Download Gravity model and the countries classification

GRAVITY MODEL AND THE COUNTRIES CLASSIFICATION
Elżbieta Czarny, Jerzy Menkes, Katarzyna Śledziewska
SUMMARY
We discuss the gravity equation based on Helpman’s theorem (1987) using the
approach of Debaere [2005]. Unlike Debaere, we do not constrain our analysis to the
split into OECD and non-OECD, but we propose an alternative division. We argue
that the division of countries: OECD and non-OECD is not an appropriate
representation of extracting developed from less developed countries. Instead, we
propose non-OECD countries division into two groups: the first one consisting of
non-OECD countries fulfilling membership criteria and the second one not fulfilling
these criteria. In our setting, the less developed countries are collected only in the last
group, whereas OECD and non-OECD countries fulfilling membership criteria are the
developed ones.
We analyze trade between pairs of countries and aggregate the results to
illustrate trade between three above mentioned groups of countries in the period of 28
years (1977-2005). We show that the Helpman’s theorem designed to explain trade
among developed countries still does not find much support among less developed
ones unless the last group is properly extracted. As expected, in our test we got
confirmation of the Helpman’s theorem for the developed countries and only a very
week confirmation in the group of the less developed ones. However, surprisingly
intensive is trade between developed and developing partners. This might be a result
of growing intensity of trade in intermediate products in the last decades and
international trade liberalization (especially under WTO auspices after 1995).
Keywords: International trade, gravity models
JEL classification: F15
1. Introduction
During the last decade international cooperation has changed. The position of
the Asian countries in world trade improved rapidly. Many countries intensified their
trade with the world thanks to international trade liberalisation (especially under
WTO auspices). Below we examine whether the changes in the world economy are
compatible to the established theoretical settings. Especially, we check whether the
Helpman’s theorem [1987] is still valid only for the developed countries, as it was
originally meant, or it can be applied (at least to some extent) for the less developed
ones1. We argue, that some surprising results indicating growing intensity of trade
between less developed countries (for non-OECD countries see e.g. Hummels and
Levinsohn [1995]) are (at least partly) consequence of improper extracting of
developed and less developed countries. Especially, the most favored division of
countries into OECD and non-OECD is not a proper approximation of a division into
developed and less developed ones. We propose an alternative. We divide the nonOECD countries into two groups: the first one consists of non-OECD countries
fulfilling membership criteria and the second one of countries not fulfilling these
criteria. We argue that the less developed countries are collected only in the last
group, whereas the first one consists of the developed ones.
In our analysis we cover all countries in the world. We analyze trade between
pairs of countries and aggregate the results to illustrate trade among three groups:
OECD-members, non-OECD countries fulfilling membership criteria and other nonOECD countries.
We proceed as follows: in section 2 we write about rationale for our countries’
classification, in section 3 we present theoretical background of our analysis, in
section 4 we discuss the results of our empirical findings, and section 6 consists of
conclusions.
2. OECD in territorial division of the world
Territorial division of the world subjects to evolutionary changes.
Geographical extension and composition of territorial units is affected by many
factors, including social, cultural and economic ones. Fundamental importance got
those, which constitute a capacity to provide a state security through participation in
setting up the international security community.
1
We divide countries into developed and less developed ones not in line with the established split into
developed and developing countries originating in the World Bank. Even ignoring this precise
definition of World Bank we find our intuitive split best fitting with analytical purpose if this paper.
Political boundaries are artificial, as they do not constitute an integral part of
the physical world map. A territory may be perceived as a unit, or as a multiplicity of
interdependent territories indistinctly divided by natural elements.
From the point of view of economics the world coherency is still visible,
though geophysical criteria of a territory differentiation are not as important as before.
Historically, criteria of marking the second level in a ternary division into: I. states, II.
continents and III. the globe, have been determined by geographical closeness,
creating interdependency on one side, and natural obstacles preventing contacts, on
the other. Geographical discoveries and resulting from them transcontinental empires
tightened intra-imperial tights, even those trans-continental, making them stronger
than tights among one continent empires. In the 18-th century Spain of Charles II, the
