UNIVERSITY OF ZIMBABWE FACULTY OF SOCIAL STUDIES DEPARTMENT OF ECONOMICS The Extent and Determinants of Intra Industry Trade in the Food Industry: The Case of Zimbabwe and its Five SADC Trading Partners (2000-2012) BY MATSURO LEON A dissertation submitted in partial fulfilment of the requirements of the Master of Science Degree in Economics (MSc Econ). June 2014
74
Embed
UNIVERSITY OF ZIMBABWEir.uz.ac.zw/jspui/bitstream/10646/3064/1/Matsuro_The... · 2019-10-16 · UN Comtrade shows evidence of two way exchange of goods within the same product category.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
UNIVERSITY OF ZIMBABWE
FACULTY OF SOCIAL STUDIES
DEPARTMENT OF ECONOMICS
The Extent and Determinants of Intra Industry Trade in the Food Industry:
The Case of Zimbabwe and its Five SADC Trading Partners (2000-2012)
BY
MATSURO LEON
A dissertation submitted in partial fulfilment of the requirements of the
Master of Science Degree in Economics (MSc Econ).
June 2014
i
DEDICATION
I dedicate this dissertation to my loving parents Jacob and Eunice, brothers, sisters, relatives and friends.
ii
ACKNOWLEDGEMENTS
Special mention goes to the University of Zimbabwe, the Economics Department in particular, for
affording me an opportunity to pursue my masters’ studies. I wish to mention the invaluable and priceless
support I received from my supervisor Dr. A. Makochekanwa, my International Trade lecturer at the
Joint Facility for Electives (JFE), Professor S. Buiguit and all Economics Department lecturers and staff.
The journey could not have been a fruitful one without my fellow college mates; Earnest, Dennis,
Runesu, Happiness, Godfrey, Cherish and Elson. To all of you I say thank you. To my friend Hillary
Makaya, thank you for being a true friend.
To my family, no words can describe all what you have been to me. I owe my success to you. You have
stood by me through all the turbulent times of my life, offering the much needed financial and emotional
support. You always make me believe in myself, indeed you motivate me to realize more than my average
potential.
My special gratitude also goes to the African Economic Research Consortium (AERC) for the financial
support and for affording me an opportunity to be part of the 2013 Joint Facility for Electives (JFE) in
Kenya (Nairobi).
Above all I praise my Lord for all that I am.
iii
ABSTRACT
Theoretical models of intra industry trade (IIT) have explained it using features of developed countries,
and to this end many studies have mainly focused on industrialised nations. The turn of the millennium
witnessed Zimbabwe reorienting its trade away from traditional partners, particularly the European
Union, towards the SADC region. Zimbabwe’s bilateral trade data with its SADC trade partners from
UN Comtrade shows evidence of two way exchange of goods within the same product category. This
study endeavors to ascertain the extent and determinants of IIT between Zimbabwe and its five SADC
trade partners (Botswana, Malawi, Mozambique, South Africa and Zambia) in the food manufacturing
industry.
The study calculated Grubel- Lloyd Indices for Zimbabwe’s bilateral trade with five of its trade partners
and found out that intra industry trade exists between Zimbabwe and its trade partners. However, IIT is
still low. Furthermore, the study employed the gravity model to find the significant country specific
determinants of IIT. Using panel data for five of Zimbabwe’ s trade partners over the period 2000 to
2012, the study estimated a pooled Ordinary Least Squares model in Stata. From the estimated results,
the study found that the product of partners’ GDP, the differences in partners’ GDP, weighted distance,
dummy variables for common boarder and common language were significant factors in explaining IIT
between Zimbabwe and its SADC partners in manufactured food products.
iv
LIST OF ACRONYMS
ASEAN Association of South-East Asian Nations
CZI Confederations of Zimbabwe Industries
COMTRADE Common Format for Transient Data Exchange
FAO Food and Agriculture Organization
FTA Free Trade Area
GDP Gross Domestic Product
HS Harmonized Commodity Description and Coding System
IDP Industrial Development Policy
IIT Intra Industry Trade
INT Inter Industry Trade
Mercosur MERcado COmún del SUR
MPS Monetary Policy Statement
PCI Per Capita Income
RISDP Regional Indicative Strategic Development Plan
RTA Regional Trade Agreement
SAARC South Asian Association for Regional Cooperation
SACU Southern African Customs Union
SADC Southern African Development Community
SAP Structural Adjustment Programme
UN United Nations
UNCTAD United Nations Conference on Trade and Development
WTO World Trade Organization
Zimstat Zimbabwe National Statistical Agency
v
TABLE OF CONTENT
DEDICATION............................................................................................................................................................... i
ACKNOWLEDGEMENTS ............................................................................................................................................ ii
ABSTRACT ................................................................................................................................................................ iii
LIST OF ACRONYMS ................................................................................................................................................. iv
TABLE OF CONTENT ........................................................................................................................................... v
LIST OF TABLES ...................................................................................................................................................... viii
LIST OF FIGURES ...................................................................................................................................................... ix
CHAPTER ONE ........................................................................................................................................................... 1
INTRODUCTION AND BACKGROUND ........................................................................................................................ 1
1.2. Statement of the problem ........................................................................................................................ 11
1.3. Study Objectives ..................................................................................................................................... 12
1.4. Research Hypothesis .............................................................................................................................. 13
1.5. Research questions ................................................................................................................................. 13
1.6. Significance of the study ........................................................................................................................ 13
1.7. Scope of the study .................................................................................................................................. 14
1.8. Organisation of the study ........................................................................................................................ 14
CHAPTER TWO ........................................................................................................................................................ 15
LITERATURE REVIEW............................................................................................................................................... 15
CHAPTER THREE ...................................................................................................................................................... 29
RESEARCH METHODOLOGY .................................................................................................................................... 29
3.0.1. Measuring IIT in the Food Manufacturing Industry ............................................................................. 29
3.0.2. The Gravity Model ............................................................................................................................... 31
3.1. The Empirical Model ................................................................................................................................... 32
3.2. Definition and Measurement of Variables ................................................................................................... 33
3.2.1. Intra-industry Trade Index ( ijkIITFM ) ............................................................................................... 33
3.2.2. Product of Gross Domestic Product between Zimbabwe and Partner k ( jkRGDP ) ........................... 34
3.2.3. Dissimilarity in Per Capita Income ( DPCIjk ) ................................................................................... 34
3.2.4. Per Capita Income ( kPCI ) ................................................................................................................... 35
3.2.7. Real Exchange Rate ( jkRER ) ............................................................................................................. 36
3.2.8. Common Border (D1) ........................................................................................................................... 37
3.2.9. Common Language (D2) ....................................................................................................................... 37
3.2.9. Free trade Area Dummy 3D ............................................................................................................. 37
3.3 Data Sources and Problems .......................................................................................................................... 38
CHAPTER FOUR ....................................................................................................................................................... 41
ESTIMATION AND RESULTS .................................................................................................................................... 41
4.1. Results of Intra Industry Trade Shares ........................................................................................................ 41
4.3. Econometric Tests and Estimation of Results ............................................................................................. 44
4.4. Estimation of the Model .............................................................................................................................. 45
CHAPTER FIVE ......................................................................................................................................................... 50
CONCLUSIONS AND POLICY RECOMMENDATIONS ................................................................................................ 50
5.1. Conclusions of the Study ............................................................................................................................. 50
5.2. Policy Implications and Recommendations ................................................................................................ 51
5.3. Study Limitations and Areas for Further Research ..................................................................................... 53
The world has increasingly become a global village, and with economic interdependence now a
common feature, there has been a proliferation of Regional Trade Agreements (RTA) (Kalaba and
Tsedu, 2008). For developing countries, the challenge has been on how to participate more
effectively in the world economy especially given their relatively small economies (Aybodji,
2008). To this end, nations in the Southern African region have sought to integrate into the world
economy through the establishment of the Southern African Development Community (SADC).
