Determinants of East African Community Trade: A Gravity Model Analysis of Trade Flows between...
Transcript of Determinants of East African Community Trade: A Gravity Model Analysis of Trade Flows between...
1
Determinants of East African Community Trade: A Gravity Model Analysis of Trade
Flows between Tanzania and Kenya
Mjema, G.*, Mahona, B.* and Bagumhe, P. **
*Institute of Finance Management, Tanzania
** Ministry of East African Cooperation, Tanzania
Paper to be presented at the 5th
International Conference organized by the Institute of Finance
Management, 23rd
-25th
August 2012
2
1 Introduction
Economists have, for a long time been preoccupied with analyzing the determinants of trade
flows between nations. During the mercantilist era for example it was argued that since gold
and other precious metals were generally accepted as global currencies, a country which was
able to accumulate large reserves of such metals was able to pay for imports from other
countries using such a reserve. Thus, trade surplus countries were synonymous to countries
with large reserves of the precious metals. The reverse was also true, that is trade deficit
countries were countries which were deficient on such metals.
It was Adam Smith and later David Ricardo who laid the foundations for analyzing the
determinants of trade flows between nations using the absolute and comparative advantage
concepts (see for example Arthur and Steven, 2003). In a nutshell, while absolute advantage
refers to a country’s ability to produce certain commodities more effectively than another
country, comparative advantage refers to a country’s ability to produce a particular
commodity with a lower opportunity cost than another.
In subsequent models that followed including the Heckscher-Ohlin approach, trade flows
between nations are said to be determined by the respective country’s state of either capital or
labour abundance. In this setting, with the exception of where the ―Leontief paradox‖ holds,
a capital abundant country will tend to produce and export capita intensive goods. The same
country will import labour intensive goods from the rest of the world.
Several variants of the Heckscher- Ohlin model have been suggested including the Linder
hypothesis which has suggested that countries with similar demands tend to develop similar
industries and would then trade with each other in similar but differentiated goods (see
Soderstein, 1980).
Other researchers including Tinbergen (1962) have attributed the flow of trade between
nations to the gravity concept borrowed from natural sciences. According to this model
factors like the sizes of countries, distance as well as other cultural factors are important
determinants of trade flows between them. This paper uses a modified version of the gravity
model of trade to analyze the determinants of trade between Kenya and Tanzania which are
members of the East African Cooperation.
Trade between EAC member states is arguably as old as the history of these states. However,
trade between Kenya and Tanzania which is the focus of this paper is of particular interest for
3
various reasons. For example, the volume of trade between the two states constitutes over
50% of the entire EAC trade. Studies including those by Ackello-Ogutu and Echessah (1989)
have revealed flourishing yet unrecorded cross border trade activities between Kenya and
Tanzania than in any other borders in the East African region. Furthermore, there is currently
more pronounced cross border investments between Kenya and Tanzania than in any other
EAC member states (EAC, 2010). Indeed, data from the Tanzania Investment Centre show
that investments from Kenya to Tanzania have increased from US $ 37.5m in 2005 to US
$ 67.25 m in 2010(an increase of over 50%). Similarly, Tanzania’s investments to Kenya
increased from US $ 0.31m in 2005 to US $ 2.54m in 2010 (an increase of over 500%).
There are indications that the on-going formalization of free movement of capital within the
auspices of the EAC Treaty trade will usher in a surge of investments between EAC member
states and more specifically between Kenya and Tanzania.
Currently EAC has five member states namely; Tanzania, Kenya, Uganda, Rwanda and
Burundi. There are potentials for other countries within the eastern and central Africa to join
this economic cooperation between nations. Historically there have been two phases in the
creation of EAC. The first phase stretches from 1967 to 1977. This period saw the signing of
the EACTreaty of cooperation between Kenya, Uganda and Tanzania in 1967 but also
witnessed its collapse in 1977 amid political and economic frictions between the three
countries.The objectives of the first phase of EAC were, inter alia, to achieve a balanced
economic growth within the EAC member states through the establishment of a common
market, common customs tariff, and common public services (see for example, Kirkpatrick
and Watanabe, 2005 and Ng’ang’a, 2006). With the demise of the first phase of the EAC in
1977, these objectives were hardly attained.
The collapse of the first phase of EAC had profound negative impacts on the economies of its
member states. Besides loosing the economies of scale advantage the region had prior to the
collapse of the scheme, each country had to re-start a costly programme of establishing
services and industries which were run somewhat efficiently at the EAC level.
Thus, in 1999 after a series of high level inter-state consultations a Treaty for establishing the
second phase of EAC was signedby the respective Heads of State of the member countries in
Arusha, Tanzania. One of the significant developments of EAC during the second phase has
been the inclusion of two new members; Rwanda and Burundi thus bringing the current
membership of EAC to five.Apart from the expansion of the EAC membership the other
4
milestones so far registered in EAC include (i) the establishment of the customs union (ii) the
establishment of the common market (iii) convertibility of Kenya, Uganda and Tanzanian
currencies (iv) harmonization of goods produced in the region (v) reduction of tariff and non-
tariff barriers and (vi) free movement of stock (EAC, 2010).