first “Empire on which the sun never sets”, economical, social and cultural distance
from Spain to its American possessions was shorter than the remoteness of Catholic
Spain from Protestant England.
Simultaneously, starting with 17-th century, alongside with establishing by a
method of coordination, a modern international community, states regard international
law norms as one of the tools of implementing their national interests corresponding
to the adopted values. Making international law norms, states agree to restrict their
liberties considering the norms to be a protective umbrella and an instrument to
engage other states in cooperation towards non-antagonistic targets. An increasing
number and scope of norms is visible in the process of international law making. But
yet what has been a novelty of 20-th century and what characterises currently entered
agreements is the increasing complexity of regulated matters. Agreements not only
specify the trade terms but also provide space for common political, social and
economic interests and generally in constituting international organisations
institutionalising cooperation at the very moment of being established. Participation in
international organisations and agreements creates new form of closeness of states.
Numerous criteria allow to isolate as level II territories in the above mentioned
ternary division such communities as: EU/EC, OECD, NAFTA and NATO.
Undoubtedly, not only political, social, cultural and economic distances from Warsaw
to Minsk are bigger than those to Lisbon, Washington or Sydney. Distance in km
from Seoul to Phenian is misleading to anyone who knows the world not only from
the globe reading.
Within Organisation for Economic Co-operation and Development (OECD),
as in other above mentioned groups of states, cooperation is a function of
communication and mutual opening of autonomous sub-systems within a pluralistic
one (Deutsch [1964,58]). Below it is shown that the fundamental OECD principles
allow to divide the world states into those meeting criteria of the organisation
membership and those who do not fulfil them, considering this division as an
approximated split into more and less developed states (below we name them
developed and developing countries).
OECD2 is a transcontinental space, in geophysical sense, being at the same
time economically, legally and politically homogeneous. From its very beginning it
made a cornerstone of peaceful order construction and economic and social prosperity
of Western European countries, being the European component of the Western
Hemisphere. Not only open market economy, relatively high level of development
and OECD legal instruments implementation but also pluralistic democracy and
human rights respect are sine qua non conditions of its membership.
OECD analysed by its aspiration and ability to broaden its membership, is
characterised by weighty time caesurae. The first stage of the Organisation
construction was the period 1961-69, although it was formally established in 1961. Its
members were: Austria, Belgium, Canada, Denmark, France, Germany, Greece,
Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands, Norway, Portugal, Spain,
Sweden, Switzerland, Turkey, United Kingdom and the United States. This stage
extension till 1969 was caused by politically conditioned Italian (1962), Japanese
(1964) and Finnish (1969) accessions. The second stage falls on years 1971-73
marked by plumping the borders meaning a “small enlargement” with Australia (in
1971) and New Zealand (1973) who became joiners confirming their uninterrupted
presence in trans-Atlantic area and choosing to directly participate in the Organisation
further developments. The third stage falling on mid 90-ies of the 20-th century
2
OECD was transformed from OEEC (Organisation for European Economic Co-operation)
established in Paris on April 16, 1948. OEEC was a frame for the European Recovery Program (ERP)
transferring resources to facilitate recovery and reconstruction of European economies and their cooperation with the USA in efforts to revitalise the continent. OEEC has connected Austria, Belgium,
Denmark, France, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal,
Sweden, Switzerland, Turkey, United Kingdom, Trieste and German western occupation zones (the
Federal Republic of Germany joined in 1949). The organisation formula has directly linked the
mentioned states with the USA and Canada. OEEC has began the process of the “Atlantic bridge”
construction. In 1959 OEEC was joined by Spain, 1954 united Yugoslavia and in 1957 Finland got the
observer statute. All the previously mentioned countries, jointly with the USA and Canada, founded
OECD by the agreement from 14 December 1960. OECD is active since September 30, 1961.
consisted in the Organisation expansion and repelling the Soviet influence space.
Under the US political pressure a long prevented OECD opening towards new
members took place. Former member states of the Council for Mutual Economic Aid
(CMEA) joined the Organisation: the Czech (in 1995), Hungary and Poland (1996)
and Slovakia (in 2000). The accession, was intentionally designed by its American
originator as a reiteration of the way of both victorious and defeated states of Western
