Performance of Intra-COMESA Trade Integration: A comparative Study
with ASEAN’s Trade Integration
Abstract
This paper aims at assessing the performance of intra-COMESA trade integration on the basis of
success of ASEAN integration, using an out-of-sample approach. The analysis employed gravity
model to estimate the coefficients of ASEAN model which are used as a benchmark to project the
potential trade for eight COMESA member states. The success of COMESA is estimated by the
ratio of potential to actual trade. The results pointed out that all countries of the selected sample
are far from their potential trade level, implying unfavorable performance of the regional trade
integration among COMESA members. The results also indicate that the gap between potential
and actual trade has been decreased in last decade, suggesting a convergence toward the potential
trade level over time. Finally, the paper ends with some policy recommendations regarding
promoting regional cooperation among COMESA member states.
Keywords: Regional integration, ASEAN, COMESA, Gravity model
JEL Classification: F14, F15.
1
1. Introduction
Regional integration (RI) has been considered as one of the prominent strategies for
development among countries and regions. It is well known that RI brings countries with
many advantages in terms of economic and political aspects. Indeed, RI promotes
economic growth and industrialization process through fostering intra-regional trade,
infrastructure and investment. Precisely, RI provides a huge market for parallel
development of new industries which reduces external vulnerability through increasing
bargaining power and in turn, increases the national income (Balassa, 1961). Moreover,
regional trade cooperation between countries is regarded as a key strategy to confront the
challenges of globalization.
The history of regional cooperation involves a number of successful regional trade
schemes in developed and developing regions. For example, the Association of Southeast
Asian Nations (ASEAN) is classified as one of the most successful trade integrations in
the World. Since its establishment in 1967, ASEAN has achieved remarkable cooperation
progress in terms of economic, social and political fronts. The intra-trade performance
also witnessed notable improvement, supported by the adoption of many trade
arrangements, including ASEAN Free Trade Area (AFTA) in 1992, with aim to remove
tariff and non-tariff barriers within the region. As result, intra-ASEAN trade has
increased from 20% in to 25% in 2009 (ASEAN Secretariat, 2009).
In Africa, the Common Market for East and Southern Africa (COMESA) is one of the
prominent regional bodies in East and South Africa. The main objective of COMESA is
promoting trade, investment and infrastructure. The COMESA integration has paid
considerable attention to cooperation in productive sectors such as, infrastructure,
agriculture, transportation and financial sector. In last decade, COMESA leaders
launched many regional cooperation arrangements, including Free Trade Area (FTA) in
2000 as well as custom union in 2009. Furthermore, the aspiration of the leaders is to
push the cooperation process toward a Common Market and economic community in the
coming decade. Having these ambitious plans for further trade cooperation, it is
2
important to assess the success of COMESA regional trade integration in comparison
with other successful trade integration. Therefore, investigating the performance of intraCOMESA trade on the basis of success of ASEAN integration, it would be useful to
gauge the prospect of COMESA for further trade integration and to uncover some
policies that can help Africa policy makers to develop the process of the integration so as
to achieve more economic cooperation. The ASEAN is selected because it is one of the
most successful regional integrations in developing regions (i.e. Eastern and Southern
Asia). ASEAN also similar to COMESA as it started with trade integration as main pillar
for economic cooperation as well as social and political integration.
Based on the above, the analysis concentrates on three questions which include: i) How
much trade could be achieved among COMESA members if their trade elasticity with
respect to economic and geographic variables (economic size, distance, common border,
language, colony links) were similar to those achieved in intra-ASEAN trade (This
indicates the potential trade)? ii) What is the nature of difference between actual trade
and potential regional trade of COMESA member states? To answer these questions, the
paper uses gravity model and out of sample approach to investigate the trend of intraCOMESA trade achievement compared with regional integration of ASEAN.
The importance of this study is to fill the gap in literature on understanding the success of
regional integration in Africa in general and COMESA region, in particular. In addition,
unlike the previous studies on measuring trade potential, this study use “out-of-sample”
approach, which did not used before to analyze the potential intra-trade of COMESA. To
our knowledge this method of projecting potential trade is only used by Patore et al.
(2009) to compare the parallel integration process of EU with the Mediterranean
countries and Central and Eastern European countries.
The paper is organized in six sections as follows. Section two outlines some stylized facts
about COMESA and ASEAN integration. Section three reviews the empirical literature
on regional integration. While section four outlines the research methodology, section
3
five presents the empirical results. Finally, section six ends with conclusion and policy
implications.
2. Some Stylized Facts about COMESA and ASEAN integration
Before analyzing the success of intra-COMESA trade integration in comparison with the
experience of ASEAN integration, it is useful to describe the features of COMESA and
ASEAN. Thus, this section outlines the main characteristics of the two agreements,
focusing on their memberships, objectives and economic size as well as their regional
trade patterns.
2.1. COMESA Regional Integration
COMESA started as a preferential trade area (PTA) in 1982 and extended in 1994 to be
one of the prominent regional integration bodies in Eastern and Southern Africa.
