121177 Research in Business and Economics Journal

The determinants of trade in the Central African Economic and

Monetary Union
Augustin Ntembe
Bowie State University
Regina Tawah
Bowie State University
ABSTRACT
Bilateral trade between the economies of the Central African Monetary and Economic
Union is low relative to trade in other economic unions in Africa. Understanding what actually
determines trade in the CEMAC regions can offer useful insights for policies that promote trade.
This study uses a modified gravity model of trade on panel data for the six member states of
CEMAC and fourteen trading partners to investigate the relationship between trade and its
determinants. Panel data unlike cross-section and time series analysis allow us to control for
unobservable individual specific characteristics that affect trade. The results of the study show
that membership of CEMAC did not increase trade among member states during the period from
1998-2008. The results highlight the importance of distance and being landlocked as obstacles to
trade within the region and between CEMAC economies and overseas trading partners.
Keywords: gravity model, economic and monetary union, pooled model, fixed effects, random
effects, Haussmann-Taylor

Copyright statement: Authors retain the copyright to the manuscripts published in AABRI
journals. Please see the AABRI Copyright Policy at http://www.aabri.com/copyright.html.

The determinants of trade, page 1

121177 Research in Business and Economics Journal

INTRODUCTION
The recent emergence of economic blocs has been attributed to economic and political
motives (Baldwin 1997). The formation of trade blocs, a form of economic union, has been
undertaken in attempts to create incentives for political cooperation and stability. Schiff and
Winters (1997) in particular, hold that by grouping their economies, countries tend to avoid
conflicts and security threats. Another reason for the creation of trade blocs is that reaching
multilateral agreements on trade liberalization under the General Agreement on Tariff and Trade
and the World Trade Organization is often more daunting than is the case within trade blocs
where countries are fewer and negotiation is easier. From the economic perspective, the
formation of economic blocs is motivated by the allocation effect and the growth effect arising
from free trade within an economic bloc (Baldwin 1997).
The allocation effect requires that in a competitive economic system resources are
allocated for the production of goods based on peoples demand for those good. Thus demand
acts as a signal for interaction between consumers and producers. When tariffs and non-tariff
barriers interfere with this signal, it becomes necessary to clear such barriers through regional
integration (UNCTAD 2009). A consequence of the allocation effect according to Baldwin
(1997) is the scale and variety effects. Regional integration could also lead to the creation of
large markets which would allow access to small firms thus enabling them to reach optimal sizes
lowering costs and prices for the consumers. As concerns the variety effect, economic integration
creates a wider market allowing consumers to choose from a variety of products at lower prices.
At the level of the firm, this creates an incentive for more efficient use of its inputs and could
increase productivity.
As regards the growth effect, economic integration expands regional markets, attract
more suppliers to these markets and this gives firms the opportunity to specialize which in turn
reduces average costs of production within the trade bloc. This has the tendency of increasing the
return to physical and non-physical factor accumulation. Regional integration accounts for an
increase in the mobility of human capital, technological spillovers, an increase in productivity
and the reduction of production costs which help to attract more investment and capital
accumulation. The location decision of foreign firms can be significantly influenced by the
formation of trade blocs. Krugman (1991) has pointed out based on the new economic geography
school that market size, the cost of production, the availability of relevant production factors and
market access are the key location variables.
Although regional integration has many positive effects on trade, supporters of free trade
argue that regional trade limits rather than promote trade (Schiff, 1997; World Bank 2000). The
argument advanced in support of this opinion is that trade blocs raise tariffs and non-tariff
barriers thus inhibiting substantial inflows of productive resources from non-members. This may
result in inefficient resource allocation and production which may diminish welfare gains from
competition.
In Africa, regional integration was pursued to enhance political unity at Pan African level
and as a means to foster economic growth and development (UNCTAD 2009). With the desire to
promote economic integration and in some cases monetary integration, there has been a
multiplicity of regional and sub-regional economic groupings across the continent (Nnanna,
2006). Compared to other regions of the world, Africas trade costs are the highest and this tends
to discourage foreign investment and also place limits on bilateral trade. The expansion of the
market through regional integration can significantly relax this constraint. Thus free movement

