Monetary Union and Bilateral Trade among CFA Franc Zone Member Countries: An Empirical Analysis

This paper examines the trade promoting effects of monetary union in the context of the CFA franc zone. Using the gravity model as a basis for predicting the volume of trade between countries, the study attempts to estimate the potential for increased trade within the CFA franc zone. The study shows that the CFA countries have experienced relatively low monetary growth, relatively strict budgetary disciplines, and consistently low inflation. However, the results of the study indicate that monetary union in the case of the CFA franc zone did not promote economic integration among member countries in the form of expanded trade. The actual trade among these countries remained small despite the use of common foreign exchange policy and free transferability of resources among these countries.


Introduction
The Franc Zone brings together fourteen African countries, grouped into two economic and monetary unions (WAEMU 1 and EMCCA 2 and the Comoros 3 , linked to France by financial cooperation agreements. It is administered by two central banks, one for each monetary union [1]. The Central Bank of West African  The centralization of foreign exchange reserves. In return for the unlimited convertibility guaranteed by France, the central banks of the CFA Zone are required to deposit at least 65% of their foreign exchange reserves (with the exception of the sums required for their current cash position and those relating to their transactions. with the International Monetary Fund) from the French Treasury, on the account of operations opened on behalf of each of them. This initial provision has been reduced for the BCEAO to 50% since the 2005 reform and to 60% (in July 2007) and 50% (in July 2009) for the BEAC. Since 1975, these assets benefit from a foreign exchange guarantee vis-à-vis the DTS. Devaradjan and Melo [2] [3] argue that the criterion of convertibility is particularly important for developing countries. Indeed, it is widely accepted that full convertibility has a positive international advertising effect and a disciplinary effect on macroeconomic policy; the reduction in the resulting uncertainty is necessary for the massive inflow of foreign investment 4  The peg to the euro did not result in a change in the parities of the CFA franc. On 31 December 1998, the Council of the European Union set the irrevocable conversion rate between the euro and the French franc (1 euro = 6.55957 FRF). This rate automatically determined the value of the euro in the CFA franc. As the CFA franc traded in French francs at the rate of 100 FCFA for a FRF 1, the parity of the CFA franc is now 1 euro = 655,957 FCFA.
The general objective of this study is to capture the impact of monetary union in force in the countries of WAEMU and EMCCA on bilateral trade flows in the context of development of trade (Appendix).
The article is organized as follows. Section 1 surveys economic performance and institutional structure of the CFA area. Section 2 sketches an econometric model. Section 3 presents and analyses the results of estimation.

Economic Performance and Institutional Structure of the CFA Franc Area
Membership in a monetary union could influence trade since it implies a reduction of the uncertainty of the exchange rate [4] [5], transaction costs [6] and simplifies costing and pricing decisions [7]. More generally, if we refer to the functions of money, it appears that a common currency facilitates trade both in its function of unit of account and means of exchange. It is therefore likely that a monetary union will strengthen the flow of trade within the zone. The ex-post analysis of the effects of monetary unions has been the subject of much empirical work. So Rose [8] [9], from a panel data study covering 186 countries in the period  showed that countries adopt a common currency have a trade volume significantly higher (from the order of three) compared with countries each issuing their own currency. Similarly, Frankel and Rose [10] indicate that the effects of a monetary union imply, beyond the growth of the bilateral trade, an increase in the overall opening rate, i.e. a net trade creation. Overall, the existence of a monetary union would have a positive effect on economic growth. Many other studies of currency unions generally emphasize the benefits and costs of using a common currency and common monetary policies [2] [8] [11] [12] [13].
Alesina, Barro and Tenreyro [14] [15] highlight the positive impact of currency unions on bilateral trade. They also highlight the significant influence of currency unions on the correlation between price movements in member countries. On the other hand, the existence of an effect resulting in a stronger correlation of business cycles in currency unions was not found.
Other more recent work focuses on experiences of monetary integration in Africa. Thus, Carrère [16] studies the impact of African regional agreements, with or without currency unions, using a gravity model, applied to panel data for 5 The Council of the European Union, by a decision of 23 November 1998, confirmed that France could "Maintain agreements on exchange rate issues that currently bind it to WAEMU, EC MAC and Comoros" (Article 1 of the Council Decision).

