Globalization and Feasibility Study of Regionalism in the Middle East and North Africa

Introduction				) is consistent and efficient but random effects
New regionalism is the point of consensus among				estimator ( ) is inconsistent.
development strategies of conflicting economic				Thus under the null hypothesis there is no systematic
doctrines. In the dominant neo-liberal paradigm,				difference between the two estimators. So we can
regionalism is assumed as a crucial phase in the				rearrange the hypothesis as
transformation of the international economic system				(4)
to globalization. The alternative paradigm also				(5)
considers regionalism as the a point of departure for				Hausman’s Statistic test (HT) is as below:
developing countries to alleviate the hegemonic				(6)
pressures of capitalism and increase their bargaining				This statistic has asymptotically distribution with K
power and as an opportunity for the formation and				degree of freedom where K is the elements of
evolution of an alternative economic system. In the				vector ? i.e.
meantime, the organization of Islamic conference				(7)
(OIC) has also viewed the regionalism as a suitable				(See: Baltagi (2001), Green (2003) and Verbeek
strategy for economic development by proposing the				(2004)
formation of an Islamic common market.				2-1-2. Modeling
In line with this strategy, this paper deals with the				To formulate an empirical model to test Linder’s
feasibility study of regionalism in the MENA region,				theory in a DID criteria emphasis is placed on two
the region where Abrahamic religions and the glorious				explanatory variables:i. the difference in the per capita
Islamic civilization, -the superior and an unparalleled				income of a particular country and the per capita
civilization of the middle ages- were born and grown.				income of the region under study (Muslim world and
However, the industrial revolution and the post				ROW). This variable, which we shall call it Linder
Renaissance environment brought about by the				variable is obtained from the square of the deviation
West's scientific boom and boost on the one hand,				per capita income of the reference country from the
and the incompetence of retrogressive Muslim rulers				average of per capita income of the region
of the time on the other hand created a deep gap				Lin=(PCGi-PCGj )2. PCG refers to per capita income
between the region and the Europe – and later				and the subscript i stands for country i in the MENA
on- the US. The geo-strategic importance of the				region and subscript j shows Muslim countries and
region and the need for the revival and				ROW.ii. The size of a country’s economy as a
reconstruction of the Islamic civilization were the main				ratio of the whole region’s GDPT. This variable is
reason behind the selection of the regions member				computed as:
countries as the statistical population.				(8)
In this paper, the Linder theory forms the underlying				In this model the volume of trade of the country to
part of the feasibility study of establishing a regional				the region under study i.e. XTij is the dependent
free trade area - and an Islamic common market, in				variable. (all figures in constant 1996 US dollars).
the later stages- as the process of globalization is				Moreover, to model the DID method, dichotomous
going ahead. In contrast with the classical trade				variables should be used. To examine the effects of
theories (such as the Heckscher- Ohlin Theory), the				globalization on both the slope and the intercept the
Linder theory focuses on the demand side and				three dichotomous variables are introduced:i. G as a
explains the trade patterns on the basis of the				proxy of globalization, a dichotomous variable for
similarities in the demand structures. As the similarities				which two values are assigned: zero for period prior
of demand structures largely depends on the				to 1991 and 1 for periods after 1991. This variable
convergence of per capita income, the following two				gives us the intercept;ii. Product of G into Linij which
hypotheses are simultaneously tested:				gives G_Lin¬ij;iii. Product of G multiplied by
Globalization brings about convergence of per capita				GDPT¬ij which yields G_GDPT¬ij¬.
income trends in the MENA region, while the North-				Thus, the empirical (Penal Data) model classified by
South per capita incomes demonstrate divergent				the trade relation of a country with the concerned
trends. Theoretically, this hypothesis is indebted to				regions (Muslim world and rest of the word) may be
Matsuyama’s (1996) symmetric breaking and				written as:
Deardroff’s (1998) multi cone theories and, in				(9)
general, suggests the presence of global divergence				For years after 1991 that the most of countries in
and hemispherical convergence. The hypothesis test				the region have launched economics reforms, G=1
is performed using the difference in differences				and for years prior to that G=0. Hence for the years
method.				before 1991 we will have
The Linder theory adequately explains the trade				(10)and for the years after 1991 we will have
behaviors of the MENA countries with the Muslim				(11)
countries and the rest of world. That is to say; with				Now, if Hausman test rejected the first hypothesis
global divergence and hemispherical convergence, the				fixed effects model should be employed as a result
MENA's intra-regional trade will increase and the				?j=0, otherwise according to random effects model
inter-regional will decrease.				?j will take a constant value in the country’s
Hence, the feasibility study of regionalization, and the				trade relations.
formation of a common market will be carried out in				According to Linder’s theory it is expected that
the three steps:i. The impact of globalization on the				?20.
convergence or the divergence of per capita incomes				2-1-3. Data and Sample
across MENA's countries;ii. The impact of globalization				As stated, this paper is mainly aimed at examining
on the convergence or the divergence of per capita				the effects of globalization on the trade relation of
incomes across Northern and southern countries,				MENA region. As a result of statistical problems, some
andiii. An Empirical Test of Linder’s hypothesis to				country had to be dropped from the sample. The
examine the trade patterns of regional countries				remaining countries are Jordan, Algeria, UAE, Iran,
before and after the trade liberalization.				Bahrain, and Tunisia, Iraq, Saudi Arabia Oman, Qatar,
Together with Slaughter (2001) studies, that support				Kuwait, Lebanon, Libya, Morocco, Egypt and Yemen.
the divergence of the per capita income trends due				These countries’ trading partners are categorized
to globalization across the world, this paper asserts				as Muslim world (all members of OIC) and the rest of
that globalization paves the way for regionalization				the world. Penn table and SESRTCIC website were
and the formation of an Islamic common market in				the sources for the raw data, which were used in
the MENA region.				this research after certain processes and calculations
The table below lists the results of all possible				of needed indexes and ratios were performed. From
scenarios for the above two hypothesis tests. The				1975- 2002 was chosen as the period under study.
first column corresponds to the first hypothesis; the				To test the stationary of variables unit root test was
second, to Slaughter’s studies; the third and				used. Given that, our null hypothesis was the
fourth, to the results of the second hypothesis test.				non-stationarity i.e.