In terms of size, EAC has a total area of 1,817,945 km2 and an estimated population of 131.9
million people. The total Gross Domestic Product (GDP) of the region was estimated (2007)
to be 61US $ bn. which translates to an average per capita income of over US $ 450 (EAC,
2010). As stated earlier, interest in this paper is to analyse factors determining the flow of
trade between Kenya and Tanzania. As the following sub-section reveals, the need to get a
better understanding of the determinants of trade flows between Kenya and Tanzania is
beneficial not only to the current trade relations in EAC but also to its future development.
2 Trade Flows between Tanzania and Kenya
Although the EAC sub-region has witnessed a surge in trade flows among member states
particularly after the establishment in 1st July, 2010 of the common market. The EAC
common market has provided for ―four freedoms‖ namely; free movement of goods, labour,
services and capital all of which are expected to boost trade and investment within the region.
Interest in analyzing the trade flows between Tanzania and Kenya is based on economic as
well as non economic factors. The essence of focusing on trade flows between Tanzania and
Kenya is that currently the two countries dominate the economy of the sub-region, in many
aspects. For example, Kenya and Tanzania share the largest border within EAC countries, an
important factor in facilitating cross border trade. While all the other EAC countries are
landlocked, Kenya and Tanzania have access to the Indian Ocean through their Mombasa and
Dar es Salaam ports respectively. This access has important implications on not only distance
but also in transport related trade aspects. The two countries have a large population of
people who can speak two languages (Kiswahili and English) and, to a certain extent, share a
common culture which is also important for trade facilitation.
Kenya and Tanzania can loosely be referred to as the economic ―power houses‖ of EAC. In
2010 for example, the combined GDP of the sub-region was estimated to be US $ 79.3 bn.
Kenya’s GDP during the same year was US $ 32.2 (equivalent to 40.6% of the total EAC
GDP). Tanzania’s GDP was US $ 22.9 bn. (equivalent to 28.9%) during the same year. This
5
implies that the combined GDP of Kenya and Tanzania was equivalent to 69.5% of the total
EAC GDP.
The combined share of GDP of the three other EAC member states (Uganda, Rwanda and
Burundi) wasonly 30.5% of the total EAC GDP. According to data from the World Bank
(290) the two countries dominate the region where GDP per capita is concern. In 2010
Kenya had the highest GDP per capita (US $ 468.7) in the EAC region compared to
Tanzania’s per capita GDP of US $ 458.4. The per capita GDP of Rwanda, Uganda and
Burundi was lower (averaged just over US $ 380) than that recorded in Kenya and
Tanzania.Trade as a percentage of GDP was highest in Kenya (65.4%) but was closely
followed up by Tanzania whose ratio in the same year was 63.8%. The trade –GDP ratio in
Rwanda, Uganda and Burundi was lower (under 50%) compared to what was recorded in
Kenya and Tanzania. Recent trade trends between Tanzania and Kenya (Table 1) suggest that
apart from their relative large GDPs they (Kenya and Tanzania) dominate the region’s trade
activities.
Table 1: Trade Flows in EAC: 2009 – 2011 (US $ m)
Year Tanzania’s Exports to
Kenya (1) (US $m)
Tanzania’s Exports to Rest of
World (2) (US $m)
(1) as % of
(2)
2009 192,904 2,982,405 6.5
2010 324,888 4,050,546 8.0
2011 221,313 4,734,960 4.7
Avg.2009-2011 246,368 3,922,637 6.2
Year Tanzania’s Exports to
Uganda(1) (US $m)
Tanzania’s Exports to Rest of
World (2) (US $m)
(1) as % of
(2)
2009 51,651 2,982,405 1.7
2010 60,205 4,050,546 1.5
2011 52,634 4,734,960 1.1
Avg.2009-2011 54,830 3,922,637 1.4
Year Tanzania’s Exports to
Rwanda (1) (US $m)
Tanzania’s Exports to Rest of
World (2) (US $m)
(1) as % of
(2)
2009 15,805 2,982,405 0.5
2010 116,802 4,050,546 2.8
2011 95,160 4,734,960 2.0
Avg.2009-2011 75,922 3,922,637 1.7
Year Tanzania’s Exports to
Burundi (1) (US $m)
Tanzania’s Exports to Rest of
World (2) (US $m)
(1)as % of
(2)
2009 24,632 2,982,405 0.8
2010 56,132 4,050,546 1.3
2011 39,848 4,734,960 0.8
Avg.2009-2011 40,204 3,922,637 0.9
Source(s): World Bank 2010, EAC 2011
6
In 2009 Tanzania’s exports to the rest of the World (ROW) was US $ 2,982,405 m compared
to the country’s exports to Kenya which were valued at US $ 192,904 m (equivalent to 6.5%).
This ratio increased to 8.0% in 2010 before leveling off to 4.7% in 2011. Tanzania’s exports
to Kenya during the 2009 – 2011 period averaged US $ 246,368 m compared to the value of
Tanzania’s exports to ROW which totaled US $ 3,922,637 m. The central message emerging
from Table 1 is that Kenya is an important destination for Tanzania’s exports than all the
other EAC countries combined. The economic trends (Table 2) indicate that the trade
scenario in EAC is not likely to change in the near future and hence Tanzania and Kenya will
continue to dominate trade activities in the sub region.
