Abundance of natural resources and its impact on regional integration: Empirical evidence

Transcription

1 Abundance of natural resources and its impact on regional integration: Empirical evidence Louai Souar 1 Mars 2017 Abstract This paper aims to explore empirically the regional integration gains in terms of trade and diversion in natural resource-rich regions. It is built essentially on Venables (2011) theoretical predications which show that the gains from regional integration are unevenly distributed between resource-rich and resource-poor countries. We used gravity model approach to explore trade creation and diversion for 13 regional trade agreements characterised by abundance of natural resource for the period Our methodology includes three levels of analysis; aggregate effects of regional integrations, gains distribution between resource-rich and resource-poor countries, impact of natural resources type on the patterns of trade creation and diversion. Estimation results are not always compatible with the theoretical predictions of Venables. We conclude that Free Trade Agreement type and remote regional agreements are the most compatible with theory of Venables. We can conclude also that resource-poor and diversified countries in most studied regional agreements made gains in terms of intra-regional trade creation; consistent with the viewpoint of Venables. For resource-rich countries, we observed that oil-rich countries are the most vulnerable to import trade diversion, then the mineral-rich and a lesser extent the agricultural-rich countries. In addition, the regional integrations induced resource-rich countries to create intra-regional trade in non-natural resource sectors with partners. But note that the mineral-rich countries and to a lesser extent the agricultural-rich countries are able to create a little intra-regional trade in natural resource sectors also especially in remote regions, while oil countries remain more oriented to global marets. Keyword: Regional integration; Natural resources; Gravity model. JEL Classification: F10; F11. 1 Introduction The present wor taes an empirical loo at the impact of the uneven allocation of natural resources on the distribution of regional integration gains between nations. The question was raised by Michel Fouquin,Rolf J. Langhammer and Rainer Schweicert (2006) in their paper entitled Natural resource abundance and its impact on regional integration: Curse or blessing?. 2 They bring forward several observations with regard to the influence of natural resources abundance on regional integration. One of them refers to post-war evidence that suggests that resource-rich countries have neither been driving forces for establishing regional integration schemes nor -once they were members of such schemespush factors for deeper integration. They also note that the jury is still out concerning the role of resource abundance as either a stumbling bloc or stepping stone for regional integration. Conditions 1 PhD student, LAREFI,University of Bordeaux, louai.souar@u-bordeaux.fr 2 Their paper presented in ELSNIT /Fundaçao Getulio Vargas Conference in San Paulo. 1

2 have changed since the first, inward looing import substitution stage of regional integration in Latin America. While at the time commodity prices were mostly flat or declining, at the present, in conditions of more outward looing regionalism, there is a high demand for commodities. And, with the process of reforms, ownership structures have changed in favor of foreign participation through foreign direct investment. The relationship between resource abundance and neighborhood relations (a somewhat broader term than integration) needs to tae into account socio-economic aspects such as the allocation and enforcement of property rights on natural resources, and the division of labor between the state and the private sector in collecting and spending revenues from resource extraction. 1 The issues paper notes that a distinction probably has to be drawn between oil-exporting countries, mineral ore producers and agricultural resource-based economies. The criteria of distinction are the processing potential, the degree of dualism, the spread between intra-regional and extra-regional demand, and the degree of exposure to exogenous shocs. Later, the question was discussed in the World Trade Report (2010) carried out by World Trade Organization (WTO). The report is entitled Trade in natural resources, and it taes a closer loo at the question in the section D (Pages ) under the title National resource abundance and regional integration. According to the report, there is two-way relationship exists between natural resources and regional integration. Regional integration affects resource-rich and resource scarce countries differently. These effects, in turn, shape the incentives for these countries to engage in regional integration. The integration of two resource-abundant countries with low tariffs and non-tariff barriers on natural resources, and similar production structures with limited manufacturing activity, is liely to lead to limited trade creation and potentially large trade diversion effects. On the other hand, regional integration may enable a resource-abundant country to diversify its production and export structure by relaxing the constraints it faces in developing a manufacturing sector. Regional integration may assuage concerns about overexploitation of natural resources and other potential negative consequences of international trade on the environment as provisions on natural resource management are sometimes included in regional and bilateral free trade agreements. However, this paper is built essentially on Venables (2011) theoretical predications which show that the gains from regional integration are unevenly distributed between resource-rich and resource-poor countries. He concludes that if the preferential trade agreement is signed by a natural resource abundant country and a natural resource poor country with a small but developing manufacturing sector, then the introduction of tariff preferences will probably lead to some trade creation in the resource poor country, as it will be able to import more natural resources from the resource rich country and extend its regional exports and reach a higher level of economic growth. On the other hand, the resource rich country may suffer from a significant amount of trade diversion as it substitutes imports from the relatively more efficient rest of the world towards the regional partner. Empirically, Céline Carrère, Julien Gourdon and Marcelo Olarreaga (2012) explore the extent to which Middle East and North Africa (MENA) different integration schemes have led to trade creation and trade diversion. Results suggest that within Pan Arab Free Trade Agreement (PAFTA), there is significant trade creation for resource poor countries associated with regional integration, and no evidence of trade diversion. In resource rich countries, however, there is evidence of pure trade diversion. In this paper, we aim to explore this question empirically for 13 regional trade agreements at the world level. The regions that we have selected are characterized principally by the uneven distribution of natural resources between member countries, the diversity of natural resources available in the region, the diversity of the regional trade agreements in terms of degree of processing, the low level of intra-regional trade and the high dependency on the outside world. 