] 
 
T  =   𝑑 𝛿1 ∗ (𝑒[𝛿2𝑐𝑜𝑛𝑡𝑖𝑗+ 𝛿3𝑙𝑎𝑛𝑔𝑖𝑗 + 𝛿4𝑐𝑐𝑜𝑙𝑖𝑗+ 𝛿5𝑐𝑜𝑙𝑖𝑗 + 𝛿6𝑙𝑎𝑛𝑑𝑙𝑜𝑐𝑘𝑖𝑗+ 𝛿7𝑅𝑇𝐴𝑖𝑗 +⋯ ]𝑖𝑗 𝑖𝑗              [5b] 
where,  
dij represents bilateral distance between country i and country j,  
64 
University of Ghana  http://ugspace.ug.edu.gh
contij represents a dummy variable for common border signifying whether the two countries 
have a common border, langij, a dummy variable representing whether the two countries speak 
a common language or not,  ccolij, a dummy variable denoting whether the country pairs had a 
common colonizer, colij, a dummy variable denoting whether one was a colony of the other at 
some point in time, landlockij, a dummy variable representing whether one of the two is a 
landlocked country (including when both countries are landlocked),  and RTAij represents 
whether the two countries are members of a regional trade agreement respectively. These 
variables have been found to be significant determinants of bilateral trade. 
 
Substituting equation [5b] in [4] gives 
ln 𝑋𝑖𝑗 =  β 0 + 𝛽1ln 𝑌𝑖 +  𝛽2 ln 𝑌𝑗 + 𝛽4 ln P𝑖   +  𝛽5 ln P𝑗 + 𝛿1 ln 𝑑𝑖𝑗 + 𝛿2𝑐𝑜𝑛𝑡𝑖𝑗 +
 𝛿3𝑙𝑎𝑛𝑔𝑖𝑗  +  𝛿4𝑐𝑐𝑜𝑙𝑖𝑗 +  𝛿5𝑐𝑜𝑙𝑖𝑗  + 𝛿6𝑙𝑎𝑛𝑑𝑙𝑜𝑐𝑘𝑖𝑗 +  𝛿7𝑅𝑇𝐴𝑖𝑗 + 𝜀𝑖𝑗   [6] 
Recent studies that have been undertaken using the gravity equation have augmented it with 
various measures of distance and country-specific variables, and other measures of trade 
facilitation, infrastructure and logistics (Perez & Wilson (2008, 2012), Turkson (2011, 2012), 
Limao & Venables (1999), Li & Wilson (2009). 
4.3.3 The multilateral trade resistance (Remoteness) problem 
Andersen and Wincoop’s gravity equation [4] comes with a peculiar challenge – the 
multilateral resistance terms Pi and Pj cannot be directly observed. In order to go around this 
obstacle, Anderson and van Wincoop (2003) used as a proxy, iterative methods to build 
estimates of the price-increasing effects of MTR barriers. This is not often preferred as it 
demands that we use the non-linear least square (NLS) method for estimating. Rather, the 
"remoteness" variable, REM is a better and simpler alternative is preferred.  
The remote variables, REMi and REMj for the exporting and importing countries can be defined 
as: 
65 
University of Ghana  http://ugspace.ug.edu.gh
𝑑𝑖𝑠𝑡𝑖𝑗 𝑑𝑖𝑠𝑡𝑖𝑗
𝑅𝐸𝑀𝑖 =  ∑        𝑎𝑛𝑑          𝑅𝐸𝑀𝑗 =  ∑                                            [7] 𝐺𝐷𝑃𝑗 𝐺𝐷𝑃⁄ 𝑖𝑗 𝑤 𝑖 ⁄𝐺𝐷𝑃 𝐺𝐷𝑃𝑤
Representing components that measure a country’s average weighted distance from its trading 
partners (Head, 2003), in which weights are the associate countries’ shares of world GDP 
(denoted through GDPj/GDPw). There are criticisms normally made from the use of this 
approach: one is that it is not theoretically correct, because the trade barrier that it captures is 
distance (Anderson and van Wincoop, 2003). The other one pertains to the precise 
measurement of distance, as the summation calls for us to specify a precise distance from itself 
(Head and Mayer, 2000 advocates the usage of the square root of country's area multiplied by 
forty percent) 
Substituting [7] into equation [6] becomes  
ln 𝑋𝑖𝑗 =  β 0 + 𝛽1ln 𝑌𝑖 +  𝛽2 ln 𝑌𝑗 + 𝛽4𝑅𝐸𝑀𝑖   +  𝛽5𝑅𝐸𝑀𝑖 + 𝛿1 ln 𝑑𝑖𝑗 + 𝛿2𝑐𝑜𝑛𝑡𝑖𝑗 +
 𝛿3𝑙𝑎𝑛𝑔𝑖𝑗  +  𝛿4𝑐𝑐𝑜𝑙𝑖𝑗 +  𝛿5𝑐𝑜𝑙𝑖𝑗  + 𝛿6𝑙𝑎𝑛𝑑𝑙𝑜𝑐𝑘𝑖𝑗 +  𝛿7𝑅𝑇𝐴𝑖𝑗 + 𝜀𝑖𝑗    [8] 
 
It is extremely important that we account for all other domestic influences affecting trade costs 
that are not captured in the Zij vector also known as the multilateral trade resistance term (MTR) 
which cannot be observed directly (Anderson and van Wincoop, 2003). Rose and van Wincoop, 
2001; Feenstra, 2004; Baldwin and Taglioni, 2006 provide the commonly used method - 
country fixed effects for exporters and importers. 
There are 3 ways to capture this effect. First, is to capture this effect, leverage the use of the 
proxy variables for “remoteness”- REMi and REMj, adding it to the augmented gravity model 
for respective country pairs i and j respectively in line with Bergstrand and Baier (2007). These 
variables are determined by computing the “weighted average of the distance to trading 
partners”, with the proportions of world GDP representing weights as shown above. Second, 
is to use an iterative method to solve the MTR as a function of the observable, Anderson and 
66 
University of Ghana  http://ugspace.ug.edu.gh
van Wincoop (2003) and thirdly, since the MTR is expected to explain the unobserved 
heterogeneity within country pairs, where MTR is fixed, we can use the country fixed effects 
to correct for this bias (Rose and van Wincoop, 2001; Feenstra, 2004; Baldwin and Taglioni, 
2006).  
In this work, I shall proceed to use the third option to correct for this unobserved heterogeneity. 
This implies that our model for this work will be equation [6] above.  
We will use the Hausman test to confirm the choice of model. However, under the fixed effects, 
the time-invariant (TI) variables such as distance, contiguity, colonization, language, amongst 
others are all absorbed by the intercept and their coefficients cannot be interpreted using the 
fixed effects thus biasing the estimates. When we observe that all the TI variables have been 
omitted, we shall take advantage of the Hausman Taylor Estimator to capture the coefficients 
of both the time varying and the time invariant estimators since we are also interested in some 
of the TI variables. 
4.4 Empirical Estimation Model  
Clearly, bilateral trade between countries is not dependent only on trade costs, country’s 
economic sizes and geographical distance. Other factors such as infrastructure, border 
efficiency, technological advancement, and regulation among others may also influence 
bilateral trade (Perez and Wilson, 2012).  
In examining the impact of trade facilitation measures on bilateral exports involving sub-
Saharan countries, this paper estimates a logarithm-transformed, augmented gravity model and 
estimates the impact of various TFIs on bilateral exports over a time period (2007-2015). 
This study follows that of Wilson et al (2002), Djankov (2006) Shepherd and Wilson (2008), 
Hoekman and Nicita (2008); Kien (2009) and Turkson (2012) who employ different variations 
of the augmented gravity models to assess the impact of various trade facilitation and logistics 
quality on bilateral exports using different estimation techniques.  
67 
University of Ghana  http://ugspace.ug.edu.gh
This work will leverage the panel estimation techniques. A panel data is represented as follows:  
Assuming we have T observations on N countries, then we have a total of NxT unique 
observations. Most panel data have time effects and firm heterogeneity that must be 
controlled for else conclusions may be biased. Similarly, heteroscedasticity [𝑽𝒂𝒓(𝜺𝒊) ≠ 𝝈𝟐 ] 
is prevalent in cross-sectional data whereas serial correlation [𝑪𝒐𝒗(𝜺𝒊𝒕,𝜺𝒊𝒔)≠ 𝟎 for all t and s] 
is a characteristic of time series data. A panel is a combination of cross-section and time series 
data so it is almost a certainty that both problems are likely to be present. These will be 
corrected by using the robust standard errors method. This method gives standard errors of 
regression coefficients that are robust to heteroscedasticity and serial correlation 
Similarly, it is possible for there to be persistent differences between countries that are not 
observable hence captured in the error terms (unobserved heterogeneity). The effect of this is 
that the error term, ε would vary more systematically across countries. A key approach to 
resolving this unobserved heterogeneity will be to add an error term that varies between 
countries but constant within countries. This new error term fi, will absorb the unobserved 
heterogeneity and leave our error term vi, randomly distributed. This is preferable as it has the 
advantage of eliminating the bias caused by heterogeneity across countries as it gives us the 
ability to control for the unobserved fixed effects. In a panel country-pair, heterogeneity can 
be controlled for by using country pair fixed effects. 
A panel data is called a random effects (RE) model if the individual heterogeneity term fit  is 
not correlated with the regressors. However, if fit is correlated with the regressors, then the 
model has to be a fixed effect model (FE) and treated as such. RE models are preferable to 
FE models since it is possible to include the time invariant (TI) variables such as distance, 
contiguity, landlocked, language, etc. In the FE model, the TI variables are absorbed by the 
intercept. In this case, we leverage the Hausman Taylor Estimation model. 
 
68 
University of Ghana  http://ugspace.ug.edu.gh
This paper, like Turkson (2012) employs either FE model, RE model or the Hausman Taylor 
model based on the assumptions about the individual effects. 
On endogeneity, Bergstrand and Baier (2003) states that because countries self-select 
themselves into trade agreements, we cannot assume that the free trade agreement variable 
fta_wto on the right-hand side is exogenous. We therefore must remedy this bias. 
  