If the global divergence and the regional convergence				H0: \| ? \| = 1 (12)
were confirmed and the Linder theory could explain				H1: \| ? \| < 1 (13)
the trade behaviors of the region with the Muslim				The result (table 2 and 3) shows that null hypothesis
world and other countries, we can expect that				is rejected. Hence, the stationary of variables and
globalization will help make the establishment of a				their convergence with the passage of time are
common market more feasible. The third (second)				confirmed.
scenario illustrates the best (worst) situation in the				2-1-4. Hypothesis Testing process
subject under discussion.				After introducing several Penal Data Models and
AS Elahi and Nahanvdian (2005) have confirmed the				examining the significance level (t-statistic) and the
convergence of per capita income trends in the				regression (f- statistic), significant variable were
region using DID method and Slaughter (2001) has				identified. Hausman test shows that the null
verified the North-South divergence using the same				hypothesis cannot be rejected in both tests (trade
method, there is no need to test the first hypothesis.				relation with Muslim countries and the rest). It means
It means that the two first stages –global				that GLS estimator is of random effects ( ) is
divergence and the hemispherical convergence- of				consistent, unbiased and efficient. See table 6, and
feasibility study of regionalization and the				7.The results of verified regressions are depicted in
establishment of an Islamic common market have				table 4 and 5.
been accomplished. However, to finalize the feasibility				2-2. Interpretations of the results
project we just need to test the second hypothesis.				Considering that one coefficient of dichotomous
The paper consists of two parts: the fist part				variable (G) leaked the due significance it was
explains Linder's theory and the second part tests				omitted from the model. According to the results
the relevance of this theory in the trade behaviors of				included in table 4, the trade behavior of countries
the Middle East and North African countries. In				under study with Muslim world is explained as below:
conclusion, considering Slaughter (2001) and Elahi				(14)
Nahavandian (2005) studies on the one hand and the				(3.47) - (33.99) (-7.01) (5.113) (2.20)
results of tests of Linder's theory on the other hand,				Subscript i in this equation and equation 15 and 16
the feasibility of regionalization in countries under				refers to Islamic countries (IC).
study is examined				As we observe all coefficients enjoy a high level of
Table 1- the possible Scenarios				significance. Based on these results trade behavior of
Scenario Per Capita Income The Test of Linder's				these countries prior to economic reforms (1991) can
Hypothesis Result				be expressed by
MENA North and South Muslim Countries The Rest of				(15)
World Intra-regional Trade Inter-regional Trade				The Negative sign Linder coefficient suggests that
Feasibility Study				with a decline in the per capita income gap across
1 Convergent Convergent Verification Verification				MENA and Muslim world for the period prior to
Increase Increase No Preference				globalization promoted trade relations. Moreover, a
2 Divergent Convergent Verification Verification				positive sign of the variable of size of economy
Decrease Increase Impossibility				shows that a faster growth of countries in region
3 Convergent Divergent Verification Verification				compared to the whole Muslim world will lead to an
Increase Decrease Highly Possibility				increase in the export of these countries. That is to
4 Divergent Divergent Verification Verification				say the greater the economic diversity the greater
Decrease Decrease No Preference				the economic relations. All these theories are fully
5-6 Convergent/ Divergent Convergent Rejection				consistent with Linder’s theory.
Verification No Judgment Increase -				To drive the equation of region’s trade behavior
7-8 Convergent/ Divergent Divergent Rejection				with Muslim world and for the period after economic
Verification No Judgment Decrease -				reforms we have to obtain ?1+ ?3 and ?2+ ?4 from
9-10 Convergent Convergent/ Divergent Verification				regression 9 i. e.
Rejection Increase No Judgment Likely possibility				(16)
11-12 Divergent Convergent/ Divergent Verification				According to coefficients obtained it becomes clear
Rejection Decrease No Judgment -				that on the one hand, globalization intensifies the
13-16 Convergent/ Divergent Convergent/ Divergent				effects of size of the economy, but on the other
Rejection Rejection No Judgment No Judgment -				hand, it decreases the effects of similarity of
I. Linder's Theory				demand patterns on trade trends. Yet, as the per
1-1. The Background				capita income gap across Muslim countries declines
Prior to 1960, trade theories were based on supply				extension of trade relations may be expected.
side and the Heckscher–Ohlin as the most popular				Consequently, we can argue that even globalization
theory placed emphasis on the factor endowment as				has not eliminated the relevance of Linder theory.
the determinant of trade model and the relative				Moreover, in view of the results of Elahi and
advantage. According this theory labor abundant				Nahavandian (2005) and Slaughter (2001) studies on
country should specialize in production and export of				the effects of globalization on the convergence in
labor-intensive goods. They should import the needed				regional incomes and divergence in global incomes in
capital-intensive commodities from countries with				process of globalization the potential for regionalism
higher per capita income.				will be strengthened paving the way for
The theory was based on a number of assumptions				establishment of FTA and – the in the later
including the similarity in consumption patterns and				stage- Islamic Common Market.
technology, the existence of constant return to scale				On the other hand, to explain the trade behavior of
and a competitive condition and the irreversibility of				MENA region with other countries two dichotomous
factor intensity in the countries involved. In addition				variables (G and G_GDPTit) were omitted from
to providing a fair explanation of relative advantage				regression No. 9 as they lacked the due significance.