Table 2: Economic Growth in EAC Region 1990 - 2010
Region 1990 –
1999
2006 2007 2008 2009 2010
Burundi -1.3 5.2 3.6 4.3 3.5 3.9
Kenya 2.1 6.3 7.0 1.5 2.6 5.6
Rwanda -0.1 8.6 7.7 11.5 6.1 7.5
Tanzania 3.1 6.7 7.1 7.4 6 7.0
Uganda 6.3 7.0 8.1 10.4 3.9 5.6
Source: EAC (2011)1
In the following sub-section the paper presents the theoretical underpinnings of the gravity
model used to analyze the determinants of trade flows between Kenya and Tanzania.
Although different authors have included different variables in the gravity model, the interest
in the current paper is on four variables notably GDP, GDP per capita, distance and trade
openness.
1 From the EAC 4
thDevelopment Strategy document published on August,2011
7
3 Modeling the Determinants of Trade Flows between Kenya and Tanzania: The
Gravity Model Approach
The gravity model used in the current paper to analyze the determinants of trade flows
between countries draws its theoretical foundations from gravity models used in the field of
natural sciences. The model predicts bilateral trade flows based on the economic sizes of
countries (using mainly GDP ) and distance between them. The model was first used in the
social sciences field by Tinbergen (1962)and p yh nen (1963) who postulated that trade
between two countries (i and j) takes the form of:
Where is the trade flow, are the economic masses of each country, is the distance
between them and G is a constant. Besides being deployed to analyze trade flows between
countries the model has also been used in international relations including evaluating the
impact of treaties and alliances on trade.
Since then (1963), the gravitymodel has become a popular instrument in empirical foreign
trade analysis such that different scholar has successfully appliedthe model to show flows of
varying types suchmigration, foreign direct investment and more specifically to
internationaltrade flows. According to this model, exports from country i to country j
areexplained by their economic sizes (GDP or GNP), their populations, directgeographical
distances and a set of dummies incorporating some kind ofinstitutional characteristics
common to specific flows.
Starting from the second half of 1970s several theoretical developments haveappeared in
support of the gravity model showing the theoretical support of the research in this fieldwhich
was originally very poor. Anderson (1979) made the first formalattempt to derive the gravity
equation from a model that assumed productdifferentiation. Bergstrand (1985, 1989) also
explored the theoreticaldetermination of bilateral trade in a series of papers in which gravity
equationswere associated with simple monopolistic competition models. Helpman and
Krugman (1985) used a differentiated product framework with increasingreturns to scale to
justify the gravity model. More recently Deardorff (1995) has proven that the gravity
equation characterizes many models and can bejustified from standard trade theories. Finally,
8
Anderson and Wincoop (2001) derived an operational gravity model based on the
manipulation of the CESexpenditure system that can be easily estimated and helps to solve
the so-calledborder puzzle. The differences in these theories help to explain the
variousspecifications and some diversity in the results of the empirical applications.
There are a huge number of empirical applications in the literature ofinternational trade,
which have contributed to the improvement of performanceof the gravity equation. Some of
them are closer related to our work. First, papers of Mátyás (1997) and (1998), Chen and
Wall (1999), Breuss andEgger (1999) and Egger (2000) improved the econometric
specification ofthe gravity equation. Second, Berstrand (1985), Helpman (1987), Wei,
(1996),Soloaga and Winters (1999), Limao and Venables (1999), and Bougheas etal, (1999)
among others, contributed to the refinement of the explanatoryvariables considered in the
analysis and to the addition of new variables.In other occasions the model has been used to
test the effectiveness of trade agreements and organizations such as the North American Free
Trade Agreement (NAFTA) and the World Trade Organization (see for example Helman,
2003, Panagariya, 1999, and Krugman 2001 among others).
While the model’s basic form consists of factors that have more to do with geography and
spatiality, the gravity model has been used to test hypotheses rooted in purer economic
theories of trade as well. One such theory predicts that trade will be based on relative factor
abundances. One of the common relative factor abundance models is the Heckscher-Ohlin
model. This theory predicts that trade patterns would be based on relative factor abundance.
According to this theory, countries with a relative abundance of one factor would be expected
to produce goods that require a relatively large amount of that factor in their production.
While a generally accepted theory of trade, comparative advantage has suffered empirical
problems. Investigations into real world trading patterns have produced a number of results
that do not match the expectations of comparative advantage theories. Notably, a study by
Wassily Leontief found that the United States, the most capital endowed country in the world,
actually exports more labor intensive commodities. Comparative advantage in factor
endowments would suggest the opposite would occur. Other theories of trade and
explanations for this relationship were proposed in order to explain the discrepancy between
Leontief’s empirical findings and economic theory. The problem has become known as the
―Leontief paradox‖ (see for example (Soderstein, 1980).