Hemisphere towards “security community” full membership. OECD accession has to
be, and in fact has been, the beginning of the “Royal Road” of the new members
towards European (EU) and transatlantic (NATO) institutions3.
Through the OECD threshold open for Central European states passed also
Mexico (in 1994) and South Korea (in 1996). Their long lasting will and readiness
had previously encountered a barrier of OECD institutional and functional inability to
adopt new members.
The OECD, who derives from and promotes vicinity, delimits with outer
boundaries the coherent space, distinguished by important institutional, legal and
economic factors, significantly different from the out of area space. Currently, OECD
members are thirty states joined by common acceptance for democratic rules and open
market economy.
Still, many countries formally meeting the membership criteria stay outside
the Organisation4. Even the EU members (meeting institutional, legal and political
terms underlying the OECD), not only new states (since January 2001) such as
Bulgaria and Romania, but also Cyprus, Estonia, Lithuania, Latvia, Malta and
Slovenia members since May 2004, have not acceded OECD. Additionally, the
Organisation maintains close relations with over 70 countries of all continents (25
states got the status of a permanent or a full participant of at least one of the OECD
Committees, while 50 states cooperate with OECD in many fields). All those states
are member candidates and for institutional and political reasons stay outside the
Organisation. Political readiness of some of the states faces lack of OECD reaction.
3
Still, public opinion and politicians in transforming states interest in participation in Western
Hemisphere institutions has focused on NATO and EU/EC accession, while other organisations and
institutions membership was considered merely as means towards the final aim of target institutions
membership and thus underestimated (that was the case of OECD as well as the Council of Europe
membership).
4
In May this year OECD decided to open for potential accession of Chile, Estonia, Israel, Slovenia and
Russia, as well as to intensify cooperation oriented towards future accession of Brasil, China, India,
Indonesia and South Africa.
Within the adopted presumptions, those non–OECD countries meeting the
formal accession criteria and not able to join because of the Organisation inability to
expand, should be in parallel to the OECD member states, treated as a separate group.
As this group is more like the OECD countries than the ones not sharing the OECD
fundamental principles and not meeting economic criteria underlying the accession, it
is a mistake to group them among the ones not sharing OECD principles and not to
the group consisting also in the OECD members.
3.
Theoretical Background
In the simplest form of the gravity equation, it is expected that bilateral trade
between two countries is directly proportional to the product of the countries’ GDPs5.
The further expectation is that bigger and more similar countries tend to trade more
intensively with each other than the smaller and differentiated ones. Such gravity
model with its more sophisticated versions has been for years “the workhorse for
empirical studies” in international economics (Eichengreen, Irwin [1997]).
In this setting, the trading countries produce differentiated goods for
international monopolistic competitive markets and trade freely with varieties of
many final products. Each country specializes in production and export of different
types of final products.
The demand is assumed to be identical and homothetic across analyzed
countries. Every good produced in any country is – by assumption – shipped to all
other countries in quantities proportionate to GDP of the purchasing country (it is in
fact similar to the standard macroeconomic idea of marginal propensity to import6).
We test the implications of the monopolistic competition model for the
aggregate international trade pattern basing on Helpman [1987]. We use multicountry
framework in which:
– i, j=1,….,C denote trading countries,
5
This model can be seen as a simplified form of a general equilibrium trade model explaining the
typical level of trade between countries. It can be supplemented with dummy variables introduced into
the gravity equation to account for deviations from the simple setting (and therefore from the typical
level of trade) that are difficult to be measured quantitatively (e.g. the impact of common language or
border, participation in a preferential trade agreements).
6
However, different marginal propensities to import express international differences in demand
characteristics and are therefore more realistic than assumption of identical demand structures across
analyzed countries. Further research of influence of differences in demand is presented in the new
works of Feenstra and Venables.
– k=1,…, N denotes products (we assume any variety of a good is a distinct product)
i
– y k denotes country i’s value of production of good k.
In our framework, GDP of country i is a sum of values of production of all
goods k (k = 1,…, N) manufactured in the analyzed country i (i = 1, …, C):
N
Y i   yki
(1)
k 1
Consequently, the global GDP (Yw) is the sum of GDPs of all trading countries