2 Common Market for Eastern And Southern Africa 3 East African Community 4 Southern African Development Community
4
1.1.1 Southern African Development Community (SADC)
SADC is a fifteen member5 regional economic grouping formerly known as Southern African
Development Coordination Conference (SADCC), which was transformed into a formal treaty
based organisation in August 1992 (SADC Secretariat). The Windhoek Extra Ordinary Summit of
March 2001 gave impetus for the formulation of the Regional Indicative Strategic Development
Plan (RISDP).
In 2003 member states adopted the RISDP, a policy document which outlines the bloc’s regional
economic agenda. The RISDP defines four6 clusters within which policies and strategies are to be
evaluated with the intention of deepening regional integration and cooperation. Intra-regional trade
in the block is influenced by the SADC Protocol on Trade whose objectives as stated in Article 2
are:
1. ‘To further liberalise intra-regional trade in goods and services on the basis of fair, mutually
equitable and beneficial trade arrangements, complemented by Protocols in other areas’.
2. ‘To ensure efficient production within SADC reflecting the current and dynamic comparative
advantages of its members’.
3. ‘To contribute towards the improvement of the climate for domestic, cross-border and foreign
investment’.
4. ‘To enhance the economic development, diversification and industrialisation of the Region’.
5. ‘To establish a Free Trade Area in the SADC Region’.
To achieve the stated objectives as is enunciated in the RISDP and outlined in the amended Trade
Protocol (2005), the region sought to establish a free trade area (FTA) by 2008, a customs union
(CU) by 2010, a monetary union (MU) by 2016, with the ultimate goal of achieving a single
currency by 20187.
The FTA was attained in August 2008 after the minimum conditions of the FTA attainment were
met (85% of intra-regional trade had attained free duty). However, the tariff phase down process
5 SADC member states are Angola, Botswana, Democratic Republic of Congo, Madagascar, Lesotho, Malawi, Mauritius, Mozambique, Namibia, South Africa, Swaziland, Seychelles, Zambia and Zimbabwe 6 Trade, Industry, Finance and Investment (TIFI); Infrastructure and Services (IS); Food, Agriculture and Natural Resources (FANR); and the Social and Human Development and Special Programmes (SHDSP) cluster. 7 http://www.sadc.int/about-sadc/integration-milestones/free-trade-area/
Source: Own calculations10 from SADC Secretariat Statistics Unit, Trade Database (2013)
Statistics from the World Bank indicate that South Africa is the dominant import source for the
SADC countries. In 2012, South Africa accounted for 62.8%, 56.2%, 35.7%, 31.4% and 25% of
Botswana, Zimbabwe, Zambia, Mozambique, and Malawi’s imports respectively. This might be a
possible explanation for the relatively high intra SADC trade shares of the considered SADC
countries. These statistics confirm Cattaneo and Fryer (2003) observation that South Africa is the
dominant source of imports for a number of SADC countries.
1.2. Statement of the problem
Most empirical studies of IIT have mainly focused on developed countries because theories of IIT
are based on features of industrialized nations (Al-Mawali, 2005). Due to that fact the much of
SSA’s exports are driven by primary products (minerals, oil and agricultural produce), the general
view has been that developing countries do not engage in IIT amongst themselves. Zimbabwe’s
IIT is severely limited in literature, however, trade statistics from UN Comtrade show that
Zimbabwe is both an exporter and importer of manufactured food products belonging to the same
statistical category, and this is especially so for its trade with the SADC region.
10 Intra SADC trade share is calculated as a proportion of a country’s total trade with the SADC block as a fraction of that country’s total trade i.e. total SADC trade and trade with the rest of the world.
12
The volume of intra SADC trade has tripled over the past decade; however it has been noted that
traded commodities lacked diversification with four sectors11 accounting for about 98% of intra
SADC trade (SADC, 2012). Despite increases in intra-regional trade volumes, studies of IIT have
found that IIT in SSA is generally low compared to other developing regions of the world
(ASEAN, Mercosur and SAARC). The South- South trade monitor reports the GL intra-industry
trade index in the SADC region at around 0.05 (UNCTAD, 2013). Thus trade in general is
dominated by INT.
Proponents of IIT argue that it is beneficial as it allows exploitation of economies of scale,
stimulates innovation and has relatively minimal reallocation effects on factors of production and
hence their returns. However, with the low IIT, SSA and Zimbabwe in particular will potentially
lose out on benefits associated with IIT. Moves towards opening up for trade may face stiff
resistance from the import competing sectors, as these sectors may be hurt. This is particularly the
case with manufacturing companies in Zimbabwe, most of which are calling for higher import
duties. The question then is, what are the determinants of IIT between Zimbabwe and its five
SADC trade partners in the food manufacturing industry?
1.3. Study Objectives
The study examines the current trade patterns of Zimbabwe’s food manufacturing industry brought
by the several developments that reshaped the industry over the past decade. Furthermore, it
identifies country specific factors that determine the degree of IIT between Zimbabwe and its
SADC trading partners over the period 2000 to 2012. The specific objectives are:
To ascertain the pattern and magnitude of IIT between Zimbabwe and its SADC trading
partners in the food manufacturing industry.