COMESA now embodies 19 members including Angola, Burundi, Comoros, the
Democratic Republic of the Congo, Djibouti, Egypt, Eritrea, Ethiopia, Kenya,
Madagascar, Malawi, Mauritius, Seychelles, Sudan, Rwanda, Swaziland, Uganda,
Zambia and Zimbabwe. The main aim of COMESA is strengthening the process of
regional economic integration, which had been initiated in order to help member states
achieve sustainable economic growth. In 2000, the members of COMESA signed the
Free Trade Area (FTA) agreement, to attain sustainable growth of the member states, to
promote joint development in all fields of economic activity, to cooperate in the creation
of an enabling environment for foreign, cross-border and domestic investment and in the
promotion of peace, security, and stability among the member states in order to enhance
the economic development in the region (COMESA, 2013). Members of the FTA include
eleven countries are: Burundi, Djibouti, Egypt, Kenya, Madagascar, Malawi, Mauritius,
Rwanda, Sudan, Zambia, and Zimbabwe. Libya and Comoros joined the FTA in 2005
and 2006 respectively, bringing the number of COMESA members included in the FTA
to 13 out of 19, other countries trade on preferential terms (COMESA, 2013).
In recent decade, the COMESA agreement witnessed an adoption of many trade
arrangements with respect to all trade sectors. In 2009 the agreement had launched the
4
custom union which expected to be extended to economic and currency union in the
coming decade. Although the growing importance of COMESA in the economy of its
members, there are many challenges hider the intra-regional trade. These constraints
include conflicting objectives of overlapping regional integrations, lack of political
commitment and inadequate infrastructure and information (Khandelwal, 2004 and
Oyejide, 1997). For example, most of COMESA countries are members in more than one
trade organization in the region such as, East African Community (EAC) and SADC.
Regarding the economic and trade performance of COMESA, Table 1 show some basic
indicators about COMESA integration. The Table reveals that COMESA region contains
about 456 million of inhabitants which they are vary widely from country to another,
ranging from less than 90 000 in Seychelles to 87 million Ethiopia. The level of
economic situation measured by GDP per capita is also varies widely among COMESA
members. For some members like Sudan and Zambia the per capita income increase by
the rate more than 200% during the period 2000-2010. While some countries like Congo
and Burundi exhibit low levels of GDP per capita, others like Libya and Seychelles
reported high level of GDP per capita.
In accordance with regional trade performance, Table 1 also shows that COMESA
undergone a sizable increase in the intra-regional exports and imports between 2000 and
2010. In 2010 the intra exports of some countries like Egypt and Kenya are very high,
while for other countries, exports to the region are far lower. For example, the intraCOMESA exports performance for Comoros, Eritrea and Seychelles did not exceed 2.5
million dollars. The share of Egypt’ exports to other COMESA members have risen from
113.8 million dollars in 2000 to 2343.7 million dollars in 2010, reporting a highest rate in
COMESA. Kenya also reported notable increase in intra-COMESA exports from about
595.7$ millions to 1658.4$ millions in 2010. Overall, Egypt and Kenya were the active
trading partners amongst COMESA states, as their contribution to total trade in 2010
accounted for 19% and 12.4%, respectively. On the other hand, Libya and Zambia were
the highest imported countries within COMESA. The increase of both exports and
5
imports for all countries between 2000 and 2010 indicate that the creation of the FTA in
2000 has resulted in improvement of intra-COMESA trade.
Table 1: COMESA Regional Integration, Selected Indicators, 2000 - 2010
Member
Population
GDP per capita
Intra-COMESA
Intra-COMESA
Share in Total
State
(million)
(USD)
Exports- in
Imports- in
COMESA trade
USD million
USD million
(%)
2010
2000
2010
2000
2010
2000
2010
2000
2010
Burundi
9.23
130.42
219.53
5.01
24.57
19.92
105.87
0.80
0.75
Comoros
20.62
583.09
1090.38
0.10
2.45
5.03
13.01
0.16
0.09
Congo
62.19
91.70
210.77
33.69
1134.3
107.12
806.13
4.52
11.17
Djibouti
0.83
762.54
1353.19
4.08
601.73
73.43
78.15
2.49
3.91
Egypt
78.08
1509.58
2803.53
113.79
2343.67
239.08
961.77
11.32
19.02
Eritrea
5.74
179.31
368.75
0.18
2.14
7.80
155.54
0.26
0.91
Ethiopia
87.10
123.89
341.08
155.14
287.30
107.58
286.24
8.43
3.30
Kenya
40.91
406.12
787.06
595.65
1658.40
77.33
504.09
21.60
12.44
Libya
6.04
6548.57 10455.57
50.41
334.78
69.29
1378.27
3.84
9.86
Madagascar
21.08
246.28
419.22
19.06
47.09
63.47
197.27
2.65
1.41
Malawi
15.01
154.00
359.58
41.51
215.56
52.78
231.83
3.03
2.57
Mauritius
1.28
3861.04
7586.97
96.82
155.74
58.30
125.31
4.98
1.62
Rwanda
10.84
206.65
519.02
35.07
82.73
28.65
415.23
2.04
2.87
Seychelles
0.09
7578.83 10842.57
2.39
2.46
12.52
46.98
0.48
0.28
Sudan
35.65
356.50
1421.09
78.71
336.49
201.21
767.93
8.98
6.36
Swaziland
1.19
1433.18
3093.54
64.98
140.25
0.53
10.67
2.10
0.87
Uganda
33.99
255.12
505.99
77.07
712.98
152.36
586.94
7.36
7.48
Zambia
13.22
322.10
1224.95
152.13
690.18
85.26
1394.23
7.62
12.00
Zimbabwe
13.08
535.04
568.43
170.72
266.96
57.70
271.20
7.33
3.10
COMESA
456.17
25283.9 44171.24 1696.5
9039.83
1419.35
8336.63
100.00
100.00
Source: Authors' calculations based on data from World Bank Indicators and websites of COMESA
(comstat.comesa.int/DataQuery.aspx).