-on the basis of factor endowment – this theory				As we see in table No 5 the trade behaviors of
ensures the absolute factor price equality drawn				countries of the region with the rest of world may
upon Samuelson’s contribution. Leontief’s				be written as:
tests (1954 and 1956) on American exports and				(17)
imports revealed that the US imports was surprisingly				(2.78) - (-3.76) (2.90) (3.08)
capital-intensive goods, whilst, US per capita capital				The coefficients of this regression equation also
was greater than any other country. Baldwin (1971)				enjoy a high level of significance. Considering the
also states that this paradox continues to exist. He				G_Lin dichotomous variable, we can drive the
found that US imports are 27% more capital				following equations for the two periods of the pre
intensive.				and post economic reforms
Tatemoto and Ichimura’s studies (1959) revealed				(18)
that this paradox exists in Japan too, -though in				(19)
another form. Japan is a labor abundant country				These two equations explicitly reveal that Linder
compared to advanced countries but a capital				theory can explain trade behaviors of MENA countries
abundant country relative to other nations. Yet				with the rest of the world. Considering Slaughter
Japan’s export was found to be capital intensive				(2001) studies which supports global divergence in
commodities and her imports were labor intensive				globalization process we can conclude that as the
goods. Whal (1961) also studied the Canadian trade				income gap increases the volume of trade in MENA
pattern and observed that although the trade with				countries will fall. The decline in region’s trade
US accounted for most of Canada’s trade				relations with rest of the world and the extension of
relation and US was a capital abundant nation				volume of trade in Islamic world sets the stage for
Canada’s export was relatively capital intensive .				the formation of a Free Trade Area.
Bharawaj (1962) also discovered the irrelevance of				Conclusions:
HO theory in the Indo-US trade relations				In view of the new wave of regionalism and
In general, one can argue that the Leontief paradox				establishment of a common market recently raised
was a turning point in the development of new trade				by OIC, this paper focused on feasibility of
theories. Leontief and others attempted to present				regionalism in the Middle East and North Africa in the
explanations to justify the inconsistency of the				age of globalization. Using a new and novel method it
Heckscher Ohlin theory with the trade patterns of				was shown that globalization not only acts as a
the countries under their study, by raising the				hindrance to the creation of an FTA in MENA rather it
difference in US labor productivity, the factor				prepares the ground for the realization of this goal.
intensity reversal, demand biasness, the abundance				The rational for feasibility study could be summarized
of natural resources in US, transportation and tariff				as under:i. According to Slaughter (2001) studies the
costs, and negligence of Heckscher–Ohlin to				per capita income trends of advanced and developing
Human capital. However, Daniel Trefler (1993)				countries are diverging;ii. According to Elahi and
stresses that in view of the excessively dispersed				Nahavandian (2005), there is a converging trend in
factor prices in different countries it is naïve to				the per capita incomes of MENA countries;iii.
talk of factor price equalization. Moreover, the				According to Linder’s theory – tested in this
existence the North-North trade shows that HO is				paper- , the more convergence (divergence) in per
incapable of explaining the trade behavior and the				capita income the greater (smaller) the trade volume.
findings of empirical studies have repeatedly				Thus, with the convergence in the region’s per
confirmed the inconsistency of HO theory with these				capita income and the confirmation of Linder’s
findings.				theory it is expected that intra regional trade with
In search for a solution to the irrelevance with HO				Islamic countries will rise and with the non-Muslim
theory with empirical studies, several efforts have				countries will decline.
been made to reform and reformulate the theory in				It should be noted that for a successful regionalism
such a way that the core idea in the theory is saved.				several conditions should be satisfied, the major
Hence, different versions such as HOV and HOC				condition being the economic complementarity of
versions of this theory have been presented to save				member countries. However, this issue was beyond
the essence of the theory even if it results to				the scope of this paper which has focused on
violation of some assumptions. However, the				examining the impacts of globalization on the volume
achievement seems to be inconsiderable.				of trade exchanges between MENA countries and
1-2. Linder’s Theory and Its Elements				Islamic world and rest of the world
Linder has examined the trade behavior from quiet a				Table 2- unit root test (variables set 1)
different approached. His approach is usually identified				Pool unit root test: Summary
with the demand side. With this approach, it seems				Date: 01/28/05 Time: 13:27
that he has succeeded to develop a theory with				Sample: 1980 2003
grater consistency with the statistical truths obtained				Series: XTIC_ALG, XTIC_BHR, XTIC_EGY,
by Leontief tests and similar tests. However, as				XTIC_IRN, XTIC_IRQ,
Leamer and Levinsohn (1995: 1383) have written,				XTIC_JOR, XTIC_KWT, XTIC_LBN, XTIC_LBY,
‘while Linder did not have a formal model, he had				XTIC_MAR,
a compelling story’. According Fillat-Castejon and				XTIC_OMN, XTIC_QAT, XTIC_SAU, XTIC_SYR,
Serrano-Sanz, this approach has caused his theory to				XTIC_TUN,
enjoy a high degree of success. In their view,				XTIC_UAE, XTIC_YEM, GDPTIC_ALG,
Linder’s theory has done better in explaining the				GDPTIC_BHR,
trade behavior “Linder considers potential trade				GDPTIC_EGY, GDPTIC_IRN, GDPTIC_IRQ,
to be explained by the so-called ‘trade-creating				GDPTIC_JOR,
forces’, whilst certain ‘brakes’ will				GDPTIC_KWT, GDPTIC_LBN, GDPTIC_LBY,
deviate real trade away from its potential, with the				GDPTIC_MAR,
pattern of trade and the trading partners of each				GDPTIC_OMN, GDPTIC_QAT, GDPTIC_SAU,
country being determined by this conjunction of				GDPTIC_SYR,
trade creating and braking forces.”				GDPTIC_TUN, GDPTIC_UAE, GDPTIC_YEM,