The determinants of trade, page 2

121177 Research in Business and Economics Journal

of factors of production as envisioned in trade agreements can help to keep production costs at
low levels, attract investments and increase competition and efficiency. This is even more
significant considering that population wise African countries are typically small with over 70
percent of them having a population of less than or equal to fifteen million people including
some with as low as three million or less. In terms of purchasing power, African countries are
also small, given that thirty-one out of the forty-six least developed countries in the world, as
classified by income level, are found in Africa (UNECA, 2010).
Despite the proliferation of regional economic groups in Africa, intra-regional trade
between African countries and especially between members of the same economic unions is
limited. Countries belonging to various economic blocs have taken limited measures in fully
reducing tariffs and eliminating non-tariff barriers, and in adopting common economic policies
that promote growth and these tend to impede meaningful economic integration that foster trade
between member states. While the problems faced by regional economic communities are
generally similar, the kinds of arrangements remain varied. This study focuses on the Central
African Economic and Monetary union (CEMAC). CEMAC effectively replaced the Customs
and Economic Union of Central African States (UDEAC) in 1999. What is particular about this
union is that it was created both as an economic and monetary union with the member countries
sharing a common currency, the franc CFA (Coopration financire en Afrique Centrale).
Free trade agreements between countries of this sub-regional group require among other
arrangements that members substantially reduce or eliminate all forms of tariffs and non-tariff
barriers to trade for greater market integration and increase in production and volume of trade
flows (Freund and Ornelas, 2009). The currency union as in the case of CEMAC is a multilateral
arrangement with far reaching economic and political implications as even barriers to labor and
capital flows are removed and individual countries give up both their fiscal and monetary
autonomy. In particular, the implementation of policies in the currency union has implications on
regional trade. A currency union enhances trade by eliminating exchange rate volatility among
member countries thereby reducing the transaction costs that hinder trade flows between
countries and regions. Celine (2003) suggests that trade creation effects mostly result from a
common currency arrangement.
Various studies carried out in the last decade show a general consensus that economic
and monetary unions enhance trade between member countries (Rose and Stanley 2005; Frankel
and Rose, 2000 Rose, 2000; Alejandro et al, 2003). In the case of Africa, however, no such
consensus was found. While some studies found positive and large as well as small effects, a few
others have found zero or negative effects. The Central African Economic and Monetary Union
is one of Africas economic blocs that raise some concern. The level of intra-regional trade
between member states of the sub-region is low. Intra-regional trade among economies of the
Economic Community of Central African Countries (ECCAS) where CEMAC countries are also
members accounts for only 2 percent or less of total trade and apparently has the lowest intraregional trade share of all regional economic communities in Africa (UNCTAD, 2009).This
raises questions on the effectiveness of economic and monetary unions as an instrument that
promotes stability and trade among member states. This study uses the gravity model to
investigate the determinants of bilateral trade between CEMAC economies and between these
economies and their major trading partners to have a clear understanding of the extent to which
economic and monetary integration contributes to the expansion of bilateral trade. Investigating
the determinants of bilateral trade in the CEMAC region is crucial in understanding why intra-

The determinants of trade, page 3

121177 Research in Business and Economics Journal

regional trade within the region is limited despite the global positive view of regional integration
on both regional and international trade.
THE CENTRAL AFRICAN ECONOMIC AND MONETARY UNIONS
A total of six countries make up the Central African Economic and Monetary Union
including Cameroon, Central Africa Republic, Chad, Republic of Congo, Equatorial Guinea and
Gabon. These countries are also member states of the Economic Community of Central African
States (ECCAS). The regional central bank of the six countries of CEMAC is the Banque des
Etats de l'Afrique Centrale (BEAC). The common currency used by CEMAC countries is the
Franc CFA which was pegged to the French franc at a fixed exchange rate of 50 franc CFA per
French franc for 45 years. In its whole history, the franc CFA was devalued once in 1994 from
the 50 franc CFA per French franc to CFA 100 per French franc. The changes require a
unanimous agreement by member states of the Union Montaireet Economique Ouest Africaine
(UEMOA), member states of the CEMAC and France. The monetary rules put in place since
1973 require that the central banks of CEMAC and UEMOA each deposit 65 % of its foreign
exchange holdings into the French Treasury while maintaining 20% foreign exchange for sight
liabilities coverage each. This in practice is intended to limit an open-ended access to the
operation accounts that the banks maintain with the French Treasury.
Following the formation of the European economic and monetary union, the CFA franc
became pegged to the Euro with the support of the French Treasury-budgetary arrangement at an
exchange rate of 1 euro = 655.957 francs CFA. Following this arrangement, the franc CFA is
guaranteed by the European Central Bank and there are no monetary implications for the latter.
Each of the central banks issuing the franc CFA maintains an operation account with the French
Treasury and the French Treasury guarantees the convertibility of the franc CFA into Euros.
How this affects the CFA franc countries in the future depends on CFA franc countries keeping
up with their efforts and commitment to better economic management. This indeed implies that
these countries have to adhere to very strict monetary and fiscal discipline imposed by the
European Central Bank.
It appears since the new arrangement that the advantages from this anchor to the euro
overshadow the potential risks in principle. In the medium to long run, the principal risks would
be linked to how strong the euro is with respect to other major currencies (pound, dollar, yen,
and Swiss franc). The CFA franc zone produces and exports a limited number of basic products
whose prices are quoted in dollars. Although CFA franc countries carry out more than 50 % of
their trade with the European Union members, the trade is denominated in dollars.
The benefits of monetary and economic union are assessed in terms of macroeconomic
stability and economic integration among member countries in the form of expanded trade and
free movement of capital within the union. Macroeconomic stability requires that member
countries agree to improve on price stability, reduce exchange rate variability, save resources
from pooling of foreign exchange reserves and the centralisation of monetary policy. A full
monetary union would ensure and support an open trading regime. When countries join monetary
unions, they lose their monetary independence so that the costs of a monetary union are
measured in terms of the loss of the exchange rate as an instrument for adjustment among
members and also the loss of seigniorage and inflation tax revenue.
Economic performance in the CEMAC zone has been quite impressive relative to that in
other economic groupings. Between 1997 and 2005, the region grew at the rate of 7.6 percent