i:
C
Y W  Y i
(2)
i 1
Further, we denote share of country’s i GDP in global GDP as si:
si  Y
i
(3)
Yw
If trade is balanced in every country i, then si denoting share of country’s i
GDP in the global GDP influences intensity of bilateral trade between i and j. As we
measure each country’s import as a fraction of its GDP, then with balanced trade it is
equal to its partner’s export7. Therefore, export of product k from country i to country
j equals to:
X kij  s j yki
(4)
Xkij in equation (4) depends on sj because export from i to j is equal to j’s import from
i, and the latter one depends on GDP of country j.
Summing up values of export of all products k manufactured in country i we
get i’s total export to j. By assumption of balanced trade it is equal to i’s import from j
and therefore to j’s export to i. We write this equation using equations (1) – (4) as:
Y jY i
X   X  s  y  s Y  w  s j s i Y w  X ji .
Y
k 1
k 1
N
ij
N
ij
k
j
i
k
j
i
(5)
7
We analyze export of all countries, avoiding therefore disturbances resulting from the non-zero costs of
international trade (e.g. tariffs and transport and insurance costs).
Summing up the first and the last elements of the equation (5) we get the total
trade of the analyzed pair of countries, defined as a sum of their exports (equal to the
partner’s import):
X ij  X
ji
 2 
  w Y iY j
Y 
(6)
Formula (6) is the simplest derivation of the gravity equation. It shows that the
intensity of bilateral trade depends on product of GDPs of both trading countries.
The re-formulation of equation (6) with the use of the formula (5) confirms
that size of the country and the difference between the trading partners matter:
X ij  X ji  2s i s jY w
(7)
A variable representing size of a country is a share of its GDP in the world’s GDP.
Different values of these shares by trading partners express the differences in their
sizes. According to the gravity approach, two countries of unequal sizes will not trade
as intensively as two countries of similar sizes would do8.
The extension of the gravity model we use, goes into evaluation of a role of
the member countries in an economic region and evaluation of economic position of a
region in the global economy. In the simplest setting we have two countries (i, j)
constituting region A. GDP of A is therefore equal to:
Y A  Yi Y j
(8)
We define the relative shares of GDPs of every country in the regional GDP and the
GDP of A relative to the global GDP as, respectively:
s iA 
8
Yi
YA
sA 
YA
Yw
E.g. if sizes of both countries are equal si = 0,1 and sj = 0,3, then their sizes influence bilateral trade
with coefficient 0,064. If they were equal in size and would have exactly the same joint share in the
world’s GDP as in the previous example si = sj = 0,2, then the respective coefficient would be bigger
than before (0,08). If two trading partners would have shares s i = 0,1 and sj = 0,01, their sizes would
influence bilateral trade with coefficient 0,024. In the last example we can see that smaller countries
trade less intensively (though if they were equal si = sj = 0,055 their trade would be more intensive –
coefficient 0,0605).
Volume of A’s trade relative to A’s GDP depends on the position of every
member country in the region and on the relative importance of the region’s GDP in
the world’s one:
( X ij  X ji ) / Y A  2s iAs jAs A
(9)
where:
s iA  s jA  1
because region A consists of just two countries.
Reformulating the right hand side of the equation (9) we get:
2s iAs jA  1  (s iA ) 2  (s jA ) 2
(10)
Equation (10) presents a simple version of Helpman’s theorem formulated under the
following assumptions: both trading countries are fully specialized in their outputs,
tastes in both countries are identical and homothetic, trade is free worldwide. The
volume of trade relative to GDP is proportional to the dispersion index defined for the
region A as:
disp  1   ( s iA ) 2
iA
(11)
The relation presented in equation (11) is visualised in equation (12), which is a
further reformulation of equations (9) and (10). In equation (12) volume of export
from region A relative to A’s GDP depends on the share of A’s GDP in global GDP
(sA) and on the size dispersion index of A’s members:
XA
 s A (1   ( s iA ) 2 )
A
Y
iA
(12)
where N – number of countries in the region A (in our setting N = 2).
The size dispersion index in brackets in equation (12) is maximized for
countries of the same relative size (for a region consisting of two identical countries
this index is equal to 0,5; in case of big dispersion the index is near to 0 - e.g. if one
country’s share is 0, 99 and the other one’s is 0,01, the index is less than 0,02).
Equation (12) shows that volume of trade in the region is also related to the
relative size of countries constituting the analyzed region. It is expected that with
increasing similarity of trading partners their bilateral trade will intensify.
Helpman [1987] tested and confirmed his theorem for OECD countries. His
method was used by different authors. Hummels and Levinsohn [1995] first chose for
their test not only OECD, but also non-OECD countries (the last group was meant as
representation of developing countries for which one can expect low intra-industry
trade (IIT)). They tested the following equation:

ln X t  X t
ij
ji
 
ij


  ln (1  dispersion tij )(Yt  Yt )   tij
i
j
(13)
They confirmed Helpman’s findings for OECD countries. However surprisingly, they
found some support for Helpman’s theorem that was designed to explain trade among
developed countries, also for trade among non-OECD countries representing the
developing ones.
Their result was revisited by Debaere [2005]. Building on Helpman’s idea,
Debaere transformed Helpman’s equation (6) using the linear form of the equation
(12):
A  i, j
 X ij  X ji
ln  i
j
 Y Y
  Yi

  ln( s i  s j )  ln 1   i
  Y  Y

2
j
  Yj
   i
 Y  Y
j



2



(14)
Debaere analyzed the surprising results for non-OECD countries obtained by
Hummels and Levinsohn. He argued that by using the equation (13) with Y on the
right hand side Hummels and Levinsohn allowed size of a country to have an impact
on the estimation.
By use of different econometric techniques and different measures of GDP
Debaere found that for OECD countries the coefficient of similarity index is
significant and positive. As in the work of Helpman [1987] and Hummel and
Levinsohn [1995] this coefficient was statistically different from its theoretical value
of 1. In Debaere [2005] it varied from 0,25 to 1,57 depending on the econometric
technique. In contradiction to Hummel and Levinsohn, Debaere proved that the index
of similarity does not play a significant role in trade of non-OECD countries.
We extend the work of Debaere testing his equation for differently constructed
groups of countries. Building on our reflections about OECD and non-OECD
countries in previous section, we divide non-OECD countries into two groups. The
first one consists of countries fulfilling OECD membership criteria. In the second one
are gathered all non-OECD countries which not fulfil these criteria.
For extraction of the last group we consider three main criteria of OECD
membership: pluralistic democracy and human rights respect, open market economy
and high level of country’s development. The first one has non-economic nature,
whereas the next ones economic.
The first condition is democratic system and respecting human rights in a
country. We treat membership in transatlantic institutions as sufficient condition of
fulfilling this criterion. Membership in Community of Democracies (2000) is also the
sufficient condition. However, because we want to cover with our analysis a relatively
long period, we can’t constrain ourselves to the above mentioned sources. We
supplemented them with the data published by the Freedom House. We have chosen
this source of data because of credibility of this institution and availability of required
data for relatively long period. The Freedom House presents a division of the world
into free, partly free and not-free countries. We treat as democratic countries “free”
and “partly free” in the classification provided by the Freedom House.
The second criterion is membership in WTO (and its GATT predecessor).
This is a proxy for a open market economy in an analyzed country.
The third criterion is a level of development approximated by GDP per capita
not lower than respective GDP of the poorest OECD – member in the year of
reference. In this context we consider GDP per capita in PPP as best fitting with an
idea of measurement of real development.
We analyze trade between pairs of countries and aggregated the results to
illustrate trade among three above named groups of countries (OECD, non-OECD
fulfilling membership criteria and others non-OECD countries). We made tests with
current GDP.
3.
Empirical research
We provide an econometric test of Helpman’s prediction for pairs of countries
extending Debaere’s analysis on the third group of countries (non-OECD country
fulfilling membership criteria). We test the gravity equation for available pairs of
countries from all three groups as dependent variable.
In our estimations we base on equation:
 X t ij  X t ji 
   ij   ln( st i  st j )   ln 1  dispersion tij
ln  i
j 
 Yt  Yt 


(15)
which is a reformulated version of Debaere’s equation (14). In formula (15) αij
represents fixed effects and can account for bilateral transportation costs, language
barriers, cultural barriers, tariffs or distances. Similarity index lnsim =


(  ln 1  dispersion tij ) represents the impact of dispersion index on bilateral trade. It
shows how the volume of trade is related to the relative size of countries (we assume
that there are C countries in the analysed region).
The size index lns = (  ln( st  st ) ) represents the impact of countries’ shares
i
j
on the intensity of trade. This index varies over time. We estimate the impact of
changing GDP shares on trade because 28 years covered by our analysis is
sufficiently long period of time for these changes to be significant.
The literature does not deliver any standard measure of GDP used to test
gravity model. The authors use different measures of GDP, only rarely explaining
why they choose just them. We use only current GDP in US$. We argue that because
exports are reported in current prices, the most appropriate measure for the share of
trade in GDP is GDP in current prices. We use this measure also for the sake of
simplicity. The other researches also use GDP in PPP in current international $ but in
our opinion this measure is improper. It is particularly unfavourable for developing
countries whose GDP in PPP is bigger than GDP in current US dollars.
We do the research using the data from Comtrade database for trade and WDI
dataset for GDP. Because of poor data from earlier years we analyze trade flows in
the period 1977 - 2005. We use the panel data techniques of estimation using STATA.
In all cases a fixed effects model specification is preferred to a random effects model
based on the Hausman test.
We estimated an equation similar to the equation (14):
A  i, j
  Yi
 X ij  X ji 
i
j