69 
University of Ghana  http://ugspace.ug.edu.gh
4.4.1 The Model 
ln 𝑒𝑥𝑝𝑜𝑟𝑡𝑠𝑖𝑗𝑡 =  β 0 +  𝛽1ln 𝑥𝑔𝑑𝑝𝑡 +  𝛽2 ln 𝑚𝑔𝑑𝑝𝑡 +  𝛽3 ln distw𝑖𝑗𝑡   +  𝛽4 ln 𝑐𝑜𝑛𝑡𝑖𝑔𝑖𝑗
+ 𝛽5 𝑙𝑎𝑛𝑑𝑙𝑜𝑐𝑘𝑒𝑑_𝑜  + 𝛽6 𝑙𝑎𝑛𝑑𝑙𝑜𝑐𝑘𝑒𝑑_𝑑 + 𝛽7𝑐𝑜𝑚𝑙𝑎𝑛𝑔_𝑜𝑓𝑓𝑖𝑗𝑡
+ 𝛽8𝑐𝑜𝑚𝑐𝑢𝑟𝑖𝑗𝑡 + 𝛽9𝑐𝑜𝑚𝑐𝑜𝑙𝑖𝑗 +  𝛽11𝑓𝑡𝑎_𝑤𝑡𝑜𝑖𝑗𝑡 +  𝛽12 ln xcostexport𝑖𝑗𝑡
+  𝛽13 ln mcostimport𝑖𝑗𝑡 + 𝛽14𝑥𝐿𝑃𝐼𝑖𝑡 +  𝛽15𝑚𝐿𝑃𝐼𝑗𝑡 +   𝛽16ln 𝑥𝑔𝑑𝑝𝑝𝑐𝑡
+  𝛽17ln 𝑚𝑔𝑑𝑝𝑝𝑐𝑡 + 𝜀𝑖𝑗𝑡 
4.4.2 Description of variables and Expected Signs 
Expected 
Variable Description 
Sign 
Exports from country i to country j at time t, dependent variable in 
exportsijt  
millions of constant US dollars 
Gross domestic product of exporting and importing countries at time 
xgdpt & mgdpt t respectively representing the size of both countries in millions of (+) 
constant US dollars 
Per capita gross domestic product of exporting and importing 
xgdppct & 
countries at time t respectively representing the effective demand of (-) 
mgdppct 
both countries in millions of constant US dollars 
Weighted distance between country i and country j at time t in 
distwcesijt  
weighted distance (pop-wt, km) CES distances with theta=-1 
Dummy variable equal 1 if there exists a common border between 
contigij (+) 
country pairs 
Dummy variable equal to 1 if country pairs have common colonial 
comcolijt (+) 
ties at time t 
Dummy variable equal to 1 if county pair share a common official 
comlang_offij (+)  
language 
landlocked_o  Dummy variable equal to 1 if the originating (exporting) country is (-) 
lDanudmlmocyk evda r iable equal to 1 if the destination (importing) country is 
landlocked_d (-) 
landlocked 
Dummy variable equal 1 if country i share a common currency with 
comcurijt (+) 
country j and 0 otherwise. 
Dummy variable equal to 1, if country pair has a trade agreement 
fta_wtoijt (+) 
(FTA/RTA) at time t 
Log of Cost to export a 20ft container from country i to country j at 
xcostexpijt (-) 
time t in constant US dollars 
Cost to import a 20ft container from country i to country j at time t in 
mcostimpijt (-) 
constant US dollars. 
70 
University of Ghana  http://ugspace.ug.edu.gh
An index from (1-5) representing the relevant logistics performance 
indicator (overall LPI, efficiency of customs procedures, ability to 
track and trace consignment, ease of arranging competitively priced 
xLPIit & mLPIjt shipment, competence and quality of logistics services, frequency  
with which shipment reach consignee, quality of trade and transport-
related infrastructure) of exporting country i and importing country j 
respectively at time t 
xlpioverallit & Aggregate LPI of exporting country i and importing country j 
(+) 
mlpioveralljt respectively at time t 
xlpicustomsit & Efficiency of customs procedures in exporting country i and 
(+) 
mlpicustomsjt importing country j respectively at time t 
xlpiatratrit & Ability to track and trace consignment in exporting country i and 
(+) 
mlpiatratrjt importing country j respectively at time t 
xlpieoacpsit & Denotes the ease of arranging competitively priced shipments in the 
(+) 
mlpieoacpsjt exporting country i and importing country j respectively at time t 
xlpicqols  & Competence and quality of logistics services in exporting country in it
mlpicqols  the exporting country i and importing country j respectively at time t 
(+) 
jt
xlpiatratr  & Ability to track and trace consignment in exporting country i and it
mlpiatratr  importing country j respectively at time t 
(+) 
jt
xlpitime  & Frequency with which shipment reach consignee within schedule for it
mlpitime  exporting country i and importing country j respectively at time t 
(+) 
jt
xlpiqinfr & Quality of Trade and transport related infrastructure in the exporting it 
mlpiqinfr  country i and importing country j respectively at time t 
(+) 
jt
an error term assumed to be uncorrelated with the independent 
εijt  
variables 
Table 4.1: Description of variables and expected signs 
 
4.5 A priori-signs 
4.5.1 Size of Country 
This variable is core to the gravity formulation since the theory states that trade is directly 
proportional to the sizes of both countries. The size of the country is measured by its economic 
size measured by the  average GDP in constant (year 2000) US dollars. Data for this variable 
is sourced from the World Bank’s WDI database and represented by xgdp and mgdp denoting 
exporting and importing country GPDs respectively. The expected sign for both exporting and 
importing countries should be positive with the destination coefficient lower than that for the 
originating country (Feenstra, 2001).  
71 
University of Ghana  http://ugspace.ug.edu.gh
4.5.2 GDP per Capita 
Linder (1961) argues that people with different income levels tend to consume different 
bundles of goods, with the poorer people demanding less and the richer ones having latent 
demand for new varieties of goods. Ramezzana (2000) also found out that for a given global 
average of income per capita as the gap in income levels increases between two country pairs, 
the volume of goods traded between the two countries is expected to decrease ceteris paribus. 
This implies that trade between developed countries is expected to be higher than that 
between developed and developing countries. This means that holding inequality in income 
levels constant between two countries, as the world income levels increase, it is expected to 
increase bilateral trade, Ramezzana (2000). Expected sign therefore is expected to be positive 
due to consumers’ love for variety. 
4.5.3 Distance 
Country pairs that are very far apart are expected to trade less than countries that have proximity 
to each other. The further the distance the higher the transportation costs which adds to the 
overall trade costs thus contributing to the “trade diversion” effect. The sign on the distance 
elasticity is expected to be negative. 
4.5.4 Infrastructure (Port Road & ICT) 
The infrastructure (ports, roads and ICT) variable will be proxied by the logistics performance 
indicators - xlpiqinfr for originating country and mlpiqinfr for destination country. This data is 
sourced from the World Bank’s LPI dataset and it captures the quality of trade and transport 
infrastructure from both countries. The expected sign on the infrastructure elasticity is positive 
as per the theory. 
  
72 
University of Ghana  http://ugspace.ug.edu.gh
4.5.5 Other Country Characteristics 
Landlocked 
In this work, the landlocked dummy variable will be represented by landlocked_o and 
landlocked_d from the CEPII dataset representing originating and destination countries 
respectively, with each variable taking the value 1 where the country is landlocked and 0 
otherwise. Expected sign on the landlock coefficient for both originating and destination 
countries is expected to be negative. 
 
Colonization 
It is generally accepted that former colonies will trade more with their former colonizer than 
with other nations, controlling for other gravity forces and multilateral resistance terms. This 
observed pattern has been credited beyond preferential trade policy or monetary agreements, 
to the existence of trade networks and institutions in the facilitation of trade between the two 
parties. This intangible asset, which is trade-enhancing has survived de-colonization, and 
stimulated the development of a business network of buyers and sellers. The reduction of the 
colonial trade ties may be explained by the depreciation of this asset.  
The variable used in this work is represented by the dummy variables comcol with a value of 
1 if there was/is a relationship and 0 otherwise. The data source is from the CEPII dataset. 
Expected sign for the elasticity is positive if there ever was or there still is a colonized-colonizer 
relationship. 
 
Currency Unions 
Promoters of currency unions have regularly touted them as the maximum credible 
commitment to zero- inflation economic policy. Some advantages often mentioned are 
improved central bank credibility, better than average inflation performance, and deeper capital 
73 
University of Ghana  http://ugspace.ug.edu.gh
markets, all of which tend to enhance productivity and as a result output. The study therefore 
focuses on the gains that a currency union offers to its individual member countries. Currency 
unions encourage bilateral trade as well as promoting overall openness (measured as trade/GDP 
ratio). The variable used for this work is the dummy comcur with its data source from the CEPII 
dataset. The values are 1 for the exporting and importing nations belonging to the same 
currency union and 0 otherwise. The expected sign on the coefficient of comcur is positive. 
 
4.5.7 Regional Trade Agreements 
The impact of free trade agreements has been found to be statistically significant on bilateral 
trade flows among member countries. The variable used to represent FTAs between country 
pairs is fta_wto and the sign is expected to be positive.  
  
74 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER 5 
RESULTS AND DISCUSSIONS 
5.1 Introduction 
This section focuses on all the results of the study – from an overview of the summary statistics, 
output of diagnostics tests leading to correction of potential issues that could have biased the 
results and also to the final choice of estimation model. Finally we discuss the results and output 
of the logistics-augmented gravity model. 
5.2 Summary Statistics 
From Table 5.1, the summary statistics shows a very large deviation in the incidence of bilateral 
exports for the overall sample involving SSA. The mean and standard deviation of the overall 
sample is US$526million and US$5.59billion respectively with 267,072 data points. The large 
standard deviation of the overall sample indicates a massive disparity between the 189 
exporting and 194 importing countries in the sample. With respect to SSA exporting to the 
ROW the average value of bilateral exports is US$52.5million with a standard deviation of 
US$595million and 45,229 observations. In comparison, the average exports and standard 
deviation are US$50.4million and US$378million for bilateral trade from ROW to SSA with 
47,943 data points.  
On the size of the economies, represented by GDP, the overall sample shows an average of 
US$423 billion with a standard deviation of US$1,520 billion and 269,672 observations. When 
SSA exports to ROW, the average GDP is US$33.1billion with a standard deviation of 
US$83.4billion and 46,554 observations. A similar trend occurs for the importing country 
GDPs 
On per capita GDP representing income levels for exporting countries, for the overall sample 
the average is US$15,254.36 with a standard deviation of US$20,716.49 for a sample size of 
269,672 as compared to US2,296.122 and US$3,518.021 for a size of 46,554 for SSA as an 
75 
University of Ghana  http://ugspace.ug.edu.gh
exporter group. For importing countries, mean income levels are US$15,047.84 with standard 
deviation of US$20,641 for a sample size of 269,117 as compared to US$20,018.39, 
US$22,498.78 and 46,387 respectively. 
The average cost of exporting a standard container is US$1,892 for SSA as an exporting 
country, way higher than the rest as the overall average is $1,369 and $1167 from ROW to 
SSA. A similar trend occurs for average cost of importing a container where ROW to SSA has 
an average of US$2,407 as compared to $1629 for the overall sample.  
The logistics performance indicator (LPI) grades 160 countries on six criteria of trade, that 
have increasingly been recognized as important to development. The data used for this exercise 
is sourced from a survey of logistics professionals who are asked questions about the foreign 
countries in which they operate. The elements  analyzed in the LPI were chosen based on recent 
theoretical and empirical research and on the practical experience of logistics professionals 
involved in international freight forwarding. (World Bank LPI, 2016). 
They include: 
1. The efficiency of customs and border management clearance (“Customs”). 
2. The quality of trade and transport infrastructure (Infrastructure”). 
3. The ease of arranging competitively priced shipments (Ease of arranging shipments”). 
4. The competence and quality of logistics services—trucking, forwarding, and customs 
brokerage (“Quality of logistics services”). 
5. The ability to track and trace consignments (“Tracking and tracing”). 
6. The frequency with which shipments reach consignees within scheduled or expected 
delivery times (“Timeliness”). 
The LPI uses standard statistical techniques to aggregate the data into a single indicator that 
can be used for cross-country comparisons. (World Bank LPI, 2016) 
76 
University of Ghana  http://ugspace.ug.edu.gh
With respect to the logistics performance indicators, for ability to track and trace consignments, 
competence and quality of logistics services, efficiency of customs clearance process, 
frequency with which shipments reach consignee within scheduled or expected time, quality 
of trade and transport-related infrastructure, the same trend occurs where averages for ROW to 
SSA are the highest followed by the overall average with SSA to ROW having the lowest 
showing how poorly SSA compares to the rest of the world in terms of trade facilitation.  
77 
University of Ghana  http://ugspace.ug.edu.gh
Table 5.1: Summary Statistics of Main Variables used in Gravity Equation     
Overall SSA-to-ROW ROW-to-SSA 
Variable 
Mean Std. Dev. Obs Mean Std. Dev. Obs Mean Std. Dev. Obs 
exportsfob 526 5,590 267072 52.5 595 45229 50.4 378 47943 
xgdp 423,000 1,520,000 269672 33,100 83,400 46554 568,000 1,750,000 49318 
mgdp 417,000 1,510,000 269117 585,000 1,790,000 46387 31,500 81,000 49153 
xgdppc  15254.36 20716.49 269672 2296.122 3518.021 46554 19752.59 22371.77 48318 
mgdppc 
15047.84 20641.03 269117 20018.39 22498.78 46387 2368.668 3613.849 49153 
(constant $) 
distwces 8.703821 0.7854012 275245 8.896495  0.4355411 47617 8.894848 0.4315381 50458 
imcostimport 1628.893 1194.755 259436 1318.369 781.0582 44263 2407.351 1606.317 49781 
excostexport 1363.219 922.4531 260178 1892.744 1132.324 47013 1167.391 717.5936 47060 
xlpiatratr 2.891466 0.6451085 231259 2.421318 0.4265409 40032 3.056723 0.627086 42883 
mlpiatratr 2.881319 0.6480661 228755 3.065724 0.6290275 40165 2.409745 0.4269345 41961 
xlpicqols 2.822164 0.6307403 231259 2.363803 0.3793143 40032 2.983657 0.6218294 42883 
mlpicqols 2.813383 0.6323332 228755 2.993256 0.623629 40165 2.358034 0.3778795 41961 
xlpieoacps 2.841558 0.5252173 231259 2.456879 0.3983475 40032 2.97596 0.4974066 42883 
mlpieoacps 2.832387 0.5283984 228755 2.981391 0.4992081 40165 2.448155 0.3986603 41961 
xlpicustoms 2.667782 0.6132508 231259 2.237543 0.3261457 40032 2.817985 0.6185212 42883 
mlpicustoms 2.659626 0.6139971 228755 2.826669 0.6209732 40165 2.233158 0.3260905 41961 
 