To identify the determinants of IIT between Zimbabwe and its trading partners in SADC.
11 unprocessed agricultural products, food manufacture, textiles and clothing
13
1.4.Research Hypothesis
The study seeks to test the following hypothesis
a) IIT does not necessarily take place between countries of small economic size and at the
same level of development.
b) Similarity in per capita income is not the main determinant of IIT
1.5.Research questions
a) Is there IIT between Zimbabwe and its SADC trading partners in the food manufacturing
sector?
b) What is the extent of IIT between Zimbabwe and SADC?
c) What are the significant determinants of IIT between Zimbabwe and SADC?
1.6.Significance of the study
Empirical studies on IIT are many in the literature; however, ‘little research has been done on
developing countries’, Al-Mawali (2005, p. 407). To the best of our knowledge, with the exception
of Sunde et al (2009), Zimbabwe – SADC trade has been neglected in empirical studies of IIT.
Comparing Sunde et al (2009) with this study is impossible for two main reasons; firstly their
study covers part of the period when the country was sliding into an economic crisis (1990-2006)
while ours covers the period when the country was sliding into the economic crisis, the crisis period
and the post crisis period (2000-2012). Hence relying on their results for policy may be misleading.
Secondly our study specifies the industry concerned (food manufacturing industry) while theirs
had a macroeconomic bias.
Furthermore, no specific study to date has been published on our research area. However,
Zimbabwe - SADC trade in processed food products has become much more important than before
in recent years due to the surge in imported food products. The study, therefore, attempts to fill
this gap by examining the recent changes in the trade patterns of the food manufacturing industry
in Zimbabwe.
14
By evaluating the existence of IIT in food products, the study determines whether IIT takes place
among countries with similar economic structures. Furthermore, it sheds light on the likely effects
of further opening up of trade on Zimbabwe’s food manufacturing sector especially considering
the envisaged SADC customs and monetary union. This study will thus contribute to the empirical
literature on Zimbabwe’s IIT especially in the food manufacturing sector. Bringing to the fore,
evidence to ascertain the nature of trade in the food manufacturing industry and thus give policy
recommendations on whether it is justified for industry to call for protection.
1.7. Scope of the study
The study utilizes panel data for the period 2000 to 2012, for Zimbabwe and five of its major trade
partners in the SADC region. The countries included are Botswana, Malawi, Mozambique, South
Africa and Zambia. The choice of these countries is motivated chiefly by the availability of data
and also the fact that they are engaged in significant trade with Zimbabwe.
1.8. Organisation of the study
This chapter gave the background and overview of the study. The remaining part of the study is
organised as follows; Chapter 2 covers the literature review, Chapter 3 outlines the methodology,
with Chapter 4 presenting the estimation and results, while Chapter 5 presents the findings and
policy recommendations.
15
CHAPTER TWO
LITERATURE REVIEW
2.0. Introduction
This chapter gives an outline of the literature review, bringing to the fore prominent aspects in
trade theory as well as empirical findings related to the field of IIT. The theoretical review focuses
on the evolution of trade theory, explaining the developments from traditional trade theory to new
trade theories (explaining IIT). The section of empirical review looks at studies that have been
done on IIT, their findings and contribution to IIT literature.
2.1. Theoretical review
International trade can be broadly classified into inter-industry trade (INT) and intra-industry trade
(IIT). The two types of trade are largely distinguished through their sources and therefore the
theories that explain them. Ates and Turkcan (2010) define IIT as, ‘the simultaneous export and
import of products which belong to the same statistical category’ (p, 16), a definition which we
will use for the purpose of this study.
2.1.1. A brief review of traditional theories of trade
This section explains the traditional theories of trade, all of which explain INT which essentially
results from supply side factors related to differences in endowments. The theories predict that
similarly endowed nations have no reason to trade particularly if the trade involves an identical
commodity.
i. Absolute advantage (Adam Smith)
The theory of absolute advantage in trade can be traced to the work of Adam Smith (1776),
(Makochekanwa, 2007). The theory postulates that two countries can benefit from trade if they
specialize in the production of products in which they are more efficient and exchanging part of
16
the output for the product in which they are less efficient. Thus, a country would specialize in
production of a good in which it has absolute advantage and export the excess to finance purchases
of the good in which it has absolute disadvantage.
ii. Comparative advantage (David Ricardo)
According to Davis (1998), the Ricardian theory is premised on the fact that, ‘technical differences
matter for trade patterns when expansion of an individual sector does not drive up marginal
opportunity costs’ (p, 203). The basic insight of the theory which departs from the absolute
advantage theory is that trade is dependent on comparative not absolute advantage. Sen (2010)
states that the Ricardian theory postulates that comparative advantage is a necessary and sufficient
condition for mutually beneficial trade as it warrants complete specialisation in the commodity
with which a country has a comparative advantage in terms of labour hours devoted per unit output.
iii. Heckscher-Ohlin (HO) Factor Proportions Theory
The HO factor proportions theory of comparative advantage postulates that international trade
offsets the uneven geographical distribution of productive resources. It is premised on the
assumption that nations have different relative factor endowments and unlike Smith and Ricardo’s
theories, it considers capital as an additional factor of production. The theory predicts that a
country would specialize and export a good which uses intensively a factor in which it is relatively
abundantly endowed, and in turn import a good which uses intensively the factor in which it is
scarcely endowed. The basic insight of the model is that traded goods constitute bundles of factors
(labour and capital), thus “international exchange of commodities is therefore indirect factor
arbitrage, transferring the services of otherwise immobile factors of production from locations
where these factors are abundant to locations where they are scarce” (Leamer, 1995; p. 1).