9
The gravity model has often been specified in econometric form as:
, ……………………..…………………………. (2)
Where represents volume of trade from country to country , and typically
represent the GDPs for countries and , denotes the distance between the two countries,
and represents an error term. The traditional approach to estimating this equation consists in
taking logs of both sides, leading to a log-log model of the form:
. ……… (3)
However, this approach has two major problems. First, it obviously cannot be used when
there are observations for which is equal to zero. Second, it has been argued by Santos
Silva and Tenreyro (2006) that estimating the log-linearized equation by least squares (OLS)
can lead to significant biases. As an alternative, these authors have suggested that the model
should be estimated in its multiplicative form, i.e.
, ………. (4)
One of the authors' more surprising findings was that, when controlling for factors like
sharing a common language, having past colonial ties does not increase trade—a finding
which contrasts with what more basic methods, such as OLS tend to indicate. In applied work,
the model is often extended by including variables to account for language relationships,
tariffs, access to sea, colonial history, exchange rate regimes, and other variables of interest.
The current paper uses a simplified version of the gravity model to analyze the determinants
of trade between Kenya and Tanzania and whether the creation of the EAC has benefitted
them. This simplified version of the gravity model examines the determinants (mainly GDP,
size of population, distance and other socio-economic factors) that could influence the flow
of trade between Tanzania and Kenya. Essentially, the gravity model traces geographical-
spatial relationship of the foreign trade between Kenya and Tanzania. Used in the economics
sense, the gravity model stipulates that the economic size of countries and geographical
distance are the two basic factors determining the bilateral trade flows between Kenya and
Tanzania.
10
The stochastic form of the equation to estimate the determinants of trade of trade flow
between Kenya and Tanzania is given as:
ijijijjijioij ADNNYYTD 654321 ……………………………………………….. (5)
Where:
ijTD Value of total trade between Kenya and Tanzania
jI andYY GDPs of the Kenya and Tanzania respectively
jiandNN Population sizes of Kenya and Tanzania respectively
ijD = The distance between the capital cities of Kenya and Tanzania
ijA Other factors (including cultural) influencing trade between two countries
ij Error term
Linearising equation (5) by taking natural log of all variables results into equation (6) as
follows:
ijijijijiij ADNYYTD loglogloglogloglog 653210 …….. (6)
The current paper uses a stochastic form of the model to estimate the determinants of the
trade flows between Tanzania and Kenya as follows: The basic, standardized empirical
gravity equation takes the following form:
ijjiijjijiij PPLnLnDPYPYLnYYLnLnTD )*(]*)/(**)/[()*( 43210 .......(7
)
While if we want to understand the home market effect one could use the following equation
ijijtijij
ijijjijiij
PEITOTOPENEAC
EXRDNNYYEXP
logloglog8
loglogloglogloglog
11107
6543210
(8)
Where:
and = GDPs of Kenya and Tanzania respectively
11
and = Population sizes in Kenya and Tanzania respectively
= Geographical distance between the respective countries’ capita cities.
= Exports of Kenya or Tanzania
= Exchange rate between Kenya and Tanzania
= Degree of openness of Kenya or Tanzania economies
= Terms of trade between Kenya and Tanzania
= per capita income of Tanzania or Kenya
The theoretical propositions of the gravity model which are tested in this paper are that
bilateral trade flows between Kenya and Tanzania tend to be directly proportional to the
product of their respective GDPs and inversely proportional to their distance (D). That is to
say, the higher the GDP ratio the higher is the trade volume and vice versa.
Similarly the geographical distance between the two countries is inversely proportional to the
trade flows between the two countries. The gist of this proposition is that the further away
integrating countries are from each other the less attractive it becomes for them to gain the
advantages of integration. Other variables which are likely to influence trade flows between
Kenya and Tanzania including exchange rate, terms of trade, per capita income and degree of
openness will also be tested. It is hoped that the outcomes of this paper will influence
positively ongoing efforts to create a stable regional integration scheme in East Africa.
This paper tests how significantly the gravity model is in explaining Tanzania’s or Kenya’s
bilateral trade flows in the EAC and tries to extract implication for Tanzania’s and Kenya’s
trade policies. The equation’s application first fixes Tanzania as i, but leaving the rest of EAC
as j=1, 2...N in a Nx1 setting, and then it does so for Kenya. In this kind of equation, all
variables are in naturallogarithm of real value terms expect dummy variables. Any variable
with relatively small numbers are usually exempted from taking the logarithm (inflation,
terms of trade and openness for this study).
In particular, since the EAC countries borders each other, use common language, share
colonial relationship or historical ties the dummy variable is not that much relevant in this
12
model. Hence, the model starts with only three explanatory variables, namely the product of
GDPs, the product of per capita GDPs and distance.
The product of GDPs serves as a proxy for the two countries’ economic size, both in terms of
production capacity and size of market. Larger countrieswith great production capacity are
more likely to achieve economies of scale and increase their export based on comparative
advantage. They also possess large domestic market able to absorbed more imports.
Therefore an increase in the product of the two Countries’ GDPs is expected to increase
bilateral trade volume. Thus, we expect the estimated coefficient of production capacity to be
positive.
As explained above, the explanatory variables, the product of GDP2 serves as a proxy for the
two countries’ economy size, both in terms of production capacity and size of the market.
The per capita GDP, however, it serves as a proxy for the income level and/ or purchasing
power of the exporting and importing countries. When the fit of the gravity model for
Tanzania was run, the results turned out that, Tanzania’s trade depend much on GDP for its
bilateral trade flow and it is negatively related to the per capita patterns of trade.
Per capita GDP3 is an explanatory variable that serves as a proxy for the incomelevel and
indeed the purchasing power of the exporting and importing countries. Since, Tanzania’s or
Kenya’s per capita is fixed, this variable will serve to predict whether Tanzania’s orKenya’s
trade flows are dependent on its trading partners’ income level.
The distance variable is mainly a trade resistance factor that represents trade barrier such
transport costs, time, cultural unfamiliarity and market access barriers. Previous literature on
this subject has tended to interpret the coefficient of distance variable as elasticity with
respect to an absolute level of geographical distance. In the gravity model, trade volume will
be larger between countries pairs that are far from the rest of the world than between
countries pairs that are close to the rest of the world (Anderson and Van Wincoop 2003;
Harrigan 2003)4.
2Products of the GDP was used since the authors did not need the home market effect analysis.
3Bergstrand (1989), combining economic geography and factor proportional theory, derived the gravity equation at the
industry level which predicts that the export of a good in bilateral trade depend on income and per capital income as well,
assuming a constant elasticity of transformation of supplies among different markets.
4 A decrease of the distance coefficient indicates that trade with far away countries increases relative to the trade with closer
countries, whereas an increase represents trade with closer countries increases faster than that with far way countries. The
13
4 Empirical Analysis
The objective of the paper was to analyze the determinants of trade flows between Kenya and
Tanzania. The gravity model of international trade was deployed for this purpose due to its
widespread acceptability and availability of data. Prior to the presentation and analysis of the
results obtained, it is important to assess the robustness of our model. For this purpose, we
have employed the Linktest to test for model specification. This command is based on the
notion that if a regression is properly specified, we should not be able to find any additional
explanatory variables that are significant except by chance. The test creates two new
variables, the variable of prediction, _, and the variable of squared prediction, _hatsq. In our
model hat is significant indicating that our model is correctly specified, and hence we have
no reason to find any additional variable (see appendix 1).
The Hausman test is used to test for the significance of the parameters. The Hausman
specification test compares the fixed versus random effects under the null hypothesis that the
individual effects are uncorrelated with the other regressors in the model. The trick is that
random effect estimator would be useful if some explanatory variable remain constant over
time. The random effect assume that group effect are uncorrelated with regressor , while
fixed effect estimator measures the relationship based on time variation within a cross section
unit. In our estimation the null hypothesis is that the preferred model is random effects V/S
the fixed effect effect. The Hausman test whether the unique error are correlated with the
regressors, the null hypothesis is, they are not correlated with the regressor. Since in our
estimation the ch2 is greater than 0.05 i.e is significant then random effect model is preferred, ,
and hence both time invariant and time invariant variable is estimated (see appendix 2).
Breusch pagan test is employed to test for the presence of heteroskedasticity, presence of
heteroskedasticity means there is constant variance which violate the assumption of OLS.
The interpretation is that you look for the p-value, if the p- value is 0.05 or small then it
means there is presence of heteroskedacity, is otherwise no potential problem of
heteroscadacity. In our case the p-value is large than 0.05 that means our model estimation is
not suffering from the potential problem of heteroskedaciticity and hence does not violate the
assumption of OLS. The point of departure was analyze empirically whether variables like
notion of relative distance remains significance at the NxN countries setting, whereas it is important drops sharply in our
Nx1 because all distances are measured absolutely from Tanzania or Kenya, we anticipate the coefficient greater than one,
but its magnitude matters.
14
GDP, per capita GDP and distance between Kenya and Tanzania have any influence on trade
flows between them..
Table 3: OLS Results for the model (equation 7)
Explanatory variables Dependent Variable (Total
volume of trade)
OLS Coefficient
Constant 47.345***
(4.984)
Product of GDPs 1.119***
(0.267)
Product of per capita GDPs -1.923**
(0.530)
Distance -3.794***
(0.609)
R2 0.863
Adjusted R2 0.74
Hausman test 0.534
Notes: (i)) the numbers in parenthesis are standard errors
(ii) ) *** and ** and * indicate significance levels 1%, 5% and 10% level, respectively
The results obtained here (Table 3) show that with the exception of the per capita GDP
variable which has an unexpected negative sign (-1.9) the other variables have the expected
signs as predicted by the gravity model. The influence of the two countries’ GDP (as
indicated by the coefficient 1) is positive (1.19) as expected. This result is not only expected but
is also consistent with the basic hypothesis of the gravity model that the trade volume
between two countries tends to increase with the increase in the economic sizes of trading
partners. This result suggests that holding constant for other variables; a 1 percent point
increase in GDP will result in about 1.119 percent increase in Tanzania bilateral trade flows
with Kenya. The seemingly high value of this estimate is consistent with Heckscher-Ohlin
type5 hypothesis basing on inter-industry trade or differentiated type of trade basing on intra-
trade model of trade (Deardorff, 1998; Grossman 1998).