6
2.2. ASEAN Regional Integration
The Association of Southeast Asian Nations (ASEAN) was established in 1967, to
accelerate the economic growth, social progress and cultural development in the region,
establish peace and stability through justice amongst its member nations, promote active
collaboration and mutual assistance on matters of common interest in the economic,
social, cultural, technical, scientific and administrative fields. The founding countries are
Indonesia, Malaysia, Philippines, Singapore and Thailand. Brunei Darussalam joined in
1984, Vietnam (1995), Laos PDR and Myanmar (1997) and Cambodia (1999), making up
what is today the ten member states of ASEAN (ASEAN Secretariat (2013).
ASEAN has adopted many trade arrangements which resulted in remarkable progress in
intra-trade performance, over the last decades (See Table 2). For example, ASEAN Free
Trade Area (AFTA) was signed in January 1992 with the aim of creating a free trade area
by 2008.
The original signatories were:
Brunei, Indonesia, Malaysia, Philippines,
Singapore and Thailand. Vietnam joined in 1995, Laos and Myanmar in 1997 and
Cambodia in 1999. This free market trade means that all ASEAN goods can be traded to
member state’s market with a minimum tariff or without tariff (ASEAN Secretariat,
2013).
Table 2 shows that the ten countries that are members of ASEAN have growing
economies with population of about 593 million. We notice that the levels of
development in two periods are very different across these countries with GDP per capita
spanning in the selected periods, from USD 23814.56 to 268.432 in 2000, and from USD
44862.8 to 706.4 in 2010, in Singapore and Myanmar, respectively. The Table shows that
the last four countries joined to ASEAN Cambodia, Laos, Myanmar and Vietnam have a
lower stage in economic development than the other six member countries. It is
interesting to note that for ASEAN countries there is a large and growing disparity
between the share of intra-regional exports and the share of intra-regional imports,
Singapore was first, followed by Malaysia and Thailand, though it should be noted that
Singapore’s share was particularly high as it handles considerable intra-regional trade,
7
and far lowest for Lao PDR and Cambodia. In addition, the share of Singapore’s in total
trade within ASEAN reporting a highest rate (34%).
Table 2: ASEAN Regional Integration, Selected Indicators, 2000 - 2010
Member
Population
GDP per capita
Intra-ASEAN
Intra-ASEAN
Share in Total
State
(million)
(USD)
Imports- in
Imports- in
ASEAN trade
USD million
USD million
(%)
2010
2000
2010
2010
2010
2010
Brunei
0.40
18086.60
29915.3
8,615.4
2,383.8
0.54
Cambodia
14.36
298.95
785.1
5,583.6
4,896.8
0.51
Indonesia
240.68
789.81
3027.2
157,779.1
135,663.3
14.34
Lao PDR
6.40
321.29
1099.9
2,432.8
2,076.4
0.22
Malaysia
28.28
4004.56
8555.5
198,800.8
164,733.5
17.77
Myanmar
51.93
268.432
706.4
7,599.5
4,198.7
0.58
Philippines
93.44
1043.46
2129.4
51,431.7
58,228.6
5.36
Singapore
5.08
23814.56
44862.8
371,194.3
328,078.9
34.18
Thailand
66.40
1968.54
4743.3
195,312.3
189,728.4
18.82
Viet Nam
86.93
401.55
1225.5
72,191.9
84,801.2
7.67
ASEAN
593.90
50997.74
3152.8
1,070,941.4
974,789.6
100.00
Source: ASEAN Finance and Macro-economic Surveillance Unit Database, websites of ASEAN Member
States' national statistics offices and World Bank Indicators.
Regarding the share of ASEAN integration in World exports, Annex I reveals that
ASEAN members contribute significantly to total world exports compared with
COMESA integration. The contribution of ASEAN to World exports is 1.55% on average
over the period 2000-2010, while the share of COMESA accounts only for about 0.03%
(See Annex I).
3. Literature Review
Having the importance of regional trade integration in economic and political relations
between countries, a huge body of literature on assessing the performance of intraregional trade has grown in last decades. Most of empirical studies have focused on
8
measuring the potential trade for trade blocs across developing and developed countries.
Although most of the empirical work has been done for European, Latin American and
Asian countries, a few studies have been done using African trade data. In this section,
however we review some of the empirical studies that are relevant for our paper.
Al-Atrash and Yousef (2000) used gravity model investigated the trade performance for
18 Arab countries with 43 others countries that represent over 90 percent of the exports
and imports of the Arab world. They found that the intra-trade within the Arab subgroups
is higher than overall intra-Arab trade. Also they found that cultural attributes measured
by language have mixed effect: while English-speaking countries tend to trade more with
each other, the French-speaking countries are not statistically significant. Moreover, their
results pointed out that ASEAN preference arrangement exerts large positive effect on
Arab trade, while the EU arrangements decrease the volume of trade.
Chauvin and Gaulier (2001) measured the potential of increasing intra-SADC trade, using
three complementary approaches. The first two ones refer to trade indices which are
export diversification indices, and revealed comparative advantages and trade
complementarity indices, while the last one is based on gravity model. They found that
there is some complementarity between SADC countries; but does not confirm the
existence of potential trade. In addition, South Africa is found to be the most significant
member in terms of export and can play further role in fostering the intra-trade in SADC
region.