The determinants of trade, page 4

121177 Research in Business and Economics Journal

making it the highest performer among all the regional groupings (Nnanna 2006). It also
outpaced Africas growth rate of 3.8 percent during the period. However, intra-regional trade for
the bloc was only 1.3 percent representing the lowest in the continent. The average inflation rate
was low and estimated at 3.1 percent making it the second best after that of UEMOA and was 7.4
percent lower than the African average. The countries of the union also had a fiscal surplus of
2.1 percent of GDP making the bloc the best performer relative to other groupings which
suffered from deficits of different sizes.
The overall performance of the franc CFA countries before the devaluation of the franc
CFA in 1994 was a dismal one in spite of the strict budgetary discipline and low inflation
compared to other African countries (Devarajan and Milo 1987). One of the main constraints
faced by these unions is the lack of and difficulty in adjusting to the external environment which
led to exchange rate misalignment and economic problems of the late 1980s and early 1990s
(Devarajan and Milo 1991). An important benefit from economic integration is the ability of the
union to promote growth in exports and cross-border trade and investment among member
countries. Yet intra-regional trade in the CEMAC region accounts for only a small percentage of
the overall trade in the region. While intra-regional trade in economic unions such as UEMOA
has been quite impressive, in the CEMAC, the percentage of intra-regional trade fell from 4.9
percent in 1970 to barely 1.6 percent in 1980. The ratio rose to 2.3 percent in 1995 and fell to 2.1
by 2000. The overall average has been maintained at below 1.5 percent and in value terms, it has
remained consistently below US$85.0 million although when the time frame is modified to
between 1995 and 2004, the average value rises above US$100.0 million. The trade estimates are
from official sources and do not include informal trade between these countries which can be
very significant.
THE GRAVITY MODEL OF TRADE
This study employs the gravity model to analyze bilateral trade flows within the CEMAC
region. It examines factors that contribute to variations in the volume of trade between these
countries and trade with major trading partners. The gravity model of international trade was first
developed by Tinbergen (1962) to explain bilateral trade flows and between two countries. The
model considered both countries as organic bodies that attract each other in proportion to their
levels of economic activity and income, and the extent of the impediment to trade. The latter
include transport costs, trade policies, uncertainty, cultural differences, and various bottlenecks.
While these factors may well determine the volume of trade flows in sub-Saharan African
countries, the high cost of transport in addition to poor transport infrastructure are important
barriers to trade.
The gravity model has been used extensively as a baseline for estimating the impact of
policy issues, such as regional trading arrangements, currency unions, political blocks, and
various impediments to trade (Havrylyshin and Prichett 1991; Frankel and Wei 1993; Bayoumi
and Eichengreen 1997; Frankel and Wei 1998; Frankel, Stein and Wei 1998; Smith 1999; Rose
2000). The deviations from the baseline gravity model are modeled by these events and policies.
Furthermore, dummy variables are used to gauge the impact of economic integration on trade.
On a more aggregate level of the analysis, various macroeconomic variables such as gross
domestic products, populations and geographical location have been used as powerful predictors
of trade potential of nations.

The determinants of trade, page 5

121177 Research in Business and Economics Journal

Model Specification
The literature on the gravity model and key elements of the trade theory has shown that
the gravity model can be derived from both the traditional and the new theory of international
trade rather than just from the new theory as suggested by Helpman and Krugman (1985).
Bergstrand (1985) has used microeconomic foundations to explain the gravity model by
specifying the supply side of the economies. Eaton and Kortum (1997) developed a Ricardian
framework while Deardorff (1998) showed that the gravity model could be derived from two
extreme cases of the Heckscher-Ohlin model with and without trade impediments. Although
there are various specifications of the gravity model, this study employs the following general
specification of the gravity model to investigate factors that account for trade within the CEMAC
countries and between these economies and overseas countries:
X ij 0 (Yi )1 (Y j ) 2 ( N i )3 ( N j ) 4 ( Dij )5 ( Aij )6 ij

(1)

where Xij is the value of the trade flow from country i to country j; Yi and Yj are the values of
real gross domestic product in the exporting and importing countries respectively and represent
the size of both economies. Ni and Nj represent the exporter and importer populations, Dij
measures the physical distance from the economic center of country i to that of country j and
captures the transport costs. Aij represents any other factors that contribute to or obstruct bilateral
trade and ij is the error term. The s are interpreted as coefficients of elasticity of exports in
respect to changes in gross domestic product, population, distance and real exchange rate. The
coefficients of the GDPs of the exporting and importing countries are expected to be positively
related to trade. The coefficient estimates for the population may either be positive or negative
depending on whether a country exports less when it has a larger population (absorption effect)
or whether a larger country exports more than a smaller country (Prewo (1978) and Bergstrand
(1986). The coefficient of the distance variable is expected to be negative since longer distance
entail higher cost of trade. Krugman (1991) emphasized on the role of geographical proximity in
the regionalization process. He showed in his analysis that a pair of countries with low
transportation costs between them will tend to have a higher volume of trade than countries
further apart. Higher transport costs are expected to reduce trade so that the effect of distance on
trade is negative.
The generalized gravity equation is rendered estimable in the following log-linear form:

ln X ij 0 1 ln Yi 2 ln Y j 3 ln N i 4 ln N j 5 ln Dij 6 Aij ij

(2)