1 The issues paper notes that a distinction probably has to be drawn between oil-exporting countries, mineral ore producers and agricultural resource-based economies. The criteria of distinction are the processing potential, the degree of dualism, the spread between intra-regional and extra-regional demand, and the degree of exposure to exogenous shocs. 2

3 As a most commonly used analytical framewor, the gravity model has been applied in a large number of empirical studies. Among them, one of the ey issues is to analyse the specific effects of trade policies by introducing dummy variables to indicate the existence of a regional trade agreement between countries. This methodology can be extended to estimate trade creation and trade diversion and thus maes an important contribution to the regionalism debate. Following the methodology proposed by the recent literature, we apply the Vinerian specification of integration effects with an extension of three different sets of RTA dummy variables representing trade creation and diversion effects in terms of export and import, as proposed by Endoh (1999), Soloaga and Winters (2001), Carrère (2006), Magee (2008) and Martínez-Zarzoso et al. (2009). With the aim of providing more insights about how the gains from regional integration will be distributed depending on the spatial allocation of natural resources between countries, we put forward augmented gravity models inspired from Céline Carrère, Julien Gourdon and Marcelo Olarreaga (2012). This study is aimed at providing additional insights in regional integration debate through the following contributions. We apply three steps of analysis in order to investigate how the gains of regional integration in terms of trade creation and trade diversion between countries for several regional trade agreements in resource-rich regions. The first step aims to estimate a basic gravity model to explore the aggregate effects for several regional integrations in resources-rich countries. Then, we aim to provide further analytical specifications in order to estimate how the effects of regional integration will be diffused across countries depending on the abundance of natural resources. The third gives more insights about how the type of natural resources plays a role in the patterns of trade creation and diversion among the integrated countries. In addition, an analytical comparison is carried out between 9 regional trade agreements in order to chec the theoretical predictions of Venables. At the technical level, augmented gravity models are estimated using data up to date with pooled regression techniques that allow to control for all bilateral factors that influence bilateral trade and are time invariant (unobserved heterogeneity), as well as for the so-called multilateral resistance factors (the effect of relative prices with respect to all trading partners). Finely, as suggested by Head and Mayer (2014), we apply multinomial PPML estimator proposed by Silva and Tenreyro (2006) to solve the zero trade issue and the presence of heterosedasticity. The rest of the paper is structured as follows. We start by providing a literature review and the contributions that form the analytical framewor of the impact of the uneven distribution of natural resource on the gains of regional integration. Subsequently, we present the characteristics of the regional integrations in resource-rich regions that need to be put at the centre of analysis. Then, we introduce our empirical models for the three steps of analysis followed by data and econometric issues. Finally, we discuss our empirical results for each regional trade agreement and carried out an analytical comparison between them. 2 The analytical framewor of the impact of natural resources on regional integration The analysis of the effects of regional agreements in terms of trade creation and trade diversion is different in the natural resource-rich regions. This is due to the specificities of these regions in terms of several aspects, lie the low tariff on trade in natural resources, and the diversity of natural resources, the existence of some economies that rely heavily on natural resources while other are resource-poor economies but have an advantage in the manufacturing sector. Based on the features of the concerned regions, there are three different cases that could be occurred between two developing countries that engage into a regional trade agreement in rich-resource region: - Case (1): Two resource rich countries specialized in different natural resources, - Case (2): Two resource rich countries specialized in the same natural resource, - Case (3): A resource-rich country and a resource-poor country. 3

4 We will discuss the theoretical contributions that could explain the potential trade for the three cases above. We aim to find out the probability of trade between the two countries, and the nature of this expansion of trade in terms of trade creation or trade diversion. 