 
78 
University of Ghana  http://ugspace.ug.edu.gh
Overall SSA-to-ROW ROW-to-SSA 
Variable 
Mean Std. Dev. Obs Mean Std. Dev. Obs Mean Std. Dev. Obs, N 
xlpitime 3.306843 0.6012586 231259 2.851887 0.4377797 40032 3.464759 0.5694892 42883 
mlpitime 3.297293 0.6038024 228755 3.472227 0.5706963 40165 2.84379 0.4369082 41961 
xlpiqinfr 2.727161 0.7075822 231259 2.191888 0.3783881 40032 2.913473 0.7010188 42883 
mlpiqinfr 2.717347 0.709559 228755 2.924924 0.7038557 40165 2.185397 0.3779948 41961 
landlocked_d 0.1896848 0.3920523 278541 0.3041432 0.4600484 48030 0.1456536 0.352762 50881 
landlocked_o 0.1903418 0.3925715 278541 0.1358734 0.3426577 48030 0.3138107 0.464045 50881 
contig 0.018403 0.134404 278541 0.0020612 0.0453542 48030 0.0021226 0.0460232 50881 
comlangoff 0.1514965 0.3585329 278541 0.177639 0.3822125 48030 0.1706727 0.3762265 50881 
comcol 0.1058121 0.3075975 278541 0.1228399 0.3282568 48030 0.115662 0.3198224 50881 
comcur 0.0160335 0.1256047 278541 0.0007495 0.0273676 48030 0.0007075 0.0265903 50881 
fta_wto 0.1377391 0.3446266 278541 0.0147824 0.1206823 48030 0.0141703 0.1181939 50881 
 
 
 
79 
University of Ghana  http://ugspace.ug.edu.gh
5.3 Diagnostic Tests 
Even though panel data comes with enormous advantages such -  more information, more 
variability, less collinearity among variables, more degrees of freedom and more efficiency, 
there are some challenges in relying on estimation results the ordinary least squares (OLS) 
regressions technique. Key amongst them is the assumption that all countries are stable over 
time (time effects) and that individual countries are the same (heterogeneity). These 
assumptions are false hence it is important that the data must be corrected for these biases.  
Heteroscedasticity is prevalent in cross-sectional data and serial correlation is common to time 
series data therefore we can safely assume that panel data – which is a combination of cross-
section and time series data will have both heteroscedasticity and serial correlation. 
5.3.1 Time-Effects 
Individual time dummies are added to the pooled regression to the estimation model and the 
model estimated using OLS. We then test whether the time dummies are jointly equal to zero. 
From the test results in Appendix 12, all the time variables are significant and the p-value =0, 
so we rejected the null hypothesis of no time effects and we conclude that the model with 
time effects is more appropriate. 
 
5.3.2 Breusch -Pagan Heterogeneity Test Results 
Breusch and Pagan Lagrangian multiplier test for random effects
        lnexports[bilateraltradeid,t] = Xb + u[bilateraltradeid] + e[bilateraltradeid,t]
        Estimated results:
                                 Var     sd = sqrt(Var)
                                                       
               lnexports     13.72675       3.704963
                       e     1.955712       1.398468
                       u     5.014599        2.23933
        Test:   Var(u) = 0
                             chibar2(01) = 60286.86
                          Prob > chibar2 =   0.0000  
From the results, the p-value=0 so we reject the null hypothesis of no heterogeneity and 
concluded that there exists heterogeneity in the data set and the RE is inappropriate to use. 
 
80 
University of Ghana  http://ugspace.ug.edu.gh
5.3.3 Correlation 
For multicollinearity to cause serious problems, the guideline is that the correlation coefficient 
among two regressors has to be greater than 0.8 (Akinkugbe, 2006). From our correlation test 
results in Tables 5.2 and 5.3, we found a very high correlation between almost all LPIs for both 
the exporting and importing variables respectively, thus necessitating the need for the 
disaggregation of the LPI variables from the model to enable us to investigate their individual 
effects on bilateral exports. 
Table 5.2: Correlation between LPI variables for exporting countries 
xlpiatratr xlpicqols xlpieoacps xlpicustoms xlpitime xlpiqinfr 
 
xlpiatratr 1.000      
xlpicqols 0.9371* 1.000     
xlpieoacps 0.8781* 0.8916* 1.000    
xlpicustoms 0.8925* 0.9317* 0.8623* 1.000   
xlpitime 0.8854* 0.8808* 0.8324* 0.8417* 1.000  
xlpiqinfr 0.9162* 0.9469* 0.8747* 0.9473* 0.8581* 1.000 
 
Table 5.3: Correlation between LPI variables for importing countries  
mlpiatratr mlpicqols mlpieoacps mlpicustoms mlpitime mlpiqinfr 
 
mlpiatratr 1.000      
mlpicqols 0.9370* 1.000     
mlpieoacps 0.8791* 0.8921* 1.000    
mlpicustoms 0.8923* 0.9317* 0.8627* 1.000   
mlpitime 0.8852* 0.8800* 0.8335* 0.8420* 1.000  
mlpiqinfr 0.9167* 0.9472* 0.8750* 0.9474* 0.8580* 1.000 
 
The correlation matrices clearly show that there exists pair-wise correlation for almost all the 
LPI variables and they are all significant and greater than 0.8 thus posing a very serious 
multicollinearity problem. To eliminate this problem, this work therefore focused on the impact 
81 
University of Ghana  http://ugspace.ug.edu.gh
of each of the individual LPI indicators as well as the overall/aggregated LPI indicator on 
bilateral export 
5.3.4 Hausman Test for Fixed Effects versus Random Effects 
                      Coefficients     
                    (b)          (B)            (b-B)     sqrt(diag(V_b-V_B))
                     fe           .          Difference          S.E.
                                                                              
      lnxgdp      .1605554     1.202939       -1.042384        .1959619
      lnmgdp       .156875     .9397095       -.7828345        .2101097
    lnxgdppc      .0746135    -.1097466        .1843601        .2082199
    lnmgdppc      .2454192    -.3049484        .5503676        .2213116
lnxcostexp~t     -.2524434    -.8105349        .5580915        .0416411
lnmcostimp~t      .1193033    -.1748287         .294132        .0437262
     fta_wto       .068117     .7167761       -.6486591        .0825069
 xlpioverall       .065698     .4479993       -.3823013        .0170261
 mlpioverall     -.0139255     .2646431       -.2785686        .0163249
 _Iyear_2008      .0864058    -.0839645        .1703703        .0102726
 _Iyear_2009     -.0046613     -.057815        .0531538        .0139199
 _Iyear_2010      .0683466    -.2043601        .2727067        .0215747
 _Iyear_2011      .2266169    -.1784982        .4051151        .0301309
 _Iyear_2012      .2774132    -.1645182        .4419314        .0353725
 _Iyear_2013      .2884626    -.2242048        .5126674        .0415456
 _Iyear_2014      .3680079    -.1790815        .5470894        .0471779
 _Iyear_2015      .3968819    -.0694116        .4662935        .0502105
                                                                              
                           b = consistent under Ho and Ha; obtained from xtreg
            B = inconsistent under Ha, efficient under Ho; obtained from xtreg
    Test:  Ho:  difference in coefficients not systematic
                 chi2(17) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                          =      926.46
                Prob>chi2 =      0.0000
                (V_b-V_B is not positive definite)  
From the Hausman results, p-value=0 which is very small therefore we rejected the null (REM 
is appropriate) and concluded that the FEM was appropriate. 
 
  
82 
University of Ghana  http://ugspace.ug.edu.gh
5.4 Durbin Wu-Hausman Endogeneity Test  
DEPENDENT VARIABLE:                   
log of exports 
  
fta_wto 7.473*** 
 (0.200) 
comcol -0.307*** 
 (0.114) 
mcostimport 0.0689*** 
 (0.0109) 
exnumdocexp -0.0922*** 
 (0.00789) 
xcostexport -0.196*** 
 (0.0143) 
xlpioverall 0.484*** 
 (0.0263) 
mlpioverall 0.160*** 
 (0.0260) 
landlocked_o -0.962*** 
 (0.0788) 
landlocked_d -1.310*** 
 (0.0781) 
comcur -0.421*** 
 (0.124) 
xgdplevel 0.412*** 
 (0.0138) 
mgdplevel 0.314*** 
 (0.0114) 
Constant 12.61*** 
 (0.120) 
  
Observations 135,702 
Number of bilateraltradeid 20,153 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
 
 
Endogeneity test results 
Wooldridge (2002) states that potential sources of endogeneity of independent variables 
normally fall under the categories of measurement error, omitted variables, and simultaneity 
bias. In this work, we assume similar to Bergstrand & Baier (2002), Trefler (1993) and Lee & 
Swagel (1997) that endogeneity occurs as a result of self-selection bias and that the free trade 
variable (fta_wto) is endogenous. In fact, the results from Bergstrand & Baier (2002) states that 
83 
University of Ghana  http://ugspace.ug.edu.gh
"traditional estimates of the effect of FTAs on bilateral trade flows have tended to be 
underestimated by as much as 75-80 percent". 
Our test result is shown below: 
Ho: All variables are exogenous 
Durbin (score) chi2(1)             =  21660.3  (p = 0.0000) 
Wu-Hausman F(1,132963)      =  25671.6  (p = 0.0000) 
The p-values of both Durbin and Wu-Hausman tests is zero implying the rejection of the null 
leading to the conclusion that our choice of the dummy variable fta_wto as the endogenous 
variable is correct. 
 
5.5 Diagnostics Results and Model Estimation Strategy 
From the results of the diagnostics, we concluded that we could not use the pooled data for the 
regression and that the coefficients will be extremely biased. There was the presence of time-
effects which had to be corrected by introducing year dummies. These dummy variables 
controlled for factors such as general inflation common to all countries as well as the overall 
rise in trade because of globalization. Another issue was the introduction of the FTA/RTA 
dummy in the model as it introduced endogeneity in the model since countries could self-select 
into various trade agreements based on the volume of trade between various countries. This 
was confirmed by the Durbin-Wu Hausman endogeneity test above.  
Heterogeneity is present hence the random effect (REM) is inappropriate and could not be used 
as this was confirmed by the Breusch-Pagan heterogeneity test. The use of the fixed effect 
model (FEM) was further confirmed by the Hausman test. However, FEM has its own 
challenges - all the time-invariant (TI) variables are dropped and absorbed into the intercept 
and this is not acceptable as we are interested in the coefficients of the TI variables as well.  
 
84 
University of Ghana  http://ugspace.ug.edu.gh
The Hausman-Taylor (HT) estimator which is a model that fuses both the consistency of an 
FEM with the effectiveness and applicability of an REM is the preferred option. Our preference 
of the HT over the REM and FEM is to ensure that the variations across countries are controlled 
as well as simultaneously incorporate time invariant variables which might be correlated with 
bilateral specific consequences within the estimation. By leveraging some instrumental 
variables that are uncorrelated to unobservable characteristics, the HT estimator has been found 
to be more efficient than the REM and FEM. Finally, the HT estimator also allows for the 
control of the endogeneity from the FTA/RTA agreement dummy within the gravity framework 
(Turkson, 2012).  
Results from the various estimation techniques can be found in Appendix 1. The first column 
shows the results from the overall pooled data without any corrections for time effects, 
heteroskedasticity and serial correlation. The second is the random effect (RE) results corrected 
for both time effects, heteroskedasticity and serial correlation. However due to the presence of 
heterogeneity, confirmed by the diagnostic tests, the REM estimation was inappropriate to use 
as the tests showed that there were some correlation between the regressors and individual 
effects. The Hausman test further corroborated the use of the FEM in the third column. 
However, as can be seen, the FE model dropped all the time invariant variables so we had to 
settle on the Hausman Taylor (HT) estimates in column four which brought back the TI 
variables and also corrected for endogeneity. 
Our focus for discussion will be centered around the HT estimation results in the last column 
labeled (4) because the estimated coefficients from all the previous results are all biased in one 
way or another. 
5.6 Main Results and Discussions 
Table 5.4 shows a summary output for results using the aggregate LPI variable and the log of 
exports as the dependent variable for bilateral trade involving sub-Saharan African (SSA) 
85 
University of Ghana  http://ugspace.ug.edu.gh
countries. The first column shows the results from the full sample, the second for trade where 
SSA is an exporter to ROW and lastly the third where SSA is an importer from ROW. 
Traditional Gravity Model 
Size (GDP)  
In line with the original gravity model Linder (1961), Tinbergen (1962), Linnemann (1966) 
Anderson and van Wincoop (2003), the size of the exporting and importing countries 
represented by their respective GDPs all have a significant impact on bilateral exports with 
their actual signs matching the expected signs. The elasticities being +1.231 and +0.973 for 
both exporting and importing countries respectively. This implies that a 10% increment of 
exports from the exporting country will lead to a positive growth of 12.31% in bilateral trade. 
Similarly, a 10% growth in the size of the importing country will lead to a corresponding 
growth of 9.73%. This is in line with Feenstra (2001), where the GDP elasticity of the 
destination country is lower. In comparison, the elasticities for country size of both exporters 
and importers within SSA to ROW are +1.127 and +1.071 respectively. Coefficients for ROW 
to SSA are +1.212 and +0.777 for exporting and importing countries respectively. These results 
are consistent with the theoretical models and similar to Nutor (2004), Akingube (2009), Perez 
and Wilson (2012) and Turkson (2012). 
 
  
86 
University of Ghana  http://ugspace.ug.edu.gh
Table 5.4: Gravity Model Results with Aggregate LPI indicator  
Dependent variable:                  I  n  volving SSA 
SSA to ROW ROW to SSA 
Log of exports (Full Sample) 
Log of exporter GDP (lnxgdp) 1.231*** 1.127*** 1.212*** 
 (0.0294) (0.0505) (0.0331) 
Log of Importer GDP (lnmgdp) 0.973*** 1.071*** 0.777*** 
 (0.0295) (0.0433) (0.0383) 
Log of exporter GDPPC (lnxgdppc) -0.231*** -0.380*** -0.256*** 
 (0.0358) (0.0635) (0.0403) 
Log of importer GDPPC (lnmgdppc) -0.360*** -0.321*** -0.305*** 
 (0.0364) (0.0538) (0.0484) 
Log of exporter’s cost to export -0.553*** -0.395*** -0.421*** 
(lnxcostexport) (0.0542) (0.0847) (0.0603) 
Log of importer’s cost to import 0.0166 -0.136* 0.0814 
(lnmcostimport) (0.0513) (0.0788) (0.0595) 
Exporter’ aggregate LPI 0.274*** 0.0238 0.423*** 
 (0.0388) (0.0555) (0.0459) 
Importer’s aggregate LPI 0.113*** 0.193*** 0.00424 
 (0.0379) (0.0632) (0.0393) 
Free Trade Agreement (fta_wto) 0.418*** 0.482*** 0.656*** 
 (0.115) (0.183) (0.134) 
Log of Distances (lndistwces) -1.010*** -1.252*** -1.091*** 
 (0.0937) (0.130) (0.101) 
Landlocked Exporter -0.746*** -1.049*** -0.798*** 
(landlocked_o) (0.108) (0.156) (0.136) 
Landlocked Importer -1.241*** -0.849*** -1.346*** 
 (landlocked_d) (0.109) (0.179) (0.122) 
Contiguity (contig) 1.913*** 1.848*** 1.676*** 
 (0.355) (0.430) (0.335) 
Common Official Language 0.953*** 1.169*** 0.580*** 
(commlang_off) (0.116) (0.172) (0.131) 
Common Colony (comcol) 0.141 0.179 0.237 
 (0.149) (0.217) (0.167) 
Currency Union (comcur) 0.906*** 0.910** 0.626** 
 (0.347) (0.406) (0.317) 
Constant -23.83*** -20.95*** -19.71*** 
 (1.430) (2.094) (1.563) 
    
Observations 50,492 27,270 29,905 
Number of bilateraltradeid 8,178 4,566 4,690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
 
  
87 
University of Ghana  http://ugspace.ug.edu.gh
GDP per Capita 
Income per enters this model significantly but with a reducing effect on bilateral trade across 
the three samples. On the overall sample, the impact is higher on the importing country (-0.360) 
as compared to the exporting country (-0.231). This implies that as per capita income increases 
in both countries, there is less consumption. A similar trend occurs when SSA exports to the 
rest of the world. In this case the impact is greater on the exporting SSA country (-0.380) versus 
-0.321 for the importing countries. The explanation here could be due to the substitution effect, 
where because of consumers’ love for variety, as incomes grow, the rest of the world substitute 
domestic varieties for imports from SSA leading to less demand for SSA’s imports hence lower 
bilateral trade. 
Distance  
The expected sign for the distance coefficient is negative. Our results corroborate Tinbergen 
(1962) and is consistent with the theory as the coefficient of the distance variable is significant 
at 1% with an elasticity of -1.01 almost the same as from ROW to SSA at -1.09. This means 
that as the distance between country pairs increase by 10%, bilateral trade between them 
reduces by 10.1% and 10.9% respectively. The highest impact is observed when SSA exports 
to ROW (12.52%). This is understandable as the average distance between SSA and the major 
trading partners are quite long and the commodities must travel further to get to their 
destinations. This makes a good case for intra-regional SSA trade.  
Landlocked 
A country being landlocked can introduce significant barriers to trade. From our results, the 
actual sign on the coefficients matches the expected sign and are significant. The coefficients 
are -0.746 and -1.241 for the exporting and importing countries respectively for the full sample. 
This means that trading with a landlocked exporting country reduces trade by 52.5% (e-0.746-1) 
and for an importing reduces trade by 71%. The biggest impact (-74%) of being landlocked is 
88 
University of Ghana  http://ugspace.ug.edu.gh
experienced when the importing country is in SSA and landlocked. On the other hand, for 
exporting countries the greatest impact is felt when SSA exports to the ROW (-65%).  
These findings corroborate Amjadi and Yeats (1995), Limao and Venables (2001), Turkson 
(2011) and Turkson (2012). Being landlocked introduces significant transportation costs of up 
to 50% (Radelet and Sachs, 1998) arising out of the long distance itself and the pass-through 
effect of the trade and transport inefficiencies of the transit country. Some countries must go 
through multiple transit countries, thus increasing their transport costs significantly further. 
Colony  
Here, we have the variable comcol referring to two countries in a colonial relationship post 
1945. The theoretical underpinnings posit that the variable should have a positive impact on 
bilateral trade supported empirically by Rauch (1999), Rose (2000) and Disdier and Mayer 
(2007). Our findings show that it has the expected positive sign, but its effect is insignificant.  
Common Language  
Country pairs having the same common official language trade more with each other. Our 
results show that the coefficient of the common language variable has a coefficient of +0.953 
at one percent significance level. This is consistent with the theory as the actual sign is the 
same as the expected. This implies that bilateral trade between country pairs with the same 
common official language trade is increased by a factor of 2.59 (e0.953) or 159% (e0.953-1). 
The biggest impact occurs where SSA exports to ROW. In this case bilateral trade increases by 
a factor of 3.22 and the least (1.79) when SSA imports from ROW. 
These findings buttress previous findings from Frankel and Wei (1993), Rauch and Trindade 
(2002), Mélitz (2008). Since languages facilitate communication and make business 
transactions simpler, two trading partners from different countries who speak identical 
language can speak and transact with each different directly whereas those without a sufficient 
grasp of a particular language need to frequently depend on a middleman or hire an interpreter. 
89 
University of Ghana  http://ugspace.ug.edu.gh
The additional complexity and costs inherent in relying on a translator may come with huge 
costs and potential for costly mistakes. These extended costs may be sufficiently large to 
prevent in any further transactions from taking place in the future. 
Contiguity 
Rauch (2001) and Wagner et al. (2002) advocates that there are information costs that may 
inhibit trade between country pairs that are far apart and that for contiguous countries, these 
costs are reduced drastically due to business operations and social networks formed as a result 
of operating across borders. These reduced costs, they claim, should promote trade and is 
consistent with our findings. From our results, country pairs sharing the same border trade on 
the average 6.8 times (e1.913) as compared to non-contiguous countries for the overall sample 
and this is the highest. The least is 5.34 times for the sample where SSA imports from the 
ROW.  
Common Currency 
There is a large effect on common currency on international trade. Rose (2000) found out that 
country pairs in the same currency union trade three times more than country pairs with 
different currencies. This is also known as the Rose effect. Our results are consistent with the 
theory as actual coefficient (+0.906). This implies that form our analysis trade between a 
country pair increases by a factor of 2.47 (e0.906) when the country pair belong to the same 
currency union. 
Free/Regional Trade Agreements (FTA or RTA) 
Through the reduction and removal of tariffs, one of the impacts of  a regional trade agreements 
(RTA) is to enable more efficient production in a region to the advantage of consumers. Vicard 
(2009) show that any kind of RTA providing trade preferences to their member countries 
significantly increases bilateral trade between them. Similarly, Vollrath and Hallahan (2011) 
found out that for agricultural trade between RTA member countries trade is boosted on 
90 
University of Ghana  http://ugspace.ug.edu.gh
average between 34% and 93%. From our results, the coefficient on the fta_wto variable is 
+0.418. This means that trade between country pairs governed by an FTA or RTA increases by 
a factor of 1.52 or grows by 52% consistent with the theoretical and empirical expectations. 
This may explain why trade has increased multiple folds in within SSA’s regional economic 
communities as evidenced in Figure 2.6 
Cost to Export/Import 
Our results show that the coefficient on export costs is significant at 1% with the expected sign 
(-0.553). This implies that increasing the costs of exporting a 20-footer container by the 10% 
will reduce exports by 5.53% consistent with the theoretical underpinnings. For SSA to ROW 
and ROW to SSA, coefficients are -0.395 and -0.421 respectively.  
On cost to import, the coefficient is only significant at ten percent when SSA exports to the 
ROW where exports are reduced by 13.6%.  
Overall Logistics Performance Indicator (LPI) 
Generally, improvement in overall logistics in a country should improve exports. The 
coefficient of the overall LPIs for exporters and importers are both significant with magnitudes 
+0.274 and +0.113 respectively. Also of interest is the fact that the exporter coefficient is larger 
than the importer. This is expected because the impact of logistics on the exports of most sub-
Saharan African countries are higher due to the nature of exports – mainly primary 
commodities. Where SSA exports to the ROW, the logistics of the importing country is very 
important (+0.193). The greatest impact (+0.423) is felt in the exporting country when SSA 
imports from ROW. 
Disaggregated Logistics Performance Indicators 
The overall Logistics Performance Index (LPI) score reflects perceptions of a country's 
logistics based on efficiency of customs clearance process, quality of trade- and transport-
related infrastructure, ease of arranging competitively priced shipments, quality of logistics 
91 
University of Ghana  http://ugspace.ug.edu.gh
services, ability to track and trace consignments, and frequency with which shipments reach 
the consignee within the scheduled time. The index ranges from 1 to 5, with a higher score 
representing better performance. The LPI uses a structured online survey of logistics 
professionals at multinational freight forwarders and at the main express carriers (Source: 
World Bank and Turku School of Economics, Logistic Performance Index Surveys) 
Here, we attempt to compare the impact of the overall LPI to the six disaggregated indicators 
notably, frequency of shipment timeliness, ability to track and trace consignments, efficiency 
of customs process, quality of logistics, infrastructure quality and ease of arranging 
competitively priced shipments because of the high correlation between the individual 
indicators. See (Tables 5.2 and 5.3). This to enable us to determine their individual impact on 
bilateral exports for all trade involving SSA whether as importers or exporters (full sample), 
SSA as exporter to ROW and SSA as importers from ROW. Our summary findings are shown 
in table 5.5 below but the full results can be found in Appendices 3-5. 
 
Table 5.5: Summary of Results from Various Estimations on Disaggregated Logistics 
Performance Indicators 
Involving SSA SSA to ROW ROW to SSA 
 xLPI mLPI xLPI mLPI xLPI mLPI 
Frequency of 0.0682*** 0.0545** -0.0233 0.131*** 0.158*** -0.0117 
Shipping Timeliness (0.0247) (0.0242) (0.0334) (0.0421) (0.0305) (0.0238) 
Ability to Track and 0.136*** -0.0224 0.0341 0.00672 0.173*** -0.0776*** 
Trace (0.0262) (0.0258) (0.0361) (0.0445) (0.0321) (0.0259) 
Competence and 0.175*** 0.0123 0.0132 0.107** 0.231*** -0.0603** 
Quality of Logistics (0.0304) (0.0294) (0.0427) (0.0496) (0.0361) (0.0300) 
0.206*** 0.108*** 0.0383 0.138*** 0.317*** 0.0719** 
Customs Efficiency 
(0.0302) (0.0300) (0.0438) (0.0495) (0.0355) (0.0314) 
0.184*** 0.0744** 0.00408 0.155*** 0.276*** -0.000755 
Infrastructure 
(0.0317) (0.0308) (0.0447) (0.0525) (0.0381) (0.0314) 
Ease of arranging 0.141*** 0.131*** 0.0282 0.117*** 0.255*** 0.0911*** 
competitively prices 
shipment (0.0267) (0.0260) (0.0375) (0.0450) (0.0327) (0.0265) 
92 
University of Ghana  http://ugspace.ug.edu.gh
 
For the overall sample involving SSA, all the exporter indicators enter the model significantly 
but for the importer, only four out of six are individually significant in explaining bilateral 
trade. Invariably, all the performance indicators of the exporter have a greater impact on exports 
as compared to that of the importer. 
On the individual LPIs, the efficiency of customs clearance process has the highest impact on 
bilateral trade flows for the exporter followed by the quality of infrastructure whilst the ease of 
arranging competitively priced shipment has the greatest impact followed by customs 
efficiency for the importer. This is not surprising as embedded in these two measures of 
logistics performance (customs efficiency and infrastructure) are most of the direct and indirect 
costs of bilateral trade involving SSA. This is because in a majority of SSA countries, the 
customs process is wrought with corrupt practices leading to the extortion effect, where 
customs officials extort bribes from exporters (Ackerman, 1997 and Bardhan, 2006),. This 
indirect tax adds to the trade costs leading to trade diversion. Similarly, SSA’s infrastructure 
quality is rated the lowest amongst its peers leading to very high transportation costs and even 
higher for landlocked countries (Limao & Venables, 2000) 
Similarly, the frequency with which shipments reach consignee on time has the lowest impact 
for both exporters and importers. For both the ability to track and trace consignments and 
competence and quality of logistics services indices, they are relevant for the exporter but have 
no influence on trade for the importer. 
Comparing the results on whether SSA is an aggregate exporter or importer presents some 
interesting findings as well. On the frequency with which shipments reach consignees in time 
index, there exists some asymmetries. Shipping timeliness and infrastructure matter for the 
importer in the SSA to ROW group and influences the exporter in the ROW to SSA group with 
the impact higher for the exporter group. This is true because if SSA is exporting to the rest of 
the world, the more frequent consignments get to the destination without undue delays, the 
93 
University of Ghana  http://ugspace.ug.edu.gh
lower the costs incurred, and the more trade will occur all things being equal. Similarly, the 
more quality the trade infrastructure, the lower the trade costs will be leading to the  trade 
enhancing effect. On the overall sample, the index influences both exporters and importers with 
the highest impact on the exporters. 
With respect to the competence and quality of logistics services index, on the overall sample, 
the greatest influence is felt on the exporter and no impact on the importer. However, for the 
SSA exporter group, the index has a positive influence on the importer group. The index has a 
positive impact on exporter and a reducing impact on the importer where SSA is an aggregate 
importer (ROW to SSA). 
On the ability to track and trace consignments index, it has an influence on the exporter for the 
overall group involving SSA and has no impact when SSA is an aggregate exporter. For the 
ROW to SSA sample, where SSA is an aggregate importer, it has a trade enhancing effect on 
the exporter and a trade diversion effect on the importer. 
Customs efficiency has a trade enhancing impact on the importer in all the three groups. On 
the exporter, the impact is felt on the overall sample and the ROW to SSA group. 
A key observation in the results is that none of the coefficients of the disaggregated LPI 
variables are significant for the exporter in the SSA to ROW group implying that none of the 
LPI variables have an impact on the exporter country in the group. This may be because of the 
nature of our exports as a majority of SSA countries export mainly raw materials and 
agricultural produce which depends largely on the logistics of the destination countries. 
   
94 
University of Ghana  http://ugspace.ug.edu.gh
CHAPTER 6 
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 
 
6.1 Introduction 
This chapter presents the summary, conclusions and policy recommendations based on the 
findings of the study. It also highlights some limitations of the study and suggested areas for 
further research. 
 
6.2 Summary and Conclusions 
This work sought to investigate why despite the increase in global trade, our share of global 
trade keeps declining. We decided to investigate impact of trade facilitation on bilateral trade 
involving SSA to enable the region to prioritize its limited resources towards reforms that will 
derive the most impact. It is not as if nothing is being done in the sub-region because a lot of 
effort and resources are being channeled towards this which has led to the growth of intra-SSA 
trade albeit a drop in the ocean in the global context.  
Our findings show that it is important to focus on the aggregate logistics impact but this does 
not tell us where to channel our resources. A lot more mileage could be gained if we 
disaggregate the indices and prioritize our investment accordingly. In the order of priority, we 
should be focusing first and foremost on initiatives that will improve border or customs 
efficiency and infrastructure. Next in line should be quality of logistics services, ease of 
arranging competitively priced shipment, ability to track and trace and frequency within which 
shipments reach consignees within schedule in that order. 
The insignificance of the logistics performance indicators for the exporters in the SSA to ROW 
group is quite interesting.  
It is also pertinent to note that none of the coefficients of the disaggregated LPI variables are 
significant in the model with the sample of exporting SSA countries to the rest of the world. 
This may be because, a majority of SSA countries exports raw materials and agriculture 
95 
University of Ghana  http://ugspace.ug.edu.gh
products which may be perishable and so it depends on the logistics of the destination countries. 
However, among SSA countries, the logistics matter since we all have similar comparative 
advantages.  Additionally, we need to gradually move away from exporting raw materials to 
value-adds and manufactures to secure our fair share of global trade in terms of value. 
6.3 Policy Recommendation 
This work, like Perez and Wilson (2012) recommends to policy makers within SSA that both 
hard and soft infrastructure are relevant for trade facilitation.  
Even though trade and transport-related infrastructure is very necessary and important for trade 
facilitation, its benefits accrue to the country in the medium to long-term and is very capital 
intensive. Whilst we invest in the hard infrastructure projects, to achieve quick-wins, 
governments should be simultaneously focusing on investing to achieve efficiency in the soft 
infrastructure - simplification, standardization, harmonization and transparency by improving 
the efficiency of customs procedures as these can be achieved with minimum investment. 
Another point to note is that if countries in SSA continue to export agriculture produce and raw 
materials to the rest of the world, improvement in our LPI indicators will have no impact on 
the volume of exports. To “catch up” with the rest of the world, SSA countries need to diversify 
its export mix as a matter of priority. When it comes to intra-SSA trade, these indicators may 
matter because most countries in the sub-region have the same comparative advantage. 
 
6.4 Limitations of the Study  
This study focuses on trade involving Sub-Saharan Africa as a homogeneous group and the 
prescription for SSA may not necessarily reflect the needs of the forty-seven individual 
countries in the overall sample.  
By using logarithm augmented gravity equations, the samples may suffer from omitted variable 
bias as logs of zero are not defined hence will be dropped (Wall, 2000). The extent of this 
96 
University of Ghana  http://ugspace.ug.edu.gh
limitation bias was impossible to investigate further since there is no separation of zero trade 
from missing records in IMF’s data.  
 
6.5 Further Research Areas 
The duration for this study did not allow for investigation into other areas of interest. 
Subsequently, further research should focus on trade facilitation for the various RECs within 
SSA, individual country priority areas controlling for major export segments – primary 
commodities, oil and non-oil. A more ambitious journey will be to investigate the impact of 
trade facilitation for major exporting products or endowments within these groups in individual 
countries. 
  
97 
University of Ghana  http://ugspace.ug.edu.gh
7.0 References 
1. Alaba, O. B. (2009). 5 ACP development, integration and the capacities of transport 
infrastructure. Beyond Market Access for Economic Development: EU-Africa 
Relations in Transition, 93. 
2. Alberto Behar & Phil Manners (2008) Logistics and Exports, University of Oxford, 
Economics Department, Discussion Paper Series, Number 439 July 2009 
3. Alesina, A., R. J. Barro and S. Tenreyro (2002), Optimal Currency Areas, NBER 
Macroeconomics Annual 2002, Vol. 17, M. Gertler and K. Rogoff (editors), 
Cambridge, MA: MIT Press, pp. 301-345 
4. Amoako-Tuffour J., Balchin N., Calabrese L. and Mendez-Parra M. (2016), Trade 
Facilitation and Economic Transformation In Africa: Source: InAfrican Transformation 
Forum, 2016 
5. Anderson, J.E, (1979), A Theoretical Foundation for the Gravity Equation, 
Source:American Economic Review, 69 (1), 106-16, Published by: American 
Economic Association 
6. Anderson, J. E. and D. Marcouiller (2002), Insecurity and the Pattern of Trade: An 
Empirical Investigation, Review of Economics and Statistics, 84 (2), 345-352 
7. Anderson and Van Wincoop’s (2003), Gravity with Gravitas: A Solution to the Border 
Puzzle, Source: The American Economic Review, Vol. 93, No. 1, (Mar., 2003), pp. 
170-192, Published by: American Economic Association 
8. Anderson, J.E. and E. van Wincoop (2004), Trade Costs, Journal of Economic 
Literature, 42(3), (Sep., 2004), pp. 691-751, published by American Economic 
Association 
9. Baier, S. L., and J. H. Bergstrand (2001), The Growth of World Trade: Tariffs, 
Transport Costs, and Income Similarity, Journal of International Economics, 53(1): 1-
27 
10. Baier, S. L., & Bergstrand, J. H. (2002). On the endogeneity of international trade flows 
and free trade agreements.  
11. Baldwin, R., & Venables, A. J. (2013). Spiders and snakes: offshoring and 
agglomeration in the global economy. Journal of International Economics, 90(2), 245-
254. 
12. Bardhan, P., & Mookherjee, D. (2006). Decentralisation and accountability in 
infrastructure delivery in developing countries. The Economic Journal, 116(508), 101-
127. 
98 
University of Ghana  http://ugspace.ug.edu.gh
13. Behar, A., & Venables, A. J. (2011). Transport costs and international trade. Handbook 
of transport economics, 97-115. 
14. Bergstrand, J. (1985), The Gravity Equation in International Trade. Source: The Review 
of Economics and Statistics 67, pp. 474-48, Published by MIT Press  
15. Bergstrand, J., & P. Egger (2011), Gravity equations and economic frictions in the 
world economy, Source: Palgrave handbook of international trade. Published by 
Palgrave Macmillan UK, 2013. 532-570.  
16. Berthouy & Hellene Ehrhartz (2013), Colonial Trade Spillovers  
17. Bougheas. S and Morgenroth E. Demetriades, P. O. (1999), Infrastructure, Transport 
Costs and Trade, Journal of International Economics: 169-189 
18. Chimilila, C., Sabuni, C., & Benjamin, A. (2014). Trade Facilitation in EAC Customs 
Union: Its Achievement and Implementation in Tanzania. Journal of Economics and 
Sustainable Development, 5(25), 1-15. 
19. Collier, P., & Gunning, J. W. (1999). Why has Africa grown slowly? The Journal of 
Economic Perspectives, 13(3), 3-22. 
20. Dee, P., Findlay, C., & Pomfret, R. (2006). Trade facilitation: what, why, how, where 
and when?. 
21. Dutt P, Traca D, (2010), Corruption and bilateral trade flows: extortion or evasion? 
Source: The Review of Economics and Statistics, Vol. 92, No. 4 (NOVEMBER 2010), 
pp. 843-860 
22. Eichengreen, B., & Irwin, D. A. (1998). The role of history in bilateral trade flows. In 
The regionalization of the world economy (pp. 33-62). University of Chicago Press. 
23. Fidrmuc J. and Fidrmuc J. (2009), Foreign Languages and Trade 
24. Global Economic Prospects (2005), Regional Trade Agreements: Effects on Trade 
25. Heady K. Mayerz T. (2013), Gravity Equations: Workhorse, Toolkit, and Cookbook 
26. Heckscher, E. F., & Ohlin, B. G. (1991). Heckscher-Ohlin trade theory. The MIT Press. 
27. Helble, M. C., Mann, C. L., & Wilson, J. S. (2009). Aid for trade facilitation. 
28. Hoekman M, and Kostecki M (2001), The political economy of the world trading 
system: the WTO and beyond  
29. Iwanow, T., & Kirkpatrick, C. (2009). Trade facilitation and manufactured exports: Is 
Africa different? World Development, 37(6), 1039-1050. 
30. Jebuni, C. D. (1997). Trade liberalization and regional integration in Africa. Regional 
integration and trade liberalization in sub-Saharan Africa, 1, 353-69. 
99 
University of Ghana  http://ugspace.ug.edu.gh
31. José de Sousay J. and Lochardz J. (2008), Trade and Colonial Status, Journal of African 
Economies, 21(3), 409-439. 
32. Kassee, D. (2014). The WTO Agreement on Trade Facilitation: status of play in 
Southern African Customs. World Customs Journal, 8(1), 101-112. 
33. Keane J, Calì M., and Kennan J. (2010), Impediments to Intra-Regional Trade in Sub-
Saharan Africa prepared for the Commonwealth Secretariat, Overseas Development 
Institute 
34. Kien T. N. (2009), Gravity Model by Panel Data Approach: An Empirical Application 
with Implications for the ASEAN Free Trade Area, Source: ASEAN Economic 
Bulletin, Vol. 26, No. 3 (December 2009), pp. 266-277, Published by: Institute of 
Southeast Asian Studies (ISEAS) 
35. Krugman, P. (1980). Scale economies, product differentiation, and the pattern of trade. 
The American Economic Review, 70(5), 950-959, Published by: American Economic 
Association 
36. Lee, J. W., & Swagel, P. (1997). Trade Barriers and Economic Growth: Theory and 
Evidence. Globalization, growth and sustainability, 15. 
37. Li Y, and Wilson J. S. (2009), Trade Facilitation and Expanding the Benefits of Trade: 
Evidence from Firm Level Data. Source: Asia-Pacific Research and Training Network 
on Trade Working Paper Series, No. 71, June 2009 
38. Limão, N., & Venables, A. (1999). Infrastructure Geographical Disadvantage and 
Costs. World Bank Policy Research Working Paper, (2257). 
39. Limao & Venables (2001), Infrastructure, Geographical Disadvantage, Transport 
Costs, and Trade, Source: The World Bank, Development Research Group, Policy 
Research Working Paper 2257  
40. Linders, G. J., & De Groot, H. L. (2006). Estimation of the gravity equation in the 
presence of zero flows. Tinbergen Institute Discussion Paper No. 06-072/3. Available 
at SSRN: https://ssrn.com/abstract=924160 
41. Lundgren, N. G. (1996). Bulk trade and maritime transport costs: The evolution of 
global markets. Resources Policy, 22(1-2), 5-32. 
42. McCallum J. (1995), National Borders Matter: US-Canada Regional Trade Patterns, 
Source: The American Economic Review, Vol. 85, No. 3, (Jun., 1995), pp. 615-623, 
Published by: American Economic Association 
43. Melitz, M. J. (2003). The impact of trade on intra‐industry reallocations and aggregate 
industry productivity. Econometrica, 71(6), 1695-1725. 
100 
University of Ghana  http://ugspace.ug.edu.gh
44. Moïsé, E. and S. Sorescu (2013), “Trade Facilitation Indicators: The Potential Impact 
of Trade Facilitation on Developing Countries' Trade”, OECD Trade Policy Papers,No. 
144, OECD Publishing, Paris. 
45. Muqayi, S., & Manyeruke, C. (2015). The Impact of the Chirundu One Stop Border 
Post in Addressing Border Protectionist Challenges. Mediterranean Journal of Social 
Sciences, 6(6 S2), 11. 
46. Nicita and Hoekman (2009), Trade Policy, Trade Costs, and Developing Country 
Trade, World Development, 39(12), 2069-2079 
47. Nkwemu M. (2011), The One Stop Border Post Concept – A Case of Chirundu Border 
Post Between Zambia And Zimbabwe, presented at the WTO Symposium in Geneva 
48. Obstfeld M. and Rogoff K. (2000), The Six Major Puzzles in International 
Macroeconomics: Is There a Common Cause? Source: NBER Macroeconomics 
Annual, Vol. 15 (2000), pp. 339-390 
49. Findlay, R., & O’Rourke, K. H. (2007). Power and plenty. Trade, War, and the World 
Economy. 
50. Perez and Wilson (2012), Export Performance and Trade Facilitation Reform: Hard & 
Soft Infrastructure, Source: World Development Vol. 40, No. 7, pp. 1295–1307, 2012, 
Published by Elsevier Ltd. 
51. Rose K. A., Lockwood B. and Quah D. (2000), One Money, One Market: The Effect 
of Common Currencies on Trade. Source: Economic Policy, Vol. 15, No. 30 (Apr., 
2000), pp. 7-45, Published by: Wiley on behalf of the Centre for Economic Policy 
Research, Center for Economic Studies, and the Maison des Sciences de l’Homme 
52. Rose, K. A. and Frankel J. A. (2000), Estimating the effect of currency unions on Trade 
and Output , Source: National Bureau of Economic Research Working Paper 7857, 
http://www.neber.org/papers/w7857 
53. Rose, K. A. and Glick (2001), Does a currency union affect trade? The time series 
evidence, qmanuscript 
54. Seck, A. (2014). Trade Facilitation and Trade Flows in Africa, Faculté des Sciences 
Economiques et de Gestion, Université Cheikh Anta Diop 
55. Shepherd, B. & Hoekman, B. (2015). Who profits from trade facilitation initiatives? 
Implications for African countries. Journal of African Trade, 2(1), 51-70. 
56. Trefler, D. (1993). Trade liberalization and the theory of endogenous protection: an 
econometric study of US import policy. Journal of political Economy, 101(1), 138-160. 
101 
University of Ghana  http://ugspace.ug.edu.gh
57. Turkson F. E. (2012), Logistics and Bilateral Exports in Developing Countries: A 
Multiplicative Form Estimation of the Logistics Augmented Gravity Equation, Source: 
Credit research paper, No.11/06 
58. Turkson F. E. (2011), Using Observable Trade Data to Measure Bilateral Trade Costs 
in Sub-Saharan Africa, (No. 12/06). CREDIT Research Paper. 
59. Tyson, J. (2015). Effect of Sub-Saharan African trade corridors on vulnerable groups. 
ODI Report. London: ODI. 
60. UNCTAD (2013), Economic Development for Africa Report, Intra-African Trade: 
Unlocking private sector dynamism, UNCTAD/ALDC/AFRICA/2013, United Nations 
Publication 
61. United Nations Economic Commission for Africa Report (2013): Trade Facilitation 
from an African Perspective, 2013, ECA Documents Publishing and Distribution Unit. 
62. Vicard, V. (2009), On Trade Creation and Regional Trade Agreements: Does Depth 
Matter? Source: Review of World Economics / Weltwirtschaftliches Archiv, Vol. 145, 
No. 2 (Jul.,2009), pp. 167-187 
63. Wooldridge, J. “M.(2002) Econometric Analysis of Cross Section and Panel Data”. The 
MIT Press 0, 5(1), 5. Chicago 
64. World Trade Report (2015), The theory and measurement of trade facilitation. Page 56-
71 
65. Wohlmuth, K. (2007). Abundance of natural resources and vulnerability to crises, 
conflicts and disasters. African Development Perspectives Yearbook, 3-48. 
66. Yi, K. M. (2010). Can multistage production explain the home bias in trade? The 
American Economic Review, 100(1), 364-393. 
 
  
102 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 1: Gravity Model Results with Aggregate LPI indicator for various Estimation 
Techniques 
Dependent variable:              (1) (2) (3) (4) 
log of exports OLS REM FEM HT 
lnxgdp 1.203*** 1.203*** 0.161 1.225*** 
 (0.0182) (0.0182) (0.197) (0.0258) 
lnmgdp 0.940*** 0.940*** 0.157 0.966*** 
 (0.0183) (0.0193) (0.211) (0.0259) 
lnxgdppc -0.110*** -0.110*** 0.0746 -0.243*** 
 (0.0240) (0.0253) (0.210) (0.0319) 
lnmgdppc -0.305*** -0.305*** 0.245 -0.394*** 
 (0.0245) (0.0259) (0.223) (0.0324) 
lndistwces -1.336*** -1.336***  -1.013*** 
 (0.0541) (0.0575)  (0.0854) 
lnxcostexport -0.811*** -0.811*** -0.252*** -0.693*** 
 (0.0469) (0.0597) (0.0627) (0.0472) 
lnmcostimport -0.175*** -0.175*** 0.119* -0.0943** 
 (0.0438) (0.0519) (0.0619) (0.0445) 
landlocked_o -0.432*** -0.432***  -0.646*** 
 (0.0714) (0.0774)  (0.0959) 
landlocked_d -1.080*** -1.080***  -1.184*** 
 (0.0721) (0.0755)  (0.0971) 
contig 1.497*** 1.497***  1.904*** 
 (0.221) (0.223)  (0.317) 
comlang_off 0.912*** 0.912***  0.964*** 
 (0.0730) (0.0715)  (0.102) 
comcol 0.160* 0.160  0.0860 
 (0.0939) (0.0976)  (0.132) 
comcur 0.861*** 0.861***  0.935*** 
 (0.217) (0.217)  (0.309) 
fta_wto 0.717*** 0.717*** 0.0681 0.439*** 
 (0.0884) (0.0840) (0.121) (0.115) 
xlpioverall 0.448*** 0.448*** 0.0657 0.269*** 
 (0.0387) (0.0478) (0.0423) (0.0377) 
mlpioverall 0.265*** 0.265*** -0.0139 0.109*** 
 (0.0379) (0.0427) (0.0413) (0.0369) 
Constant -18.63*** -18.63*** 4.191 -21.43*** 
 (0.776) (0.847) (5.234) (1.142) 
     
Observations 50,492 50,492 50,492 50,492 
Number of bilateraltradeid 8,178 8,178 8,178 8,178 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
 
  
103 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 2: Results for Disaggregated LPI for all trade involving SSA (Full Sample) 
 
Disaggregated Logistics Performance Indicators 
Dependent 
Overall 
variable:              Ability to 
Ease of 
LPI Customs 
log of exports Time Track and Logistics Infra 
Arranging 
Efficiency 
Trace Shipment 
lnxgdp 1.231*** 1.259*** 1.249*** 1.247*** 1.241*** 1.244*** 1.226*** 
  (0.0294) (0.0294) (0.0294) (0.0295) (0.0292) (0.0295) (0.0295) 
lnmgdp 0.973*** 0.985*** 0.994*** 0.991*** 0.973*** 0.983*** 0.954*** 
  (0.0295) (0.0295) (0.0295) (0.0296) (0.0293) (0.0296) (0.0296) 
lnxgdppc -0.231*** -0.209*** -0.214*** -0.227*** -0.223*** -0.234*** -0.198*** 
  (0.0358) (0.0357) (0.0356) (0.0358) (0.0357) (0.0360) (0.0358) 
lnmgdppc -0.360*** -0.357*** -0.348*** -0.354*** -0.359*** -0.367*** -0.344*** 
  (0.0364) (0.0363) (0.0363) (0.0365) (0.0363) (0.0365) (0.0364) 
lnxcostexport -0.553*** -0.568*** -0.544*** -0.547*** -0.580*** -0.558*** -0.594*** 
  (0.0542) (0.0540) (0.0542) (0.0541) (0.0539) (0.0540) (0.0539) 
lnmcostimport 0.0166  0.0233  0.0159  0.0203  0.0065  0.0221  0.0042 
  (0.0513) (0.0512) (0.0513) (0.0512) (0.0512) (0.0512) (0.0512) 
Exporter LPI 0.274*** 0.0682*** 0.136*** 0.175*** 0.206*** 0.184*** 0.141*** 
  (0.0388) (0.0247) (0.0262) (0.0304) (0.0302) (0.0317) (0.0267) 
Importer LPI 0.113*** 0.0545** -0.0224 0.0123 0.108*** 0.0744** 0.131*** 
  (0.0379) (0.0242) (0.0258) (0.0294) (0.0300) (0.0308) (0.0260) 
fta_wto 0.418*** 0.413*** 0.392*** 0.412*** 0.420*** 0.434*** 0.395*** 
  (0.1150) (0.1150) (0.1150) (0.1150) (0.1150) (0.1150) (0.1140) 
lndistwces -1.010*** -0.945*** -0.947*** -0.954*** -1.040*** -0.975*** -1.066*** 
  (0.0937) (0.0939) (0.0940) (0.0942) (0.0934) (0.0941) (0.0940) 
landlocked_o -0.746*** -0.735*** -0.740*** -0.743*** -0.738*** -0.733*** -0.728*** 
  (0.1080) (0.1080) (0.1080) (0.1080) (0.1080) (0.1080) (0.1090) 
landlocked_d -1.241*** -1.245*** -1.237*** -1.236*** -1.241*** -1.232*** -1.233*** 
  (0.1090) (0.1090) (0.1090) (0.1100) (0.1090) (0.1090) (0.1100) 
contig 1.913*** 2.012*** 2.023*** 2.005*** 1.861*** 1.961*** 1.816*** 
  (0.3550) (0.3560) (0.3560) (0.3570) (0.3550) (0.3570) (0.3590) 
comlang_off 0.953*** 0.980*** 0.971*** 0.970*** 0.950*** 0.958*** 0.962*** 
  (0.1160) (0.1160) (0.1160) (0.1160) (0.1160) (0.1160) (0.1170) 
comcol 0.1410  0.1300  0.1380  0.1470  0.1320  0.1430  0.0857 
  (0.1490) (0.1500) (0.1500) (0.1500) (0.1490) (0.1500) (0.1510) 
comcur 0.906*** 0.957*** 0.959*** 0.935*** 0.886** 0.945*** 0.884** 
  (0.3470) (0.3480) (0.3480) (0.3490) (0.3470) (0.3480) (0.3510) 
Constant -23.83*** -24.83*** -24.88*** -24.75*** -23.36*** -24.24*** -22.45*** 
  (1.4300) (1.4370) (1.4380) (1.4400) (1.4240) (1.4400) (1.4170) 
Observations 50492 50492 50492 50492 50492 50492 50492 
Number of 
8178 8178 8178 8178 8178 8178 8178 
bilateraltradeid 
Standard errors in parentheses           *** p<0.01, ** p<0.05, * p<0.1 
Appendix 3: Gravity Model Results for Disaggregated LPI for all SSA to ROW trade 
104 
University of Ghana  http://ugspace.ug.edu.gh
 
Disaggregated Logistics Performance Indicators  
Dependent 
Overall 
variable:              Ability to 
Ease of 
LPI Customs Timeliness Track and Logistics Infra Arranging log of exports Efficiency 
Trace Shipment 
lnxgdp 1.127*** 1.161*** 1.148*** 1.146*** 1.118*** 1.138*** 1.112*** 
  (0.0505) (0.0502) (0.0505) (0.0506) (0.0505) (0.0509) (0.0504) 
lnmgdp 1.071*** 1.095*** 1.102*** 1.089*** 1.074*** 1.076*** 1.065*** 
  (0.0433) (0.0428) (0.0432) (0.0433) (0.0431) (0.0436) (0.0432) 
lnxgdppc -0.380*** -0.378*** -0.385*** -0.391*** -0.379*** -0.395*** -0.372*** 
  (0.0635) (0.0631) (0.0635) (0.0636) (0.0637) (0.0639) (0.0640) 
lnmgdppc -0.321*** -0.319*** -0.297*** -0.313*** -0.312*** -0.326*** -0.291*** 
  (0.0538) (0.0527) (0.0531) (0.0533) (0.0537) (0.0541) (0.0530) 
lnxcostexport -0.395*** -0.381*** -0.361*** -0.376*** -0.408*** -0.382*** -0.414*** 
  (0.0847) (0.0845) (0.0851) (0.0846) (0.0842) (0.0843) (0.0844) 
lnmcostimport -0.136* (0.1160) (0.1250) (0.1260) -0.151* (0.1260) -0.162** 
  (0.0788) (0.0786) (0.0788) (0.0787) (0.0787) (0.0788) (0.0785) 
Exporter LPI 0.0238 -0.0233 0.0341 0.0132 0.0383 0.00408 0.0282 
  (0.0555) (0.0334) (0.0361) (0.0427) (0.0438) (0.0447) (0.0375) 
Importer LPI 0.193*** 0.131*** 0.00672 0.107** 0.138*** 0.155*** 0.117*** 
  (0.0632) (0.0421) (0.0445) (0.0496) (0.0495) (0.0525) (0.0450) 
fta_wto 0.482*** 0.406** 0.409** 0.470** 0.453** 0.503*** 0.431** 
  (0.1830) (0.1840) (0.1840) (0.1830) (0.1830) (0.1830) (0.1830) 
lndistwces -1.252*** -1.188*** -1.187*** -1.203*** -1.292*** -1.223*** -1.295*** 
  (0.1300) (0.1290) (0.1300) (0.1310) (0.1310) (0.1310) (0.1300) 
landlocked_o -1.049*** -1.066*** -1.085*** -1.071*** -1.041*** -1.048*** -1.045*** 
  (0.1560) (0.1550) (0.1560) (0.1560) (0.1560) (0.1560) (0.1560) 
landlocked_d -0.849*** -0.844*** -0.837*** -0.837*** -0.857*** -0.844*** -0.851*** 
  (0.1790) (0.1780) (0.1790) (0.1790) (0.1800) (0.1800) (0.1800) 
contig 1.848*** 1.990*** 1.990*** 1.943*** 1.800*** 1.885*** 1.800*** 
  (0.4300) (0.4260) (0.4300) (0.4300) (0.4310) (0.4330) (0.4320) 
comlang_off 1.169*** 1.194*** 1.194*** 1.187*** 1.167*** 1.173*** 1.176*** 
  (0.1720) (0.1700) (0.1710) (0.1720) (0.1720) (0.1730) (0.1730) 
comcol 0.1790  0.1910  0.1930  0.1990  0.1630  0.1900  0.1450 
  (0.2170) (0.2150) (0.2170) (0.2180) (0.2180) (0.2190) (0.2190) 
comcur 0.910** 0.982** 0.976** 0.933** 0.885** 0.937** 0.915** 
  (0.4060) (0.4020) (0.4050) (0.4060) (0.4070) (0.4080) (0.4080) 
Constant -20.95*** -22.90*** -22.78*** -22.19*** -20.20*** -21.44*** -19.88*** 
  (2.0940) (2.0890) (2.1060) (2.1050) (2.0840) (2.1090) (2.0730) 
Observations 27270 27270 27270 27270 27270 27270 27270 
Number of 
4566 4566 4566 4566 4566 4566 4566 
bilateraltradeid 
  
105 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 4: Gravity Model Results for Disaggregated LPI for all ROW-SSA trade 
 
Disaggregated Logistics Performance Indicators  
Dependent 
Overall 
variable:              Ability to 
LPI Customs Ease of 
log of exports Timeliness Track and Logistics Infra Efficiency Arranging 
Trace 
lnxgdp 1.212*** 1.242*** 1.236*** 1.226*** 1.234*** 1.227*** 1.208*** 
  (0.0331) (0.0331) (0.0332) (0.0341) (0.0327) (0.0332) (0.0337) 
lnmgdp 0.777*** 0.784*** 0.793*** 0.787*** 0.773*** 0.788*** 0.749*** 
  (0.0383) (0.0385) (0.0385) (0.0394) (0.0379) (0.0384) (0.0389) 
lnxgdppc -0.256*** -0.216*** -0.214*** -0.232*** -0.243*** -0.247*** -0.209*** 
  (0.0403) (0.0401) (0.0401) (0.0410) (0.0400) (0.0405) (0.0405) 
lnmgdppc -0.305*** -0.304*** -0.310*** -0.311*** -0.302*** -0.315*** -0.285*** 
  (0.0484) (0.0487) (0.0487) (0.0497) (0.0483) (0.0486) (0.0496) 
lnxcostexport -0.421*** -0.446*** -0.430*** -0.413*** -0.438*** -0.423*** -0.457*** 
  (0.0603) (0.0602) (0.0603) (0.0604) (0.0601) (0.0602) (0.0601) 
lnmcostimport 0.0814  0.0865  0.0733  0.0923  0.0821  0.0900  0.0866 
  (0.0595) (0.0594) (0.0597) (0.0596) (0.0592) (0.0593) (0.0596) 
Exporter LPI 0.423*** 0.158*** 0.173*** 0.231*** 0.317*** 0.276*** 0.255*** 
  (0.0459) (0.0305) (0.0321) (0.0361) (0.0355) (0.0381) (0.0327) 
Importer LPI 0.00424 -0.0117 -0.0776*** -0.0603** 0.0719** -0.000755 0.0911*** 
  (0.0393) (0.0238) (0.0259) (0.0300) (0.0314) (0.0314) (0.0265) 
fta_wto 0.656*** 0.613*** 0.663*** 0.656*** 0.644*** 0.674*** 0.631*** 
  (0.1340) (0.1350) (0.1350) (0.1340) (0.1340) (0.1340) (0.134) 
lndistwces -1.091*** -1.056*** -1.039*** -1.033*** -1.118*** -1.055*** -1.126*** 
  (0.1010) (0.1010) (0.1020) (0.1040) (0.1000) (0.1010) (0.103) 
landlocked_o -0.798*** -0.785*** -0.782*** -0.800*** -0.797*** -0.794*** -0.806*** 
  (0.1360) (0.1360) (0.1370) (0.1400) (0.1350) (0.1360) (0.139) 
landlocked_d -1.346*** -1.352*** -1.352*** -1.359*** -1.343*** -1.347*** -1.333*** 
  (0.1220) (0.1230) (0.1230) (0.1250) (0.1220) (0.1230) (0.125) 
contig 1.676*** 1.749*** 1.756*** 1.758*** 1.640*** 1.725*** 1.622*** 
  (0.3350) (0.3370) (0.3380) (0.3460) (0.3330) (0.3370) (0.343) 
comlang_off 0.580*** 0.616*** 0.604*** 0.602*** 0.577*** 0.588*** 0.596*** 
  (0.1310) (0.1320) (0.1320) (0.1350) (0.1300) (0.1320) (0.134) 
comcol 0.2370  0.2170  0.2270  0.2300  0.2400  0.2370  0.186 
  (0.1670) (0.1680) (0.1680) (0.1720) (0.1660) (0.1680) (0.171) 
comcur 0.626** 0.657** 0.661** 0.636* 0.617* 0.663** 0.610* 
  (0.3170) (0.3190) (0.3190) (0.3270) (0.3160) (0.3190) (0.326) 
Constant -19.71*** -20.38*** -20.41*** -20.37*** -19.74*** -20.20*** -18.68*** 
  (1.5630) (1.5720) (1.5770) (1.6020) (1.5480) (1.5690) (1.566) 
Observations 29905 29905 29905 29905 29905 29905 29,905 
Number of 
4690 4690 4690 4690 4690 4690 4,690 
bilateraltradeid 
106 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 5: Gravity Model Results for Frequency with which shipments reach consignee on 
time logistics measure – “lpitime” for various groupings/samples 
Dependent variable:              
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.259*** 1.161*** 1.242*** 
  (0.0294) (0.0502) (0.0331) 
lnmgdp 0.985*** 1.095*** 0.784*** 
  (0.0295) (0.0428) (0.0385) 
lnxgdppc -0.209*** -0.378*** -0.216*** 
  (0.0357) (0.0631) (0.0401) 
lnmgdppc -0.357*** -0.319*** -0.304*** 
  (0.0363) (0.0527) (0.0487) 
lnxcostexport -0.568*** -0.381*** -0.446*** 
  (0.0540) (0.0845) (0.0602) 
lnmcostimport 0.0233  (0.1160) 0.0865  
  (0.0512) (0.0786) (0.0594) 
xlpitime 0.0682*** -0.0233 0.158*** 
  (0.0247) (0.0334) (0.0305) 
mlpitime 0.0545** 0.131*** -0.0117 
  (0.0242) (0.0421) (0.0238) 
fta_wto 0.413*** 0.406** 0.613*** 
  (0.1150) (0.1840) (0.1350) 
lndistwces -0.945*** -1.188*** -1.056*** 
  (0.0939) (0.1290) (0.1010) 
landlocked_o -0.735*** -1.066*** -0.785*** 
  (0.1080) (0.1550) (0.1360) 
landlocked_d -1.245*** -0.844*** -1.352*** 
  (0.1090) (0.1780) (0.1230) 
contig 2.012*** 1.990*** 1.749*** 
  (0.3560) (0.4260) (0.3370) 
comlang_off 0.980*** 1.194*** 0.616*** 
  (0.1160) (0.1700) (0.1320) 
comcol 0.1300  0.1910  0.2170  
  (0.1500) (0.2150) (0.1680) 
comcur 0.957*** 0.982** 0.657** 
  (0.3480) (0.4020) (0.3190) 
Constant -24.83*** -22.90*** -20.38*** 
  (1.4370) (2.0890) (1.5720) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
 
Appendix 6: Gravity Model Results for the Ability to Track and Trace Consignments logistics 
107 
University of Ghana  http://ugspace.ug.edu.gh
measure–“lpiatratr” for various groupings/samples 
Dependent variable:     
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.249*** 1.148*** 1.236*** 
  (0.0294) (0.0505) (0.0332) 
lnmgdp 0.994*** 1.102*** 0.793*** 
  (0.0295) (0.0432) (0.0385) 
lnxgdppc -0.214*** -0.385*** -0.214*** 
  (0.0356) (0.0635) (0.0401) 
lnmgdppc -0.348*** -0.297*** -0.310*** 
  (0.0363) (0.0531) (0.0487) 
lnxcostexport -0.544*** -0.361*** -0.430*** 
  (0.0542) (0.0851) (0.0603) 
lnmcostimport 0.0159  (0.1250) 0.0733  
  (0.0513) (0.0788) (0.0597) 
xlpiatratr  0.136*** 0.0341 0.173*** 
  (0.0262) (0.0361) (0.0321) 
mlpiatratr  -0.0224 0.00672 -0.0776*** 
  (0.0258) (0.0445) (0.0259) 
fta_wto 0.392*** 0.409** 0.663*** 
  (0.1150) (0.1840) (0.1350) 
lndistwces -0.947*** -1.187*** -1.039*** 
  (0.0940) (0.1300) (0.1020) 
landlocked_o -0.740*** -1.085*** -0.782*** 
  (0.1080) (0.1560) (0.1370) 
landlocked_d -1.237*** -0.837*** -1.352*** 
  (0.1090) (0.1790) (0.1230) 
contig 2.023*** 1.990*** 1.756*** 
  (0.3560) (0.4300) (0.3380) 
comlang_off 0.971*** 1.194*** 0.604*** 
  (0.1160) (0.1710) (0.1320) 
comcol 0.1380  0.1930  0.2270  
  (0.1500) (0.2170) (0.1680) 
comcur 0.959*** 0.976** 0.661** 
  (0.3480) (0.4050) (0.3190) 
Constant -24.88*** -22.78*** -20.41*** 
  (1.4380) (2.1060) (1.5770) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
  
108 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 7: Gravity Model Results for Competence and Quality of Logistics Services 
measure–“lpicqols” for various groupings/samples 
 
Dependent variable:     
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.247*** 1.146*** 1.226*** 
  (0.0295) (0.0506) (0.0341) 
lnmgdp 0.991*** 1.089*** 0.787*** 
  (0.0296) (0.0433) (0.0394) 
lnxgdppc -0.227*** -0.391*** -0.232*** 
  (0.0358) (0.0636) (0.0410) 
lnmgdppc -0.354*** -0.313*** -0.311*** 
  (0.0365) (0.0533) (0.0497) 
lnxcostexport -0.547*** -0.376*** -0.413*** 
  (0.0541) (0.0846) (0.0604) 
lnmcostimport 0.0203  (0.1260) 0.0923  
  (0.0512) (0.0787) (0.0596) 
xlpicqols 0.175*** 0.0132 0.231*** 
  (0.0304) (0.0427) (0.0361) 
mlpicqols 0.0123 0.107** -0.0603** 
  (0.0294) (0.0496) (0.0300) 
fta_wto 0.412*** 0.470** 0.656*** 
  (0.1150) (0.1830) (0.1340) 
lndistwces -0.954*** -1.203*** -1.033*** 
  (0.0942) (0.1310) (0.1040) 
landlocked_o -0.743*** -1.071*** -0.800*** 
  (0.1080) (0.1560) (0.1400) 
landlocked_d -1.236*** -0.837*** -1.359*** 
  (0.1100) (0.1790) (0.1250) 
contig 2.005*** 1.943*** 1.758*** 
  (0.3570) (0.4300) (0.3460) 
comlang_off 0.970*** 1.187*** 0.602*** 
  (0.1160) (0.1720) (0.1350) 
comcol 0.1470  0.1990  0.2300  
  (0.1500) (0.2180) (0.1720) 
comcur 0.935*** 0.933** 0.636* 
  (0.3490) (0.4060) (0.3270) 
Constant -24.75*** -22.19*** -20.37*** 
  (1.4400) (2.1050) (1.6020) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
  
109 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 8: Gravity Model Results for Efficiency of Customs Procedures measure–“Customs 
Efficiency” for various groupings/samples 
Dependent variable:     
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.241*** 1.118*** 1.234*** 
  (0.0292) (0.0505) (0.0327) 
lnmgdp 0.973*** 1.074*** 0.773*** 
  (0.0293) (0.0431) (0.0379) 
lnxgdppc -0.223*** -0.379*** -0.243*** 
  (0.0357) (0.0637) (0.0400) 
lnmgdppc -0.359*** -0.312*** -0.302*** 
  (0.0363) (0.0537) (0.0483) 
lnxcostexport -0.580*** -0.408*** -0.438*** 
  (0.0539) (0.0842) (0.0601) 
lnmcostimport 0.0065  -0.151* 0.0821  
  (0.0512) (0.0787) (0.0592) 
xlpicustoms 0.206*** 0.0383 0.317*** 
  (0.0302) (0.0438) (0.0355) 
mlpicustoms 0.108*** 0.138*** 0.0719** 
  (0.0300) (0.0495) (0.0314) 
fta_wto 0.420*** 0.453** 0.644*** 
  (0.1150) (0.1830) (0.1340) 
lndistwces -1.040*** -1.292*** -1.118*** 
  (0.0934) (0.1310) (0.1000) 
landlocked_o -0.738*** -1.041*** -0.797*** 
  (0.1080) (0.1560) (0.1350) 
landlocked_d -1.241*** -0.857*** -1.343*** 
  (0.1090) (0.1800) (0.1220) 
contig 1.861*** 1.800*** 1.640*** 
  (0.3550) (0.4310) (0.3330) 
comlang_off 0.950*** 1.167*** 0.577*** 
  (0.1160) (0.1720) (0.1300) 
comcol 0.1320  0.1630  0.2400  
  (0.1490) (0.2180) (0.1660) 
comcur 0.886** 0.885** 0.617* 
  (0.3470) (0.4070) (0.3160) 
Constant -23.36*** -20.20*** -19.74*** 
  (1.4240) (2.0840) (1.5480) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
  
110 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 9: Gravity Model Results for Quality of Trade and Transport Infrastructure 
measure– “Infrastructure” for various groupings/samples 
Dependent variable:        
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.244*** 1.138*** 1.227*** 
  (0.0295) (0.0509) (0.0332) 
lnmgdp 0.983*** 1.076*** 0.788*** 
  (0.0296) (0.0436) (0.0384) 
lnxgdppc -0.234*** -0.395*** -0.247*** 
  (0.0360) (0.0639) (0.0405) 
lnmgdppc -0.367*** -0.326*** -0.315*** 
  (0.0365) (0.0541) (0.0486) 
lnxcostexport -0.558*** -0.382*** -0.423*** 
  (0.0540) (0.0843) (0.0602) 
lnmcostimport 0.0221  (0.1260) 0.0900  
  (0.0512) (0.0788) (0.0593) 
xpliqinfra 0.184*** 0.00408 0.276*** 
  (0.0317) (0.0447) (0.0381) 
mpliqinfra 0.0744** 0.155*** -0.000755 
  (0.0308) (0.0525) (0.0314) 
fta_wto 0.434*** 0.503*** 0.674*** 
  (0.1150) (0.1830) (0.1340) 
lndistwces -0.975*** -1.223*** -1.055*** 
  (0.0941) (0.1310) (0.1010) 
landlocked_o -0.733*** -1.048*** -0.794*** 
  (0.1080) (0.1560) (0.1360) 
landlocked_d -1.232*** -0.844*** -1.347*** 
  (0.1090) (0.1800) (0.1230) 
contig 1.961*** 1.885*** 1.725*** 
  (0.3570) (0.4330) (0.3370) 
comlang_off 0.958*** 1.173*** 0.588*** 
  (0.1160) (0.1730) (0.1320) 
comcol 0.1430  0.1900  0.2370  
  (0.1500) (0.2190) (0.1680) 
comcur 0.945*** 0.937** 0.663** 
  (0.3480) (0.4080) (0.3190) 
Constant -24.24*** -21.44*** -20.20*** 
  (1.4400) (2.1090) (1.5690) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
  
111 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 10: Gravity Model Results for Ease of Arranging competitively priced shipment –“ 
lpieoacps” for various groupings/samples 
 
Dependent variable:      
Involving SSA SSA to ROW ROW to SSA 
log of exports 
lnxgdp 1.226*** 1.112*** 1.208*** 
  (0.0295) (0.0504) (0.0337) 
lnmgdp 0.954*** 1.065*** 0.749*** 
  (0.0296) (0.0432) (0.0389) 
lnxgdppc -0.198*** -0.372*** -0.209*** 
  (0.0358) (0.0640) (0.0405) 
lnmgdppc -0.344*** -0.291*** -0.285*** 
  (0.0364) (0.0530) (0.0496) 
lnxcostexport -0.594*** -0.414*** -0.457*** 
  (0.0539) (0.0844) (0.0601) 
lnmcostimport 0.0042  -0.162** 0.0866  
  (0.0512) (0.0785) (0.0596) 
xlpieoacps 0.141*** 0.0282 0.255*** 
  (0.0267) (0.0375) (0.0327) 
mlpieoacps 0.131*** 0.117*** 0.0911*** 
  (0.0260) (0.0450) (0.0265) 
fta_wto 0.395*** 0.431** 0.631*** 
  (0.1140) (0.1830) (0.1340) 
lndistwces -1.066*** -1.295*** -1.126*** 
  (0.0940) (0.1300) (0.1030) 
landlocked_o -0.728*** -1.045*** -0.806*** 
  (0.1090) (0.1560) (0.1390) 
landlocked_d -1.233*** -0.851*** -1.333*** 
  (0.1100) (0.1800) (0.1250) 
contig 1.816*** 1.800*** 1.622*** 
  (0.3590) (0.4320) (0.3430) 
comlang_off 0.962*** 1.176*** 0.596*** 
  (0.1170) (0.1730) (0.1340) 
comcol 0.0857  0.1450  0.1860  
  (0.1510) (0.2190) (0.1710) 
comcur 0.884** 0.915** 0.610* 
  (0.3510) (0.4080) (0.3260) 
Constant -22.45*** -19.88*** -18.68*** 
  (1.4170) (2.0730) (1.5660) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
  
112 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 11: Gravity Model Results for the Overall logistics performance indicator – 
“lpioverall” for various groupings/samples 
Dependent variable:             
Full Sample  SSA to ROW  Row to SSA  
log of exports 
lnxgdp 1.231*** 1.127*** 1.212*** 
  (0.0294) (0.0505) (0.0331) 
lnmgdp 0.973*** 1.071*** 0.777*** 
  (0.0295) (0.0433) (0.0383) 
lnxgdppc -0.231*** -0.380*** -0.256*** 
  (0.0358) (0.0635) (0.0403) 
lnmgdppc -0.360*** -0.321*** -0.305*** 
  (0.0364) (0.0538) (0.0484) 
lnxcostexport -0.553*** -0.395*** -0.421*** 
  (0.0542) (0.0847) (0.0603) 
lnmcostimport 0.0166  -0.136* 0.0814  
  (0.0513) (0.0788) (0.0595) 
xlpioverall 0.274*** 0.0238 0.423*** 
  (0.0388) (0.0555) (0.0459) 
mlpioverallI 0.113*** 0.193*** 0.00424 
  (0.0379) (0.0632) (0.0393) 
fta_wto 0.418*** 0.482*** 0.656*** 
  (0.1150) (0.1830) (0.1340) 
lndistwces -1.010*** -1.252*** -1.091*** 
  (0.0937) (0.1300) (0.1010) 
landlocked_o -0.746*** -1.049*** -0.798*** 
  (0.1080) (0.1560) (0.1360) 
landlocked_d -1.241*** -0.849*** -1.346*** 
  (0.1090) (0.1790) (0.1220) 
contig 1.913*** 1.848*** 1.676*** 
  (0.3550) (0.4300) (0.3350) 
comlang_off 0.953*** 1.169*** 0.580*** 
  (0.1160) (0.1720) (0.1310) 
comcol 0.1410  0.1790  0.2370  
  (0.1490) (0.2170) (0.1670) 
comcur 0.906*** 0.910** 0.626** 
  (0.3470) (0.4060) (0.3170) 
Constant -23.83*** -20.95*** -19.71*** 
  (1.4300) (2.0940) (1.5630) 
Observations 50492 27270 29905 
Number of bilateraltradeid 8178 4566 4690 
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 
113 
University of Ghana  http://ugspace.ug.edu.gh
Appendix 12 Time Effects Diagnostics - Pooled Regression correcting for Time 
 
114