Cases where an increase in bilateral trade volume is less proportionate to the increase in GDP
can be explained in various ways. One is the existence of relatively large home-market effect
5 Frankel (1998) showed the coefficient lying in the range of 0.75-0.95. Our estimate more or less fit the range
but in a higher level.
15
(Trifler 1995, McCallum6, 1995). The existence of a large home market for a particular good
tradable between two countries implies that only a small amount of goods can be available
for trade in the other markets7. The other possibility is the presence of lower level of intra-
industry trade, whereby the volume of trade is higher in sectors characterized by a
monopolisticcompetition or scale economies (Harrigan, 2003).Thus, a country enjoying a
lesser scale economies will trade a smaller volume. Finally, this situation can be explained by
the existence and extent of trade barriers. Accordingly, the higher and wider the trade
barriersare, the smaller will be the trade volume.
Theoretically the per capita GDP variable is supposed to be a significant factor in
determining Tanzania’s bilateral trade in not only Kenya but in theEAC bloc as well.
However, results in Table 3 indicate that this variable is negatively related to the bilateral
trade flows between Kenya and Tanzania. The estimated coefficient ß2 has a magnitude of -
1.923, suggesting that predicting that a 1 percent increase in per capita GDP leads to about 2
percent decrease in bilateral trade flows between Kenya and Tanzania. The empirical result
is different from the regression analyses of Frankel (1997) that predicted that a 1% increase
in per capita GDP leads to about 0.1% increase in bilateral trade flows. Our findings implies
that Tanzania trade patterns follow a GDP pattern rather than a per capita pattern, relying
more on its trading partners’ overall economic size than its income level. The implications
from our results suggests that Tanzania trade can be characterized asa country which depends
more on exporting of quantity-based standardized products that are sensitive to the overall
market size instead of exporting quality-based high value-added products that are sensitive to
the trading partners’ income levels.
The estimation of the distance is significant with the expected negative sign. The results show
that geographical distance is an important factor for Tanzania’s trade. The coefficient ß3is
estimated to be -3.794, the result that is higher compared to previous studies (Frankel 1997;
Wall 1999; Buch et al. 2003), but in the line of Grossman (1998) who pointed out that most
of empirical gravity studies show a surprising larger size of the estimated coefficient on the
6 Home-biased effect, such as localized taste of local distribution network, plays a greater role in trade.
7 This is true for a small market, theoretically the home-market effect can be interested as an elasticity of export
with respect to domestic income that exceeds the importing country’s income elasticity of export and import for
a country, thus engendering the gravity analysis utilizing Y1and Y2 as a separate terms more appropriate.
16
distance variable. Furthermore, the coefficient implies higher cost of trading in the EAC bloc
because it reflects transport costs, time and market access barriers8and that the data used are
of nearby countries because a decrease in of the distance coefficient indicates that trade with
far away countries increase relative to the trade with closer countries, whereas an increase
represents trade with closer countriesincrease faster than that with far countries.
We also tested the influence of Tanzania’s trade openness (defined as total exports plus total
imports divided by GDP) on the trade flows to Kenya. The results indicated that the
estimatedcoefficient on the openness index is 0.99and is
robustwiththepresentspecificationofthegravityequation. This result implies that when
Tanzania tariff and policy liberalization change by 0.9% the country’s export tend to
increase by 1%. This suggests further that there is some correlation between Tanzania
tariff liberalization in EAC and her export performance. This finding is in line with
the standard international trade theories. In the presence of compensatory and
complimentary policies, openness to international trade tends to accelerates
development of poor countries (Dollar and Kraay, 2000).
Although high openness index is widely accepted an indicator of the degree of
trade liberalization it is certainly not a sufficient indicator of the welfare p art of
trade policy reform. The World Development Indicator (2006) for example show
that Sub Saharan Africa as a region managed to increase its openness index from
56% in 1975 to 66% in 2004. China on the other hand increased its openness index
from just 9% in 1975 to 65% in 2005. While Sub Saharan Africa has not managed
to eradicate object poverty, China has succeeded to do so.
This finding has important policy messages for Tanzania. It calls for policy makers
to focus more on the fundamentals of economic growth, investment,
macroeconomic stability, human resources and good governance as a strategy for
addressing poverty. Flexibility is highly needed while pursing this policy. Since
complete openness might expose a country to greater risk from external shocks.
Poor country like Tanzania may find it hard to buffer these shocks and to bear the
costs they incur.
5 Concluding Remarks
8Butch et al,. 2003 argue that changes in distance coefficient do not carry much information on changes in
distance costs over time. Change in distance costs are to a large extent picked up solely in the constant term of
the gravity models.
17
The purpose of this paper was to contribute to the on-going debate on the determinants of
trade between nations which began during the mercantilist era and was later shaped by the
introduction of absolute and comparative cost advantage theories. The Heckscher – Ohlin
and later the Linder hypotheses and other modern trade theories also contributed significantly
to the understanding of trade flows between nations.
The current paper has adopted a modified version of the gravity model used originally by
Tinbergen (1962) as improved by other researchers including Helman (2003) and Krugman
(2001) among others to analyze factors underpinning trade flows between Kenya and
Tanzania. Interest in analyzing the trade flows between the two countries has come as a
result of the observation that the two countries can loosely be referred as the ―economic
powerhouses‖ within the auspices of the East African Cooperation. The combined GDPs of
Kenya and Tanzania constitute over 60% of the total EAC GDP. Likewise there are
significant recorded and unrecorded trade activities between Kenya and Tanzania than in
other EAC countries. The importance of the two countries in influencing not only trade
activities but also economic development of the region is likely to increase with the
intensification of the EAC integration especially towards the formation of the common
market and monetary union. That is why a better understanding of the trade flows between
Kenya and Tanzania is essential in enabling the EAC countries to formulate appropriate
macroeconomic policies for the development the region.
Different forms of the gravity model were tested in the current paper. However, the most
plausible model for this paper was one where the total trade volume between Kenya and
Tanzania was influenced by four factors namely, the product of Kenya and Tanzania GDPs
( ), the product of their respective per capita incomes ( ) to degree of trade
openness and distance between Kenya and Tanzania.
The results obtained generally support the basic hypothesis of the gravity model that the
economic size of the respective countries (denoted as ) exerts a positive impact
on the trade between Kenya and Tanzania. This result is hardly surprising. GDP is a
reflection of not only the production capacity of a country but also the consumption and
export ability of countries. Thus the results indicate that the commodities produced in both
Kenya and Tanzania not only sustains consumption levels in the two countries but also is
exported to markets in the respective countries.
18
The results obtained with respect to the combined per capita income in Kenya and Tanzania
( ) are unexpected. The impact of this variable was supposed to ―mirror‖ that of
combine GDPs. Instead, the result shows this variable has a negative impact on the trade
flows between the two countries. One possible explanation for this result is that Tanzania
tends to depend more on exporting quantity products that satisfy the broad EAC market
instead of producing goods that are sensitive to the EAC partners’ income levels.
The distance variable was expected to have a negative impact on trade flows between Kenya
and Tanzania. The coefficient on this variable is – 3.794 and implies among other things the
existence of high costs of trading between Kenya and Tanzania. The costs of trading refereed
to here include transport costs (freight etc.), time related costs and costs related to market
access.
The results obtained with regard to trade openness showed that trade liberalization measures
have significantly improved trade flows between Kenya and Tanzania.The results obtained
herein have policy implications for both Kenya and Tanzania. They point to the importance
of the two countries focusing more on policies aimed at promoting economic growth with a
view to promoting trade flows between their countries and more significantly within the EAC
flows. Furthermore there is need to encourage the industries which cater specifically for the
needs of over 120 million people the region is a ready market for products from such
industries.
Efforts need to be made especially on measures aimed at reducing trade related costs in EAC
and specifically Kenya and Tanzania. The two countries have to come up with concrete
measures aimed at reducing tariff and non-tariff restrictions to trade.
Finally, the study has pointed at the need for EAC countries and specifically Kenya and
Tanzania to undertake further trade liberalization of their economies with a view to
promoting trade activities within the region.
19
References
Anderson, J.E..(1979), ―A theoretical foundation for the gravity equation‖, American
Economic Review, 69: 106-116
Anderson J.E. and E. van Wincoop (2001), ―Gravity with Gravitas: A Solutionto the
Border Puzzle,‖ NBER Working Paper 8079.
Anderson, J., van Wincoop, E. "Gravity with Gravitas: A Solution to the Border
Puzzle." AmericanEconomic Review (2003).
Baier, JH Bergstrand (2009)."Bonus Vetus OLS:"A Simple Method for
Approximating Trade-Cost Effects Using the Gravity Equation." Journal of
International Economics (2009)
Bergstrand, J.H. (1985), ―TheGravity Equation in International Trade:
SomeMicroeconomic Foundations and Empirical Evidence‖. The Review
ofEconomics and Statistics 71: 143-153.
Bergstrand,J.H, (1989), ―The Generalized Gravity Equation,
MonopolisticCompetition, and the Factor-Proportions Theory in International
Trade,‖The Review of Economics and Statistics 67: 474-481.
Bougheas S. et al. (1999), ―Infrastructure, Transport Costs and Trade,‖ Journalof
International Economics 47: 169-189.
Breuss, F., and Egger, P. (1999), ―How Reliable Are Estimations of East-West Trade
Potentials Based on Cross-Section Gravity Analyses?‖Empirica 26 (2): 81-95.