We augment the gravity model shown in equation (1) in order to account for as many factors as
possible that affect trade. These variables potentially account for cultural phenomenon,
geographical nature of the two countries, and the historical nature of the relationship between the
two countries. Further, we extend the gravity equation by considering the preferential trade
policies which entails unilateral reduction in trade barriers granted by developed countries to
developing countries. These are expected to stimulate exports from developing countries to the
preference-giving country yielding a higher flow of trade than that which would normally be
expected. We also take into account the location of the country with particular interest in whether

The determinants of trade, page 6

121177 Research in Business and Economics Journal

they are coastal countries or landlocked countries. Given the total trade flow or exports, we
augment equation (2) as follows:

ln X ij 0 1 ln Yi 2 ln Y j 3 ln N i 4 N j 5 ln Landlij 6 ln Dij 7 ln Aij ij (3)

where 0 is the portion of the intercept that is common to all years and the trading countries.
Landl is a landlocked dummy which is equal to 1 if the exporting or the importing country has a
coast, and zero, otherwise. Ni and Nj are the sizes of population in both countries; Aij is any other
factor that either enhances or hinders trade among i and j; and uij is a normally distributed
random error term with E (lnij) = 0.
We also introduce the bilateral real exchange rate (RERji) as a proxy for relative prices.
An increase in RER reflects a depreciation of the importing countrys currency against that of the
exporting country which should reduce imports and as a result, one would expect 9 in equation
(4) and 11 in equation (5) to be less than zero. The intra-union gravity equation is a restricted
gravity model applicable to trade within the CEMAC custom and monetary union augmented by
country pair-specific variables as follows:
ln(X ijt ) 0 1lnYit 2 lnYjt 3lnD ij 4 lnN i 5 lnN j 6 BOR ij
7 Landl ij 8 CCOLij 9 ln RERijt ijt

(4)

where BOR is a dummy that takes the value 1 if countries i and j share a common border and 0 if
otherwise, Dij is the lateral distance that separate the economic (or capital) centers of countries i
and j in kilometers. And CCOLij is a dummy that takes the value 1 if country i and country j have
a common colonizer and zero, otherwise. Equation (4) can also be augmented by including a
variable that measures preferential trade agreements between countries of the CEMAC region
and their major trade partners as follows:
ln(X ijt ) 0 1lnYit 2 lnYjt 3lnD ij 4 lnN ti 5 lnN jt 6 Pij 7 BOR ij
8 Landl ij 8 CEMAC 9 CCOLij 10 EU 11RER ji ijt

(5)
where, Pij is a dummy equal one if the exporting and importing country have a preferential trade
agreement such as the case of the European Union and most Sub Saharan African countries and
zero otherwise. EU is dummy that takes the value 1 if the country trading with a CEMAC
country is a member of the European Union and 0, otherwise. CEMACij is a dummy that takes
the value one if countries i and j belong to the CEMAC economic and monetary union and zero
otherwise. By including CEMAC as a dummy in the specification, it enables us to investigate if
belonging to the monetary and economic union is beneficial to trade or not. The arrangements
require that member states put in place trade-friendly monetary policies and remove tariff and
non-tariff restrictions on trade. Given that the monetary union is well structured compared to
others in Africa, we expect a significant and positive effect of the CEMAC dummy on trade.

The determinants of trade, page 7

121177 Research in Business and Economics Journal

Data and Estimation Procedure
In this study, we construct a panel dataset in case of total exports. The set of importing
countries consist of France, Italy, Germany, Spain, Belgium, Switzerland, Holland, Britain, USA
, India, China, Morocco, Japan and Canada while the exporting countries include all CEMAC
countries. These countries are listed in Appendix A. To ensure that there is sufficient coverage,
we use data from 1998 to 2008. Each annual bilateral export flow is an observation. We consider
only positive trade values thus dealing with the problem of zero-trade and missing values. In this
analysis we use export data obtained from the International Monetary Fund Direction of Trade
Statistics Yearbooks 2005 and 2009, the IMF World Economic Outlook 2009 for the population
and GDP data. The geographical distance is the distance between the capital cities of two
countries. Distances can be estimated by browsing on http:/www.wcrl.ars.usda.gov/cecl/java/latlong.htm.
To successfully estimate equations (4) and (5), we employ panel data which enables us to
control for unobservable factors specific to both the exporting and the importing countries that
determine bilateral trade between both countries. Thus we include bilateral effects in the model
to account for heterogeneity in bilateral trade relations. These effects are modeled as random
effects judging that the fixed effects model is grossly inadequate since the within
transformation removes variables such as distance, common border, common colonizer or some
other time-invariant regional dummies. Other common estimation approaches such as pooled
ordinary least squares (OLS) and random effects produce coefficients that are biased when the
time-variant variable is correlated with the unit fixed effects. Since unit effects are unlikely to be
uncorrelated with time-variant variables and they are also unlikely to be normally distributed,
estimating the gravity model by way of either pooled OLS or random effects may lead to biased
results.
However both approaches may be recommended when time-invariant variables are
uncorrelated with the unit fixed effects. Random effects specification performs better than
pooled OLS in the presence of time-invariant variables. When time-invariant variables hinder the
estimation of unit fixed effects, random effects become preferable to pooled OLS estimation.
The Hausman test based on the differences between within and generalized least squares
estimates reveals correlation between the bilateral specific effects and the explanatory variables.
Hence, this problem is overcome by way of instrumental variables estimation proposed by
Hausman and Taylor (1981) which suggest the use of instruments for the variables that are likely
to be correlated with the random effects. The Hausman-Taylor approach performs better than the
random effects and pooled OLS when time-variant variables are correlated with unit effects.
ESTIMATION RESULTS OF THE BILATERAL TRADE MODEL
We estimate the pooled or ordinary least squares model, the fixed effects, random effects
and the Hausman-Taylor models and with various tests, determine which of these models is most
appropriate when estimating the flow of trade within the CEMAC region and between CEMAC
economies and randomly selected trade partners.