The differences in endowments reflect differences in factor prices and hence the ultimate product
prices, thus in autarky differently endowed countries will have different terms of trade which forms
the basis for trade. Opening up for trade will result in relative commodity price equalization as
countries eliminate the excess supply and demand in their countries. Assuming there are no barriers
to trade, consumers will purchase goods from a cheaper source as long as there are price
17
differentials, until relative prices equalize in the two countries. At such a point there will not be
any excess demand or supply as export supply even out import demand (Markusen et al, 1995).
iv. The Factor Price Equalization Theorem (FPET)
The relative abundance of productive factors within a country determines those factors' relative
costs (Hanink, 1988). As countries reorganize production upon opening up for trade, there will be
more demand for the abundant factor and less demand for the scarce factor. This comes about as
countries specialize in the production of the product using the abundant factor intensively. The
prices of the abundant factor in both countries will increase while those of the scarce factor falls
until the relative factor prices in both nations are equalized as countries simultaneously realize
relative commodity price equalization.
v. Stolper - Samuelson Theorem (SST)
The theorem explains the distributive effects of trade on returns to factors of production. It predicts
that opening up for trade would reward the abundant factor at the expense of the scarce factor. A
capital abundant country would thus demand more capital to increase its output of a capital
intensive good to meet increased demand from the international market. It will draw some of this
additional capital from the labour intensive industry and in so doing increase the marginal
productivity of capital in both sectors (the labour and capital intensive industries), simultaneously
reducing that of labour. Since factors are paid their marginal productivities, capital in this case is
rewarded, and labour is at the losing end.
vi. The specific factors model (SFM)
According to Markusen et al (1995) work on the SFM, can be traced to Jones (1971) and
Samuelson (1971). The assumptions building this model are the same as those of the HO theorem,
with the only difference being the existence of capital specific to sectors. ‘With sector-specific
capital and mobile labour, the model shows that not all units of a factor have the same interest in
the opening or restriction of international trade’ Cattaneo and Fryer (2003, p. 8). Upon opening
18
up for trade, the real income of capital used in the production of the export good increases, while
the real income for the specific capital associated with the production of the import competing
good declines. The nominal wage for the mobile factor (labour) increases, however, the real wage
falls in terms of the export good and increases in terms of the import good. The theory implies that
with specific factors, there is an ambiguous gain of the specific factor used in the expanding sector
and a fall in the specific sector used in the contracting sector.
Summary
All the above theories are supply oriented and predict that trade is only feasible between countries
with different endowments. Thus they are plausible in explaining trade between industries, what
is often termed ‘inter-industry’, however they fail in explaining trade as well as its related effects
between countries with the same endowment characteristics; ‘intra-industry trade’.
2.1.2. New Trade Theories
According to Al-Mawali (2005), recent trade theories depart from traditional trade theories by
relaxing the assumptions of the HO theorem. The new theories take into account imperfect
competition, product differentiation and economies of scale in world trade. These theories explain
IIT.
i. Linder Theory
The theory was proposed by a Swedish economist Staffan Burenstan Linder in 1961. He took a
different perspective in explaining trade and argues that trade should not be addressed from the
supply side as in comparative advantage models, but should be viewed as an interrelationship
between similar markets (Hanink, 1988). Linder argues that trade amongst countries with similar
endowments results from overlapping demands. The theory postulates that consumer tastes and
preferences are influenced by their level of income, thus the per capita income level of a country
will yield a particular pattern of tastes. Producers differentiate their products to meet the demands
of domestic consumers, thus in essence income levels determine a pattern of tastes which in turn
trigger a production response (Appleyard, 2009).
19
In explaining the Linder theory, Appleyard (2009), hypothesised a three country scenario; country
I with a relatively low per capita income compared to II, and country III having the highest income
of the three. Due to lower per capita income in country I, consumers will demand products A, B,
C, D and E arranged in ascending order in terms of quality. A higher per capita income in country
II will yield demand of say products C, D, E, F and G, while an even higher per capita income in
country III will yield demand of E, F, G, H and I. Given the pattern of production in the three
countries, trade will only be observed for goods that have ‘overlapping demand’. Country I and II
will trade in goods C, D and E; country II and III will trade in goods E, F and G; while country I
and III will trade in product E. An important implication of this theory is that international trade
will be more intense between countries with similar per capita incomes.
ii. The Krugman Model (1979)
The model was put forward by Paul Krugman to explain three seemingly paradoxes or rather
stylised facts about modern day trade. He relaxed traditional theories assumptions of perfect
competition and constant returns to scale, opting for economies of scale and monopolistic
completion. He further assumed two factors of production (Labour 1 and Labour 2) which are
specific to industry 1 and 2, and whose wage rates are given by W1 and W2. Each industry
consisting of a number of firms specialising in differentiated products, operating on the portion
where average costs are falling. Due to fixed costs in production, a firm wishing to exploit
economies of scale will specialise in a certain line of products within an industry effectively
reducing it’s per unit cost. Thus economies of scale necessitate differentiation; which amongst
similarly endowed nations results in simultaneous exports and imports within an industry.
Opening up for trade will result in factor price equalisation; however, the pattern of production is
left unchanged. Two effects on welfare arise. First, real wage remain unchanged in terms of the
products of its industry, but may fall or rise in terms of the other depending on whether it’s a scarce
or abundant factor. The second is associated with an expanded market and hence increased variety
and it works to everyone’s benefit. However, overall IIT benefits all factors as the fall in wage of
the scarce factor is offset by the gains derived from a larger market. According to Krugman (1979)
20
‘…if economies of scale are sufficiently important both factors gain from trade’ (p. 14). It can
thus be argued that IIT is associated with fewer adjustment problems as compared to HO trade.
iii. The Falvey Model (1981)
Rodney Falvey (1981) developed a model of IIT, embodying elements of the Heckscher-Ohlin
model with the concept of product differentiation. The model assumes two factors of production
(capital and labour) and two countries (A and B) with different factor endowments. The model
further assumes that any given variety of a product X can be produced in the two countries,
however, the products differ in terms of quality. The quality differences arise from the differing
capital to labour (K/L) ratios used in the production processes. A higher K/L ratio indicates a
higher quality variety of good X and a lower K/L ratio is associated with a lower quality variety
of good X.
The differences in factor endowments implies that there will be different K/L ratios between
countries. If for simplicity we assume that country A is capital abundant and B is labour abundant;
given the endowment characteristics, A would have a comparative advantage in producing the
higher quality variety with B having a comparative advantage in the lower quality variety. In
autarky, the higher quality variety will be cheaper in country A, but expensive in B. Conversely
the lower quality variety will be cheaper in B but expensive in A. This forms the basis for trade.