Caruso, R. "The impact of International Economic Sanctions on Trade, An Empirical
Analysis,"Peace Economics, Peace Science and Public Policy, vol.9, no.2
(2003).http://www.bepress.com/peps/vol9/iss2/1/
Chen, I-H., and H.J. Wall (1999), ―Controlling for Heterogeneity in GravityModels of
Trade,‖Federal Reserve Bank of St. Louis Working Paper 99-010A
Deardorff, A.V. (1995), ―Determinants of Bilateral Trade: Does Gravity Workin a
Neo-Classic World?‖ NBER Working Paper 5377
Deardorff, Alan V. ―Determinants of Bilateral Trade: Does Gravity Work in a
Neoclassical World?‖ In The Regionalization of the World Economy, edited by
J.A. Frankel. Chicago: University of Chicago Press. 1998, 21.
Egger, P. (2000), ―A Note on the Proper Econometric Specification of theGravity
Equation,‖ Economics Letters 66: 25-31.
20
Feenstra, Robert C., James R. Markusen, and Andrew K. Rose.―Using the Gravity
Equation to Differentiate among Alternative Theories of Trade.‖The Canadian
Journal of Economics, Vol. 34, No. 2. (May, 2001), pp. 431
Frankel, Jeffery A. Regional Trading Blocs: In the World Economic System.
Washington, DC: Institute of International Economics. October 1997.
Hacker, R Scott, Einarsson, Henrik.―The Pattern, Pull, and Potential of Baltic Sea
Trade.‖ The Annals of Regional Science, Vol. 37, No. 1, pp. 15–29.
Helpman, E. and Krugman, P. (2003) ―Market Structure and Foreign Trade:
Increasing Return, Imperfect Competition and International Economy‖,The MIT
Press, Cambridge, MA/London.
Isard, W., "Location Theory and Trade Theory: Short-Run Analysis". Quarterly
Journal of Economics, vol. 68, 1954, p. 305- 322.
Krugman, P and Obstfeld, E. (2003) ―International Economic Theory and Policy‖,
6th
edition,
Limao, N., and A.J. Venables, (1999), ―Infrastructure, GeographicalDisadvantage and
Transport Costs,‖ Policy Research Working Paper 2257,World Bank.
Mátyás, L. (1997) ―Proper Econometric Specification of the Gravity Model‖, The
World Economy, Vol. 20, Issue No 3.
McPherson, M.A et al (2000). ―A Re-examination of the Linder Hypothesis: a
Random-Effects Tobit Approach.‖ International Economics Journal Vol.14 No. 3
Paas, T. (2000) ―Gravity Approach for Modeling Trade Flows between Estonia and
the Main Trading Partners‖, Working Paper, No. 721, Tartu University Press
Panagariya, A. (2000). ―Preferential Trade Liberalization: The Traditional Theory and
New Developments,‖ Journal of Economic Literature 38, June, 287-331
Poyhonen,P. (1963), ―A Tentative Model for the Volume of Trade
betweenCountries,‖WeltwirtschaftlichesArchiv 90: 93-99.
Santos Silva, J.M.C. and Tenreyro, Silvana (2006), ―The Log of Gravity,‖ The
Review of Economics and Statistics, 88(4), pp. 641–658.
Santos Silva, J.M.C. and Tenreyro, Silvana (2011), Further simulation evidence on
the performance of the Poisson pseudo-maximum likelihood estimator,
Economics Letters, 112(2), pp. 220-222.
Soloaga, I., and Winters, A. (1999) ―Regionalism in the Nineties: What Effectson
Trade?‖ Development Economic Group of the World Bank, mimeo.
21
Summary, Rebecca M. ―A Political-Economic Model of U.S. Bilateral Trade.‖ The
Review of Economics and Statistics, Vol. 71, No. 1. (Feb., 1989), pp. 179–182.
Tinbergen, J. (1962), ―Shaping the World Economy. Suggestions for anInternational
Economic Policy,‖ New York.
Wei, S.-J. (1996), ―Intra-national versus International Trade: How StubbornAre
Nations in Global Integration?,‖ NBER Working Paper 5531.
Yeats, A. (1998), Ò DoesMercosurÕ s Trade Performance Raise Concerns Aboutthe
Effects of Regional Trade Arrangements?Ó The Word Bank EconomicReview 12
(1): 1-28.
Viner, J. (1950) The Customs Union Issue. New York: Carnegie Endowment for
International Peace
22
Appendix 1
(Appendix Table 1: Test for Model Specification)
_cons .0012679 .0332237 0.04 0.970 -.0666822 .0692181 _hatsq -.0065264 .0143904 -0.45 0.654 -.035958 .0229052 _hat 1.014454 .0371544 27.30 0.000 .9384647 1.090443 EXPTZA Coef. Std. Err. t P>|t| [95% Conf. Interval]
Total 45.0939847 31 1.45464467 Root MSE = .12757 Adj R-squared = 0.9888 Residual .471952524 29 .016274225 R-squared = 0.9895 Model 44.6220322 2 22.3110161 Prob > F = 0.0000 F( 2, 29) = 1370.94 Source SS df MS Number of obs = 32
Appendix 2
Appendix Table 2: Test for Heteroscekedasticity
Prob > chi2 = 0.1685 chi2(1) = 1.90
Variables: fitted values of EXPTZ Ho: Constant varianceBreusch-Pagan / Cook-Weisberg test for heteroskedasticity