The determinants of trade, page 8

121177 Research in Business and Economics Journal

Estimation Method and Results
A common problem encountered when modeling trade flows using panel data is that
there are unobservable time invariant random variables which are difficult to be quantified and
which may have influence on some explanatory variables and even the volume of trade. Ordinary
least squares (OLS) or the pooled model does not allow for the heterogeneity of countries, it also
fails to estimate country specific effects and assumes that all countries are homogenous. Table 1
and Table 2 show the estimation results of the two equations used in this study. In both estimated
equations, the Breusch-Pagan test rejects the null hypothesis in favor of the random effect model
implying that the OLS is not adequate for this study.
Table 1. Estimation Results: Intra-CEMAC Countries Trade Flows
Variables
Intercept
RGDP exporter
RGDP importer
Pop exporter
Pop importer
Distance
Border
landlocked
Common colonizer

Fixed Effects
- 2.865
(-1.74)
-0.003
(0.01)
-0.285
(-0.77)
2.319
(0.83)
0.5654
(0.19)

Random Effects
9.396
(1.02)
0.249
(1.54)
-0.055
(-0.38)
1.408***
(3.80)
-0.10
(-0.25)
-2.373*
(-1.87)
2.848***
(2.43 )
-2.340***
(-2.49)
1.1155
(1.54)
330
0.765
3.53
978.49***

Hausman-Taylor
9.680
(0.94)
0.252
(1.13)
-0.073
(-0.34)
1.4178***
(3.31)
-0.124
(-0.33)
-2.399*
(-1.73)
2.829***
(2.15)
-2.332***
(-2.62)
1.113
(1.05)
330

Number of obs
330
R-squared
0.1179
Hausman test
Breusch and Pagan test
Heteroskesdasticity test
23615.31***
Test of over-identification
3.571
(Sargan-Hasen statistic)
***, **, and * significant at 99%, 95% and 90% respectively (t-student are in brackets)

The fixed effect model allow for unobserved factors that explain the volume of trade between
two countries and leads to unbiased and efficient results (Bair and Bergstrand 2005; Carrere
2006; Rose 2000). The model also assumes that there are time-invariant characteristics that are
unique to the individual and should not be correlated with other individual characteristics. The
fixed effect from the econometric standpoint is preferable to the random effects since it is
The determinants of trade, page 9

121177 Research in Business and Economics Journal

unlikely that the unobserved specific random effects are uncorrelated with the explanatory
variables (Baier and Bergstrand 2005). The fixed effect model is no longer effective if the error
terms are correlated. In this case, the random effect model provides a better method of estimating
the model. The random effect assumes that the entitys error term is not correlated with the
explanatory variables which allows for time-invariant variables to play a role as explanatory
variables.
Table 2. Estimation Results: Bilateral flows from CEMAC Countries to Trading Partners
Variables
Intercept
RGDP exporter
RGDP importer
Pop exporter
Pop importer
Real exchange rate

Fixed Effects
-1.527
(-0.79)
0.942***
(3.21)
1.054***
(3.12)
1.461
(1.22)
-2.232**
(-1.88)
-1.664***
(-4.65)

Distance
Preferential Trade
Agreement
CEMAC
Common Language
Landlocked
EU

0.649***
(2.31)
1254
0.027
9.43

Random Effects
8.236
(2.66)
1.092***
(5.64)
0.704***
(3.88)
0.014
(0.07)
0.246
(0.90)
-1.818***
(-7.59)
-2.735***
(-4.26)
-0.568*
(1.28)
-1.706***
(-2.42)
0.311*
(1.66)
-0.664***
(-2.71)
0.451**
(1.79)
1254
0.540

Hausman-Taylor
8.706
(1.87)
1.112***
(6.50)
0.622**
(2.38)
0.066
(0.19)
0.1919
(0.51)
-1.830***
(-7.21)
-2.710**
(-2.49)
-0.742
(-1.32)
-1.822*
(-1.83)
0.310
(1.03)
-0.668**
(-2.04)
0.555***
(2.50)
1254

Number of observations
R-squared
Hausman Test
Breusch&Pagan test
3136.86***
Heteroskesdasticity
832.49***
Test of over-identification
7.731
***, **, and * significant at 99%, 95% and 90% respectively (t-student are in brackets)