G. Sirpe
the period 1962-1996. The existence of a monetary union seems to increase in a particularly important way the intensity of the bilateral trade between the member countries 6 . In addition, the Carrère study highlights the trade creation and diversion effects inherent in regional trade agreements. Moreover, Tsangarides, Ewenczyk and Hulej [17] establish that the positive effects of membership in a monetary union are not region-specific and affect both Africa and the rest of the world. Moreover, they indicate that monetary unions induce a net creation of trade and a stability of the exchanges. At the same time, they favor the correlation of price movements, without affecting the covariance of outputs.
However, when a country is a member of a monetary union, it loses its monetary independence. It therefore follows that the costs of monetary union will be measured in terms of: -Loss of the exchange rate instrument as an adjustment variable and; -Loss of seigniorage.
What about in the WAEMU and ECCAS zones? However, greater attention was paid to the relative performance of CFA countries in the 1980s and 1990s. GDP growth rates fell below those of 6 The coefficients found are of the order of 3.25 for the ECCAS and 3.13 for the WAEMU (against 0.22 for the ECOWAS).

G. Sirpe
sub-Saharan African countries outside the zone. CFA [20]. This period corresponds to the oil crisis and the deterioration of the terms of trade. Devarajan and Melo [3] note that the lack of adjustment to the external environment has given way to some misalignment of the exchange rate and a lower economic performance. On the other hand, Medhora [21] points out that the poor performance of the 1980s can be attributed in part to poor loan practices and inadequate su- Sub-Saharan African exports accounted for only 2.8% of total exports in the Africa region and 7.1% in 1992. Inter African imports accounted for 5.8% of the total in 1984 and 7.3% in 1992 [24]. The direction of the trade matrix of the CFA zone countries also shows the same thing. Except for three countries (Burkina Faso, Mali and Chad), inter-African imports hardly exceed 10% of total imports [25].
Other studies [26] argue the contrary, highlighting the fact that the inter-African trade potential is enormous. However, these studies still recommend more trade liberalization and accelerated economic cooperation to realize this potential.
In this work, we are interested in only one of the factors, that of the impact of the monetary union in the promotion of bilateral trade between the countries members of the CFA franc zone. In addition to the standard factors affecting trade, the study will try to estimate the determinants of bilateral trade within the countries of the CFA franc zone.

The Econometric Model
There are several techniques and methods for evaluating regional trade 7 . Among these is the gravity model. This is a simple tool and often gives very good results in predicting bilateral trade volumes. Inspired by Newton's theory 8 , the gravity model expresses trade flows between two countries as proportional to their economic weight and inversely proportional to the geographical distance separating them. 7 These include monitoring of macroeconomic indicators such as growth and inflation [27], flows revealed commercial advantages and comparative advantages [28] [29], etc. 8 Newton's theory of gravitation refers to a physical law according to which the gravitational force between two objects is proportional to their respective masses and inversely proportional to the square of the distance between these two objects.

Theoretical Foundations and Empirical Justification of the Gravity Model
The gravity model is a generic name for the family of quantitative models developed by the astronomer Stewart in 1940 [38]. As we can see and as Martinez-Zarzoso and Nowak-Lehmann [39] say, the theoretical support for this model was originally poor and it was not until the mid-1970s that many theoretical developments emerged as the basis the gravity model. If the first attempt at theoretical explanation was given by Anderson [35], who derived the expression of gravity from a model that assumes the differentiation of products, other work consisted not only in strengthening the prior theoretical framework but also to propose other extensions. These include, among others, Bergstrand [36], Helpman [40], Oguledo and MacPhee [37] [41] Deardorff [37], Hummels and Levinsohn [42], Anderson and Wincoop [43]. The work of these authors consisted in taking into account two main determinants that characterize the models of the new theory of trade, namely: economies of scale combined with product differentiation and transport costs.
Empirically, the gravity model has proved to be a particularly useful tool for analyzing bilateral trade between countries [44]. Moreover, since Rose [8] [46], they are quite rare in African countries. Foroutan and Pritchett [21] were the first to apply an improved version of this model to Sub-Saharan African (SSA) countries in order to quantify the level of potential intra-SSA trade and to compare it with the level of current. The results of the work of Foroutan and Pritchett [21] confirm that, in general, the observed intra-ASS trade (current) is weaker than expected. The reasons are, according to these authors of structural order 9 . Other authors like Naudet [47], think rather that the low level of exchanges result from the fact that the countries of the re- 9 For all 19 SSA countries, the average proportion of intra-SSA imports observed was 3.5% compared to 3.6% predicted by the model. For exports the figures were 4.6% and 3.9%.