Country A exports the higher quality variety to B in exchange for the lower quality version of
product X from B. The end result is a pattern of IIT based on factor endowments. Contrary to the
Krugman model, IIT is not explained by economies of scale and imperfect competition, but is
rather ‘…a result that reflects a linking of factor endowments and intensities to the phenomena of
product differentiation’ Appleyard et al (2009, p. 184)
iv. Explanations for IIT
There is general consensus among trade economists that comparative advantage theories of factor
endowments are of little help in explaining trade between countries with relatively the same
endowments (capital and labour). Theoretical work on IIT can be traced to the work of Grubel and
21
Lloyd (Al-Mawali 2005). In an extract from Grubel and Lloyd seminal discussion, Appleyard et
al (2009) identifies the following as some of the plausible explanations for IIT;
a) Product differentiation
According to Appleyard et al (2009), ‘product differentiation refers to products that are
seemingly the same but which are perceived by the consumer to have real or imagined
differences’ (p. 180). Consumers in both countries will thus purchase imports within the same
product category for the ‘perceived differences’ even though the products may be the same,
for instance different brands of rice or soft drinks.
b) Transport costs
When transport cost are significant it may necessitate consumers to import rather than buy
from within their boundaries and this is specially so for low value bulky products. For instance
a miller in Victoria Falls may find it cheaper to buy grain from Livingstone (Zambia, a distance
of less than 30km) than to drive a distance of 900km to Harare. Thus despite the fact that a
country produces enough for domestic consumption, consumers may find it desirable to import
the same product after factoring in transport costs.
c) Dynamic economies of scale
He termed this, “learning by doing’ which results in per unit cost reduction due to experience
in the production of a particular good. This explanation is related to the product differentiation
argument. Each producer produces for both domestic and export market, thus the cost reduction
necessitates an increase in sales of each of the versions of the product over time, which
enhances intra industry trade. Krugman (1981) argues that economies of scale in production
necessitates countries to concentrate on a ‘subset of products’ within a particular industry and
this gives rise to intra industry specialisation and trade.
d) Degree of product aggregation
Some economists argue that IIT is a ‘statistical artefact’ arising from the degree of aggregation
used (the way trade data is recorded and analysed). If a product category is broad for instance
beverages and tobacco we may have a higher degree of IIT compared to when its narrowed
22
down to a specific product like a certain brand of wine. Davis (1995) cites Finger (1975) and
Chipman (1985) as having argued that, ‘…existing classifications place goods of
heterogeneous factor proportions in a single industry, and so intra-industry trade is
unremarkable’ (p. 205).
e) Differing income distributions in countries
The explanation is akin to the Linder hypothesis and it was propounded by Herbert Grubel
(1970). His argument was that even though countries may have similar per capita incomes, the
distribution may be different. Hypothetically, consider two countries I and II, country I has a
larger proportion of low income consumers and II has a relatively even distribution of income.
Country I will be pre occupied with the need to produce a version of a product which suits the
desires of the majority (low income households), and country II will produce that which suits
a majority of its household’s (middle income). However there are some consumers in country
I who are richer and can afford the version of the country II product, at the same time there are
some within country II whose income is lower and hence are consumers of the version
produced in country I. This brings about IIT as both countries trade within the same industry.
2.2. Empirical Review
Studies of IIT can be broadly classified into two groups. According to Sharma (1999), one group
focused on developing theoretical explanations of IIT in the presence of ‘product differentiation
and increasing returns to scale’, while the other was concerned with the determinants of IIT within
an econometric setup (p. 1). There has been a general consensus amongst most international trade
theorists that conventional trade theory cannot explain trade in goods of similar factor content i.e.
intra-industry trade (Davis, 2005). Krugman (1979) and Lancaster (1980) are often acknowledged
for pioneering a theoretical framework explaining IIT in terms of economies of scale in production
and product differentiation (Neum, 2012).
Helpman (1986), contended that despite the success of theoretical models of IIT, focusing on
‘monopolistic competition and differentiated products’, and their seemingly plausible explanation
of larger trade volumes amongst similarly endowed nations, it was necessary to examine their
23
consistency with data (p. 63). He tested three hypothesis which were drawn from Helpman and
Krugman (1985). Two of these concern the behaviour of IIT and the other, trade volume. Using
annual data for fourteen industrial nations12, over the period 1956 to 1981, the study found that
with greater similarity in factor endowments, there will be a corresponding larger share of IIT. The
study also found out that as countries became similar over time their IIT shares increased.
Furthermore he also found that changes in country size over time can plausibly explain the
escalating trade- income ratios. These findings are consistent with the new trade theories, further
confirming their usefulness in explaining the IIT phenomenon.
In a significant departure from the work of other international economists who argued for
economies of scale and imperfect competition as the plausible explanations of IIT, Davis (1995)
refuted the idea and argued that, ‘…to the contrary, intra-industry trade arises quite naturally in
a constant returns setting’ (p. 223). In his paper, ‘Intra-industry trade: A Heckscher-Ohlin-
Ricardo approach’, Davis (1995) argues that empirical studies have produced mixed results on the
role of scale economies in IIT. While acknowledging that scale economies would result in intra-
industry specialisation (a key ingredient of IIT), he argues that they are not the only reason for
such specialisation. Technical differences can as well result in specialisation especially when it is
feasible to expand one sector without driving up ‘marginal opportunity cost’. According to him,
the basic characteristics of IIT, that is, ‘trade in goods of similar factor intensities’ and ‘large
number of goods produced and traded’ are synonymous with the Ricardian model (p. 222)
Greenway, Hine and Milner (1995); Al-Mawali (2005) and Abdd-el Rahman (1991) argue that it
is pertinent to distinguish between horizontal intra industry trade (HIIT) and vertical intra industry
trade (VIIT), arguing that different forces drive these two. For instance, VIIT is driven by
differences in factor endowments whereas similarities in endowments explain HIIT (Al-Mawali,
2005). It has been argued that trade liberalisation bring about adjustment costs, however the
distribution varies depending on the type of IIT, with costs being considerably lower for HIIT
compared to VIIT (Kandogan, 2003a)
12 Belgium, Canada, Denmark, France, Germany, Japan, Ireland, Italy, Luxembourg, Netherlands, Sweden, Switzerland, UK and USA
24
Ates and Turkcan (2010) examined the extent and composition of IIT in the automobile sector and
sought to determine the country specific factors influencing IIT between US and its 37 trading
partners. They decomposed trade into INT and IIT with its components; horizontal intra industry
trade (HIIT) and vertical intra-industry trade (VIIT). The study employed the adjusted Grubel-
Lloyd index for computing the overall IIT, and Greenway et al (1995) method for HIIT and VIIT.
An analysis of the indices showed that automobile trade is dominated by INT; however, the trend
exhibited an increase in share of IIT from 15% to 20% for the period 1989 to 2006 and the increase
in IIT was driven by substantial increases in VIIT. The study employed the logit transformation of
the gravity model for determining the determinants of IIT as well as its components VIIT and
HIIT. In addition to the traditional variables13, the study included foreign direct investment (FDI),
and used the Random Effects Model for estimation. Regression results for IIT and VIIT were
almost the same, however they were significantly different with those for HIIT. Factors aligned to
differences in endowment explained VIIT while those related to product differentiation and
similarity in per capita incomes explained HIIT.