We decide between fixed and random effects models by employing the Hausman (1978)
test for each of the samples. The test indicates whether the specific effects are correlated or not
with the explanatory variables. A high value of the Hausman test (or p-value 0.05) indicates
that the errors are correlated with the independent variables and the fixed effect model is
The determinants of trade, page 10

121177 Research in Business and Economics Journal

preferred to the random effect model. Consequently, the random effects model would be
inconsistently estimated in the presence of such correlation. The absence of such correlation
implies that the random model should provide consistent estimates of the coefficients. Besides, it
allows for time invariant variables to play a role as explanatory variables. The results of the
Hausman tests of the differences between random effects and within estimates are presented in
both tables.
The test results (H=1.35 distributed as Chi2(5), with p-value = 0.9302) accepts the null
hypothesis that there would be correlation between the bilateral specific effects and the
explanatory variables for the intra-CEMAC sample hence the random effects model (GLS )
estimator is unbiased and efficient. The result of the test for the whole sample (H=9.90,
distributed as Chi2 (6), with p-value = 0.1288) equally show that the null hypothesis of no
correlation is accepted and the random effect model of panel estimation is the appropriate
estimation strategy. Furthermore, the fixed effect model uses only the within variance for the
estimation and disregards the between variance, and does not allow the estimation of timeinvariant variables so that bilateral effects are modeled as random variables (Baltigi 2001; Hsiao
2003; Wooldridge 2002).
Homoscedasticity in panel data is always assumed and needs to be tested. We use the modified
Wald test for group-wise heteroskedasticity in fixed effect regression to test for
heteroskedasticity. Table 1 and Table 2 report a p-value of 0.000 for both samples indicating the
presence of heteroskedasticity. We control for heteroskedasticity in both the fixed and the
random effects models by using the option robust in the estimations.
Although the Hausman test supports the choice of the random effects over the fixed effect
model, variables such as gross domestic product for the exporting and the importing countries
may be correlated with bilateral specific effects. This problem can be dealt with using
instrumental variables estimation such as Hausman-Taylor (1981). The Hausman-Taylor method
also incorporates time-invariant variables correlated with bilateral specific effects. We performed
a Hausman-Taylor test of over-identification based on the comparison of the Hausman-Taylor
estimator and the fixed effect model. The results indicate the choice of the Hausman-Taylor
estimator because it is more efficient, allowing for estimating time-invariant effects.
Discussion of Regression Results on Panel Data
The results discussed here are based on the Hausman-Taylor estimation presented in the
fourth columns of Table (1) and Table (2). The estimated results of the intra-CEMAC equation
indicate that most of the key variables specified in the traditional gravity model of trade are
significant with the expected signs. The results indicate the GDP for the exporting economies is
positively related to bilateral trade between CEMAC countries but is not a significant
determinant. Surprisingly and contrarily to economic theory, the GDP of the importing country is
negatively related to bilateral trade within the region suggesting lower imports by CEMAC
countries from other members of the union. The population of the exporting country has a
positive and significant effect on bilateral trade (1.42, t = 3.31) while that of the importing
country is negatively related to trade as expected (-0.12, t = -0.33). The larger the population of
the importing country, the more diverse is its production thus making the country more selfsufficient and it is normally expected to be negatively related to trade. The results also show that
geographical distance has a negative and significant effect (-2.40, t = (-1.73)) on bilateral trade
between the economies of the custom and economic union. The results also show that land

The determinants of trade, page 11

121177 Research in Business and Economics Journal

border with the expected positive coefficient is a significant determinant of bilateral trade (5
percent level). The significance of land border in explaining trade within the CEMAC region
highlights one of the key concerns for trade within the region which is transport infrastructure
and proximity. Bilateral trade between countries in the region tends to be higher when the
country-pair share a common border than those that do not share a common border.
Another variable of key interest which is also a robust determinant of bilateral trade in
the CEMAC region is landlocked. Out of the six countries that make up the membership of the
economic and monetary union, two are landlocked and have access to the coast through land
transit via other countries which is costly. The estimated results are robust and highly significant
at 1 percent indicating the difficulties faced by landlocked countries in bilateral trade. The real
exchange rate variable is significant at 10 percent and has a negative coefficient implying that
real exchange rate of CEMAC countries has a significant but negative impact on trade within the
region.
In estimating the second equation, variables such as the real exchange rate, the EU and
CEMAC dummies, preferential trade agreements between some of major trading partners and the
CEMAC economies, and common language have been added. These are all dummies which take
the value 1 if the variable is observed during the period under consideration and 0 otherwise. The
estimates of the whole sample in Table (2) show interesting results.
The coefficient of the GDP of the exporter countries is highly significant and has a
positive sign implying that an increase in the GDP of CEMAC economies will increase their
exports to trading partners. The results also indicate that an increase in the GDP of the importing
country leads to an increase in imports from the CEMAC region because of the high absorbing
capacity of the importing country. Like in intra-CEMAC trade, geographical distance has a
negative and significant (-2.71, t = -2.49) impact on bilateral trade. Equally significant is the
landlocked variable with estimated coefficient of -0.67 and t-value of -2.04. Landlocked
countries have the disadvantage that their products must pass land transit via other countries to
reach the nearest seaports. These products are subject to custom regulations, roadblocks,
checkpoints, and fares that render trade expensive for landlocked countries.
The real exchange rates of the CEMAC countries is strongly significant with a negative
sign (-1.83, t = -7.21) implying that a depreciation of the currencies of the trading partners
against that of the CEMAC economies reduce exports from the region. Distance is also a
significant variable that accounts for bilateral trade flows between CEMAC countries and other
countries (-2.71, t = -2.49). The variable has the expected negative sign and is significant in the
Hausman-Taylor estimates with elasticity of -2.70.The CEMAC dummy is a significant
determinant of bilateral trade but is negative implying that belonging to CEMAC region does not
increase bilateral trade for member countries and is thus not an advantage for trade. Contrarily,
the EU variable is positive and also significant (0.55, t = 2.50) at 5 percent indicating that
membership of European Union countries having bilateral trade with CEMAC countries has a
positive impact on trade between those countries and those of the CEMAC region. The trade
preference variable is only significant in the random effect model and has a negative sign in both
models. This may probably explain the fact that these trade preference agreements do not
actually provide any trade advantages to CEMAC countries.

The determinants of trade, page 12

121177 Research in Business and Economics Journal

CONCLUSION
We have used panel data on trade between CEMAC economies and between these
economies and a set of randomly selected countries with which they trade bilaterally. The use of
panel estimation methods provides acceptable results because it allows for controlling the
individual heterogeneity of the countries to avoid results that are biased and inconsistent. Based
on the tests for the choice of the appropriate model, we settled on the random effect model which
appears convenient for our sample compared to the often appropriate fixed effect model and the
Hausman-Taylor estimator which was tested to perform better than the fixed effect model. We
estimated the bilateral trade model after correcting for income, country size and distance and we
equally introduced dummy variables to capture effects specific to countries and to the groups.
The results show the relevance of the time invariant variable distance as an important
determinant of bilateral trade between CEMAC countries and between these countries and other
trading partners. The results also show the significant role of variables such as GDP, common
border, common language, real exchange rate and being landlocked or not in explaining bilateral
trade in the CEMAC region. Surprisingly and in contrast to results obtained by Carrere (2002),
regional trade agreements increase trade, the CEMAC dummy is not a significant determinant of
trade between countries of the region. This finding supports the evidence that trade between
these countries is the least among all regional trading groups within African (UNCTAD, 2009).
REFERENCES
Amjadi, Azita and Yeats, Alexander. 1995, Have Transport Costs Contributed to the
Relative Decline of African Exports? Some Preliminary Empirical Evidence, World Bank
Policy Research Working Paper No. 1559. Washington DC
Amjadi, Azita; Reincche, U.; and Yeats Alexander. 1996. Did External Barriers Cause the
Marginalization of Sub Saharan Africa in World Trade? World Bank Discussion Paper
No. 348. Washington DC
Baldwin, Richard, 1997. The Causes of Regionalism, CEPR Discussion Papers 1599, C.E.P.R.
Discussion Papers.
Baltigi, B. H. 2001. Econometric Analysis of Panel Data. 2nded. New York: John Wiley $ Sons.
BayoumiTamin and Eichengreen, Barry 1997. Is Regionalism Simply a Diversion? Evidence
from the Evolution of the EC and EFTA, NBER Chapters, in: Regionalism versus
Multilateral Trade Arrangements, NBER-EASE Volume 6, pp. 141-168 National Bureau
of Economic Research, Inc.
Bergstrand, J. 1985. The Gravity Equation in International Trade: Some Microeconomic
Foundations and Empirical Evidence, The Review of Economics and Statistics 67.
Bergstrand Jeffrey H. 1986. United States-Japanese trade: predictions using selected economic
models, New England Economic Review, Federal Reseave Bank of Boston, issue May,
Pp 26-37.
Caroline Freund and Emanuel Ornelas, 2009. Regional Trade Agreements, CEP Discussion
Papers dp0961, Centre for Economic Performance, LSE.
Celine Carrere, 2003. Revisiting the effects of Regional Trading Agreements on trade flows
with Proper specification of the Gravity Model, Working Papers 200310, CERDI.
Celine Carrere, 2003. African Regional Agreements: Their Impact on Trade With or Without
Currency Unions, Working Papers 200311, CERDI.