G. Sirpe
gion do not fully exploit their commercial potential. For the latter trade West Africa to take this example could represent 25% of total trade by 2020.
In order to circumscribe intra-African trade, Elbadawi [48] will also use the traditional gravity model by including in his sample some African regional groupings. He insists particularly on the impact of currency unions on the flow of regional trade. This work is in line with those made by Rose [8] [9]. The latter author has shown that at comparable levels of development the countries belonging to a monetary union would trade 3.3 times (e1, 2 = 3.3) more than those with their own currency. This result will be nuanced by that obtained by Nitsch [49] on the same sample but corrected by multiple variations 10. Longo and Sekkat [50] show that, except for the traditional variables of the gravity model, poor infrastructure, mismanagement of economic policy, and internal political tensions have a negative impact on trade between African countries. Looking at the period 1962-1996 in the context of an improved gravity model, Carrère [15] uses the Hausman-Taylor method to show that African regional trade agreements have generated a significant increase in trade between member countries.
Musila [51] in the case of COMESA (Common Market for Eastern and Southern Africa), ECCAS (Economic Community of Central African States) and ECOWAS (Economic Community of West African States) finds that the intensity of trade creation is stronger in the ECOWAS countries followed by those of COMESA. Avom [52], in the case of CEMAC, also used the Hausman-Taylor estimator [53], as part of an improved gravity model, to reveal that participation in the monetary union did not take place very little effect on regional trade.
Other interesting work has been done [15]  The originality of our work compared to the studies carried out lies in the empirical approach adopted. The study is based, in this case, on the estimation of a gravity model, intended to circumscribe the determinants of bilateral trade between WAEMU and ECCAS member countries and especially to capture the commercial potentialities of within these two groups since the implementation of the two currency unions in the mid-1990s.

Specification of the Model
Empirical analysis is based on an augmented form of the traditional gravity model. The use of this augmented model makes it possible to determine the effect of distance and belonging to the same monetary zone on the intensity of trade between CFA Franc Zone member countries. This distance is usually measured between the economic centers or capitals of the two countries considered. Formally, the gravity equation, in its simplest form, is given by: capita GDP variable has also been introduced to measure the level of development of each country, as it is assumed that as a country develops, it tends to become more specialized, and to trade more [28]. The effect of the monetary union on trade is measured using the method used in the article by Rose [8]  where XIJCOR ij is the flow of exports between countries i, and j at period t, GDP is the total real GDP, GDPT is the real GDP per capita, D ij is the distance between i and j from the CEPII website. UM is a dummy variable that is 1, when i and j share the same currency area. LAND is a dummy variable that is 1, if i and j share a border. OIL is a dummy that takes into account the oil producing countries, COTTON is a dummy that takes into account cotton countries, ε ij is the error term.

Presentation and Analysis of Results of Estimation
The sample taken into account includes all eight (8)  takes into account the quality as well as the availability of data. Indeed, the data relating to imports are generally reported with more precision, with regard to the duties and taxes to be collected [15]. In addition, the available information does not distinguish between cases or exports to a given country are zero cases where they are not reported. As a result, the summation of imports and exports could lead to biased results. The data for this study (annual exports and imports) are in constant millions of US dollars from 1988 and come from the IMF's "Direction of Trade" and the World Bank's "African Development Indicators". Data for real and per capita GDP cover the period from 1990 to 2006, which is 17 years. They are extracted from the World Bank Economic and Social Data Bank (ESDP) and AfDB African Development Reports. GDP in constant millions of US dollars in 1988 is at market price and GDP per capita is the standard of purchasing power. Data on the distance (in kilometers) between the coexchangers comes from the CEPII website (Center for Prospective Studies and International Information) ( Table A1 in Appendix).
One of the characteristics of trade between African countries is the scarcity of data for a large number of bilateral relations. Thus, the value of trade between two countries can be zero. With a specification in logarithm, such an observation will become indeterminate. To solve this problem, there are two possibilities. Zero values can be eliminated if their percentage in the observations is small and subsequently used the ordinary least squares (OLS) method for estimating the model. If the proportion of nil observations is high, the use of OLS leads to biased results. This can be verified by averaging the error term. It is common in this case to use a non-linear estimation technique such as Tobit, which explicitly recognizes the existence of the null values of the dependent variable and treats G. Sirpe them as non-registered trade flows while normalizing the distribution. In addition, if estimating from a Tobit is not a problem, the measurement of the model's performance is not the subject of a consensus. The measure generally used in the literature, Pseudo-R2, comes in several forms (Veall and Zimmermann, 1994). We use in this work the Mac Fadden Pseudo-R2 which is the most widespread. The results of the estimates for the two sub-periods are summarized in Tables.
The results of our estimates are presented in Table 1 and Table 2.