2.1 The case of a regional bloc composed of two resource rich countries specialized in different natural resources In a region characterized by abundance of natural resources, we may find countries specialized in production and export of different natural resources such as oil, minerals or agricultural raw materials. On one hand, according to the classical theory of international trade and more precisely the Hecscher - Ohlin (H - O) model, these countries and with the elimination of internal tariff barriers between them, they will have trade creation. The countries with different factor endowments are expected to trade with each other the most, and that the prices of the factors of production will tend to converge between the two countries after trade or integration occurs. On the other hand, Meade (1955) and Hillmann (1957) have concluded that the higher the initial tariff rates between countries entering a customs union, the larger are the expected gains of economic integration between these countries. In this case, the tariffs imposed on resource commodities are unliely to constitute a major barrier initially to trade within the region which will be integrated (Fouquin et al., 2006). In addition, resource abundant countries especially oil and minerals-rich countries are world marets oriented, so they are targeting principally global marets because of the rising demand in industrialized countries and large emerging marets rather than developing countries marets that are created after the regional integration. We conclude that establishing trade agreement between rich countries in different natural resources does not contribute to trade creation or trade diversion in natural resources commodities between them. But in fact, this conclusion does not apply to all inds of natural resources. The imposed tariffs on agricultural products actually are still high. WTO rules allow countries to impose such tariffs on agricultural imports or exports in order to protect the local production. Thus, if the regional trade agreement covers agricultural commodities, we expect some trade creation for agricultural-rich countries after the elimination of intern tariffs, where the preferential agreement will enhance agricultural countries (especially small countries) to export to other rich partners, and we expect trade diversion for other resources-rich countries, where they will replace their agricultural imports from the rest of world by agricultural imports from the regional partners. As well, as argued by Abdel Jaber (1971), the category of primary products itself is too large, and once disaggregated, so potential benefits may arise between countries that are specialized to export natural commodities. However, agricultural commodities are considered the most diversified of natural resources and therefore the possibility of trade creation with other rich countries is larger. 2.2 The case of a regional bloc composed of two resource rich countries specialized in the same natural resources We note with the features of regions that will be studied, there are countries may share the same abundant natural resources. We recall that Viner (1950) and Lipsey (1960) suggested that more benefits of economic integration will accrue to competitive countries (countries producing similar products) than to complementary countries (countries producing dissimilar products). This analysis tends to be in favor of countries that specialize in exports of the same primary products, therefore these countries may be said to be competitive in the Vinerian way. But this is not true, the fact that these countries do not need to import goods that are already produced domestically, therefore they will have few incentives to trade with each other, as there is very little product differentiation in the same resource commodity. Beside, most of resources-rich countries target developed counties marets. And that hinders the benefits of 4

5 economic integration between developing countries that are specialized in same natural resources (El- Naggar 1964). Thus, trade creation and trade diversion effects in natural resource commodities are liely to be negligible between countries specialized in the same natural resource. The regional integration can contribute to create trade in agricultural product between agricultural countries as a result of eliminating or reducing the high tariffs on agricultural products on the one hand,, because of the diversity of agricultural products on the other hand. In the same context, Lipsey (1960) concluded that a customs union will more liely produce welfare gains to a country the higher is the proportion of trade with its partner in the union, and the lower the proportion with the rest of the world. In this case, since the intra-regional trade between resources rich countries specialized in the production of the same natural resource is low, the implementation of a regional agreement would not have a major impact in the development of intra-regional trade. 2.3 The case of a regional bloc composed of a resource-rich country and a resourcepoor country We begin to review the traditional view on the effects of regional integration between two countries with different production structures; such as a country produces a commodity with a high resourceintensive another country produces an industrial commodity. There are two basic views on this subject according to the classical theory of economic integration. On one hand, Viner (1950), Meyer (1956) and Lipsey (1960) claimed that the gains will arise between countries if they are competitive countries in producing similar products than to complementary countries which are producing dissimilar products. On the other hand, other developments concluded that integration between potentially complementary countries may be welfare increasing (Meade s, 1955), instead of the general proposition that integration should be between competitive countries. More recently, papers lie Heimenz and Langhammer (1990), Inotai (1991) and Shams (2003) have continued to argue that complementarity or dissimilarity of economic structures would be better to the case of economic integration among developing countries. Greenway and Milner (1990) argue that a union among similar (competitive) countries assumes that trade will come from intra-industry specialization. Such trade expansion has been evident in the case of developed industrialized countries, where maret size and incomes may support such specialization. However, this is obviously less possible in the case of smaller poorer marets that characterize developing country marets. We note from previous classic contributions that the impact of regional integration in this case is ambiguous. However, the issue has been addressed recently by Venables (2009). He presents a theoretical model to investigate how the gains from regional integration in resource-rich regions are distributed between countries. He concluded that if the preferential trade agreement is signed by a natural resource-rich country and a natural resource-poor country with a small but developing manufacturing sector, then the gains from regional integration are unevenly distributed between them. Regional integration implies a reduction in tariffs on imports. This enables resource-poor country to extend their exports to the resource-rich partner of manufacturing goods, as it will benefit from privileged access to marets inside the agreement and will be able to import more natural resources from the resource rich country. Hence, regional integration will probably lead to some trade creation in the resource-poor country and to earn more foreign exchange. This extra foreign exchange accruing to resource-poor country raises income, thereby bidding up the prices of these regionally traded goods, increasing wages, creating a terms-of-trade gain, importing more capital goods from the rest of the world and reach a higher level of economic growth. On the other hand, resource-rich economies lose (or at best experience very modest gains) from regional integration. First, a terms-of-trade gain for the resource-poor country is necessarily a termsof-trade loss for the resource-rich economy. In addition, the resource-rich country may suffer from a significant amount of trade diversion as it substitutes imports from the relatively more efficient rest of the world towards the regional partner. 5

6 In contrast, multilateral trade liberalization will be beneficial for the resource-rich country as lower tariffs on more cost-efficient imports from non-member countries will entail trade creation, but no trade diversion. Moreover, external trade liberalization implies a reduction in tariffs on imports from the rest of the world. Since intra-regional trade taes the form of exports of manufactured goods from the resource-poor country to the resource-rich country, this reduction in the price of imports from the rest of the world is a terms-of-trade gain for the resource-rich economy. Hence, while trade is a way for the resource-rich economy to relax the constraint causing diminishing returns in the use of its resource revenues, these gains come from non-preferential opening. In sum, there appears to be a two-way relationship between natural resources and regional integration. The analysis points to the potential for conflicting interests between resource-poor countries seeing preferential regional integration, and resource-rich countries seeing nonpreferential trade opening. However, this effect is often contingent upon the location of the countries concerned and the ind of natural resource in which they are abundant. Hence, relative resource abundance in these different contexts, in turn, may shape the incentives for countries to engage in regional integration. This model points to the importance of regional integration as a way of spreading the benefits of unevenly distributed resource wealth. Céline Carrère, Julien Gourdon and Marcelo Olarreaga (2012) checed the Venables (2009) theoretical predictions empirically for different integration schemes in MENA region. They explore the extent to which Middle East and North Africa (MENA) different integration schemes have led to trade creation and trade diversion. Results suggest that there is trade creation in most agreements, and that trade diversion may only be a problem in the Pan-Arab Free Trade Area (PAFTA), in particular when considering non-oil imports. As predicted by Venables (2011) trade diversion seems to be concentrated in resource rich importers. These are generally countries that export only a few products and with a highly concentrated export bundle. Interestingly, these countries have also significantly increased their exports of non-oil goods to resource poor countries, but these increases were not accompanied by trade diversion in resource poor countries. Hence, while further intraregional trade integration is an important avenue for enhancing diversification of resource-poor MENA countries, resource-rich countries have no strong incentive for further preferential regional integration from a purely economic standpoint, and this may explain their relative reluctance to engage in this type of scheme. 3 Principals characteristics of natural resources and regional trade agreements We provide in the section the principal aspects that characterize the relationship between natural resources and regional integration. First, natural resources are distributed unevenly between countries. Furthermore, there are different types of natural resources in the same regional trade agreement. Thirdly, we can see different levels of regional integration in the same region. Finally, regional trade agreements established in resource-rich regions are characterized by the low level of intra-regional trade and by the dependency on the outside world. All these characteristics mae it natural to thin that regional integration in the context of abundance of natural resources in the region might be particularly valuable. 3.1 Uneven spatial allocation of natural resources between member countries Natural resources tend to be concentrated in relatively few locations around the world. They are concentrated in a small number of countries in each region, while others have limited domestic supplies. This has implications for the economic structure of economies, for the trade policies between them and with the outside world. Therefore, depending on the abundance of natural resources, countries in the selected regional trade agreements can be classified into three main categories. The first group is Resource-rich countries; it includes countries that have high share of natural resource exports of total exports; including agricultural raw materials, minerals and fuels. The second group is Resource-poor countries, it includes countries that have a few share of natural resources of total 6

7 exports. Their exports are dominated principally by food products and manufactured products. 1 The final group is Diversified countries; it includes countries that export both resource-intensive products and non-resource products. 2 Table B in Appendix provides the classification of countries for 13 selected regional trade agreements according the abundance of natural resources. Columns 3 and 4 in Table B indicate the contribution of natural resource exports and non-natural resources exports of total exports for each country respectively; hence this percentage is used principally to classify countries between Resource-rich country Resource-poor country and Diversified country in each selected regional trade agreement Diversity of natural resources available Another feature that characterizes the regional integration in resource-rich regions is the diversity of natural resources. We distinguish between three types of natural resource sectors: minerals, fuel, agricultural raw materials as opposed to the non-resource sectors: manufacturing and food. In Table B in Appendix, columns from 5 to 9 indicate the share of sectors for each country in the selected regional trade agreements. 5 According to particular criteria also, we can conclude the sectoral specializations through the contribution of sectors for each country in each selected regional trade agreement (see column 10 in Table B). 6 We note that there are countries that depend entirely on the export of one type of natural resources, the share of one resource sector exceeds 60 percent of total exports, such as Venezuela in CAN, MERCOSUR and LAIA agreements (the average share of fuel sector for the period is 86.8% of total exports), Botswana in SADC and SACU agreements (the average share of minerals sector for the period is 87.5% of total exports) and Russia Federation in CIS (the average share of fuel sector for the period is 66.6% of total exports).while other countries are rich in various types of natural resources, they specialize in exporting more than one type of natural resource commodities, such as Burina Faso (the average share of agricultural raw materials sector for the period is 38% of total exports and the average share of minerals sector for the same 1 According to the World trade rapport (2010), food is not classified as natural resource sector for many reasons. To begin with, their production requires other natural resources as inputs, particularly land and water but also various types of fertilizer. More importantly, agricultural products are cultivated rather than extracted from the natural environment. 2 The classification between Resource-rich countries and Resource-poor countries belongs mainly to the model of Venables (2011). The same classification has been used by Carère and al (2012) in order to study the impact of the uneven distribution of natural resources between countries in terms of trade creation and trade diversion in the region of MENA. In our study, we have a worldwide sample of regional trade agreements in resource rich regions. After the analysis of the economic structures of member countries, we conclude that there are some countries that don t have extreme specialization in a sector, whether in the export of natural resource commodities or in the export of non-resource goods, they possess natural resources from one hand and they have developed their manufacturing sectors on the other hand. Therefore, we have added third category which is Diversified countries in order to classify these countries. 3 According to the literature, the criterion is determined according to the objective of study, and thus there is no approved method to classify between countries in terms of the abundance of natural resources. However, we follow a criterion of classification which has been inspired from several references, such as: Fouquin et al (2006) and classification of UNCTAD. Criteria of classification : - Country is rich in natural resources if the share of natural resource exports exceeds 60 percent (Resource exports 60%) of total exports including agricultural raw materials, minerals and fuels. - Country is poor in natural resources if the share of non-natural resource exports exceeds 60 percent (Non-resource exports 60%) of total exports. - Country is diversified if the shares of natural resource exports vary from 40 percent to 50 percent (40% Resource exports 50%) of total exports. 4 SITC(Standard International Trade Classification) under Revision 3 codes for : -Resource exports = Agricultural raw materials (2 - ( )) + Minerals ( ) + Fuels (3) -Non Resource Exports = Food ( ) + Manufactured ( (667+68)) 5 We inspired this distinction between the types of natural resources basically from Fouquin et al (2006). He suggested this distinction according to the relation between the type of natural resources and economic growth. Data analysis shows that there is a strong correlation between the ind of international specialization and growth. The group of countries which show the strongest specialization in manufacturing reach the highest unweighted average annual GDP per head growth over the period with 2.9%, followed by countries specialized in agriculture with 1.50%, then by those specialized in minerals with 1.47% and finally by oil-rich countries with the lowest average growth of 0.76%. 6 Criteria of sectoral specialization: -For resource-rich country: the country is specialized in a resource sector, if its share exceeds 30 ( 30%) of total exports. -For resource-poor country: the country is specialized in a non-resource sector, if its share exceeds 30percent ( 30%) of total exports. -For diversified country: the country is diversified by the sector that exceeds 20 ( 20%) of total exports, the diversified country is characterized by the prominence of several sectors of both resource and non-resource sectors. 7

8 period is 36.9% of total exports), and Bahrain in GCC and PAFTA agreements (the average share of fuel sector for the period is 38.9% of total exports and the average share of minerals sector for the same period is 31.4% of total exports). Concerning resource-poor countries, there are countries that are heavily specialized in one nonresource sector, lie Uraine in CIS agreement (the average share of manufacturing sector for the period is 64.3% of total exports ), Jordan in PAFTA (the average share of manufacturing sector for the period is 73.1% of total exports ) and Paraguay in MERCOSUR and LAIA (the average share of food sector for the period is 70.9% of total exports) and Malawi in SADC and COMESA (the average share of food sector for the period is 79% of total exports). While there are resource-poor countries are specialized in several non-resource sectors, for example, Brazil in MERCOSUR and LAIA agreements ( the average share of manufacturing sector for the period is 40.6% of total exports and the average share of food sector for the same period is 30.5% of total exports), Senegal in ECOWAS and WAEMU agreements (the average share of manufacturing sector for the period is 32.2% of total exports and the average share of food sector for the same period 32.2% of total exports) and Republic of Moldova in CIS agreement (the average share of manufacturing sector for the period is 54.9% of total exports and the average share of food sector for the same period is 40% of total exports). For diversified countries, analysis of the shares of sectors indicate that there are some countries that are specialized in two sectors, one resource sector and one non-resource sector, for example Sierra Leone in ECOWAS agreement (the average share of minerals sector for the period is 47.5% of total exports and the average share of food sector for the same period 33% of total exports) and Uzbeistan in CIS agreement (the average share of manufacturing sector for the period is 37% of total exports and the average share of fuel sector for the same period 20.4% of total exports). On the other hand, there are some diversified countries that are specialized in more than two sectors, three or four sectors, such as Syria Arab Republic in PAFTA agreement (the average share of manufacturing sector for the period is 30.6% of total exports, the average share of fuel sector for the same period is 37.8% of total exports and the average share of food sector for the same period is 27.4% of total exports) and Zimbabwe in SADC and COMESA agreements (the average share of manufacturing sector for the period is 31% of total exports, the average share of minerals sector for the same period 26.6% of total exports and the average share of food sector for the same period 22.5% of total exports). 3.3 Diversity of regional trade agreements The third regional characteristic is the diversity of regional agreements in terms of the degree of processing. Table A in Appendix provides a list of the selected regional trade agreements; it includes Member countries, type of agreement and date of entry into force. The regional agreements that we have selected are covering most rich regions in natural resources in the world. We note that the type of agreement varies from Partial Scope Agreement (PSA), and Free Trade Agreements (FTA), and Customs Union (CU) to Economic Integration Agreement (EIA). We note also that each region may contain several different regional conventions, lie CAN, MERCOSUR and LAIA in South America, ECOWAS and WAEMU in West Africa and SADC and SACU in South Africa. Furthermore, some countries participate in several regional agreements, lie countries of Gulf; they are members in both PAFTA and GCC, and some African countries, lie Democratic Republic of the Congo that participates in COMESA and SADC, and Niger which is member in WAEMU and ECOWAS and etc. We suppose that the type of the regional trade agreement in such natural resource-rich regions has an impact on the distribution of gains of the regional integration between differentiated countries in terms of the abundance of natural resources and the sectorial specialization. 8

9 3.4 Low intra-regional trade and high dependency on outside world The fourth regional feature concerns the trade structure for the selected regional trade agreements. We note that the contribution of total intra-regional exports of total exports in the selected regional trade agreements is low (see Table C in Appendix column 2), as we can see that the contribution of intraregional exports for selected regional trade agreements ranging from 2 to 18 percent of total exports. Nevertheless it slightly elevated in some conventions such as CIS (17.4%), EAC (18.7%), LAIA (16.3%), MERCOSUR (13.6%) and SADC (15.4%), this due to the presence of some large developing countries in these conventions, such as Russia in CIS, Brazil in LAIA and in MERCOSUR and South Africa in SADC. Meanwhile, we see that the contribution of extra-regional exports for the selected regional trade agreements range from 80% to 97% of total exports (Table C column 3). On the other hand, we note that the contribution of natural resource exports of total exports in most cases is high (see Table C in Appendix column 4). In addition, most of natural resource exports are oriented towards the rest of the world as we can see in Table C columns 5 and 6. However, the contribution of non-resource exports is higher in some agreements despite that they are directed mostly to outside of the region. These agreements are: LAIA and MERCOSUR agreements, where they include some large developing countries lie Brazil, Argentina, Mexico, Paraguay and Uruguay, and SACU agreement, where it includes South Africa as a large developing country and there is only one resource-rich country which is Botswana. With regard to the structure of imports for the regional trade agreements, imports are characterized that they are imported heavily from outside of region (see Table D in Appendix, column 3). As well, they are dominated largely by non-resource imports (see Table D, column 7). On the other hand, the contribution of intra-regional imports of total imports is low (see Table D, column 2), even the nonresource imports does not account for a high share (see Table D, column 8). For imports in the natural resource sectors, we note that the low contribution of intra-regional resource imports in all regional agreements is low (see Table D, columns 5 & 6). We conclude from above that the level of intra-regional trade in selected regional trade agreements is low. This reflects the fact that the regional integrations in resource-rich areas do not appear as major exports marets. Therefore, the trade structure of the regional integrations in resource-rich regions is rather world maret oriented. In order to study the relation between the abundance of natural resource and regional integration, we selected a world wide sample of regional trade agreements that are characterized mainly by the abundance of natural resources: CAN (Andean Community) LAIA (Latin American Integration Association), MERCOSUR (Southern Common Maret), ECOWAS (Economic Community of West African States), WAEMU (West African Economic and Monetary Union), SADC (Southern African Development Community), SACU (Southern African Customs Union), COMESA (Common Maret for Eastern and Southern Africa), EAC (East African Community), CEMAC (Economic and Monetary Community of Central Africa), CIS (Commonwealth of Independent States), PAFTA (Pan-Arab Free Trade Area), GCC (Cooperation Council for the Arab States of the Gulf). 9

10 4 Empirical Models The gravity model is one of the common methods used to estimate trade creation and trade diversion in the empirical studies. We will introduce in the first place a generalized gravity model with an extension of three different sets of regional trade agreement (RTA) dummy variables representing trade creation and diversion effects in terms of export and import. We aim to see how gravity models are built and the interpretation of RTA dummy variables. Then, we will provide our three gravity models in order to investigate how the gains of regional integration in terms of trade creation and trade diversion between countries for several regional trade agreements in resource-rich regions. The first model aims to estimate a basic gravity model that explores the aggregate effects of regional integration in terms of trade creation and diversion for each regional trade agreement. Second model is set to estimate how the effects of regional integration will be diffused across countries depending on the abundance of natural resources. Finally, we put forward third gravity model to estimate if the extent impact of the type of natural resources on the patterns of trade creation and diversion among the integrated countries. 4.1 Gravity models with three dummy variables In a generalized gravity model, trade between country i and country j is positively related to the size of the economies and negatively related to the distance, a proxy for transportation costs, between them. In addition a number of bilateral factors that foster or impede trade are usually included as explanatory variables. Hence, adding the time dimension it can be specified as: X ijt = (a + gdp it + gdp jt + pop it + pop jt + Dist ij + F ij + RTA ijt + ε ijt ) (1) where X ijt is trade flows or exports from country i to country j in year t. gdp it is GDP for country i, year t and gdp jt is GDP for partner j in year t. pop it and pop jt denotes the respective population size for country i in year t and partner j in year t. Dist ij denotes geographical distance between the two countries. Finally F ijt denotes other bilateral factors such as a common border, colonial relationship or a common language that can foster trade. To analyze the effects of regional trade agreements, a dummy variable RTA ijt is introduced to indicate the existence of a regional trade agreement between countries i and j. However, this methodology can be extended to estimate trade creation and trade diversion and thus maes an important contribution to the regionalism debate. The first extended use of regional dummy to capture an RTA s effect on intra-regional trade was done by Aiten (1973). Later Bayoumi and Eichengreen (1995) and Franel (1997) added a second dummy to test the RTA effect on the trade of bloc members with non-members. Recently, Soloaga and Winters (2001) incorporated three dummy variables in order to offer a simple and clear distinction between trade creation and trade diversion. The first dummy captures trade creation, while the second and the third dummy variables capture export trade diversion and the import trade diversion, respectively. They argued that both are needed because bloc member s imports and exports could follow different patterns after the formation of an RTA. Equation (1) becomes after introducing the three regional trade agreements dummy variables: X ijt = a + gdp it + gdp jt + pop it + pop jt + Dist ij + 1 RTAintra ijt 3 RTAimp ijt + ε ijt + 2 RTAexp ijt + (2) The three regional trade agreements variables, RTAintra ijt, RTAexp ijt variables that measure the trade effects of agreement, where: and RTAimp ijt are binary RTAintra ijt is a dummy variable taes a value of one if both countries i and j in year t belong to the same agreement and zero otherwise. A positive and statistically significant coefficient of 1 represents trade creation effects and indicates that intra-regional trade has been promoted more by the free trade agreement and is higher than normal trade levels. 10

11 RTAexp ijt is a dummy variable taes a value of one if exporter i belongs to the same agreement in year t and destination country j does not and zero otherwise. A positive and statistically significant coefficient of 2 is defined as an Export Trade Creation (XTC) effect and indicates that regional integration leads to a switch of export activities from agreement member countries to non-member countries. Conversely, a negative and statistically significant coefficient of 2 indicates a decrease in exports from member countries to non-member countries and is defined as an Export Trade Diversion (XTD) effect. RTAimp ijt taes a value of one if exporter i is a non-member in year t and destination country j belongs to the agreement and zero otherwise. A positive and statistically significant coefficient of 3 is defined as an Import Trade Creation (MTC) effect and indicates expanded imports from nonmember countries to member countries. Conversely, a significantly negative of 3 indicates an Import Trade Diversion (MTD effect. Trotignon (2010) draw a typology of trade creation and diversion for the three regional trade agreements dummy variables. Table (1) summarizes the interpretations of the respective signs and relatives values of RTAintra ijt,rtaexp ijt and RTAimp ijt Table 1 Interpretation of regional trade integration dummy variables Sign of Regional Coefficients RTAintra ijt RTAexp ijt RTAimp ijt Differences in Absolute Size Trade Creation and Diversion ITC, XTC and MTC RTAintra > RTAimp ITC, XTC, MTD RTAintra < RTAimp XTC, MTD RTAintra > RTAexp ITC, XTD, MTC RTAintra < RTAexp XTD, MTC RTAintra > RTAexp + RTAimp ITC, XTD, MTD RTAintra < RTAexp + RTAimp XTD and/or MTD Source: Trotignon (2010), Table 4, p.242 and Craig R. MacPhee and Wanasin Sattayanuwat (2014), Table 3, p.71. ITC : Intra-bloc trade creation : Stimulating effect on trade between partners XTC : Export trade creation : Stimulating effect on exports to the rest of the world MTC : Import trade creation : Stimulating effect on imports from the rest of the world XTD : Export trade diversion : Exports to the rest of world are replaced by intra-bloc trade MTD : Import trade diversion : Imports from the rest of world are replaced by intra-bloc trade 4.2 Basic gravity model (1): Assessing Regional Trade Agreements in natural resource-rich regions We follow the Vinerian specification of integration effects with an extension of three different sets of RTA dummy variables representing trade creation, export diversion and import diversion effects, as proposed by Endoh (1999), Soloaga and Winters (2001), Carrère (2006), Magee (2008) and Martínez- Zarzoso et al. (2009), so that we can test whether the creation of 13 regional trade agreements has facilitated international trade among the member countries at the expense of non-member countries. The baseline gravity model is given by: X ijt = a + lngdp it + lngdp jt + lnpop it + lnpop jt + lndis ij + lang ij + contig ij + col ij + comcol ij + RTAintra ijt + RTAexp ijt + RTAimp ijt + γ i + σ j + δ t + ε ijt (3) 11

12 where: a X ijt lngdp jt lngdp jt lnpop it lnpop jt lndis ij lang ij contig ij col ij comcol ij RTAintra ijt RTAexp ijt RTAimp ijt γ i σ j δ t ε ijt is constant, is the is the value of exports from country i to country j in year t at current US$ is the GDP of country i in year t at current US$ in logs, is the GDP of country j in year t at current US$ in logs, is the population size of country i in year t in logs, is the population size of country j in year t in logs, is the distance between country i and country j in logs, is a dummy variable indicating that country i and country j have a common language, is a dummy variable indicating that country i and country j have a common border, is a dummy variable indicating that country i and country j have ever in colonial relationship, is a dummy variable indicating that country i and country j have ever common colonizer post 1945, is a dummy variable taes a value of 1 if both countries i and j in year t belong to the same agreement and zero otherwise, is a dummy variable taes a value of one if exporter i belongs to the same agreement in year t and destination country j does not and zero otherwise, taes a value of one if exporter i is a non-member in year t and destination country j belongs to the agreement and zero otherwise, Exporter fixed effects, Importer fixed effects, Time fixed effects, is assumed to be a log-normally distributed error term. 4.3 Augmented gravity model (2): Impact of uneven distribution of natural resources on regional integration gains 1 In order to provide further insights to explain how the effects of the regional trade agreements will be distributed across countries in the context of uneven allocation of natural resources, we explore within the same gravity setup how patterns of trade creation and trade diversion vary across bilateral pairs depending on whether the exporter or importer are resource-rich country (R), resource-poor country (P) or diversified country (D). The new gravity equation is composed by four main groups of variables, more formally: X ijt = a + lngdp it + lngdp jt + lnpop it + lnpop jt + lndis ij + lang ij + contig ij + col ij + comcol ij + β 1 [R i R j RTA ijt ] + β 2 [R i P j RTA ijt ] + β 3 [R i D j RTA ijt ] + β 4 [P i R j RTA ijt ] + β 5 [P i P j RTA ijt ] + β 6 [P i D j RTA ijt ] + β 7 [D i R j RTA ijt ] + β 8 [D i P j RTA ijt ] + β 9 [D i D j RTA ijt ] + β 10 [R i W j RTA ijt ] + β 11 [P i W j RTA ijt ] + β 12 [D i W j RTA ijt ] + β 13 [W i R j RTA ijt ] + β 14 [W i P j RTA ijt ] + β 15 [W i D j RTA ijt ] + γ i + σ j + δ t + e ijt (4) The first structure of the equation (4) captures the effects of the traditional explicative variables as equation (3). The second group includes intra-regional trade dummies for each agreement (β 1 to β 9 ), where exporter and importer are both members of agreement.the exporters are decomposed into three categories according to the abundance of natural resources, resource-rich country R i, P i resource-poor country and diversified country D i, as well as the importers R j, P j and D j. The intra- 1 This approach is inspired essentially from Céline Carrère, Julien Gourdon & Marcelo Olarreaga (2012). They explore how patterns of trade creation and trade diversion vary across bilateral pairs depending on whether there are resource-rich or resource-poor countries. This could be done by introducing dummy variables which captures all combinations of intra-regional import between resource rich countries and resource poor countries to capture the trade creation among countries, as well as the combinations of extra-regional import with the rest of the world in terms imports in order to capture the trade diversion. 12

13 regional trade dummy variables represent all the possible combinations of intra-regional bilateral trade in agreement as follows: R i R j, R i P j, R i D j, P i R j, P i W j, P i P j, D i R j, D i P j and D i D j. The objective of this decomposition of intra-regional trade flows according to the contribution of natural resources in economies is to explore how the effects of trade creation and trade diversion are distributed within agreement depending on whether the exporter or importer is resource-rich country, resource-poor country or diversified country. More formally: R i R j RTA ijt is unity when both exporter and importer are rich in resource within agreement or zero otherwise, R i P j RTA ijt is unity when exporter is resource-rich country and importer is resource-poor country within agreement or zero otherwise, R i D j RTA ijt is unity when exporter is resource-rich country and importer is diversified country within agreement or zero otherwise, P i R j RTA ijt is unity when exporter is resource-poor country and importer is resource-rich country within agreement or zero otherwise, P i P j RTA ijt is unity when both exporter and importer are poor in resource within agreement or zero otherwise, P i D j RTA ijt is unity when exporter is resource-poor country and importer is diversified country within agreement or zero otherwise, D i R j RTA ijt is unity when exporter is diversified country and importer is resource-rich country within agreement or zero otherwise, D i P j RTA ijt is unity when exporter is diversified country and importer is resource-poor country within agreement or zero otherwise, D i D j RTA ijt is unity when both exporter and importer are diversified countries within agreement or zero otherwise. The third group of binary variables captures the effects of regional integration on extra-regional trade of members in each agreement in terms of extra regional exports. It includes three dummy variables where the exporter is member in agreement and the importer is not (β 10 to β 12 ). Exporters are decomposed again into resource-rich country R i, resource-poor country P i or diversified country D i, and then combined with the rest of world: R i W j, P i W j and D i W j.we aim to explore the extent of heterogeneity in export trade creation or diversion in agreement whether the exporter is resource -rich member, resource-poor member or diversified member. More formally: R i W j RTA ijt is unity if exporter is resource-rich member within agreement and importer is non-member or zero otherwise, P i W j RTA ijt is unity if exporter is resource-poor member within agreement and importer is non-member or zero otherwise, D i W j RTA ijt is unity if exporter is diversified member within agreement and importer is non-member or zero otherwise. The last group of binary variables captures the effects of regional integration on extra-regional trade of members in each agreement in terms of extra-regional imports. It includes three dummy variables where the exporter is rest of the world (non-member in agreement) and the importer is member in it (β 13 to β 15 ). Importers are decomposed also into resource- rich country R j, resource-poor country P j or diversified country D j, and then combined with the rest of the world: W i R j, W i P j and W i D j. We aim to explore the extent of heterogeneity in import trade creation or diversion in agreement whether the importer is resource-rich member, resource poor member or diversified member. More formally: W i R j RTA ijt is unity if importer is resource rich member within agreement and exporter is non-member or zero otherwise, W i P j RTA ijt is unity if importer is resource poor member within agreement and exporter is non-member or zero otherwise, W i D j RTA ijt is unity if importer is diversified member within agreement and exporter is non-member or zero otherwise. 13