The determinants of trade, page 13

121177 Research in Business and Economics Journal

Deardorff, Alan, 1998. Determinants of bilateral trade: Does gravity work in a neoclassical
world?, in Jeffrey Frankel (ed.), The regionalization of the world economy (Chicago:
The University of Chicago Press).
Devarajan, S. and J. De melo. 1987. Evaluating Participation in African Monetary Unions: A
Statistical analysis of the CFA Zones. World Development. 15(4):483-496.
Economic Commission for Africa (ECA). 2010. Assessing Regional Integration in Africa IV:
Enhancing Intra-African Trade, Addis Ababa
Economic Commission for Africa (ECA). 2004. Trade Facilitation. African Trade Centre
Briefing No. 1
Frankel J., Stein E., Wei S. (1995), Trading blocs and the Americas: the natural, theunnatural,
and the supernatural, Journal of Development Economics, 47, 61-95.
Frankel Jeffrey A. and Wei Shang-Jin., 1995.European Integration and the Regionalzation of
World Trade and Currencies: the Economics and Politics, Center for International and
Development Econiomics Research (CIDER) Working Papers C95-053, University of
California at Berkeley.
Ghebreyesus, S. Ghirmay 2007. Impact of Monetary union in Promoting Trade among Franc
Zone Economies: An Empirical Analysis. African Economic and Business Review Vol. 5
No. 2.
Glick, Reuven, and Andrew K. Rose Does a currency Union Affect Trade? European
Economic Review, Vol. 46, No. 6, pp. 1125-51.
Hadjimichael, M. T. and M. Galy. 1997. The CFA Franc Zone and the EMU. WorkingPaper of
the International Monetary Fund. WP156.
Hadjimichael, Michael T and Michel Galy. 1997. The CFA Franc Zone and the EMU. IMF
Working Paper (WP/97/156), Washington D.C.
Hausman Jerry A. 1978. Specification Tests in Econometrics.Econometrica, vol. 46, issue 6, pp
1251-71.
Hausman Jerry. A. and W.E. Taylor. 1981. Panel Data and Unobservable Individual Effects.
Econometrica 49, 1377-1398.
Helpman and Krugman (1985).Market Structure and Foreign Trade, MIT Press
Hsiao, Cheng. 2003. Analysis of Panel Data, 2nd edition, Cambridge University Press, Ch. 4, 1-6.
IMF (International Monetary Fund). 1999. World Economic Outlook. Washington, D.C.
Kortum Samuel S. 1997. Research, Patenting, and Technological Change, Econometrica,
Econometric Society, vol. 65(6), pages 1389-1420, November.
Krugman, P. 1991. The move to free trade zones, in Symposium on Policy Implications of
Trade and Currency Zones, sponsored by the Federal Reserve Bank of Kansas City
Lederman D., zden . (2004), U.S. Trade Preferences: All Are Not Created Equal.
Unpublished, World Bank, Washington DC.
Snow T, Faye M, McArthur J, Sachs J. 2003. Countries Case Studies on Challenges Facing
Landlocked Developing Countries. Background Paper for HDR 2003, UNDP.
McCallum, J. National Borders Matter: CanadaUS Regional Trade Patterns, American
Economics Review, vol.85, No.3 (1995).
Nnanna, O.J. (2006) Economic and Monetary Integration in Africa. A Paper presented at the G24
Summit in Singapore on September 14, 2006.
Pyhnen, Pentti. 1963. A Tentative Model for the Volume of Trade between Countries,
Weltwirtschaftliches Archive, Vol. 90 (1963), pp. 93-100.

The determinants of trade, page 14

121177 Research in Business and Economics Journal

Prewo, Wilfred, 1978. Determinants of the Trade Pattern among OECD Countries from 1958 to
1974, Jahrbucher fur Nationalekonomie und StatistiK 193 341-358.
Rose, A. K. 2000. One Money, One Market: The Effect of Common Currencies on
Trade.Economic Policy, April 30:
Sanso, Marcos, Rogelio Guairan, and Fernando Sanz,1993. Bilateral Trade Flows, the Gravity
Equation, and Functional Form, Review of Economics and Statistics 75 (1993), 266-275.
Schiff, M. 1997. Small is Beautiful: Preferential Trade Agreements and the Impact of
CountrySize, Market Share, and Smuggling. Journal of Economic Integration 12 (3):
359-87.
Smith Pamela J. 1999. Are weak patent rights a barrier to US exports? Department of Applied
Economics, University of Minnesota, St. Paul, Minnesota 55108, USA
Tinbergen, Jan. Shaping the World EconomySuggestions for an International Economic Policy,
The Twentieth Century Fund (1962).
UNTCTAD (United Nations Conference on Trade and Development). 1999. Agreed
Conclusion, Plenary Meeting, 29 October 1999
UNCTAD (United Nations Conference on Trade and Development). 2003. Review of
Maritime Transport.A Report by the UNTCTAD Secretariat.United Nations, New York and
Geneva.
UNCTAD (2009).Trade and Development Report, 2009: Responding to the global crisis.
Climate change mitigation and development.United Nations publications, Sales No.Sales
No.E .09.II .D.16 New York and Geneva.
Wooldridge, J. MJ. 2002. Econometric Analysis of Cross Section and Panel Data. Cambridge,
MA MIT Press.
World Bank . 1998 Global Development Indicators. Washington DC. P. 135.
World Bank. 2003. Private Participation in Infrastructure: Trends in Developing Countries in
1990-2001. Washington D.C.
Yeats, Alexander. 1989. Do Caribbean Exports Pay Higher Freight Costs? World Bank
Discussion Paper No. 348. Washington D.C.
Yeats, J. Alexander; Amjadi, Azita; and Reinche, Ulrich and Ng, Francis. 1996. What
Caused Sub Saharan Africas Marginalization in World Trade? Finance and Development.
Washington D.C.

The determinants of trade, page 15