Khandelwal (2004) studied the prospects and challenges for trade expansion in COMESA
and South African Development Community (SADC). He finds that the COMESA FTA
has taken a market-liberalization approach to regional integration, but has been hampered
by country-level implementation issues. On the other hand, SADC has taken the approach
of addressing infrastructure and supply constraints but has also had implementation
problems. He also finds that possibilities of growth in intraregional trade may be limited,
but that the two arrangements provide opportunities for their member states to adopt
9
policy credibility for trade reforms and trade liberalization and to address structural
problems.
Pastore et al. (2009) examined the performance of EU members with the Mediterranean
(MED) countries and the new EU members using a gravity model of intra-EU trade
including 13 members during the period 1995-2002. Employing out of sample
methodology they reports that There is a sizeable and unexploited trade potential with
both groups of partners, but the ratio of potential to actual trade with the MED countries
is much larger more dispersed and stable compared to that with the new EU members.
Their results also indicate that the potential tends to converge to actual trade in a much
longer time in the case of MED countries.
Simwaka (2011) used a gravity model to assess the success of SADC FTA over the
period 1998-2007. He separated the data sample into two periods; pre-integration that
before the adoption of FTA (1998-2000) and post-integration that after SADC FTA came
into operation (2003-2007). He found that the predicted trade is higher than the observed
intra-regional trade, suggesting an existence of trade potential among member states.
Therefore, they conclude that SADC FTA leads to trade creation and enhance the trade
capabilities of member countries. His results, however contradicts the findings by
Chauvin (2002) who found that SADC trade potentials are rather small or negative,
especially for South African exports. Finally, comparing SADC with other regional
integrations, Simwaka argued that ASEAN and NAFTA perform better than SADC.
Stack and Pentecost (2011) used a gravity model of new trade theory determinants for a
for twenty OECD trading partners countries with EU countries over the period 19922003. Based on out-of-sample approach they project the trade volumes for ten new
member states and ten associated countries. Their findings revealed that the projected
trade ratios for the ten new member states are multiples of actual 2003 levels,
indicating that trade expansion between these countries tend to continue. On other
hand, for the Mediterranean countries the ratio of potential to actual trade is found to be
near unity value, implying fewer opportunities for further trade integration with the EU.
10
4. Methodology and Data
To examine the performance of intra-COMESA regional trade on the basis of success of
ASEAN integration, the analysis proceeds via two steps: first, we estimate the
coefficients of a gravity model of intra-ASEAN trade, and then we apply them into
equations of trade between the selected sample of COMESA members to calculate the
potential trade relative to intra-ASEAN trade. The calculated potential trade volume will
be compared with the actual volume of intra-COMESA trade.
4.1. Gravity Model Specification
The gravity model of bilateral trade is widely used in the literature to investigate the
determinants of bilateral trade flow. The gravity model was firstly used by Tinbergen
(1962) and Linneman (1966) and later developed by (Anderson, 1979). In recent decades
this model has been developed further and used extensively in trade literature. The paper
employs the gravity model as benchmark to compare between ASEAN and COMESA
intra-trade integration.
The gravity model of trade is derived from Newton’s gravity law in Physics, which
postulate that there is a gravitational pull between two physical bodies as proportional to
their mass and inversely proportional to their distance. This theory is analogous to the
international trade as follows: the trade flow between two countries (exporter and
importer) is proportional to the product of each country’s (economic mass) commonly
measured by GDP, divided by the distance between the country’s respective centers of
gravity. Thus, trade between two countries depends on their Gross Domestic Product
(GDP), population size and the distance between them. Hence, the estimable gravity
model that used in our analysis could be specified as follows:
𝐿𝑛 𝑇𝑖𝑗𝑡 = 𝛼𝑖𝑗 + 𝛽1 𝐿𝑛 𝐺𝐷𝑃𝑖𝑡 + 𝛽2 𝐿𝑛 𝑃𝑂𝑃𝑖𝑡 + 𝛽3 𝐿𝑛 𝐺𝐷𝑃𝑗𝑡 + 𝛽4 𝐿𝑛 𝑃𝑂𝑃𝑗𝑡 + 𝛽5 𝐿𝑛 𝐷𝐼𝑆𝑖𝑗
+ 𝛽6 𝐶𝐵𝑖𝑗 + 𝛽7 𝐶𝐿𝑖𝑗 + 𝜇𝑖𝑗𝑡
(1)
Where i indicates the reporter countries, j are the trading partners and t is period under
consideration, i.e. 1998-2010, Xij is the trade variable between country i and country j;
11
POPi and POPj are the populations at time t of country i and j respectively; GDPi and
GDPj are gross domestic product of country i and j at time t; Dij is the geographical
distance in kilometers between the capital city of country i and of country j; CL is a
dummy variable to capture common language or colonial history, taking value of 1 if the
two countries speak same language or have ex-colony links, and zero otherwise. CB is a
dummy taking a value of one if the trade partners share a common land borders or sea
borders, and zero otherwise; finally, 𝜇𝑖𝑗𝑡 is the error term. All the variables are expressed
in natural logarithms except dummy variables.
According to the theory, the coefficient of GDP is expected to be positive, as an increase
of national income indicates more imports demand and exports supply. The impact of
population size (POP) is mixed as suggested by most of previous empirical studies.
Markheim (1994) argues that a country with large population size entails a large domestic
market and high degree of self-sufficiency and less need to trade (absorption effect).
Other argument show that, a large population means more progress in specialization and
division of labour and increase of the production, which are generally associated with a
larger need for trading (scale effect). The coefficient of distance is expected to be
negative, as the larger physical distance between two countries’ economic centers, the
higher is the cost of transporting goods between them. Finally, the dummy variable CL
and CB are expected to be positive, as sharing borders, ex-colony links, and same
language indicate geographical closeness, better information, same cultures and
institutions as well as legal systems.