Linder (1961) challenged some beliefs on the theory				LINIC_ALG,
of international trade at that time, particularly the HO				LINIC_BHR, LINIC_EGY, LINIC_IRN, LINIC_IRQ,
theorem. According to HO, relative endowments of				LINIC_JOR,
productive factors theory provided an explanation for				LINIC_KWT, LINIC_LBN, LINIC_LBY, LINIC_MAR,
the model, which placed emphasis on the differenced				LINIC_OMN,
goods in terms of factor intensity and countries in				LINIC_QAT, LINIC_SAU, LINIC_SYR, LINIC_TUN,
terms of factor endowment. Therefore, trade of				LINIC_UAE,
capital-intensive goods with labor-intensive goods				LINIC_YEM
between capital abundant and labor abundant				Exogenous variables: Individual effects, individual linear
countries should be established. While trade				trends
exchanges were largely established in goods of similar				Automatic selection of maximum lags
characteristics and between countries with				Automatic selection of lags based on SIC: 0 to 4
comparable levels of development enjoying a very				Newey-West bandwidth selection using Bartlett
high rate of growth. In fact, Linder’s theory tried				kernel
to provide an answer to these two aspects, namely				Cross-
the pattern of trade and the trading partners.				Method Statistic Prob.** sections Obs
Another novel point in Linder’s theory is the				Null: Unit root (assumes common unit root process)
emphasis it places on the dynamic aspects of the				Levin, Lin & Chu t* -15983.7 0.0000 51 1012
relationships between trade and development. The				Breitung t-stat -1.88254 0.0299 51 961
growth experienced by a country modifies its				Null: Unit root (assumes individual unit root process)
demand structure and, thereby, the range of both				Im, Pesaran and Shin W-stat -2000.53 0.0000 51 1012
potential and real exports, explaining how the pattern				ADF - Fisher Chi-square 997.442 0.0000 51 1012
of trade changes over time. Within this scope of				PP - Fisher Chi-square 987.951 0.0000 51 1031
potential trade, actual trade is determined from a set				Null: No unit root (assumes common unit root
of factors that tend to strengthen it, the so-called				process)
trade creating forces, and others which tend to limit				Hadri Z-stat 18.0027 0.0000 51 1061
it, the so-called trade braking forces. All these offer				** Probabilities for Fisher tests are computed using
an underlying theoretical basis for prediction of trade				an asympotic Chi
behaviors. Those in greater demand within the				-square distribution. All other tests assume
country will be exported—the so-called expansion				asymptotic normality.
thesis—while those in less demand will be				Table 3- unit root test (variables set 2)
imported. This approach involves a form of a priori				Pool unit root test: Summary
reasoning which illustrates the intra-industry trade and				Date: 01/28/05 Time: 13:24
Linder’s approach. Gray (1988) considers				Sample: 1980 2003
Linder’s approach as a key element in the				Series: XTRW_ALG, XTRW_BHR, XTRW_EGY,
intra-industry trade a paradigm’.				XTRW_IRN,
1-2-1. The Trade-Creating Forces				XTRW_IRQ, XTRW_JOR, XTRW_KWT,
As stated above and unlike supply side theories of				XTRW_LBN, XTRW_LBY,
trade, Linder turned his attention towards demand				XTRW_MAR, XTRW_OMN, XTRW_QAT,
when seeking to explain trade. According to his				XTRW_SAU,
thinking, the demand characteristics of two countries				XTRW_SYR, XTRW_TUN, XTRW_UAE,
would act as decisive factors in explaining potential				XTRW_YEM,
trade, and it is Linder’s core idea that has				GDPTRW_ALG, GDPTRW_BHR, GDPTRW_EGY,
developed a significant part of the subsequent				GDPTRW_IRN, GDPTRW_IRQ, GDPTRW_JOR,
literature. In this theory, monopolistic competition is				GDPTRW_KWT, GDPTRW_LBN, GDPTRW_LBY,
considered to be as a possible factor in the growth				GDPTRW_MAR, GDPTRW_OMN, GDPTRW_QAT,
of intra-industry trade.				GDPTRW_SAU, GDPTRW_SYR, GDPTRW_TUN,
The relationship between demand and international				GDPTRW_UAE, GDPTRW_YEM, LINRW_ALG,
trade can be established in two ways, that is to say,				LINRW_BHR,
through the complementarity of the demand				LINRW_EGY, LINRW_IRN, LINRW_IRQ,
structures of two countries and through the degree				LINRW_JOR,
of representativity of the demand for common				LINRW_KWT, LINRW_LBN, LINRW_LBY,
products. In the first way, emphasis is placed on the				LINRW_MAR,
trade for the satisfaction of necessities. The second				LINRW_OMN, LINRW_QAT, LINRW_SAU,
approach emphasizes on the characteristic of				LINRW_SYR,
demand.				LINRW_TUN, LINRW_UAE, LINRW_YEM
In both methods, the production substantially occurs				Exogenous variables: Individual effects, individual linear
for the satisfaction of internal necessities of a				trends
country and - in view of the production possibilities -				Automatic selection of maximum lags
the surplus to internal needs are exported to				Automatic selection of lags based on SIC: 0 to 4
countries with similar demand structures.				Newey-West bandwidth selection using Bartlett
Linder rightly argues that the determinants of the				kernel
Demand structure are the modal or median per capita				Cross-
income and the average per capita income is not a				Method Statistic Prob.** sections Obs
good determinant of demand structure - particularly				Null: Unit root (assumes common unit root process)
for countries with a income dispersion. However, its				Levin, Lin & Chu t* -22022.1 0.0000 51 998
too difficult to get the per capita income distribution				Breitung t-stat -3.10329 0.0010 51 947
of different countries hence, income average is used				Null: Unit root (assumes individual unit root process)
to determine the demand structure (Linder, 1961:94).				Im, Pesaran and Shin W-stat -3454.40 0.0000 51 998
To justify the role of per capita income in a				ADF - Fisher Chi-square 748.777 0.0000 51 998
country’s demand structure, Linder draws on the				PP - Fisher Chi-square 727.799 0.0000 51 1027
concept of the income elasticity of demand (Linder,				Null: No unit root (assumes common unit root
1961: 94).				process)
As it is deduced from Engel’s law, by increasing				Hadri Z-stat 14.9077 0.0000 51 1065
per capita income, higher quality and luxury goods are				** Probabilities for Fisher tests are computed using
regarded as necessity and former necessity fall in				an asympotic Chi
inferior goods basket resulting in an increase in the				-square distribution. All other tests assume
demand for luxury goods. With respect to the				asymptotic normality.