1   i
ln  i


ln(
s

s
)


ln
j 
j
  Y  Y
 Y Y 
2
  Yj
   i
j
 Y Y



2

   ij

(16)
We analyze trade between pairs of countries and aggregate the results to illustrate
trade among all groups of countries.
We took into account three OECD membership criteria discussed above:
democracy, WTO (earlier GATT) membership, GDP per capita measured in PPP at
least as high as in the poorest OECD member (in last years it was almost always
Mexico). We run regression for the OECD and countries fulfilling their membership
criteria with democratic classified as “free” and “partly free” countries jointly
constituting the group of developed countries.
In table 1 we present the results for bilateral trade of developed and less
developed countries. The results are organized as follows. In column (1) we present
aggregated results for bilateral trade among developed countries. In column (2) we
present trade among developed and less developed countries. In column (3) we show
trade among countries not fulfilling OECD membership criteria. In the last column we
analyse trade among less developed countries.
Table 1. Estimation results
(1)
(2)
0.336
0.356
(11.83)** (15.73)**
Lnsim
1.010
0.872
(46.56)** (57.72)**
Constant
9.317
8.507
(65.67)** (68.02)**
Observations
22929
55470
Number of group
1843
5684
R-squared
0.35
0.32
Absolute value of t statistics in parentheses
 ** significant at 1%; * significant at 5%
Lns
(3)
-0.173
(4.81)**
0.081
(2.43)*
3.890
(15.74)**
23536
3478
0.02
Table 1 evidences that we got full support for Helpman’s theorem. It shows
confirmation in the fact that β coefficient from equation 16 (Lnsim) is near to 1 for
developed countries defined as OECD – members and countries fulfilling the three
main membership criteria (with democratic defined countries “free” and “partly free”
in the setting of the Freedom House). This way to extract developed countries seems
to fit perfect with the theory of international trade.
We got surprisingly high β coefficient in column (2) containing results for
trade between developed and less developed countries, though significantly lower
than we got in trade of developed partners. It can be explained with intensification of
production fragmentation and consequently growing trade in intermediate goods.
International trade liberalisation (especially under WTO/GATT auspices) results also
in intensification of North – South type of international trade.
Results in columns (3) also support the Helpman’s theorem as we got very low
values of β for trade among less developed countries.
As far as the size of a country is concerned, we got results compatible both
with the theoretical predictions and the common sense. For the developed countries
and the countries with different level of development, the trading country size is
positively correlated with the intensity of bilateral trade (see Lns in columns (1) and
(2)). Opposite we note in the case of the less developed countries. Their size is
negatively correlated with the volume of bilateral trade (negative Lns in column (3)).
It means that in the case of two less developed countries their size (measured with
current GDP of respective countries) negatively affects potential trade.
We obtained relatively big constant effects (αij). It can be the evidence of
influence of price changes on intensity of trade.
We additionally proved our thesis about division of the world into OECD and
non-OECD countries as improper proxy for extracting developed and less developed
countries testing the equation (16) for OECD and non-OECD countries. Results are
reported in the table 2.
Table 2. Results of estimation for OECD and non-OECD countries
(1)
Lns
Lisim
Constant
Observations
Number of group(k1 k2)
R-squared
0.154
(4.33)**
0.955
(28.57)**
9.368
(61.91)**
8268
434
0.25
(2)
0.485
(21.31)**
0.963
(77.14)**
9.501
(85.51)**
51215
3825
0.45
(3)
-0.013
-0.48
0.383
(15.51)**
5.29
(29.38)**
42452
4915
0.12
Absolute value of t statistics in parentheses
** significant at 1%; * significant at 5%
In table 2 we present results for trade among OECD members (column (1)),
trade of one OECD member with a non-OECD partner (column (2)) and trade among
non-OECD partners (column (3)). In all cases we got results less fitting the theory.
Especially, β coefficient for developed countries (Lnsim in column (1)) is lower than
for developed and less developed countries (Lnsim in the column (2)). In this case we
got confirmation of Helpman’s theorem not only for developed countries and trade