Li et al (2003) examined the extent of IIT in insurance services between the United States and its
trade partners for the period 1995 and 1996. The study employed two-stage least squares (2SLS)
and two-stage nonlinear logit (2SNL) models to determine the significant determinants of IIT in
the insurance sector. The study found that both FDI and trade intensity have a positive relationship
with IIT. Differences in per capita income, differences in financial market size, differences in
trade openness and trade imbalance in goods and services were all found to be negatively related
with IIT. These empirical findings confirm the new theoretical trade models that take trade and
foreign direct investment as compliments rather than substitutes. Furthermore, the results confirm
predictions by theoretical models of IIT for instance Krugman (1979), in that similarly endowed
nations are likely to engage in more IIT, as all proxies for dissimilarity were found to be negatively
related to IIT.
The study by Li et al (2003) lends credence for extending the analysis of IIT beyond the usual
country specific determinant of trade to include the differences in trade openness and differences
13 GDP, differences in market size, geographical distance, exchange rate
25
in trade imbalances as explanatory variables. Furthermore, the study included industry specific
variables to capture differences in financial market size and insurance market. However, despite
this being the case our study will focus only on nation specific determining factors of IIT. This is
necessitated by the unavailability of data to capture food manufacturing industry specific variables.
Kocyigit and Sen (2007) analysed the extent and patterns of IIT for Turkey’s trade with the
European Union and the rest of the world for the period 1997 to 2005. They used 3 digit level data
of the Harmonised System (HS3) for Turkey’s leading export and import commodities to calculate
GL indices of IIT. The study found that different industries have different levels of IIT. IIT was
high for sophisticated manufactured products where economies of scale play an important role.
However, it was considerably lower for labour intensive industries where Turkey has a
comparative advantage relative to its trade partners. Furthermore, the study found out that Turkey’s
IIT has been on the increase over the years with both the EU and the rest of the world. This came
at a time when Turkey’s industry evolved to be more aligned to the industrial structure of the
developed countries. It was also observed that IIT with the EU rose significantly after the signing
of a custom union between Turkey and the EU.
Among the studies which empirically analyzed intra-trade in the SADC region is one by Sunde et
al (2009) who investigated the determinants of intra- industry trade (IIT) between Zimbabwe and
its seven trade partners in SADC. The partners included in the study were Botswana, DRC,
Malawi, Mauritius, Namibia, South Africa and Zambia. The study utilized panel data for bilateral
trade shares between Zimbabwe and its trade partners for the period 1990 to 2006. The authors
calculated the unadjusted G-L index of IIT, which was used as the dependent variable. Furthermore
they employed a modified gravity model equation to model the determinants of IIT between
Zimbabwe and its trade partners using pooled ordinary least squares. The empirical results
confirmed that IIT is explained by per capita income, trade intensity, distance, exchange rate and
gross domestic product.
In Zambia, a related study by Mulenga (2012) was done on the determinants of IIT between
Zambia and the SADC region for the period 1998 to 2006. The study employed the adjusted GL
index for determining the extent of IIT, and the modified gravity model to evaluate the
26
determinants of IIT. The empirical results confirmed the existence of IIT. Diagnostic tests noted
the presence of heteroskedasticity, however this was corrected by estimating a generalized least
square regression (GLS) of the random effects model (REM). The REM results showed that IIT
is explained by traditional gravity model variables; GDP, per capita incomes and distance.
However, the results further revealed that IIT is also explained by dissimilarity in per capita income
(DCPI), common border and common language. Despite having the expected signs, exchange rates
and trade intensity were found to be insignificant in explaining IIT between Zambia and its SADC
trade partners. GDP, common language and common border were found to have positive signs
while DCPI had a negative sign as is postulated by the Linder hypothesis.
Mulenga (2012) and Sunde et al (2009) are some of the notable researchers to empirically
investigate the country specific determinants of IIT for the Southern African region. Both studies
employed the modified gravity model to empirically model IIT, and they used the same
explanatory variables. However, Sunde et al (2009) used the unadjusted GL index, while Mulenga
(2012) used the adjusted GL index. The use of the unadjusted GL index is often argued to result
in estimates that are biased downwards, thus Sunde et al (2009) study suffers from this problem.
To avoid this potential bias in our study we are going to use the adjusted GL index for calculating
the IIT shares, this index was also used by Ates and Turkcan (2010). Despite both Sunde et al 2009
and Mulenga (2012) finding a positive GDP variable in their study, we find it awkward to explain
Zimbabwe’s IIT in terms of its partners’ GDP only. This may represent a possible spurious
regression problem, where an increase in the GDP of South Africa for instance results in an
increase in Zimbabwe’s IIT. To this end we borrow the GDP variable from Al-Mawali (2005) who
defined it as a product of the reporting country’s GDP and that of each of its trade partners.
Cattaneo and Fryer (2003) sought to establish the effects of opening up of trade on poverty, through
its impact on income distribution. To achieve this the authors investigated the effects of trade
liberalisation on IIT and INT, with special focus on the developing SADC region. Using 4 digit
SITC14 level trade data for SACU15, Mauritius, Malawi, Mozambique, Tanzania, Zambia and
Zimbabwe, the authors constructed Brullharts (1994) marginal intra-industry trade (MIIT) indices
14 Standard International Trade Classification 15 Southern African Customs Union
27
for the manufacturing sector of SACU’s trade with the mentioned trade partners for the period
1994 to 2004. The empirical results confirmed that there has been considerable trade liberalisation
as reflected by increases in intra-SADC trade. However, despite the increases in intra SADC trade
during this period, much of the trade has been inter industry in nature, implying that trade
liberalisation was associated with higher adjustment costs (potential unemployment and political
resistance). It was suggested that deliberate policy stance to stimulate IIT is required, with an
emphasis also aimed at not only reducing tariff related barriers but also addressing non-tariff
barriers (infrastructural inadequacies, border delays and administrative constraints as well as
supply constraints).
Al-Mawali (2005) investigated the country specific determinants of IIT for the South African
economy using the gravity model. He used Kandogan (2003a and 2003b) methodology for
computing HIIT and VIIT, and employed the gravity model using panel data for 50 trade partners
over the 1994 to 2000 period. Diagnostic test results confirmed the fixed effects model (FEM) as
the most appropriate in modeling IIT between South Africa and its trade partners. The study found
out that South Africa conducts much of its IIT with larger economies, furthermore geographical
distance was found to have a significant negative effect on IIT. Empirical findings of the
augmented gravity model showed that market size and standard of living variables had a significant
effect on IIT, VIIT and HIIT, while geography was found to repel IIT in all its forms. Overall,
political risk, technology gap and integration variables were found to be insignificant in explaining
both IIT, VIIT and HIIT.