degree of representativity of demand, when a				Table 4- trade behaviors of the MENA countries with
country’s production potential is greater, the				the Muslim countries
probability of exporting will be higher otherwise; these				Dependent Variable: XTIC?
demands will be satisfied with imports.				Method: GLS (Variance Components)
An original point in this analysis is that the trading				Date: 11/02/04 Time: 10:25
countries enjoy a similar level of demand structure				Sample: 1975 2002
and hence similar per capita income distribution.				Included observations: 28
So to speak, one can claim that the closer the per				Number of cross-sections used: 17
capita income average the higher the possibility of				Total panel (unbalanced) observations: 364
occurrence of trade. Theoretical developments in the				Variable Coefficient Std. Error t-Statistic Prob.
analysis of demand using models inspired by Linder				C 771.8152 351.2558 2.197302 0.0286
have concentrated on three topics. (Fillat-Castejón				GDPTIC? 172.0235 33.64398 5.113055 0.0000
and Serrano-Sanz, 2004: 326-7) the association				LINIC? -3.12E-09 4.45E-10 -7.006903 0.0000
between the level of income and the demand for				G_GDPTIC? 278.5233 33.99053 8.194144 0.0000
quality, on the basis of consumer preferences				G_LINIC? 1.75E-09 5.04E-07 3.471607 0.0006
expressed in terms of the characteristics of the				Random Effects
goods and not just in terms of quantities. This				_ALG--C -816.3480
approach allows us to explain why economic growth				_BHR--C -170.2156
leads to a higher horizontal differentiation of products				_EGY--C -932.7171
and to an increase in the average quality or				_IRN--C -1182.359
sophistication that is demanded. There are				_IRQ--C 448.6195
non-homothetic preferences and the growth in				_JOR--C -314.7449
income affects the demand for different goods in				_KWT--C 249.3942
different ways, giving rise to structural changes.				_LBN--C -456.1962
Markusen (1986) tested Linder’s observation that				_LBY--C -348.9692
people with similar per capita incomes consume similar				_MAR--C -678.7201
sets of goods. Non-homothetic preferences, which in				_OMN--C -82.74419
Markusen’s analysis have taken the form of an				_QAT--C -826.9708
assumption, are formalized and empirically tested in				_SAU--C 3883.275
Hunter and Markusen (1988). According to this logic,				_SYR--C 300.9173
the change in the structure of demand will have				_TUN--C -509.9924
implications over the composition of trade, in that the				_UAE--C 1491.185
greater the non-homothetic nature of demand, the				_YEM--C -732.7498
more intense will be the trade between two				GLS Transformed Regression
countries with similar per capita incomes. The				R-squared 0.806204 Mean dependent var 1260.885
distribution of income and preferences within				Adjusted R-squared 0.804045 S.D. dependent var
countries is an essential point when considering the				1740.713
possible overlapping of demands and defining the				S.E. of regression 770.5584 Sum squared resid
varieties or qualities of a good to be traded. The				2.13E+08
usual models of international Trade neglect the				Durbin-Watson stat 1.938733
details, but Linder’s ideas allow us to be more				Unweighted Statistics including Random Effects
exact. Hence, in countries with an even income				R-squared 0.813467 Mean dependent var 1260.885
distribution and with a similar level of per capita				Adjusted R-squared 0.811389 S.D. dependent var
income, we expect that the overlapping of demand				1740.713
increases.				S.E. of regression 755.9813 Sum squared resid
However, with an even per capita income and an				2.05E+08
uneven distribution around an average, a range of				Durbin-Watson stat 1.959710
qualities will be demanded for each type of product				Table 5- trade behaviors of the MENA countries with
and a form of vertical differentiation, will emerge.				ROW
When the income is concentrated at higher level				Dependent Variable: XTRW?
better quality products will be traded otherwise lower				Method: GLS (Variance Components)
quality product will gain greater significance.				Date: 11/02/04 Time: 10:41
Henceforth, dispersion of income distribution will exert				Sample: 1975 2002
an influence over trade in an aggregate form, and by				Included observations: 28
way of the range of varieties that are susceptible to				Number of cross-sections used: 17
trade. However, as mentioned above access to				Total panel (unbalanced) observations: 349
dispersion of income of countries and fitting them in				Variable Coefficient Std. Error t-Statistic Prob.
model is beyond the scope and space of this paper.				C 7740.308 2512.330 3.080929 0.0022
According to Linder (1961: 110), small sized countries				GDPTRW? 116986.6 40382.56 2.896959 0.0040
establish greater trade with larger ones rather than				LINRW? -2.78E-08 7.39E-09 -3.763694 0.0002
smaller countries. Therefore, product differentiation is				G_LINRW? 1.93E-08 6.92E-07 2.784526 0.0057
another trade-creating force, although this aspect				Random Effects
was hardly developed in his work. However, the				_ALG--C 1791.944
volume of trade is positively associated with the size				_BHR--C -5145.395
of economy and market. It has subsequently				_EGY--C -7095.146
received a great deal of attention, above all in				_IRN--C 2628.343
relation to the size of the market. Moreover, this is				_IRQ--C -413.8024
the reason of greater trade exchanges between				_JOR--C -8215.257
small sized -large sized countries compared to small-				_KWT--C 827.2209
small countries. Other studies carried out on the basis				_LBN--C -9016.838
of Linder ‘s theory suggest that size also				_LBY--C 131.9288
conditions the possibilities of diversification and				_MAR--C -5479.285
manifests itself in volume and specialization; that is to				_OMN--C -4193.818
say, it has not only a quantitative but also a				_QAT--C -3752.049
qualitative influence.				_SAU--C 33539.94
For example, Keesing (1968) demonstrated how the				_SYR--C -6665.189
larger size of a country led to a higher exports and				_TUN--C -5458.853
lower imports, whilst both depended positively on				_UAE--C 10743.85
income. Balassa Balassa, B. and Bauwens, L. (1988)				_YEM--C -8078.223
confirmed the need for large internal markets for the				GLS Transformed Regression
export of manufactures, due to scale economies,				R-squared 0.738121 Mean dependent var 9530.519
consequently, large countries find themselves in an				Adjusted R-squared 0.735844 S.D. dependent var
advantageous situation. Fillat-Castejón and				12827.13
Serrano-Sanz (2004) given the higher				S.E. of regression 6592.647 Sum squared resid
income-elasticity of industrial goods, the exports of				1.50E+10
large countries at any level of per capita income will				Durbin-Watson stat 1.885858
be systematically biased towards industry in				Unweighted Statistics including Random Effects
comparison with the average of small countries. ,				R-squared 0.751595 Mean dependent var 9530.519
Perkins and Syrquin (1989) observed that large				Adjusted R-squared 0.749435 S.D. dependent var
countries present exports which specialize in				12827.13
manufactures, whilst the exports of small countries				S.E. of regression 6420.810 Sum squared resid
are specialized in minerals.				1.42E+10
These studies along with an assessment of the				Durbin-Watson stat 1.890042
possible influence of size in specialization leads us to				References
propose the hypothesis that, with respect to each				? Balassa, B. and Bauwens, L. (1988), "The
level of per capita income, size causes the country of				determinants of intra-European trade in manufactured
reference to export standardized goods to and				goods", European Economic Review, 32(7), pp.
import differentiated goods from its small sized				1421–1438
trading partners, and to export differentiated goods				? Baldwin, Robert (1971), “Determinants of the
to and imports Standardized goods from its large				Commodity Structure of U.S. Trade,” American
sized trading partners. Hence, as the size of the				Economic Review 61
trading partner grows, exports are stimulated and				? Baltagi, Badi H. (2001), "Econometric Analysis of
imports inhibited in the differentiated product sectors				Panel Data", Second Edition, West Sussex, England:
of the reference county, while the opposite occurs				John Wiley and Sons
with standardized products. Thus, the greater the				? Bergstrand, J. H. (1989), “The Generalized Gre
difference in size of economy the greater the				vity Equation, Monopolistic Competition, and the
potential for trade. Specifically, product differentiation				Factor Proportions Theory in International
is not relevant.				Trade,” Review of Economics and Statistics,
1-2-2. The Trade-Braking Forces				February, 143-53.
Brakes on trade are considered as those factors that				? Bergstrand, J. H. (1990), "The
cause real trade to deviate from potential trade. The				Heckscher–Ohlin–Samuelson Model, the Linder
three factors explicitly recognized by Linder are the				Hypothesis and the Determinants of Bilateral
use of scarce factors in the demanded goods,				Intra-Industry Trade", The Economic Journal, 100, pp.
distance and human-made trade obstacles. The use				1216–1229.
of scarce factors is the main connecting point with				? Bharawaj, R. (1962), "Factor Proportions and the
the Heckscher–Ohlin theorem; Linder argued that				Structure of India-U.S. Trade" Indian Economic Journal,
the intensive use of a scarce factor in a variety				October pp.57-84
included within the overlapping of demand causes the				? Casas, François and E. Kwan Choi (1984), "Trade
efficiency to decrease and the trade to deviate and				Imbalance and the Leontief Paradox," Manchester
imposes deadweight loss. This loss arises from				School Vol. 52, pp. 391-401
implying the expensive domestic factor of production				? Casas, François and E. Kwan Choi (1985),"The
instead of cheap foreign factors of production. Thus,				Leontief Paradox: Continued or Resolved?" Journal of
the main source of difference between Linder’s				Political Economy Vol. 93, No. 3 Jun, 610-615.
and HO theory is the latter’s emphasis on the				? Deardorff, Alan V. (1998), “Does Growth
endowment of productive. In other words, this				Encourage Factor Price Equalization?”, The
theory is developed with a supply side approach				University of Michigan, Research Seminar In
whilst, in Linder’s demand side approach				International Economics, Discussion Paper No. 431,
overlapping of demand shapes the pattern of trade.				available at:
However, the use of scarce use of a scarce factor in				? Elahi, Nasser and Mohammad Nahavandian (2005),
a variety included within the overlapping of demand is				"The Impact of Globalization on the Per Capita
recognized as a trade breaking force; that is to say,				Income Convergence Across the Islamic Countries of
that what is considered as an opportunity in HO				the Middle East and North Africa (MENA),
theory its absence is thought to be an obstacle to				Nameh-ye-Mofid [Persian], 45(10), pp. 81-110
trade in Linder’s theory. As far as distance is				? Fillat-Castejón, Carmen and José Ma
concerned, firms cannot extend their trading horizons				Serrano-Sanz (2004), "Linder Revisited: Trade and
without costs, given that they have to face				Development in the Spanish Economy" International
transport and organization costs because of				Review of Applied Economics, July Vol. 18, No. 3,
diminishing rate of return. Tariffs and the other				323–348
obstacles imposed by people are also considered as				? Fortune, J. N. (1972), "Income distribution as a
the third trade-braking force.				determinant of imports of manufactured consumer
Subsequent developments of Linder’s work have				commodities", Canadian Journal of Economics, 5, pp.
similarly placed emphasis on the role played by				257–267
information flows. Vahlne and Wiedersheim-Paul				? Fortune, J. N. (1979), "Income distribution and
(1977) have attempted to reflect this with the				Linder’s thesis", Southern Economic Journal, 46,
concept of “psychological distance”, which				pp. 158–167
takes the form of differences in the level of				? Gray, H. P. (1988) "Intra-Industry Trade: An
development. The different levels of technical				“Untidy” Phenomenon", Weltwirtschaftliches
education are considered to be another trade				Archiv, 124, 211-29
breaking force. A final brake on the potential trade of				? Greene, William (2003), "Econometric Analysis", Fifth
a country could take the form of its economic				Edition. Prentice Hall
isolation, the result of a divergence in its growth path				? Greytak, D. and McHugh, R. (1977),
from that of its neighboring countries. A country that				“Linder’s Trade Thesis: An Empirical
finds itself isolated for this reason will have limited				Examination,” Southern Economic Journal,
trading horizons. Hufbauer (1970) was the first to				January, 1386-89.
declare the divergence in growth path among the				? Greytak, D. and Tuchinda, U. (1990), “The
trade-braking forces.				Composition of Consumption and Trade Intensities:
2. Empirical Tests				An Alternative Test of the Linder Hypothesis,”
Several empirical studies suggest that Linder’s				Weltwirtschaftliches-Archiv 126, , 50 - 57.
theory provides a good explanation of trade				? Hallak,J. C. (2003), "The Effect of Cross-Country
behaviors of countries and that the countries with				Differences in Product Quality on the Direction of
similar demand structure experience a higher trade				International Trade", February, University of Michigan
flows. In early years of the introduction of				? Hanink, D. M. (1988), “An Extended Linder Model
Linder’s theory Hufbauer (1970), Fortune (1979),				of International Trade,” Economic Geography,
Sailors et al (1973), Hirsch (1973), and Kohlhagen				October, 322-34.
(1977) provided reliable evidences that supported the				? Hanink, D. M. (1990), “Linder, Again,”
model. However, when the role of geographical				Weltwirtschaftliches-Archiv 126, , 257-67.
distance was recognized as another determinant of				? Hirsch, Z. and Lev, B., “Trade and Per Capita
trad patterns; Hoftyzer (1975), Greytak and McHugh				Income Differentials: A Test of the Burenstam-
(1977), Kennedy and McHugh (1983), Qureshi et al				Linder Hypothesis,” World Development,
(1980), showed that Linder's theory needs serious				September 1973, 11-17.
reforms. In the recent years a number of studies,				? Hoftyzer, J. (1975), “Empirical Verification of
with different approaches have tried to test				Linder’s Trade Thesis: Comment,” Southern
lender’s theory. Some studies have followed the				Economic Journal, April, 694-98.
gravity model. Bergstrand (1989, 1990), played a				? Hoftyzer, J. (1984), “A Further Analysis of the
significant role in Linder’s theory and gravity				Linder Trade Thesis,” Quarterly Review of
model and studies by Thursby and Thursby (1987),				Economics and Business, Summer, 57-70.
Hanink (1988), Greytak. and Tuchinda (1990),				? Houthaker, H.S. (1957), "Can Speculators forecast
McPherson et al. on Linder's theory are noteworthy.				prices?", Review of Economicsand Statistics, 39,
Schott (2004), Hummels, and Klenow (2002), Hallak				73-87
(2003), and Fillat-Castejón and Serrano-Sanz (2004)				? Hufbauer, G. C. (1970), "The impact of national
carefully tested the structure and the concept of				characteristics and technology on the commodity
similarity of demand patterns with more and found				composition of trade in manufactured goods", in R.
that it could appropriately explain a country’s				Vernon (Ed.) "The Technology Factor in International
trade behavior.				Trade", New York: National Bureau of Economic
2-1. Methodology of Research				Research
Considering that in this study the trade behaviors of				? Hummels, D. and Levinsohn, J. (1995), "Monopolistic
Muslim countries in MENA with the other Muslim and				competition and international trade: reconsidering the
Non-Muslim countries pre and post economic reforms				evidence", Quarterly Journal of Economics, 110(3), pp.
are examined, the panel Data econometric technique				799–836
was used in the framework of				? Hummels, D. and P. Klenow (2002), “The
Difference-In-Differences (DID) method. With The				Variety and Quality of a Nation’s Exports,”
DID technique, we may classify the results in two				available at:
-pre and post reform- periods. The Penal Data				? Hunter, L. C. and Markusen, J. R. (1988), "Per capita
econometric technique enjoys a number of				income as a determinant of trade", in R.C. Feenstra
advantages compared to the cross section or time				(Ed.) "Empirical Methods for International Trade",
series data . Some of these advantages are us				Cambridge, MA:MIT Press, pp. 89–109
under:i. It allows for a larger number of accessible				? Jones, R., and P. Kenen (eds.) (1984), "Handbook of
data as it uses both cross section and time series				International Economics", Vol. 1, North-Holland
data;ii. Unlike the method that uses one-dimensional				? Keesing, D. B. (1968), "Population and industrial
data, Penal Data enables the researcher to test				development: some evidence from trade patterns",
dynamic and behavioral hypotheses with a higher				American Economic Review, 58, pp. 448–455
level of certainty;iii. Penal Data provides a better				? Kennedy, T. E. and McHugh, R. (1983), “Taste
instrument to analysis the nature of disturbance,				Similarity and Trade Intensity: A Test of the Linder
unobserved and latent terms (Nerlove, 2002: 3-4)				Hypothesis for U.S. Exports,”
As data sets in panel data are greater than time				Weltwirtschaftliches-Archiv 119, 84-96.
series and cross sectional data sets and on the other				? Kohlhagen, S. W. (1977), “Income Distribution
hand explanatory variables usually vary –both				and ‘Representative Demand’ in International
with time and from one individual to another- panel				Trade Flows—An Empirical Test of Linder’s
data estimators are more efficient that other				Hypothesis,” Southern Economic Journal, July,
estimators (Verbeek, 2004: 343)				167-72.
The general form used in Penal Data models has the				? Leamer, E. E. and Levinsohn, J. (1995), "International
following expression:				trade theory: the evidence", in G. Grossman and K.
(1)where Yit is dependant variable i in time t, Xit				Rogoff (Ed.) "Handbook of International Economics",
represents a k-dimensional vector independent				Vol. III, Amsterdam: North-Holland
variable and ?it is assumed to be related to individuals				? Leontief, W. W. (1953), “Domestic Production
and time ? represents fixed effects ?i stands for				and Foreign Trade: The American Capital Position Re-
cross effects and ? Yt represents special time				Examined,” in J. Bhagwati, ed., "International
effects.				Trade: Selected Readings", Middlesex, England:
In Panel Data, coefficients are estimated in two				Penguin Books
different ways: fixed effects model and random				? Linder, S. B. (1961), "An Essay on Trade and
effects model. In the fixed effects model the				Transformation" New York: Wiley and Sons,
intercept of liner regression vary from individual to				? Markusen, J. R. (1986), "Explaining the volume of
another. Whilst in the second model intercept are the				trade: an eclectic approach", American Economic
same but there is random error for all individuals in				Review, 76(5), pp. 1002–1011
this model ?i + ?it acts as the error term, which				? Matsuyama, Kiminiori (1995), “Economic
consists of two parts: individual specific part and the				Development as Coordination Problems, ” 1123,
common part. It assumed that the ?i ? ?it is				Northwestern University, Center for Mathematical
independent of each other and of Xit. Although this				Studies in Economics and Management Science
suggests that OLS estimators are still unbiased and				? Matsuyama, Kiminori (1996), “Why Are There
consistent but due to the compound structure of				Rich and Poor Countries?: Symmetry-Breaking In The
error terms ?i + ?it these estimators will not be too				World Economy” , NBER Working Paper 5697
efficient if ?2 ? =0. On the contrary, GLS estimators				? McPherson, M., M. Redfearn, and M. Tieslau (2001),
are more efficient-despite being consistent and				“International Trade and Developing Countries: An
unbiased. Therefore, in the fixed effects Models GLS				Empirical Investigation of the Linder Hypothesis,”
estimators are used See: (Verbeek, 2004: 345-51).				Applied Economics, 33, 649-657.
2-1-1. The Specification of An Optimal model				? Nerlove, M. (2002), "Essays in Panel Data
A highly important decision in using Penal Data				Econometrics." New York, NY: Cambridge University
technique is selection of an optimal model. The				Press,
appropriate interpretation is that the fixed effects				? Perkins, D. H. and Syrquin, M. (1989) Large countries:
approach is conditional upon the values for ?i. That is,				the influence of size, in H. Chenery and T. N.Srinivasan
it essentially considers the distribution of Yit given ?i,				(Eds) "Handbook of Development Economics", Vol. 2,
where the ?i s can be estimated. This makes sense				(Amsterdam: Elsevier Science Publishing) pp.
intuitively if the individuals in the sample are ‘one				1691–1753
of a kind’, and cannot be viewed as a random				? Qureshi, U. A., French, G. L. and Sailors, J. W. (1980),
draw from some underlying population. This				“Linder’s Trade Thesis: A Further
interpretation is probably most appropriate when i				Examination,” Southern Economic Journal,
denotes countries, (large) companies or industries,				January, 933-36
and predictions we want to make are for a particular				? Sailors, J. W., Qureshi, U. A. and Cross, E. M. (1973),
country, company or industry. Inferences are thus				“Empirical Verification of Linder’s Trade
with respect to the effects that are in the sample.				Thesis,” Southern Economic Journal, October,
In contrast, the random effects approach is not				pp.262-68
conditional upon the individual ?i s, but ‘integrates				? Schott, P. (2004), “Across-Product versus
them out’. In this case, we are usually not				Within-Product Specialization in International
interested in the particular value of some				Trade,” The Quarterly Journal of Economics, Vol.
person’s ?i; we just focus on arbitrary individuals				119 pp. 646-677
that have certain characteristics. The random effects				? Slaughter, Matthew J. (2001), “Trade
approach allows one to make inference with respect				Liberalization and Per Capita Income Convergence: A
to the population characteristics. One way to				Difference-In-Differences Analysis” , Journal of
formalize this is noting that the random effects				International Economics, Volume 55, Issue 1, October,
model states that				203-228
(2)				? Stolper, Wolfgang F. and Karl Roskamp (1961),
While the fixed effects model estimates				"Input-Output Table for East ermany, with
(3)				Applications to Foreign Trade" Bulletin of the Oxford
The estimated coefficients are equal only when the				Institute of Statistics, November Vol. 23, pp. 379-392.
Holds. Accordingly, given the time period if there is a				? Tatemoto, M. and Ichimura, S. (1959) "Factor
small number of individuals and the identification of				proportions and foreign trade: the case of Japan",
each individual has a particular significance that fixed				Review of Economics and Statistics 41, 442–46
effect model is used otherwise random effects				? Thursby, J. G. and Thursby, M. C. (1987),
model also may be an appropriate approach.				“Bilateral Trade Flows, the Linder Hypothesis, and
However, even when random effects framework				Exchange Risk,” Review of Economics and
are appropriate because of relative frequency the				Statistics, August, 488-95.
fixed effects might be more appropriate; as there				? Trefler, Daniel (1993),"International Factor Price
may be a correlation between ?i ? Xit but this				Differences: Leontief Was Right!," Journal of Political
correlation is neglected in random effects and this				Economy, Vol. 101 pp. 961-87.
may lead to inconsistency of estimators. . (Verbeek,				? Vahlne, J.-E. and Wiedersheim-Paul, F. (1977),
2004: 351-2)				"Psychic distance—an inhibiting factor in
The best way to specify the optimal model is				international trade", Vol. 2, Working Paper,
Hausman test. The null Hypothesis in this test states				Department of Business Administration, University of
that there is no correlation between ?i and Xit. Hence				Uppsala.
the fixed effects (GLS estimators of ( )is not				? Verbeek, Marho (2004), "A Guide to Modern
efficient –although it is consistent. In the				Econometrics" , Washington: Wiley
alternative hypothesis (H1) correlation between ?i and				? Wahl, D. F.
Xit is accepted. As a result fixed effects estimators (

Global Trade

Globalization and Feasibility Study of Regionalism in the Middle East and North Africa