between them but also (even more!) for unevenly developed countries. At the same
time both coefficients are lower than respective coefficients in Table 1. The lowest β
coefficient – as expected – we got for trade between less developed countries
(however, it is significantly higher than in the Table 1).
These results confirm our opinion that OECD is not an appropriate proxy for
developed countries because it leaves aside many countries fulfilling membership
criteria and not accepted by the OECD because of political and organisational
reasons. The traditional division of countries makes both groups of countries more
similar than the developed and less developed countries are in reality. This results in
false confirmation of the Helpman’s theorem also for the less developed countries.
Table 3. Tests of the gravity equation with value of (Trade/GDP) for pairs of countries as
dependent variable (results reported for countries grouped depending of their level of
development)
Developed
economies &
Developed
economies
lns (GDP - current USD)
0.829
(25.83)**
lnsim (GDP - current USD)
Constant
Observations
Number of groups
R-squared
Absolute value of t statistics in parentheses
significant at 5%; ** significant at 1%.
Source: own calculation.
1.639
(61.36)**
-0.544
(3.50)**
12566
1124
0.42
Developing
economies &
Developed
economies
Developing
economies
&
Developing
economies
0.414
0.087
(11.23)**
(2.58)**
0.934
0.721
(34.17)**
(22.65)**
-4.939
-7.291
(23.33)**
(30.47)**
27424
41235
3688
7670
0.41
0.15
Finally, in Table 3 we analyse bilateral trade between developing and
developed countries (we use the respective World Bank classification). The results
reported in Table 3 give confirmation of Helpman’s theorem not only for developed
countries and trade between them. Interesting and compatible to the theoretical
predictions is the β coefficient much bigger for developed countries then for other
trade partners. Also the lowest β coefficient for trade between developing countries is
important. That means that Helpman’s theorem is more meaningful for developed
then for developing countries. However, also in this analysis we got results less
compatible with the theoretical prediction than in our alternative test (Table 1).
4.
Conclusions
We got full confirmation of Helpman’s theorem for developed countries.
However, in almost all analyzed cases we can confirm increasing similarity of trading
partners (positive lnsim). Therefore we can expect trade expansion. This result is not
surprising for trade between developed countries and less developed ones. Of course,
our research proves that the developed countries become much more similar than
other groups of partners and therefore they are expected to trade more intensively than
any other pairs of countries do.
In our opinion, increasing similarity even of very different countries can be the
result a. o. of intensified cooperation in production between countries. It may change
the characteristics and reasons for trade with growing intensity of trade in semi
products. As for the theory, it can result in necessity of substituting the gravity
approach with a new theory or supplemented gravity model with additional elements.
Our division of countries proxing developed and less developed ones turned out to be
better than the most used division into OECD and non-OECD countries.
REFERENCE
Debaere [2005], Monopolistic Competition and Trade Revisited: Testing the Model without Testing for
Gravity, Journal of International Economics, Vol. 66, pp. 249-266
Deutsch K. [1964] Communication Theory and Political Integration, w: The Integration of Political
Communities.
Egger P. (2002) An Econometric View on Estimation of Gravity Models and the Calculation of Trade
Potential, The World Economy, Vol. 25, Nr 2, pp. 297-312.
Eichengreen B., Irwin D. [1997], The role of history in bilateral trade flows, in: Frankel J. (ed.) [1997,
pp. 33˛-57],
Frankel J. (ed.) [1997], The Regionalization of the World Economy, University of Chicago Press,
Helpman E. [1987], Imperfect Competition and International Trade: Evidence from Fourteen Industrial
Countries, Journal of the Japanese and International Economies, vol. 1, pp. 62-81
Hummels D., Levinsohn J. (1995), Monopolistic Competition and International Trade: Reconsidering
the Evidence, Quarterly Journal of Economics, vol. 110, nr 3, pp. 799-836.
The Integration of Political Communities [1964], P. E. Jacob, J. V. Toscano (eda.), Philadelphia.