4.2. Data Sources
The data used in the gravity model of intra-ASEAN trade spans over the period 19982010 and involves eight ASEAN countries1. On the other hand the data for COMESA
member states covers eight countries over the period 2004-20112. The trade data for
ASEAN members are extracted from IMF’ Direction of International Trade statistics,
1
The analysis of intra-ASEAN includes: Brunei, Indonesia, Laos, Malaysia, Philippines, Singapore,
Thailand and Vietnam. The remaining countries were excluded, because they joined ASEAN integration
after 1998.
2 Due to the lack of data for some COMESA’ countries, the projection of intra-COMESA trade integration
involve eight countries: Egypt, Ethiopia, Kenya, Malawi, Mauritius, Uganda, Zambia and Zimbabwe.
12
while the data for COMESA are collected from COMSTAT Database website. The data
about current GDP and population size are collected from World Bank’ Development
Indicators. Data on distance between countries is calculated based on the country location
provided by the CIA World Fact-book. Finally, Information about common language, excolony history and common border were sourced from the CIA World Fact-book.
4.3. Estimation Methodology
The gravity model in equation (1) is estimated via the panel data methods namely,
pooled, fixed effects (FE) and random effects (RE) models. As our regression models
involve individual effects, it is important to decide whether they are fixed of random;
thus we focus on the fixed and random effects models. When estimating the trade flows
between a randomly selected sample of trading partners from a large population a random
effects is more appropriate, while fixed effects model is better when estimating the flows
of trade between an ex ante predetermined selection of countries (Egger, 2006). This
paper aims at identifying the determinants of regional trade between eight ASEAN
members; thus fixed effects would be appropriate than random effects model. The eight
members were selected based on the availability of trade statistics during the period
1998-2010. However, the Hausman test statistic is applied to check further whether the
fixed effects model is more appropriate than the random effects model. If the null
hypothesis of no correlation between the individual effects and regressors is rejected,
then fixed effects model is better than the random effects model.
Following (Simwaka (2011) and Pastore et al. (2009)), the gravity model estimators are
used as a benchmark to assess the potential trade of COMESA regional integration. That
is, to assess the performance of intra-COMESA trade, the estimated coefficients of
gravity model relative to the intra-ASEAN trade model will be applied to the similar
specification of intra-COMESA trade model. The success of intra-COMESA trade
integration is measured by the ratio of potential to actual trade. As the projected potential
trade is the amount of trade that would be achieved if COMESA can achieves the intraregional trade integration similar to the ASEAN. The ratio of potential trade to actual
trade measures the success of intra-COMESA trade integration relative to intra-ASEAN
13
trade integration. A ratio of one indicates that the potential trade equals actual trade,
implying that intra-COMESA trade is at successful level compared to intra-ASEAN. The
higher is the ratio, the higher is the gap that needs to be filled by creation more trade. A
ratio less than one indicate that actual trade is close to its potential level.
5. Empirical Results
5.1. Estimation of Gravity Model
The results of estimation of gravity model in equation (1) using pooled, fixed effects and
random effects models are presented in Table 3. The results in the second column of
Table 3 are those of the pooled panel data model. The shortcoming of this model is that it
does not consider for heterogeneity of countries, and no country specific effects are
estimated, hence assumes that all countries are homogenous in terms of cross-section and
time.
Column three and five presents the results of the fixed effects models which take into
account the heterogeneity by estimating country specific effects. To support the
efficiency of fixed effects, the F-test was performed to check the poolability of the data3.
The result of the F-test shows that the null hypothesis of equality of the individual effects
is rejected, suggesting that a model with individual effects must be selected (i.e. fixed
effects or random).
Finally, the results in column 4 and 6 are those of the random effects model. The main
advantages of random effects model are: like fixed effects it allows heterogeneity in the
cross section, but it avoids the loss of degrees of freedom which occurs in fixed effects.
To choose between the FE and the RE models, the Hausman test rejects the hypothesis
that the coefficients of the FE models and the RE models are equal, suggesting that FE
estimates are more consistent. Nevertheless, the RE estimates are more efficient than the
FE estimates, but they are inconsistent (Pastore et al., 2009).
3
The pooled model is the (αij = α ) and same parameters over time and across trading partners, while the
unrestricted model allows an intercept and other parameters to vary across trading partners ()
14
One problem with a fixed effects model is that variables that do not change over time
(e.g. distance, common border or common language) cannot be estimated directly
because they are fixed effects and are therefore removed in estimates at the difference.
Since both time varying and time invariant effects must be taken into account in the
gravity model, we follow (Pastore et al., 2009) in relying on the RE models instead of FE
model. Precisely, we rely on RE model of column 6 which takes into account all
variables specified in our gravity model.
Table 3: Results of Gravity model estimation for intra-ASEAN trade (1998-2010)
Variable
Constant
LOG(GDPI)
LOG(POPI)
LOG(GDPJ)
LOG(POPJ)
LOG(DIS)
CL
CB
R2
F
Hausman Test
No of
Observation
No of Groups
Obs per group
The dependent Variable is the total trade
2
3
4
5
6
Pooled
FEM
REM
FEM
REM
-50.26***
2.20
-32.67***
-80.64
-18.12***
(-12.36)
(0.06)
(-3.63)
(-0.58)
(-4.07)
1.32***
-2.72***
0.94***
-2.65**
0.09*
(10.16)
(-3.03)
(4.29)
(-2.94)
(1.76)
0.36***
3.96
0.08***
(3.34)
(0.75)
(2.88)
1.49***
3.40***
1.46***
3.42***
0.17***
(11.76)
(3.88)
(6.77)
(3.87)
(5.73)
0.14
0.86
0.03
(1.32)
(0.16)
(0.48)
-1.29***
-1.12
-1.41
(-3.92)
(-1.19)
(-1.47)
-0.21
-0.23
-0.38
(-0.42)
(-0.16)
(-0.27)
2.51**
2.12
2.59*
(5.36)
(1.61)
(1.93)
0.64
0.78
0.58
0.78
0.61
94.14
34.74
24.05
33.70
18.22
31.95 (0.0000)
25.42 (0.0000)
728
728
728
728
728
56
7
56
7
56
7
Note: *, **, *** indicate significance at 10, 5 and 1 per cent respectively
-t-statistics in parentheses
15
56
7
56
7
The Table shows that all the estimated coefficients of the preferred model in column 6
carry their expected signs, and in line with the theory, except the dummy variable of
common language and ex-colony links. The coefficient of GDP of reporter country is
positive and significant as expected, implying that an increase in national income of the
exporter country encourages trade flow to the trading partner. The impact of reporter’s
population is also positive and significant, confirming most of the previous empirical
studies (e.g. Simwaka (2011) and Pastore et al. (2009)).
The GDP and population size of the trading partner have positive impact on trade flow
from the source countries. This finding indicates that a trading partner with large
economy and market size measured by its GDP and population size exerts positive effect
on the volume of trade with the reporter country. In addition, the coefficient of
geographical distance is negative as expected, implying that high transportation cost
between member countries negatively affect trade flow.
Unexpected, the coefficient of the dummy variable of common language or common
border is negative, but it is not statistically different from zero. This result suggests that
members speak same language or have ex-colony links tend to reduce the size of trade
between them. This finding could be explained by the fact that there is few countries in
ASEAN speak the same language and most of them are occupied by different colonizers.
Finally, the impact of common border is found to be positive, suggesting that members
sharing a common land or sea borders enjoy more trade activities between them.
5.2. Estimating Potential Trade of COMESA Integration
Having estimated the gravity model of ASEAN members, the next step is to project the
potential trade by applying the coefficients of the gravity model in column 6 of Table 3 to
a sample of eight COMESA countries, over the period 2004-2011. We excluded the
coefficient of common language or colony links, because it is not consistent, nor is
statistically significant. Thereafter, the potential trade will be compared with actual trade
in order to assess the success of intra-COMESA trade. The ratios of potential to actual
trade between each of eight COMESA members and other partner are presented in Figure
16
1 through 8 in Annex V. Since our sample involves eight countries, each of them has
seven country pairs which presented in seven graphs.
Figures 1 to 8 show that the ratio of potential to actual trade is greater than one, implying
that potential trade among these eight COMESA members is higher than actual trade
level. Though, a big difference between observed and predicted level of trade for all
trading members, there is a variation across countries. Figure 1 reveal that the actual trade
of Egypt with all trade partners is far from its potential, over the period 2004-2011. The
ratio of potential to actual trade exceeds more than one thousand for three country pairs:
Malawi, Zambia and Zimbabwe. This is referred to the far distance between Egypt and
these countries. On the other hand, Egypt has a low potential trade gap with Ethiopia,
Kenya and Mauritius. However, for all country pairs the gap between Egyptian potential
trade and actual one has been decreased in recent years,
As for the performance of Ethiopian regional trade with other selected COMESA
members, Figure 2 points out that Ethiopian actual trade with Egypt and Kenya is closer
to its potential compared to other countries in the sample. However, in the last decade
there is remarkable progress toward decreasing the gap between potential and actual trade
with all members, implying that Ethiopia has scope to increase its trade with other
COMESA members, benefited from the adoption of FTA in recent decade.
Regarding the regional trade performance of Kenya, Figure 3 reveals that Kenya is most
successful country among the selected COMESA member states. As shown in Figure 3
the actual trade of Kenya is less distant from its potential with all countries, especially
Uganda. Among all countries Kenya-Uganda trade relation is most successful, as the ratio
of potential to actual trade is less than 1. This situation is due to the common border,
language and ex-colony links between the two countries. Overall, compared to all
remaining eight countries, Kenya has a successful regional trade performance within
COMESA countries.
17
Figure 4 presents the performance of Malawi regional trade with other selected
COMESA countries. Like other members, Malawi is far from its potential with most of
the countries. However, with some countries like Egypt, Zambia and Zimbabwe, the
Malawian trade undergone improvement in last five years. This is referred to the
closeness of Zambia and Zimbabwe beside the huge economy of Egypt. The similar
results are reported for Mauritius in Figure 5. The Figure show that Mauritius’ trade is
also far from its potential with other countries during the period 2004-2011. With some
countries like Kenya, Zambia and Zimbabwe, the actual trade of Mauritius is somehow
close to its potential.
As for Ugandan regional trade performance, Figure 6 indicates that with most of the
country pairs in the sample the actual trade of Uganda is more distant from its potential.
Nevertheless, Uganda is closer to its potential with Kenya, supporting the results in the
Kenyan case. Also the results of Zambia and Zimbabwe trade potential in Figure 7 and 8
indicate that their potential trade is far from the actual trade for all selected COMESA
members, except the neighbouring countries. The figures show that the ratio of potential
to actual trade for Zambia-Zimbabwe and Zimbabwe-Zambia is close to one. On the
other hand, for non neighbouring countries, the ratio of potential/actual trade is very high,
reported more than thousand for some far distant countries such as, Ethiopia and Uganda.
This finding suggests that distance and transport costs have a negative impact on intraCOMESA trade performance.
Overall, all selected COMESA countries are far from their potential trade level, although
some countries like Egypt, Kenya and Ethiopia experienced a developed performance of
trade integration. Only Kenya-Uganda trade relation is the most successful country pair
in selected COMESA members, confirming the significant impact of common language,
borders and ex-colony links on regional trade. However, all country pairs have undergone
a decreasing trend of the ratio of potential to actual trade in last ten years, implying that
the adoption of FTA has result in removing trade barriers between the COMESA member
states.
18
6. Conclusion and Policy Implications
This paper aims at assessing the performance of intra-COMESA regional integration on
the basis of success of ASEAN trade integration, using an out-of-sample approach. The
analysis used a gravity model for a sample of ASEAN countries over the period 19982010. The estimated coefficients of the gravity equation were applied to intra-COMESA
trade model to calculate the potential trade. Then, the performance of intra-COMESA
trade integration is measured by the ratio of potential to actual trade.
The empirical results show that the actual trade between most of the country pairs is too
far from the estimated potential trade level, implying unfavorable performance of intraCOMESA trade compared to ASEAN trade integration. However, the ratio of potential to
actual trade has a downward trend for most of country pairs, suggesting a decreasing
difference between actual and potential trade over time. This result also indicates that
there is some progress towards the potential trade, indicating the success of
implementation of COMESA FTA in last decade. The results also show that actual trade
for some relatively developed courtiers like Egypt and Kenya is close to its potential,
especially Kenya which is found to be the most effective member in the sample.
Moreover, Kenya-Uganda pair has been found as most successful bilateral trade
integration among the COMESA members. Further, the results point out that the actual
bilateral trade between countries with common language, border or ex-colony links is
close to its potential level. This finding confirms the significant impact of distance and
cultural aspects on trade among COMESA members.
Based on the above findings, many policy implications can be drawn. As COMESA
regional cooperation is far from its potential integration level, policy makers in member
countries should adopt effective trade promotion measures to achieve trade potential
level. The downward trend of potential/actual trade ratio also means a progress toward
full trade cooperation which needs further actions to speed the economic and trade
integration process.
19
Precisely, exports diversification should be at the top of policy agenda for COMESA
countries. Therefore, member countries need to give special attention to industrialization
so as to enhance trade integration, since industrialization is the major reason behind the
success of ASEAN integration. In addition, increasing regional trade needs promoting
transport and communication infrastructure networks between the members. Further
efforts should be exerted for attracting Foreign Direct Investment (FDI) as well as
promoting the trade sectors. Finally, as COMESA future plans are to implement an
economic union, member countries should adopt comprehensive trade liberalization
measures ranging from removing tariff barriers to improve customs ports procedures at
the borders.
References
Al-Atrash, H. and Yousef, T. (2000), “Intra-Arab Trade: Is it Too Little?”, IMF Working
Paper, No. 10.
Anderson, J. (1979), “A Theoretical Foundation for the Gravity Equation”, The American
Economic Review, Vol. 9, No. 1.
ASEAN
Secretariat
(2013),
About
ASEAN,
available
online
at:
http://www.asean.org/news/asean-secretariat-news
Balassa, B. (1961), The Theory of Economic Integration, Richard Irwin, Homewood
(Ill.).
Chauvin, S. and Gaulier, G. (2002), “Regional Trade Integration in Southern Africa”,
CEPPI, Working paper No. 12.
COMESA Profile (2013): available: www.comesa.int
COMESA statistics database (COMSTAT) (2013). http://comstat.comesa.int/Home.aspx
20
Egger, P (2000), “A Note on the Proper Econometric Specification of the Gravity
Equation”, Economic Letters, 66, 25-31.
IMF (2013) World Economic Outlook
Khandelwal, P. (2004) “COMESA and SADC: Prospects and Challenges for Regional
Trade Integration”, IMF Working Paper, No. 227.
Linnemann, H. (1966), An Econometric Study of International Trade Flows. Amsterdam:
North-Holland.
Markheim, D. (1994) “A note on Predicting the Trade Effects of Economic Integration
and other Preferential Trade Agreements: An Assessment”, Journal of Common Market
Studies, Vol. 32, No.1.
Oyejide, A., Elbadawi, I and Yeo, S. (1999), Regional Integration and Trade
Liberalization in Sub-Saharan Africa, Vol. 3, Regional Case Studies (London: Macmillan
Press).
Pastore, F., Ferragina, A. and Giovannetti, G. (2009) “A Tale of Parallel Integration
Processes: A Gravity Analysis of EU Trade with Mediterranean and Central and Eastern
European Countries”, Review of Middle East Economics and Finance 5 (2).
Simwaka, K. (2011), “An Empirical Evaluation of Trade Potential in Southern African”,
AERC Research Paper, No. 235. African Economic Research Consortium, Nairobi.
Stack, M. and Pentecost, E. (2011) “Trade Integration A Gravity Model Approach to
Estimating Prospective Trade Gains in the EU Accession and Associated Countries”,
Nottingham University Discussion Paper No. 2010/11.
21
Tinbergen, J. (1962), Shaping the World Economy-Suggestions for an International
Economic Policy. New York: The Twentieth Century Fund.
UNCTAD, (various issues), available online at: http://unctad.org/en/Pages/Statistics.aspx
World Bank’ Development indicators, (various issues).
22
Annexes
Annex I: The share of COMESA and ASEAN in World Exports: 2000-2010
REGION
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
COMESA
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.04
0.05
0.06
0.03
ASEAN
1.54
1.41
1.43
1.56
1.55
1.58
1.58
1.55
1.56
1.59
1.73
1.55
Sources: UNCTAD Statistics and IMF Direction of Trade CD-ROM.
Annex II: List of Countries Considered for the Study
ASEAN Members
Brunei Darussalam
Indonesia
Laos
Malaysia
Philippines
Singapore
Thailand
Vietnam
COMESA Members
Egypt
Ethiopia
Kenya
Malawi
Mauritius
Uganda
Zambia
Zimbabwe
Annex III: Descriptive Statistics of the Variables Used in the Analysis- ASEAN Integration
Variable
Trade
Mean
SD
Skewness
Kurtosis
Normality
Observations
19.047
5.548
-2.135
7.464
157.75 (0.000)
728
Reporter’ GDP
24.733
1.581
-0.897
2.659
101.30 (0.000)
728
Partner’ GDP
24.680
1.630
-0.846
2.483
94.94 (0.000)
728
Reporter’ population 16.792
1.979
-0.767
2.527
78.30 (0.000)
728
Partner’ population
16.713
1.985
-0.660
2.385
64.34 (0.000)
728
Annex IV: Descriptive Statistics of the Variables Used in the Analysis- COMESA
Variable
Trade
Reporter’ GDP
Partner’ GDP
Mean
16.084
23.401
23.196
SD
2.403
1.081
0.763
Skewness
-0.440
0.925
0.447
Kurtosis
2.627
3.362
3.686
Jarque-Bera
17.043(0.000)
66.24 (0.000)
23.69 (0.000)
Observations
447
447
447
Reporter’ population
Partner’ population
16.779
16.792
1.253
1.2649
-0.939
-0.915
3.241
3.205
66.84 (0.000)
63.21 (0.000)
447
447
23
Annex V: Intra-COMESA-Potential Trade/Actual Trade (2004-2011)
Figure 1: Egypt-COMESA: Potential Trade/Actual Trade (2004-2011)
Ethiopia
Kenya
Potential Trade /Actual Trade
300
200
100
0
60
40
20
0
Malawi
Mauritius
6000
4000
2000
0
300
200
100
0
Uganda
Zambia
1500
3000
2000
1000
0
1000
500
0
Zimbabwe
8000
6000
4000
2000
0
24
Figure 2: Ethiopia- COMESA: Potential Trade/Actual Trade (2004-2011)
Kenya
Egypt
400
300
200
100
0
400
300
200
100
0
Potential Trade /Actual Trade
Malawi
6000
Mauritius
4000
3000
4000
2000
2000
1000
0
0
2005 2006 2007 2008 2009 2010
2005200620072008200920102011
Uganda
Zambia
6000
6000
4000
4000
2000
2000
0
0
Zimbabwe
6000
4000
2000
0
25
Figure 3: Kenya-COMESA: Potential Trade/Actual Trade (2004-2011)
Egypt
Potential Trade /Actual Trade
8
6
4
2
0
Ethiopia
20
15
10
5
0
Malawi
30
Mauritius
80
60
40
20
0
20
10
0
Zambia
Uganda
1.5
15
1
10
0.5
5
0
0
Zimbabwe
300
200
100
0
26
Figure 4: Malawi-COMESA: Potential Trade/Actual Trade (2004-2011)
Kenya
Egypt
80
60
40
20
0
20
15
10
5
0
Mauritius
Ethiopia
800
600
400
200
0
10000
Potential Trade /Actual Trade
5000
0
Zambia
Uganda
25
20
15
10
5
0
1200
1000
800
600
400
200
0
Zimbabwe
12
10
8
6
4
2
0
27
Figure 5: Mauritius-COMESA: Potential Trade/Actual Trade (2004-2011)
Egypt
Ethiopia
8000
6000
4000
2000
0
1000
800
600
400
200
0
Kenya
Potential Trade /Actual Trade
100
Malawi
6000
4000
50
2000
0
0
Uganda
Zambia
1500
600
1000
400
500
200
0
0
Zimbabwe
300
200
100
0
28
Figure 6: Uganda-COMESA: Potential Trade/Actual Trade (2004-2011)
Ethiopia
Egypt
3000
1000
800
600
400
200
0
2000
1000
0
Malawi
Kenya
1200
1000
800
600
400
200
0
Potential Trade /Actual Trade
8
6
4
2
0
Mauritius
Zambia
6000
4000
2000
0
1500
1000
500
0
Zimbabwe
6000
4000
2000
0
29
Figure 7: Zambia-COMESA: Potential Trade/Actual Trade (2004-2011)
Egypt
300
200
100
0
Ethiopia
2000
1000
0
Potential Trade /Actual Trade
Kenya
Malawi
40
30
20
10
0
8
6
4
2
0
Mauritius
600
400
200
0
Uganda
1000
500
0
Zimbabwe
6
4
2
0
30
Figure 8: Zimbabwe-COMESA: Potential Trade/Actual Trade (2004-2011)
Egypt
Ethiopia
1000
6000
4000
500
2000
0
0
Potential Trade /Actual Trade
Kenya
Malawi
400
300
200
100
0
15
10
5
0
Uganda
Mauritius
150
4000
3000
2000
1000
0
100
50
0
Zambia
5
4
3
2
1
0
31