1990-1994 Period
This phase corresponds to the period during which most of the CFA zone countries have been applying (for some years) structural adjustment programs following macroeconomic imbalances. The results obtained above (Table 1) show that the estimates made over the period 1990-1994 are quite robust.
The explanatory power of the model is 94.3%, and the model is globally significant. All variables except GDP per capita are significantly different from zero. The estimates obtained are consistent with the empirical results obtained in previous work. The remoteness of two countries reduces trade by 0.73%, while increases in real GDP and GDP per capita intensify them. The sharing of a common border is also one of the determining factors that explains the increase in bilateral trade. GDP and monetary union dummy variables (CEMAC, UEMOA) and the common border (Land) contribute most to the increase in export flows. The Countries in the Franc Zone with a common border trade three times more than other countries.
In addition, the results show that bilateral trade increases in the WAEMU zone (ex CEAO) by 22.27 times and in the CEMAC zone (formerly UDEAC) by 3.28 times. In other words, trade in the WAEMU zone is 6.78 times more intense than in the CEMAC zone (see Table A2 in Appendix). The analysis of IMF statistics confirms the results obtained. During this sub-period, the share of intra-zone trade in UEMOA's total trade is greater than in the CEMAC zone. This is between 8% and 11% in the UEMOA, while in the CEMAC zone, it fluctuates between 0.90% and 3.51%. In view of these results, it can be said that the objective of CEAO, which was, inter alia, to promote trade between these countries in response to market problems, has been achieved.
Oil producing countries trade more with each other than cotton producers. In the 1990-1993 sub-period, trade between oil-producing countries increased by 1.38%, while that of cotton producers increased by only 0.87%. This can be explained by the fall in export prices of agricultural commodities (especially cotton) during this period, following the deterioration of the terms of trade and the competitive devaluation policies of neighboring countries not belonging to the region, not at the CFA zone.   These results conceal the weakness of intra-zone trade. According to the IMF, intra-WAEMU trade is still hampered by significant non-tariff barriers (national standards, quantitative restrictions on certain imports, treatment discrimination of national and regional products, etc.). As for the CEMAC zone, the preferential tariff adopted in 1994 on intra-Community trade is applied unequally.

Period 1994-2006
As can be seen in the sub-periods under consideration, the oil producing countries trade more with each other than the cotton producing countries, certainly because most of these countries are landlocked.

Conclusions
The purpose of this paper was to analyze the impact of currency unions on bilateral trade.
From an overall point of view, the main expected effects of the adoption of a common currency and monetary policy fall into two main areas. On the one hand, the reduction in transaction costs related to currency differences allows the expansion of trade and the growth of activity. On the other hand, the strengthening of the credibility of the Monetary Authority resulting from its regional status contributes to the stabilization of the macroeconomic framework.
On the trade side, most ex-ante analyzes have highlighted the weakness of intra-regional trade potential, mainly related to structural barriers. Based on gravitational models, this study has attempted to highlight the determinants of bilateral trade and, in particular, the impact of the existing monetary union, in this case UEMOA and CEMAC.
It thus appeared that the geographical and structural factors, but also the membership of the monetary union, determine the intensity of bilateral trade flows within these two unions. In addition, the impact of the common currency is reflected in significant trade creation. However, the potential for intra-regional trade could be strengthened by putting more emphasis on the structural reforms needed to diversify economies and thus promote complementarities, develop infrastructure and enhance convergence of macroeconomic performance and policies.
In a panel of bilateral trade between CEMAC and WAEMU countries over the period analysed, it appears that economic size, geographical and political factors are the major drivers of bilateral trade between UEMOA and CEMAC members countries. More importantly, our results show that a generalized model which includes all the possible dimension of trade effects (both the main and interaction effects) is more appropriate for the analysis of bilateral trade in UEMOA and CEMAC. Therefore, ignoring any of these effects gives misleading inferences as suggested by the results of the analysis carried out.
In terms of policy implications, we recommend that concerted efforts should be geared at increasing the productive capacity and value addition in countries in the ECOWAS region. This will not only promote trade and output but result into more employment opportunities, increased revenue and attract the much needed capital inflow into the region. Also, despite the fact that infrastructure is inevitable for growth in intra-regional trade; it is currently insufficient and dilapidating in nature in ECOWAS as a whole. This calls for adequate attention.
Giving the importance of political stability to trade, we recommend that all stakeholders should strive for the prevention and prompt resolution of conflict and political instability in the region. Finally, analysis of bilateral trade in UEMOA and CEMAC should always take cognizance of all the dimensions of the panel, especially since countries in the region differ or change over time and space.

Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper.