Received: 13 October 2022 Accepted: 2 November 2022 DOI: 10.1111/twec.13365 O R I G I N A L A R T I C L E Regional integration and non-tariff barriers to Intra-Sub-Saharan Africa trade Ebo Turkson1  | Abena D. Oduro1 | Priscilla Twumasi Baffour1 | Peter Quartey2 1Department of Economics, School of Social Science, University of Ghana, Accra, Abstract Ghana The paper assesses the ex-post trade effect of sub-regional 2Institute of Social, Statistical and trade agreements (RTAs), financial integration and other Economic Research, University of Ghana, Accra, Ghana non-tariff barriers on intra-regional trade involving 43 Sub-Saharan Africa (SSA) countries. The objective is to Correspondence find out if RTAs within SSA had increased trade flows to Ebo Turkson, Department of Economics, School of Social Science, University of inform current efforts of establishing a successful conti- Ghana, Accra, Ghana. nental free trade area in SSA. Estimating a gravity model Email: feturkson@ug.edu.gh augmented with measures of trade agreements and finan- Funding information cial integration, the paper made use of bilateral trade flows Consortium pour la recherche économique and key gravity covariates from CEPII database over the en Afrique period 1960–2015. After controlling for the endogeneity of the trade agreement dummy, multilateral price resistance and zero-valued trade flows, we find RTAs within SSA and especially among Economic Community of West Afri- can States and Southern Africa Development Community members to have had a positive and significant impact on bilateral trade. Financially integrated trading partners also traded more, while we also found distance, landlocked- ness, common currency and colonial link to have an impact on trade costs and bilateral trade flows within SSA. The results indicate the need to focus on policies to expand and integrate regional markets in SSA by removing impedi- ments to trade and improving on trade facilitation measures to significantly improve trade performance under the newly established African Continental Free Trade Area. World Econ. 2022;00:1–19. wileyonlinelibrary.com/journal/twec © 2022 John Wiley & Sons Ltd. 1 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2 TURKSON eT al. K E Y W O R D S intra-African trade, regional integration, trade agreements, trade policy 1 | INTRODUCTION Recently, regional economic integration has received much attention globally on account of the bene- fits it generates among participating countries. The relative success of the monetary and currency integration by the European Union (EU) has played a significant role in the accelerated commitment by other regional integration efforts in the developing world. As noted by Ackah et al. (2013), it is the enormous potential trade benefit of greater integration that has encouraged most economies to seek to be part of the greater interdependence generated by integration efforts. Trade is widely accepted as an important engine of economic growth and development, and through trade, many regions and coun- tries have achieved tremendous success in lifting people out of poverty to prosperity. Sub-Saharan Africa (SSA), however, remains the only region globally that has not been able to use trade as a potent instrument for the achievement of rapid and sustainable economic growth and development. The sub-region remains at the periphery of global trade as it continues to lag in terms of contribution to global trade, contributing on average about 2% between 2010 and 2019 (IMF, 2021). The African Union Programme on Boosting Intra-African Trade indicated that intra-African trade as 2012 stood at about 13%, and at most not more than 20% (if allowance is made for unrecorded informal cross-border trade). Although it could be argued that the major reason for the low levels of intra-SSA trade is that most SSA countries have similar comparative advantage (export similar commodities), the relatively high trade costs within SSA have also contributed to the low levels of intra-SSA trade flows. Non-policy-induced non-tariff barriers (NTBs) have been recognised to consti- tute the major contributors to higher trade costs within SSA and thus the most significant hindrance to trade and integration. These barriers have been recognised to be more important in affecting the volume and direction of trade flows than price factors such as tariffs and currency exchange cost. Within SSA inadequate trade facilitation including the poor state of infrastructure, excessive delays at borders mainly because of burdensome administrative procedures for clearing goods for import and export, a lack of coordination and uniformity between countries within the sub-region regarding technical regulations, rules of origin and standards, policies on licences and permits, higher transport costs, remoteness and long overland distances continue to hinder trade and integration efforts. Studies such as Ackah et  al.  (2012), Turkson et  al.  (2020), Wilson et  al.  (2008), Mbabazi et al. (2006) and Limao and Venables (2001) have confirmed that the relatively low level of African trade flows is largely due to the poor state of infrastructure and inadequate trade facilitation measures. It is suggested that reducing such NTBs through regional integration will go a long way in increasing SSA trade flows. Over the last three decades, regions and countries globally have increasingly sought to make use of trade policy in the form of regional trade agreements (RTAs) to reduce NTBs to trade. With regard to SSA, there has been an increase in the number of RTAs among countries within the sub-region, with several cross-membership reported among countries belonging to different regional economic communities (RECs; Turkson, 2015). Very little is known about what motivates the increasing interest in RTAs in SSA, especially when the trade-enhancing impact has not been clearly positive (see for instance, Turkson, 2015 for a discussion about the trade-enhancing impact of RTAs in Africa), and trade policy measures have not been effective in reducing NTBs and increasing bilateral trade flows within the sub-region. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 3 Theoretically, it has been argued that given the similarities of comparative advantage and struc- tural supply-side characteristics, trade agreements in SSA should not be expected to contribute significantly to bilateral trade flows (Turkson,  2015). Empirically, institutional bottlenecks, the lack of political will, high external trade barriers (both Tariff and non-Tariff barriers) and the nature of the complex web of overlapping trade agreements have been cited as factors that hamper intra-regional trade (ECA, 2004; Gunning, 2001; Johnson, 1995; Lyakurwa et al., 1997; Yang & Gupta, 2005). The need to reduce these NTBs has seen a bouquet of trade policy reforms that have been pursued to boost trade from SSA. The various RECs within the sub-region have variously sought to eliminate these NTBs to encourage intra-sub-regional trade, yet some challenges still remain. As far back as 2012, the AU decided to establish a Pan-Africa Continental Free Trade Area (CFTA) by 2018 and to pursue an AU Action Plan for Boosting Intra-Africa Trade. It was envisaged that boosting intra-Africa trade will deepen Africa's market integration while at the same time serve as an effective instrument for the attainment of rapid and sustainable socio-economic development. The newly established African Continental Free Trade Agreement (AfCFTA) is an agreement that purposely creates a common African market, which is expected to contribute to the bigger vision of deepening economic integration (facilitated by movement of persons) to provide greater opportu- nities for Africa's structural transformation, economic diversification and development. The agree- ment aims to create a liberalised market for goods and contribute to the movement of natural persons and capital. The realisation of the AfCFTA however depends to a large extent, on the way various obstacles hampering intra-regional trade on the African continent are addressed, and how the exiting regional integration efforts under the various RECs can serve the needed building blocks in fostering intra-Africa trade. While an attempt to unpack the NTBs that hinder bilateral trade within SSA will inform the debate of how to get AfCFTA to take off successfully, an estimation of an intra-Africa trade cost function will enable estimates of the contribution of policy and behind the border trade measures to SSA trade costs to be obtained. Indeed, previous attempts at reducing NTBs have failed in boosting intra-regional trade in Africa partly as a result of data and measurement problems relating to most NTBs to trade within SSA (Turkson, 2012). Data on many components of trade frictions in Africa are either not readily available or limited if available. Though this is not peculiar to SSA, data issues relating to absence or unavailability, the improper combination of available data with other fragmentary or missing data to make it useful and aggregation bias are more profound in the SSA case relative to other regions within the global trading system. It is against this background that the study seeks to examine the important role that sub-regional integration efforts, mostly through trade agreements and NTBs have played in impacting on trade costs and intra-regional bilateral trade flows in SSA. Specifically, we seek to answer the question: a. To what extent have trade agreements and NTBs within SSA impacted on bilateral trade flows within the sub-region? b. To what extent have trade agreements and NTBs explain variations in trade cost for SSA? A clear understanding of the role of trade costs in impacting trade flows is very important in order to promote deeper integration of the economies across the sub-region. Such an exercise will have impor- tant policy relevance in the context of current efforts by the African Union to boost intra-SSA trade through the newly established Africa Continental Free Trade Area (AfCFTA). 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 4 TURKSON eT al. 2 | REGIONAL ECONOMIC AND TRADE INTEGRATION IN SSA Prior to the establishment of AfCFTA in 2018, SSA had five main RECs in West Central, East, Horn of Africa and Southern Africa.1 The origins of regional integration on the continent can be traced to the establishment of SACU in 1910, through to the historic Pan African Congresses, to the first regional federations. African trade integration has included numerous arrangements at regional and sub-regional levels. The RECs were established mainly to foster and promote economic integration between member states within the various sub-regions and through the wider African Economic Community (AEC), which was established under the Abuja Treaty in 1991. The RECs include Economic Community of West African States (ECOWAS), Economic Community of Central African States (ECCAS), Southern Africa Development Community (SADC), East African Commu- nity (EAC) and Intergovernmental Authority on Development (IGAD). Although the AU recognises a sixth community—the Common Market for Eastern and Southern Africa (COMESA), it involves three RECs in South, East and Central Africa and Egypt. 2.1 | Economic integration in West Africa-Economic Community of West African States Economic Community of West African States was founded on 28 May 1975 by a sub-regional group of 15 West African countries, with the signing of the Treaty of Lagos to foster and promote economic integration. To achieve ‘collective self-sufficiency’ it was envisaged that by establishing ECOWAS a single large trading bloc through an economic and trading union will be created. As a result of the inability of member states in harmonising policies, most of the ideals of ECOWAS remained unre- alised at the end of the 1980s. However, in 1993 there was a revision of the Treaty of Lagos towards establishing a common market in West Africa by creating a free trade zone and customs union, remov- ing trade mobility restrictions and giving the right of residence and establishment to member states. Within ECOWAS, tariffs have been removed on unprocessed goods and handicrafts, and there has been a complete elimination of tariffs on industrial goods and abolishing of entry and exit visas for residents travelling within the sub-region. The 2019 Africa Regional Integration Index (ARII),2 shows that the overall integration index of ECOWAS scored an average of 0.425, reflecting a moderately integrated community. The scores on the five dimensions of integration show that ECOWAS' strongest dimension is free movement of people (0.733), whereas its weakest dimension is productive integration (0.220). The high score of free movement of people demonstrates the vision and fulfilment of ECOWAS, thus creating a border- less region and open visa policies for its member countries. 2.2 | Economic integration in Central Africa-Economic Community of Central African States Economic Community of Central African States was formed out of the Customs and Economic Union of Central Africa (UDEAC of UAEAC) in 1981, which was established by the Brazzaville Treaty in 1 The five (5) RECs and member countries and states are listed in Appendix 5. 2 ARII uses 16 indicators, grouped into five dimensions (including free movement and trade integration), to measure how well each country and region in Africa is integrated with its neighbours. ARII also measures the state of regional integration for the continent as a whole. See https://www.integrate-africa.org/about-the-index/. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 5 1966 as a customs union with a free trade area between member states and a common external tariff for imports from non-member countries. Economic Community of Central African States began func- tioning in 1985 but remained inactive due to several financial difficulties which lead to a delay in the formal contact between the AEC and ECCAS up till October 1999. To promote economic integration among francophone states of Central Africa,3 UDEAC in June 1999 signed a treaty for the establishment of the Economic and Monetary Community of Central Africa (CEMAC). On 24 January 2003, the European Union (EU) concluded a financial and unifica- tion agreement between ECCAS and CEMAC, with ECCAS taking responsibility for the peace and security and CEMAC for issues regarding customs and monetary union. The association of CEMAC to AEC is mainly through ECCAS. CEMAC's objectives are trade promotion, instituting a genu- ine common market, and greater solidarity among people and towards underprivileged countries and regions. ARII 2019 indicates that ECCAS is moderately integrated with an overall score of 0.442. Its strongest dimension is macroeconomic integration with a score of 0.684 and its weakest dimension is productive integration with a score of 0.323. The disparities among member states in terms of trade integration, free movement of people, infrastructure integration, macroeconomic integration and productive integration are large. 2.3 | Economic integration in South Africa: Southern Africa development community Southern Africa Development Community was founded out of the Lusaka declaration in 1980 by Angola, Botswana, Lesotho, Malawi, Mozambique, Swaziland, Tanzania, Zambia and Zimbabwe to reduce the economic dependence on South Africa and promote economic integration. Initially known as the Southern African Development Coordination Conference (SADCC), SADC was formalised in 1992 under the Windhoek declaration to deepen economic cooperation and cross-border trade integra- tion. In 1996, SADC member states signed a protocol of trade to eliminate trade barriers and harmo- nise trade procedures and documentation. Southern Africa Development Community scores 0.337 based on the 2019 ARII score, with 9 of its 16 members performing at a level that is average for the community. Its score indicates the commu- nity is relatively poorly integrated. Its strongest regional integration dimension is free movement of people with a score of 0.490, and its weakest regional dimension is infrastructural integration with a score of 0.214. Southern Africa Development Community's country rankings appear to reflect the current state of socio-economic integration, where the best performers are flourishing and experienc- ing a good standard of living. 2.4 | Economic integration in East Africa: East African community The East African Community, an intergovernmental organisation, currently comprises five east Afri- can countries Burundi, Kenya, Rwanda, Tanzania and Uganda. Originally founded in 1967 by Kenya, Tanzania and Uganda, the EAC collapsed in 1977 but was revived by 7 July 2000. In 2008, after negotiations with the SADC and the COMESA, the EAC agreed to an expanded free trade area including the member states of the three Regional Trade Agreements (RTAs). 3 Cameroon, Central African Republic, Chad, Republic of Congo, Equatorial Guinea and Gabon. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 6 TURKSON eT al. The ARII (2019) shows that the REC performs most strongly on the free movement of people dimension, where its countries average 0.644. However, its macroeconomic integration dimension is not far behind with a score of 0.660. No EAC member has a bilateral investment treaty in force, and EACs' score on trade integration is not high (0.440). This is mostly attributed to the low share of regional exports, although the community has eliminated tariffs among its member countries. 2.5 | Economic integration on the horn of Africa: Intergovernmental authority on development Initially established as the Intergovernmental Authority on Drought and Development (IGADD) in 1986, IGAD was intended to support member states to cope with the problems of recurring and severe droughts and other natural disasters that had caused ecological degradation and economic hardship in the Horn of Africa between 1974 and 1984. On 18 April 1995, IGAD was revitalised by expanding its areas of regional cooperation through the creation of a full-fledged regional political, economic, development, trade and security entity similar to the SADC and ECOWAS. The IGAD assembles eight countries that range from Kenya, a continental powerhouse to the economically small Somalia. Its overall regional integration score is 0.438 indicating that the community is moderately integrated with its best performers being Uganda and Kenya and its worst performers being Eritrea and South Sudan. Intergovernmental authority on development performs best in the free movement of people (0.540), as most of its members have committed to liberalising mobility on the continent. The productive integration dimension (0.321) is where it has the most to improve. For the dimension of free movement of people, Djibouti and Somalia are the top perform- ers. Uganda leads IGADs in integrated production and integrated trade. It has the highest score for the export of intermediate goods in the community. It has the best trade share and the second-best export share. 2.6 | Multiplicity of trade agreements in SSA One of the interesting aspects of African regional trade agreements is the array of such trade agreements and the extent of overlapping membership. Authors such as Gunning (2001), Yang and Gupta (2005) and Chacha  (2013) have argued that the overlapping memberships undermine the effectiveness of African RTAs because it adds to the financial burden of members states and makes it difficult for customs officials to deal with different tariff reduction rates, rules of origin, trade documentation and statistical nomenclatures. Invariably, these have been counterproductive to the trade facilitation goals of these agreements. Interestingly, the multiplicity of trade agreements within SSA gives an impression of the trade crea- tion effect of these agreements, otherwise what explains the dense web of such agreements within SSA. What is unknown is whether the existing trade agreements/customs unions have facilitated increased trade and any welfare gains for member states. Indeed, the conclusions from previous studies that have sought to assess the ex-post trade effects of RTAs in SSA have remained inconclusive. For instance, while Johnson (1995), Lyakurwa et al. (1997), Gunning (2001), ECA (2004), Yang and Gupta (2005) and Chacha (2013) confirmed the conventional belief that RTAs in SSA have not enhanced trade among member countries, studies such as Deme (1995), Elbadawi  (1997), Cernat  (2001), Carrère  (2004), Coulibaly (2007), Afesorgbor and van Bergeijk (2011) and Turkson (2015) found SSA RTAs to have been trade-creating. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 7 3 | METHODOLOGY 3.1 | Empirical strategy The empirical approach adopted in this paper is the estimation of a two-part analytical framework. In the first part, a multiplicative gravity equation is estimated using a panel of bilateral trade flows and standard gravity covariates to unpack the NTBs that impede or promote intra-Africa trade and also to estimate the trade effect of regional trade integration on intra-African trade flows. We adopt the Anderson and van Wincoop (2003) theory-based gravity model assumed in a multi- plicative form as: 𝜑𝜑1 𝜑𝜑 𝛾𝛾 𝛼𝛼 𝛽𝛽𝑋𝑋𝑖𝑖𝑖𝑖𝑖𝑖 = 𝜑𝜑 20.𝑌𝑌 .𝑌𝑌 .𝑑𝑑 .𝑍𝑍 𝑘𝑘.𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 .𝜂𝜂𝑖𝑖𝑖𝑖 .𝜐𝜐𝑖𝑖.𝜖𝜖𝑖𝑖𝑖𝑖 .𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖𝑖𝑖 (1) The econometric specification in logs is of the form: ( ) ( ) ( ) ( ) ln 𝑋𝑋𝑖𝑖𝑖𝑖𝑖𝑖 = 𝛿𝛿0 + 𝛿𝛿1ln (𝑌𝑌𝑖𝑖𝑖𝑖) + 𝛿𝛿2ln 𝑌𝑌𝑖𝑖𝑖𝑖 + 𝜑𝜑𝜑𝜑𝜑𝜑 𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛼𝛼1ln (𝑌𝑌 𝑌𝑌𝑌𝑌𝑖𝑖𝑖𝑖) + 𝛼𝛼2ln 𝑌𝑌 𝑌𝑌𝑌𝑌𝑖𝑖𝑖𝑖 +𝛼𝛼3𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝐴𝐴𝐴𝐴𝜑𝜑𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖 + 𝛼𝛼4𝑌𝑌𝐶𝐶𝜑𝜑𝐶𝐶𝜑𝜑𝑖𝑖𝐴𝐴𝜑𝜑𝑖𝑖𝑖𝑖 + 𝛼𝛼5𝑌𝑌𝐶𝐶𝐶𝐶𝐴𝐴𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛼𝛼6𝑌𝑌𝐶𝐶𝐶𝐶𝜑𝜑𝐴𝐴𝜑𝜑𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛼𝛼7𝐿𝐿𝐴𝐴𝜑𝜑𝑑𝑑𝜑𝜑𝐶𝐶𝐴𝐴𝐿𝐿𝐴𝐴𝑑𝑑 𝑖𝑖𝑖𝑖 (2) +𝛼𝛼8𝑅𝑅𝐴𝐴𝐶𝐶𝐶𝐶𝑖𝑖𝐴𝐴𝜑𝜑𝐴𝐴𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛼𝛼9𝐹𝐹 𝑖𝑖𝜑𝜑𝐹𝐹𝜑𝜑𝑖𝑖𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛼𝛼10𝐹𝐹 𝑖𝑖𝜑𝜑𝐹𝐹𝜑𝜑𝑖𝑖𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛽𝛽1𝐸𝐸𝑌𝑌𝐸𝐸𝐸𝐸𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛽𝛽2𝐸𝐸𝑌𝑌𝑌𝑌𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖 +𝛽𝛽3𝐸𝐸𝐴𝐴𝑌𝑌𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛽𝛽4𝐸𝐸𝐴𝐴𝑆𝑆𝑌𝑌𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛽𝛽5𝐹𝐹𝐼𝐼𝐴𝐴𝑆𝑆𝑖𝑖𝑖𝑖𝑖𝑖 + 𝜂𝜂𝑖𝑖𝑖𝑖 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖, where Xijt is bilateral exports at time t and is specified to be a function of GDP of origin and destination (Yit and Yjt), GDP per capita of origin and destination (YPCit and YPCjt), the distance between bilateral trading partners (dijt), a vector Z of controls thought to proxy for other aspects of bilateral and country characteristics (Zij) including adjacency, colonial link, common currency, common official language, landlockedness, remoteness and trade agreement dummies ( ) 𝐴𝐴 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖, 𝐸𝐸𝐸𝐸𝐸𝐸𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖, 𝐸𝐸𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖, 𝐸𝐸𝐴𝐴𝑆𝑆𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖 𝑎𝑎𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎𝐴𝐴𝑆𝑆𝑖𝑖𝑖𝑖𝑖𝑖 . We also include degree of financial ( ) integration of the origin and destination cou𝐴𝐴ntries 𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑎𝑎𝐹𝐹𝑎𝑎 𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑗𝑗𝐹𝐹 to capture the extent of export diversification into higher value addition. As noted by Kose et al. (2006), financial integra- tion can help capital-poor countries diversify away from their production of agricultural or primary products for exports into higher value addition and this diversification should increase exports and thereby reduce macroeconomic volatility. As shown in equation 2, we also control for bilateral fixed ( ) e𝐴𝐴ffects 𝜂𝜂𝑖𝑖𝑖𝑖 and𝐴𝐴 time (𝜗𝜗𝑡𝑡) . Most empirical studies have assumed an exogenous dummy variable to represent the effect of belonging to a trade agreement. As noted by Trefler (1993) and Baier and Bergstrand (2007), in real- ity trade agreement dummies are not exogenous, because, for unobservable reasons (such as political factors, notariff barriers and domestic policies that inhibit bilateral trade), countries can endogenously select into a trade agreement. According to Baier and Bergstrand (2007), the potential endogeneity of the trade agreement (either RTA or PTA) dummy could thus be attributed to omitted variables and sample selection bias. Thus, the trade agreement dummy coefficient is generally underestimated because the dummy within the gravity framework is correlated negatively with the error term leading to the classical ‘attenuation bias’ of the FTA coefficient towards zero. To correct for omitted variables and selection bias arising from the endogeneity of regional trade agreement dummies (included to capture the impact of the various intra-SSA RTAs), we will adopt the Hausman–Taylor panel technique to estimate the gravity equation as stated in equation (2). Based on instrumental variables, the Hausman–Taylor panel estimator in dealing with the endogeneity bias from the trade agreement dummies assumes the existence of correlation between 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 8 TURKSON eT al. some explanatory variables and the individual heterogeneity effect. As a fixed effect estimator, the Hausman–Taylor panel estimator has proven to be more efficient than the panel random and fixed effects model (REM and FEM) techniques. Although the fixed effects vector decomposition (FEVD) technique also provides solutions to the econometric concerns raised against the REM and FEM, the choice of the Hausmann-Taylor over the FEVD is premised on the evidence that the FEVD is generally not well-suited for relatively large samples (see Plümper & Troeger, 2004). The second part analytical framework that will seek to answer research question (2) estimates a bilateral trade cost function for SSA without relying on the traditional measures based on the gravity-related quantifiable trade costs components. Recent improvements in empirical methodology in the trade literature with respect to trade costs have made it possible to estimate bilateral trade cost without relying on the traditional measures. Novy  (2013) following closely the work of Head and Ries (2001) derived an explicit analytical solution for the multilateral trade resistance variables, which help to solve the trade cost function of the form; / / 1 1 ( ) ( ) 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 2 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 2 (𝜎𝜎 − 1)𝜏𝜏 = − 1 = − 1, (3)𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 where τij is the total trade cost (i.e. measures bilateral trade costs relative to domestic trade costs),4 tijtji is the bilateral trade costs of countries i and j, and tiitjj is the domestic trade costs of countries i and j. The measure of the international component of trade costs net of distribution costs in the destination country ( ) is given as 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖𝐴𝐴 . Intuitively, equation (3) indicates that when bilateral trade costs decrease relative 𝑡𝑡𝑖𝑖𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖 to domestic trade costs, total trade costs (τij) will decrease, making it easier for countries i and j to trade relative to domestic trade. This will therefore imply that bilateral trade flows will increase relative to domestic trade flows. Similarly, if bilateral trade flows increase relative to domestic trade flows, one can infer that it has become easier for the two countries to trade (possibly because bilateral trade costs have declined relative to domestic trade cost), and this will be reflected in a decline in total trade costs. Based on the micro-founded measure obtained from equation (3) above, we propose to estimate a multiplicative tariff equivalent bilateral trade cost function of the form: ( ) 𝜏𝜏𝑖𝑖𝑖𝑖 = 𝛾𝛾0 + 𝛾𝛾1ln 𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛾𝛾2𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖 + 𝛾𝛾3𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐴𝐴𝑖𝑖𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛾𝛾4𝐶𝐶𝐶𝐶𝐶𝐶𝐴𝐴𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛾𝛾5𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐴𝐴𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 +𝛾𝛾6𝐿𝐿𝐴𝐴𝐴𝐴𝑑𝑑𝐶𝐶𝐶𝐶𝐴𝐴𝐿𝐿𝐴𝐴𝑑𝑑𝑖𝑖𝑖𝑖 + 𝛾𝛾7𝐹𝐹 𝑖𝑖𝐴𝐴𝐹𝐹𝐴𝐴𝑖𝑖𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛾𝛾8𝐹𝐹 𝑖𝑖𝐴𝐴𝐹𝐹𝐴𝐴𝑖𝑖𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖 + 𝛾𝛾9𝐸𝐸𝐶𝐶𝐸𝐸𝐸𝐸𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛾𝛾10𝐸𝐸𝐶𝐶𝐶𝐶𝐴𝐴𝐸𝐸𝑖𝑖𝑖𝑖𝑖𝑖 (4) +𝛾𝛾11𝐸𝐸𝐴𝐴𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛾𝛾12𝐸𝐸𝐴𝐴𝑆𝑆𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖 + 𝛾𝛾13𝐹𝐹𝐼𝐼𝐴𝐴𝑆𝑆𝑖𝑖𝑖𝑖𝑖𝑖 + 𝜂𝜂𝑖𝑖𝑖𝑖 + 𝜐𝜐𝑖𝑖 + εijt . For SSA where measures of trade costs components are limited, and in some cases unavailable, the micro-founded measure of trade cost offers an opportunity to track changes in trade costs in SSA and to use such a measure to assess the trade effect of SSA RTAs on intra-sub-regional trade flow. This will make a meaningful contribution to the integration and trade cost literature on Africa. 3.2 | Data Data on bilateral exports covered the period 1960 to 2015 and are obtained from the ‘square’ grav- ity data set for all world pairs of countries by the Centre d'Etudes Prospectives et d'Informations 4 See Appendix 1 for the derivation of the Micro-founded measure of trade cost. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 9 Internationales (CEPII).5 This data set also contains some variables relating to the standard gravity covariates (mostly, country-specific characteristics). Data on bilateral exports for the same period are sourced from the UN-COMTRADE data set. In order to focus on SSA, we concentrate mainly on bilateral trade relations within this sub-region involving 43 countries. With regard to data on the micro-founded trade cost measure, we will make use of the database used by Arvis et al. (2012). This data set has been subsequently updated to 2014. The bilateral meas- ure of trade costs featured in this database is truly comprehensive in the sense that it captures trade costs in its wider sense, including not only international transport costs and tariffs but also other trade cost components discussed in Anderson and van Wincoop (2004), such as direct and indirect costs associated with differences in languages, currencies and cumbersome import or export procedures. Known as the ESCAP—World Bank-trade costs-data set, it spans the period from 1995 to 2014 and covers the primary and manufacturing sectors and involves bilateral trade involving 196 countries out of which 44 are SSA countries. Refer to Appendices 2 and 3 for the definition, source and summary statistics for the main variables. 4 | ANALYSIS OF RESULTS AND POLICY IMPLICATIONS 4.1 | Results from gravity model The results obtained from estimating the gravity equation  2 are presented in Table  1. The results presented include the estimates of the fixed effects model (columns 1 and 2) and Hausman–Taylor (columns 3 and 4). The choice of the fixed effect estimator is based on the Hausman test results. As shown in Appendix 4, the test rejects the null hypothesis suggesting the presence of a correlation between individual heterogeneity and the covariates of the estimated gravity model, implying the inappropriateness of the random effects estimator. To control for the endogeneity of the trade agree- ment variable, we concentrate on the estimates of the Hausman–Taylor in columns 3 and 4. As a result of the existence of zero-valued trade flows in the sample, we perform sensitivity anal- ysis to examine the effect of the trade agreement variables. The inclusion of the zero-valued trade flows is to ensure that we do not omit from the analysis, the possibility of the trade agreement variable explaining in part why countries do not trade within SSA. The results in columns 5 and 6 of Table 1 are obtained by estimating equation 2 using the negative binomial pseudo-maximum likelihood esti- mator (NBPML), a member of the family of Poisson pseudo-maximum likelihood estimators (see Burger et al., 2009). The use of the NBPML technique allows for the inclusion of zero-valued trade flows and also for the unobserved heterogeneity between countries to be accounted for. The results from the estimation in columns 3 and 4 of Table 1 generally perform satisfactorily as most of the standard covariates of the gravity model conform with expectations. Importer and exporter GDP, common colonial link, common language, use of a common currency, remoteness and adjacency (i.e. sharing a common border) are expected to exert a statistically significant positive effect on bilateral exports. Distance and the number of landlocked countries in a pair were found to exert a negative effect on bilateral exports, confirming prior expectations. This supports support the argument that within SSA, high transport costs resulting from trading over longer distance and with landlocked countries impose significant trade costs, implying that with SSA trade will be diverted from landlocked and longer distance partners. 5 As generated and used by Head et al. (2010). 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 10 TURKSON eT al. T A B L E 1 Impact of SSA RTAs on bilateral trade within SSA Dependent variable Dependent variable (log of exports ijt) (exports ijt) Fixed effects Hausman–Taylor NBPML Variables (1) (2) (3) (4) (5) (6) GDPit 0.275 0.373 0.497*** 0.487*** 0.391*** 0.405*** (0.511) (0.504) (0.111) (0.110) (0.010) (0.010) GDPjt 0.546 0.665 0.266** 0.258** 0.312*** 0.328*** (0.478) (0.472) (0.121) (0.120) (0.010) (0.010) Gross Financial Integrationit 0.039** 0.044*** 0.035** 0.041** 0.017*** 0.018*** (0.017) (0.017) (0.017) (0.016) (0.002) (0.002) Gross Financial Integrationjt 0.021 0.015 0.017 0.011 0.005** 0.005** (0.016) (0.016) (0.016) (0.016) (0.002) (0.002) GDP per capitait 0.097 −0.011 −0.119 −0.113 −0.157*** −0.116*** (0.515) (0.508) (0.094) (0.092) (0.008) (0.009) GDP per capitajt −0.338 −0.469 −0.097 −0.093 −0.103*** −0.067*** (0.466) (0.460) (0.092) (0.089) (0.008) (0.009) Remotenessijt 0.379* 0.446** 0.500*** 0.547*** −0.073*** −0.101*** (0.203) (0.202) (0.179) (0.178) (0.016) (0.017) Common Currencyijt 0.405* 0.399* 0.384* 0.389* 0.026 0.216*** (0.213) (0.225) (0.197) (0.207) (0.023) (0.024) Distanceij −1.469*** −1.448*** −0.761*** −0.735*** (0.100) (0.102) (0.015) (0.015) Common Colonyij 0.362** 0.299* 0.764*** 0.661*** (0.158) (0.157) (0.018) (0.018) Common Languageij 0.632*** 0.722*** 0.260*** 0.325*** (0.143) (0.141) (0.018) (0.018) Adjacencyij 1.346*** 1.420*** 0.350*** 0.457*** (0.212) (0.208) (0.026) (0.026) Number Landlockedij −0.726*** −0.699*** −0.531*** −0.525*** (0.094) (0.094) (0.012) (0.012) RTAijt 0.211* 0.218* 0.139*** (0.126) (0.124) (0.019) ECOWASijt 0.476*** 0.482*** 0.239*** (0.165) (0.163) (0.024) IGADijt −0.515 −0.500 0.289*** (0.453) (0.449) (0.097) SADCijt 0.705*** 0.692*** 0.504*** (0.180) (0.174) (0.028) EACijt −0.263 −0.283 0.215** (0.491) (0.485) (0.104) 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 11 T A B L E 1 (Continued) Dependent variable Dependent variable (log of exports ijt) (exports ijt) Fixed effects Hausman–Taylor NBPML Variables (1) (2) (3) (4) (5) (6) ECCASijt −1.289*** −1.306*** −0.982*** (0.375) (0.372) (0.046) Constant included Yes Yes Yes Yes Yes Yes Control for bilateral & time effects Yes Yes Yes Yes Yes Yes Observations 25,782 25,782 25,782 25,782 50,587 50,587 Number of bilateral pairs 1433 1433 1433 1433 1433 1433 Note: Country i and j refer to exporter and importer country respectively. Robust standard errors in parentheses (clustered at the bilateral country pair). ***p < .01, **p < .05, *p < .1. Generally, the results in columns 5 and 6 of Table 1 (inclusion of zero-valued bilateral exports) are quite comparable to those found in columns 3 and 4 of Table  1. Importer and exporter GDP, common colony, common currency, common language and sharing a common border were found to exert a statistically significant positive effect on bilateral exports. Distance, the number of landlocked countries, GDP per capita for both exporter and importer countries, and remoteness exerted negative effects on bilateral exports. The negative impact of GDP per capita on bilateral trade within SSA confirms the low-income elasticity of demand for primary products (agricultural exports tend to be income inelastic). While we expect that bilateral trading partners that are remote from the rest of the world should trade more, after controlling for zero-valued trade flows, we find remote countries trad- ing less, and this is more likely because of similarity of comparative advantage. With regard to financial integration, we find the higher degree of financial integration of the exporting partner to increase diversification and thus increase the flow of bilateral exports within SSA, although the extent of importer financial integration does not. When we include zero-valued trade flows (columns 5 and 6), we find the extent to which exporter and importer countries are finan- cially integrated as positively impacting on bilateral trade flows within SSA, implying that the success of the AfCFTA will depend on the extent to which SSA countries are financially integrated within the global financial architecture. We find the presence of RTAs to be statistically significant in enhancing trade. On average, membership of an RTA in SSA has a positive trade effect under the different specifications. As shown in column 3 of Table 1, bilateral exports between member countries in any of the RTAs in SSA was 24% (exp 0.218 = 1.24) higher relative to bilateral exports between non-member countries. This finding is robust when we include zero-valued trade flows (shown in column 5), as we find members trading by 15% (exp 0.139 = 1.15) higher than trade among non-members. This implies that on average SSA RTAs have been trade-enhancing. Comparatively within specific RTAs, SADC and ECOWAS countries did better in intra-REC trade. Based on the Hausman–Taylor estimator (column 4), we find that for ECOWAS and SADC countries, bilateral exports increased by 62% (exp 0.482 = 1.62) and 100% (exp 0.692 = 2.0) respectively among member countries and states, relative to trading with non-members. Conversely, ECCAS countries traded less within their trade bloc relative to the non-members. The estimated decline in bilateral exports was about 269% (exp 1.306  = 3.69). The statistically insignificant EAC and IGAD 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 12 TURKSON eT al. trade agreements could be due to the low degree of integration for the periods under consideration. When we include zero-valued bilateral trade flows, we find (see column 6) that 4 out of the 5 RTAs have been trade-enhancing. For ECOWAS, IGAD, SADC and EAC countries, bilateral exports were higher by 27% (exp 0.239 = 1.27), 36% (exp 0.289 = 1.36), 65% (exp 0.504 = 1.65) and 24% (exp 0.215 = 1.24) relative to trade with non-members respectively. For ECCAS, the negative trade effects found for positive-valued trade flows is robust when we control for zero-valued flows, although the reduction in exports is of a smaller the magnitude at 167% (exp 0.982 = 2.67). 4.2 | Micro-founded measures of trade cost in SSA The micro-founded measure of trade cost is a relative measure that seeks to estimate the total non-tariff trade barrier costs for a country with its partners relative to domestic trade costs. To explain the vari- ation in the micro-founded trade cost measure for regions involved in the global multilateral trading system, Table 2 shows the estimates of the average relative trade costs for Africa and those for the rest of the world between 1995 and 2014. As can be observed from Table 2, the average relative trade cost for African countries with all their trading partners compared with the rest of the world is the highest. Even within Africa, the estimates for SSA remain higher than the continent's average. This clearly indicates that SSA has the highest trading cost, and this confirms evidence from the Doing Business and Trading Across Borders Data- bases etc. Table 3 presents the regression results examining the effects of integration and/or trade agreement and other NTBs on trade cost in SSA. The dependent variable is the logarithmic transformation of the micro-founded relative trade cost measure. Columns (1) and (3) of Table 3 present the pooled results without controls for fixed effects. In order to account for multilateral resistance. Columns (2) and (4) present the pooled results with controls for country-fixed effects to ensure completeness. In terms of the extent to which financial integration and/or trade agreement explain variation in the micro-founded trade cost measures for Africa, the empirical estimations suggest that countries in ECOWAS, IGAD, SADC and EAC tend to trade at a lower trade cost relative to trading with partners from other regional blocks in Africa. At a disaggregated level where we distinguish manufactures, the pattern is similar to what we find for all goods. For the latter, the trade cost is higher for ECOWAS and ECCAS countries. For financial integration, it was observed that it generally lowers trade cost in SSA though statistically the impact is negligible. The main implication of these results is that SSA countries need to focus on expanding and integrating regional markets to significantly improve trade performance. This can facilitate proper entry and dominance in the international market which tends to be very much competitive. T A B L E 2 Estimates of overall average relative trade cost (period averages)-all goods. SSA AFRICA AMERICA ASIA EUROPE 1995–1999 297.4 295.2 256.0 258.1 203.4 2000–2004 322.8 320.3 292.6 283.5 225.5 2005–2009 340.2 335.6 311.8 283.0 231.9 2010–2014 324.8 319.9 306.6 276.5 214.9 1995–2014 325.1 321.3 295.6 278.2 221.5 Source: Authors construct- estimates obtained from estimating equation (2). Abbreviation: SSA, Sub-Saharan Africa. Bold Values are Regions. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 13 T A B L E 3 Pooled OLS estimates of the average relative bilateral trade cost function. All goods Manufacturing Variables (1) (2) (3) (4) Log of Distanceij 0.283*** 0.531*** 0.340*** 0.582*** (0.009) (0.011) (0.010) (0.012) Sharing Common Borderij −0.393*** −0.246*** −0.372*** −0.195*** (0.014) (0.013) (0.017) (0.016) Common Languageij −0.136*** −0.129*** −0.124*** −0.148*** (0.011) (0.012) (0.012) (0.015) Common Currencyij −0.157*** −0.136*** −0.163*** −0.137*** (0.013) (0.020) (0.015) (0.025) Common Legal Originij −0.006 −0.079*** −0.024** −0.083*** (0.010) (0.009) (0.012) (0.012) Number Landlockedij 0.274*** 0.369*** 0.296*** 0.004 (0.007) (0.043) (0.008) (0.083) Net Foreign Assetsit −0.019*** −0.005 −0.067*** 0.013 (0.006) (0.004) (0.023) (0.036) Net Foreign Assetsjt −0.019*** −0.005 −0.067*** 0.013 (0.006) (0.004) (0.023) (0.036) ECOWASijt −0.102*** 0.012 −0.061*** −0.055** (0.013) (0.020) (0.015) (0.024) IGADijt −0.233*** 0.022 −0.122** 0.035 (0.045) (0.030) (0.048) (0.036) SADCijt −0.407*** −0.105*** −0.328*** −0.105*** (0.015) (0.017) (0.017) (0.020) EACijt −0.481*** −0.228** −0.302*** −0.161 (0.104) (0.091) (0.117) (0.111) ECCASijt 0.164*** 0.027 0.166*** 0.031 (0.022) (0.022) (0.025) (0.025) Constant included Yes Yes Yes yes Control for time effects No Yes No Yes R-squared 0.450 0.692 0.454 0.667 Observations 12,490 12,490 10,068 10,068 Note: Country i and j refer to exporter and importer country. Robust standard errors in parentheses (clustered at the bilateral country pair). ***p < 0.01, **p < 0.05, *p < 0.1. 5 | CONCLUSION AND POLICY IMPLICATIONS This paper presents an empirical estimation of the relationship between regional and financial inte- gration, trade costs and intra-SSA trade. It is motivated by the importance of trade cost in affecting trade flows among nations and the extent to which greater regional and financial integration and NTBs can help efforts in reducing trade cost in Africa. A clear understanding of the role of trade costs in 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 14 TURKSON eT al. enhancing trade is thus very important in order to promote deeper integration of the economies across the sub-region. Using bilateral trade flows between 43 SSA countries over the period 1960–2015, we find trade agreements to increase bilateral exports within SSA. Comparatively, SADC and ECOWAS coun- tries did better in intra-REC trade than other RTAs and the results are robust for Hausman–Taylor and NBPML estimations. The inclusion of zero-valued trade flows produces statistically significant results for all RTAs in SSA. Intra-regional trade increased in all the RECs with the exception of coun- tries in ECCAS. Generally, this paper contributes to the literature on the potential impact of trade agreements on trade flows within SSA. By making use of a panel of observations and controlling for bilateral fixed and time effects enabled for the correction of the endogeneity of the trade agreement variable unlike most previous studies on SSA. We also accounted for multilateral price resistance by the inclusion of the bilateral remoteness indicator (as in Frankel & Wei, 1998) and explicitly dealt with the zero-valued flows. The results show the need for policymakers to further identify and reduce trade barriers in SSA as average trade costs are still higher on the continent relative to the rest of the world. Integration efforts to reduce trade policy barriers and deepen financial integration within the various RECs within SSA should be vigorously pursued as the evidence from our results show clearly that countries that belong to a RTA or are financially integrated trade more as a result of lower trade costs. Further policies to ensure the use of common currencies could reduce trade costs and increase bilateral trade flows within SSA, and this would be the ‘sine qua non’ for the success of the newly established African Continental Free Trade Area (AfCFTA). Indeed, current efforts at ensuring a smooth take-off of the CFTA for SSA should commence with the removal of all NTBs within RECs. There is no doubt that the simi- larity in comparative advantage explains why intra-SSA trade is relatively low; however, the increase in the production of higher value addition exports will provide enormous opportunities for increased trade within the sub-region. ACKNOWLEDGEMENT This paper is an updated version of the paper submitted to the African Economic Research Consor- tium (AERC) in 2019 as one of the collaborative papers under ‘Rethinking Regional Integration in Africa for Inclusive and Sustainable Development’. The Authors appreciate the research grant from AERC to carry out this research effort. FUNDING INFORMATION This research was supported by the African Economic Research Consortium (AERC) under the Collaborative Research Project on Rethinking Regional Integration in Africa. DATA AVAILABILITY STATEMENT Data sharing is not applicable to this article as no new data were created or analyzed in this study. ORCID Ebo Turkson https://orcid.org/0000-0002-5366-5167 REFERENCES Ackah, C., Turkson, F. E., & Opoku, K. (2012). Trade costs and intra-regional trade flows in ECOWAS. Journal of West African Integration, 1(1), 1–35. EPAU, ECOWAS. Ackah, C., Turkson, F. E., & Opoku, K. (2013). Measuring trade costs in ECOWAS. Modern Economy, 4, 56–65. doi:10.4236/me.2013.41007 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 15 Afesorgbor, S., & van Bergeijk, P. A. (2011). Multi-membership and the effectiveness of regional trade agreements in western and southern Africa: A comparative study of ECOWAS and SADC. (ISS Working Paper Series No 520). Institute of Social Studies (ISS). Anderson, J. E., & van Wincoop, E. (2003). Gravity with gravitas: A solution to the border puzzle. American Economic Review, 93(1), 17092. Anderson, J. E., & van Wincoop, E. (2004). Trade costs. Journal of Economic Literature, 42(3), 691–751. Arvis, J.-F., Duval, Y., Shepherd, B., & Utoktham, C. (2012). Trade costs in the developing world: 1995-2010. (ARTNeT Working Papers No. 121/December 2012). Baier, S. L., & Bergstrand, J. H. (2007). Do free trade agreements actually increase Members' International trade? Jour- nal of International Economics, 71(1), 72–95. Baier, S. L., & Bergstrand, J. H. (2009). Estimating the effects of free trade agreements on international trade flows using matching econometrics. Journal of International Economics, 77(1), 63–76. https://doi.org/10.1016/j. jinteco.2008.09.006 Burger, M., van Oort, F., & Linders, G. J. (2009). On the specification of the gravity model of trade: Zeros, excess zeros and zero-inflated estimation. Spatial Economic Analysis, 4(2), 167–190. Carrère, C. (2004). African regional agreements: Impact on trade with or without currency unions. Journal of African Economies, 13(2), 199–239. Cernat, L. (2001). Assessing regional trade arrangements: Are south-south RTAs more trade diverting. Global Economy Quarterly, 2(3), 235–259. Chacha, M. (2013). Regional integration and the challenge of overlapping memberships on trade. Journal of Interna- tional Relations Development, 17, 522–544. https://doi.org/10.1057/jird.2013.13 Coulibaly, S. (2007). Evaluating the trade effect of developing regional trade agreements: A semi-parametric approach. Journal of Economic Integration, 24, 709–743. Deme, M. (1995). The impact of ECOWAS on intraregional trade flows: An empirical investigation. The Review of Black Political Economy, 23(3), 113–129. ECA. (2004). Assessing regional integration in Africa. Economic Commission for Africa. Elbadawi, I. A. (1997). The impact of regional trade and monetary schemes on intra-sub-Saharan Africa trade. In A. Oyejide, I. Elbadawi, & P. Collier (Eds.), Regional integration and trade liberalization in sub-Saharan Africa. Volume 1: Framework, issues and methodological perspectives. Macmillan. Frankel, J., & Wei, S. (1998). Regionalization of world trade and currencies: Economics and politics? In J. A. Frankel (Ed.), The regionalization of the world economy (pp. 189–226). University of Chicago Press. Gunning, J. W. (2001). Trade blocs: Relevant for Africa? Journal of African Economies, 10(3), 311–335. Head, K., Mayer, T., & Ries, J. (2010). The erosion of colonial trade linkages after independence. Journal of Interna- tional Economics, 81(1), 1–14. Head, K., & Ries, J. (2001). Increasing returns vs National Product Differentiation as an explanation for the pattern of US-Canada trade. American Economic Review, 91(4), 858–876. IMF. (2021). Direction of trade statistics. Johnson, O. E. (1995). Regional integration in sub-Saharan Africa. Journal of European Integration, 18(2–3), 201–234. Kose, A. M., Prasad, E., Rogoff, K., & Wei, S.-J. (2006). Financial globalization: A reappraisal. (IMF Working Paper No 2006/189). https://doi.org/10.5089/9781451864496.001 Limao, N., & Venables, A. J. (2001). Infrastructure, geographical disadvantage, transport costs and trade. World Bank Economic Review, 15, 451–479. Lyakurwa, W., McKay, A., Ng'eno, N., & Keynes, W. (1997). Regional integration in sub-Saharan Africa: A review of experiences and issues. In A. Oyejide, I. Elbadawi, & P. Collier (Eds.), Regional integration and trade liberaliza- tion in sub-Saharan Africa. Volume 1: Framework, issues and methodological perspectives. Macmillan. Mbabazi, J., Milner, C., & Morrissey, O. (2006). Trade openness, trade costs and growth: Why sub-Saharan Africa performs poorly. (CREDIT Research Paper No 06/08). Novy, D. (2013). Gravity Redux: Measuring international trade costs with panel data. Economic Inquiry, 51(1), 101–121. Plümper, T., & Troeger, V. E. (2004). The estimation of time-invariant variables in panel analyses with unit fixed effects. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.565904 or https://ssrn.com/abstract=565904 Trefler, D. (1993). Trade liberalisation and the theory of endogenous protection: An econometric study of U.S. import policy. Journal of Political Economy, 101(1), 138–160. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 16 TURKSON eT al. Turkson, E. F. (2012). Using observable trade data to measure bilateral trade costs in developing countries: Evidence from sub-Saharan Africa. (CREDIT Research Paper 12/06), Nottingham School of Economics, University of Nottingham, UK. Turkson, F. E. (2015). Integration and regional trade in sub-Saharan Africa. In O. Morrissey, R. A. López, & K. Sharma (Eds.), Handbook on trade and development. Edward Elgar Publishing. Turkson, F. E., Mensah, R. A., & Barimah, A. (2020). Does trade facilitation promote bilateral trade in sub-Saharan Africa? Policy and issues. In G. O. A. Odularu, M. Hassan, & M. Babatunde (Eds.), Fostering trade in Africa - trade relations, business opportunities and policy instruments. Advances in African Economics, Social and Polit- ical Development, Springer Publications. ISBN/EAN 3030366316 / 9783030366315. Wei, S. (1996). Intra-National Versus International Trade: How Stubborn are Nations in Global Integration? NBER Working Paper Series. Working Paper 19285. http://www.nber.org/papers/w19285 Wilson, J. S., Njinkeu, D., & Fosso, B. P. (2008). Expanding trade within Africa: The impact of trade facilitation. (Policy Research Paper 4790), The World Bank Development Research Group. Yang, Y., & Gupta, M. S. (2005). Regional trade arrangements in Africa: Past performance and the way forward. In Regional trade arrangements in Africa: Past performance and the way forward (EPub) (Vol. No. 5-36). Interna- tional Monetary Fund. How to cite this article: Turkson, E., Oduro, A. D., Baffour, P., & Quartey, P. (2022). Regional integration and non-tariff barriers to Intra-Sub-Saharan Africa trade. The World Economy, 00, 1–19. https://doi.org/10.1111/twec.13365 APPENDIX 1: DERIVATION OF THE MICRO-FOUNDED MEASURE OF TRADE COST Novy (2013) specifies country i's domestic trade flow as: ( ) 𝑦𝑦2 1−𝜎𝜎 𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 = , (A1) 𝑦𝑦𝑤𝑤 Π𝑖𝑖𝑃𝑃𝑖𝑖 where xii and tii are domestic (intra-national) trade flows and trade costs respectively of country i. From equation (A1), the product of the multilateral resistance variables can be solved as; / 1 / 𝜎𝜎 − 1 ⎛𝑋𝑋𝑖𝑖𝑖𝑖 ⎞ ⎜ 𝑦𝑦𝑖𝑖 ⎟Π 𝑃𝑃 = 𝑡𝑡 , (A2)𝑖𝑖 𝑖𝑖 / 𝑖𝑖𝑖𝑖 ⎜ ⎟ 𝑦𝑦𝑖𝑖 ⎜ 𝑦𝑦𝑤𝑤 ⎟ ⎝ ⎠ As indicated in equation (A2), if domestic trade flows in country i (tii) is known, then given nomi- nal income in country i (yi), world income (y w) and the elasticity of substitution (σ), the multilateral trade resistance variables Пi and Pi would be known. Similarly for country j: / 1 / ⎛𝑋𝑋 𝜎𝜎 − 1𝑗𝑗𝑗𝑗 ⎞ ⎜ 𝑦𝑦𝑗𝑗 ⎟Π (A3)𝑗𝑗𝑃𝑃𝑗𝑗 = / 𝑡𝑡𝑗𝑗𝑗𝑗 . ⎜ ⎟ 𝑦𝑦𝑗𝑗 ⎜ 𝑦𝑦𝑤𝑤 ⎟ ⎝ ⎠ Clearly equations  (A2) and (A3) show that multilateral trade resistance does not depend on time-invariant proxies but rather easily observable time-varying trade data. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 17 The explicit solution for the multilateral resistance variables can be used to solve for bilateral trade costs from the general equilibrium model. To do this Novy (2013) obtained a bidirectional gravity equation by multiplying corresponding gravity equations for trade flows from the opposite direction (i.e. XijXji). That is: [ ] ( )1−𝜎𝜎 ( )1−𝜎𝜎 ( )2( )1−𝜎𝜎 𝑦𝑦𝑖𝑖𝑦𝑦𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 𝑦𝑦𝑖𝑖𝑦𝑦𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 𝑦𝑦𝑖𝑖𝑦𝑦𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 = = . (A4) 𝑦𝑦𝑤𝑤 Π𝑖𝑖𝑃𝑃 𝑤𝑤 𝑤𝑤𝑖𝑖 𝑦𝑦 Π𝑖𝑖𝑃𝑃𝑖𝑖 𝑦𝑦 Π𝑖𝑖𝑃𝑃𝑖𝑖𝑃𝑃𝑖𝑖Π𝑖𝑖 From equations (A2), (A3) and (A4): 1−𝜎𝜎 ⎡ ⎤ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ( )2 ( )𝜎𝜎−1 𝑦𝑦𝑖𝑖𝑦𝑦𝑖𝑖 ⎢ 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 ⎥ 𝑡𝑡𝑖𝑖𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖𝑋𝑋 𝑋𝑋 (A5)𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 = = 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 . 𝑤𝑤 ⎢ ⎥/ /𝑦𝑦 1 1 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 ⎡ / ⎤ ⎡ / ⎤ ⎢ 𝑋𝑋 ⎥𝑖𝑖𝑖𝑖 𝜎𝜎 − 1 𝑋𝑋𝑖𝑖𝑖𝑖 𝜎𝜎 − 1 ⎢⎛ ⎞ ⎥ ⎢⎛ ⎞ ⎥ ⎢ 𝑦𝑦𝑖𝑖 𝑦𝑦 ⎥𝑖𝑖 ⎢⎜ ⎟ ⎥ ⎢⎜ ⎟ ⎥ ⎢ / 𝑡𝑡𝑖𝑖𝑖𝑖 / 𝑡𝑡𝑖𝑖𝑖𝑖 ⎥ 𝑦𝑦 𝑦𝑦 ⎢⎜ 𝑖𝑖 ⎟ ⎥ ⎢⎜ 𝑖𝑖 ⎟ ⎥ ⎢ 𝑦𝑦𝑤𝑤 𝑦𝑦𝑤𝑤 ⎥ ⎝ ⎠ ⎝ ⎠ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦ Re-arranged 1/ ( ) 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 𝜎𝜎 − 1= . (A6) 𝑡𝑡𝑖𝑖𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 As bilateral and domestic trade costs can be asymmetric (i.e. tij ≠ tji and tii ≠ tjj), the tariff equiva- lent total trade costs (τij) could be obtained by taking a geometric mean of trade costs in both directions minus one: / / 1 1 ( ) ( ) 𝑡𝑡𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖 2 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 2 (𝜎𝜎 − 1)𝜏𝜏 = − 1 = − 1. (A7)𝑖𝑖𝑖𝑖 𝑡𝑡𝑖𝑖𝑖𝑖𝑡𝑡𝑖𝑖𝑖𝑖 𝑋𝑋𝑖𝑖𝑖𝑖𝑋𝑋𝑖𝑖𝑖𝑖 APPENDIX 2: MAIN VARIABLES: DEFINITION AND SOURCE • Data on Bilateral exports (Xij and Xjt) was sourced from sourced from UN-COMTRADE • Measures of Economic Size: GDP measured at Current US dollars • Measures of distance and other country characteristics such as contiguity, common language, common currency and common legal origin are sourced from CEPII • Number landlocked is constructed based on the number of bilateral pairs of country that are landlocked • Remoteness of country pair: a proxy variable for multilateral resistance to trade. It is calculated based on the approach used in Wei (1996) and Baier and Bergstrand (2009) as a function of bilat- eral distance and Gross Domestic Product. • Gross Foreign Assets is used as a measure of financial integration and is sourced from the World Development Indicators of the World Bank. • T rade agreement dummies generated for each RTA as follows: • RTA ijt: Dummy = 1 if i and j are SSA countries and at the same time t belonged to the same RTA, 0 if otherwise. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 18 TURKSON eT al. • ECOWAS RTA ijt: Dummy = 1 if i and j are both members of the ECOWAS RTA a time t, 0 if otherwise. • SADC RTA ijt: Dummy = 1 if i and j are both members of the SADC RTA a time t, 0 if otherwise. • EAC RTA ijt: Dummy  =  1 if i and j are both members of the EAC RTA a time t, 0 if otherwise. • ECCAS RTA ijt: Dummy = 1 if i and j are both members of the ECCAS RTA a time t, 0 if otherwise. • IGA RTA ijt: Dummy = 1 if i and j are both members of the IGAD RTA a time t, 0 if otherwise. APPENDIX 3: AVERAGE VALUES OF THE MAIN VARIABLES USED FOR THE ESTIMATION. All No RTA ECOWAS IGAD SADC EAC ECCAS Bilateral Exportij (US$ Million) 5.18 2.12 10.40 49.30 41.40 13.40 2.88 GDPi (US$ Billion) 11.60 11.30 10.80 16.80 20.40 3.07 9.44 GDPj (US$ Billion) 11.70 11.40 10.80 16.80 20.20 3.07 9.44 Distanceij (1000 pop-wt, km) 3.46 3.96 1.34 1.01 1.74 0.72 1.43 Number Landlockedij 0.69 0.71 0.38 1.34 0.84 0.78 0.93 GDP per capitait 1.15 1.17 0.53 0.40 1.85 0.20 2.30 GDP per capitajt 1.13 1.14 0.53 0.40 1.83 0.20 2.29 Remoteness Indexij (1000) 3.14 3.19 2.83 3.87 3.34 3.93 2.65 Gross Foreign Assetsit 262.00 243.00 0.00 0.00 1420.00 0.00 0.00 Gross Foreign Assetsjt 259.00 240.00 0.00 0.00 1420.00 0.00 0.00 APPENDIX 4: HAUSMAN SPECIFICATION TEST RESULTS (EQUATION 2) . Fixed effects Random effects Difference Standard errors (b) (B) (b-B) sqrt(diag(Vb-VB)) GDPit 1.930894 1.48847 0.442425 0.16538 GDPjt 3.400718 1.380291 2.020427 0.165267 Common Currencyijt 0.316098 0.228222 0.087876 0.045349 GDP per capitait −2.18755 −1.23183 −0.95572 0.172567 GDP per capitajt −3.45969 −1.00864 −2.45105 0.173298 Remotenessijt 0.562935 −0.29998 0.862912 0.051381 Gross Financial Integrationit −0.04802 −0.00152 −0.0465 0.003015 Gross Financial Integrationjt 0.034203 0.070014 −0.03581 0.002905 ECOWAS −1.0603 −0.79822 −0.26207 0.043431 IGAD −1.48827 −1.10022 −0.38806 0.088221 SADC 2.52618 2.24937 0.27681 0.044153 EAC 1.669856 1.626775 0.043081 0.053384 ECCAS −1.86922 −1.69739 −0.17183 0.048338 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TURKSON eT al. 19 b, consistent under Ho and Ha; obtained from xtreg; B, inconsistent under Ha, efficient under Ho; obtained from xtreg. Test: Ho: difference in coefficients not systematic. [ ] 𝐴𝐴 𝐴𝐴2(13) = (𝑏𝑏 − 𝐵𝐵)′ (𝑉𝑉 − 𝑉𝑉 )−1𝑏𝑏 𝐵𝐵 (𝑏𝑏 − 𝐵𝐵) = 850.37 . 𝐴𝐴 Prob > 𝜒𝜒𝟤𝟤 = 0.0000 . (Vb−VB is not positive definite). APPENDIX 5: REGIONAL ECONOMIC COMMUNITIES (RECS)-MEMBER COUNTRIES. No. ECOWAS (15) ECCAS (11) SADC (16) EAC (6) IGAD (8) 1 Benin Angola Angola Burundi Djibouti 2 Burkina Faso Burundi Botswana Kenya Eritrea 3 Cabo Verde Cameroon Comoros Rwanda Ethiopia 4 Ivory Coast Central African Rep. Congo, Dem. Rep. South Sudan Kenya 5 The Gambia Chad Eswatini Tanzania Somalia 6 Ghana Congo Lesotho Uganda South Sudan 7 Guinea Congo, Dem. Rep. Madagascar Sudan 8 Guinea-Bissau Equatorial Guinea Malawi Uganda 9 Liberia Gabon Mauritius 10 Mali Rwanda Mozambique 11 Niger Sao Tome and Principe Namibia 12 Nigeria Seychelles 13 Senegal South Africa 14 Sierra Leone Tanzania 15 Togo Zambia 16 Zimbabwe Note: Based on the 2019 Africa Regional Integration Index Report. 14679701, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/twec.13365 by University of Ghana - Accra, Wiley Online Library on [19/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License