The study by Al-Mawali (2005) as well as that by Ates and Turckan (2009) went a step further to
investigate the determinants of VIIT and HIIT. They however differed in their methodological
approach in estimating VIIT and HIIT, Al-Mawali (2005) employed the Kondogan (2003a)
methodology while Turckan (2009) used Greenway et al (1995) method. However, despite these
differences in the measures of VIIT and HIIT, both studies employed the modified gravity model
to investigate the determinants of IIT. The gravity model was also employed by Sunde et al (2009)
and Mulenga (2012). In our study we will employ the gravity model to investigate the determinants
of IIT between Zimbabwe and its five trade partners.
28
Dhakal, Pradhan and Upadhyaya (2009), tested the empirical validity of the Linder Hypothesis for
five Asian trade partners (Indonesia, Malaysia, Philippines, South Korea and Thailand) using a
modified gravity model. The study employed panel data pooled for the years 1997, 1999, 2001,
2003 and 2005; and estimated three models in fixed effects. The study found statistically
significant per capita income, ASEAN and distance variables, with the expected positive priori
signs for the first variable and a negative sign for the distance variable. However, the Linder
variable which was measured by differences in per capita income was found to have a significant
positive coefficient for one model and a negative (expected) coefficient but statistically
insignificant for the other. The study thus found no statistical evidence to support the Linder
hypothesis in explaining trade between the five countries.
2.3 Conclusion
From the empirical literature, there is consensus that an increasing proportion of world trade takes
the form of IIT. Furthermore, traditional theories of trade though plausible in explaining INT, fall
short in explaining IIT. While there is agreement that opening up for trade will result in factor
price equalization, effects on returns to factors of production vary remarkably depending on the
nature of trade, with costs being considerably higher for INT compared to IIT.
Most of the studies have sought to determine the country specific determinants of IIT, regressing
variables such as GDP, DCPI, distance, trade intensity, common language and boarder amongst
other variables on IIT using the gravity model. Findings from most studies confirm a positive
relationship between GPD, TI, common language and common boarders with IIT, while DCPI and
distance have been found to be negatively related to IIT.
Despite IIT being a widely researched topic, evidence suggests that not much has been done for
the African region and Zimbabwe in particular. Therefore, guided by empirical literature this study
seeks to establish the country specific determinants of IIT in food products between Zimbabwe
and some of its trade partners in the SADC region using the gravity framework.
29
CHAPTER THREE
RESEARCH METHODOLOGY
3.0. Introduction
This chapter presents the specification of the model as well as the econometric framework within
which the determinants of IIT between Zimbabwe and its five trade partners in the SADC region
are estimated. Firstly Grubel-Lloyd indices of IIT will be calculated to establish the nature16 of
trade in the food manufacturing industry. Furthermore, the chapter describes and justifies the
variables used in the model, linking some of the ideas raised in the previous chapter to the empirical
model.
3.0.1. Measuring IIT in the Food Manufacturing Industry
Trade flows can be identified as either IIT or INT. This section of the study presents the empirical
methodology on the measurement of intra industry trade. A number of measures of IIT have been
proposed in the empirical literature, including the Balassa index, the Aquinto index and the Grubel-
Lloyd index (Ates and Turkson, 2010).
a. The Balassa Index
Balassa (1966), is often credited for pioneering the work on a numerical measurement of IIT,
through his formulation of what is often termed the Balassa index. The algebraic formulation of
this index which expresses trade balance as a proportion of total trade is represented as follows;
jj
jj
jmx
mxINT
(1.0)
Where, jINT is inter industry trade index, while jx and jm are the values of exports and imports
of commodity j respectively.
16 Is it dominated in inter industry trade or intra-industry trade?
30
The index takes value of zero to indicate complete IIT and one to indicate complete INT. The zero
value for IIT is argued to be the major source of the index’s weakness, especially when it comes
to empirical analysis. This led to the development of yet another index; the Grubel-Lloyd index.
b. The Grubel Lloyd Index of IIT
The index was proposed by Grubel and Lloyd (1975) and is now commonly used for determining
the extent of IIT. The algebraic formulation of the index is given by;
ijkijk
ijkjki
ijkMX
MXIIT 1 (1.2)
where;
ijkIIT - is the intra-industry trade index in industry i between country j and partner k
Xijk - are country j ’s exports of industry i to country k
Mijk - are country j ’s imports of industry i from country k
However, Ates and Turkson (2010) argue that, over and above the aggregation bias, the unadjusted
G-L index is negatively related with an overall trade imbalance; a problem often cited in empirical
literature. Kocyigit and Sen (2007) argue that large trade imbalances result in GL indexes which
are biased downwards, hence the indexes are most likely to be underestimated. To correct for the
deficiencies, Grubel and Lloyd proposed the adjusted GL index which incorporates trade
imbalances (Kocyigit and Sen, 2007). This study follows Ates and Turkson (2010) formulation
of the adjusted G-L index modified by multiplying by 100 and computed as follows;
100*
1
11
n
i
ijkijk
n
i
ijkijk
n
i
ijkijk
ijk
MX
MXMX
IITFM (1.3)
31
Where, ijkIIIFM is the intra industry trade between Zimbabwe and trade partner k in the food
industry while ijkX and ijkM are Zimbabwe’s exports and imports of product category i in the food
industry to partner k respectively.
According to Ates and Turkson (2010), ‘the index computes the export and import flows with
country k in industry j , adjusted or weighted according to the relative share of the trade flows in
the i products included in industry j ’ (page, 18). It assumes values between 0 and 100, with zero
indicating complete international specialization (INT) and 100 signifying purely IIT. A GL index
of more than 50% signifies that the sector is IIT driven and a value less than 50% implies that it is
INT driven.
The first step in computing the G-L index is to select manufactured food products from UN
Comtrade bilateral trade data of Zimbabwe’s five trade partners. The bilateral trade flows utilized
in this study are classified at the 2 digit level of the Harmonized System (HS2). The food
manufacturing industry is represented by 11 product categories and the detailed description is
given in appendix 1.
Estimating the Determinants of IIT
This section outlines the framework within which country specific determinants of IIT are to be
modeled using the gravity model.
3.0.2. The Gravity Model
The basic gravity model explains trade between countries as depending upon their GDP,
population size and geographical distance. According to Bergstrand (1985), the model essentially
predicts that trade flows between countries depend on each other’s trade potential and ‘economic
forces either aiding or resisting the flow's movement from origin to destination’ (p. 447).
According to Deardorff (1998), two authors; Tienbergern (1962) and Poyhonen (1963) are often
credited for pioneering the use of the Gravity Model to analyze economic flows. However, it is
often argued that they did not give the theoretical justification of the use of the gravity model
32
prompting certain scholars to challenge its usefulness as an empirical device capable of explaining
trade flows.
The typical specification of the basic gravity model takes the form of;
ij
ji
ijD
MMAF (1.4)
Where ijF represents the trade flow, A is a constant, M is the economic mass of country i and
partner j and ijD is the physical distance between the two countries.
Despite early criticism of Tienbergen’s (1962) application of the gravity model in terms of its lack
of ‘theoretical underpinnings’, recent developments17 in trade theory have strengthened the
theoretical basis for the gravity model confirming its usefulness in empirical testing of bilateral
trade flows (Baldwin & Taglioni, 2006, p. 1). A number of studies in the IIT literature have
recently applied this model with success (Al-Mawali (2005); Simwaka (2006); Sunde at el (2009);
Ates & Turkcan (2010) and Mulenga (2012)). This study applies the Gravity Model, to estimate
the determinants of IIT between Zimbabwe and its SADC trade partners.
3.1. The Empirical Model
The empirical model is borrowed from studies by Sunde et al (2009) and Mulenga (2012).
However, we add two more variables in our model, (differences in gross domestic product and free
trade area dummy).
The dependent variable is intra industry trade ( ijkIITFM ) and the explanatory variables are the
product of gross domestic product ( jkRGDP ), differences in gross domestic product ( jkDGDP
), dissimilarity in per capita income ( jkDPCI ), trade intensity ( jkTII ), distance ( jkDIS ), real
exchange rate ( jkRER ), dummy variable for common border 1D , dummy variable for common
17 A number of authors have theoretically justified the use of the gravity model in empirical analysis including Anderson and Van Wincoop (2003), Bergstrand (1985) and Deardoff (1995)
33
language (2D ) and dummy variable for free trade area (
3D ). The general model and the expected
signs of the explanatory variables is as defined below.
Zimbabwe’s trade volume with a trade partner k to its total trade volume and it is computed as
follows;
jj
jkjk
jkMX
MXTI
(1.9)
where;
jkX - Zimbabwe’s exports to country k
jkM - Zimbabwe’s imports from country k
jX - Zimbabwe’s world exports
jM -Zimbabwe’s world imports
3.2.7. Real Exchange Rate ( jkRER )
The exchange rate is the price of a currency in terms of another currency. We calculate the real
exchange rate as the cross exchange rates between trading partners’ currencies adjusted for
inflation. However, since we are concentrating on two periods; the pre and post Zimbabwean dollar
($ZW), we use the United States Dollar as the local currency for the period under which the country
has been using multi-currency18. The use of the real exchange rate is justified on the basis that it
gives a measure of the economic competitiveness in terms of import and exports. However, there
is no consensus in the empirical literature as to the priori sign of exchange rate on IIT (Ates and
Turkson, (2010). The real exchange rate is calculated as;
j
kjkjk
p
pERER (2.0)
where;
jkRER = real exchange rate between Zimbabwe and trading partner k
jkE = nominal exchange rate between Zimbabwe and trading partner k ’s currency
18 The multi-currency system is the current monetary regime in Zimbabwe where a basket of currencies are used for everyday transactions, however, they are dominated by the United States dollar (US$).
37
jp = Zimbabwe’s GDP deflator
kp =Trading partner k’s GDP deflator
3.2.8. Common Border (D1)
The existence of a common border represents possibility of IIT due to locational advantages
(Balasa & Bauwens, 1987). Ceteris paribus IIT between countries sharing a common border is
likely to be higher compared to countries which do not share a common border.
otherwise
bordercommonasharecountriesifD
0
11
3.2.9. Common Language (D2)
The inclusion of the variable is justified on the basis that it aids information flow and lowers
transaction costs, thus it increases IIT between countries. It is measured as a dummy variable
specified as below
otherwise
languagecommonasharecountriesifD
0
12
We propose a positive relationship between common language and IIT
3.2.9. Free trade Area Dummy 3D
It is often hypothesized that trade increase with the openness of a country, a free trade area is thus
expected to be positively related to IIT. Kocyigit and Sen (2007) found a significant increase in
IIT between Turkey and the EU after the signing of a custom union agreement between Turkey
and the EU. The FTA dummy variable is defined as follows:
otherwise
statusFTAattainedhadSADCD
0
13
38
We include the variable to examine if the attainment of the FTA in SADC had any effect on intra
industry trade.
3.3 Data Sources and Problems
In constructing our empirical model, we employ panel data from a sample of six countries for a
thirteen year period spanning from 2000 to 2012. We justify the inclusion of these countries on
the basis that they conduct considerable trade with Zimbabwe. Most of the omitted SADC
countries conduct insignificant trade with Zimbabwe in food products. Apart from this, for some
countries, there is no reported bilateral trade flows between the omitted countries and Zimbabwe.
The use of panel data is justified on the superiority of panel data over both time series and cross
sectional data, as it enables us to identify and estimate effects that would otherwise not be
detectable in pure cross section or time series data (Koutsoyannis, 1977). Al-Mawali (2005),
argues that panel data has the advantage of controlling for individual heterogeneity, as ignoring
these unobserved individual specific effects lead to bias in the estimates.
We use two digit Harmonized System (HS) data from UN Comtrade for obtaining the bilateral
trade shares of manufactured food products; World Bank, World Development Indicators data for
GDP, per capita incomes, exchange rates and GDP deflators. For the bilateral geographical
distance we used http://www.cepii.fr/ website.
In the IIT literature HS2 level data is relatively aggregated, and this may present challenges of
aggregation bias (overstatement of IIT extent). However, there was insufficient data at higher
levels of the HS (HS4), and for this reason we had to rely on HS2 data for calculating IIT.
3.4 Diagnostic Tests
For the purpose of assessing the adequacy and relevance of the model, the study will carry out the
following diagnostic tests in Stata; multicollinearity, heteroskedasticity, Breusch and Pagan
Lagrange multiplier test for random effects, Poolability test (F-test) and the Hausman test for