Viner’s (1950) seminal theory of CU serves as the foundation for welfare, macroeconomic and trade analysis of Preferential Trade Agreements (PTAs). Subsequent refinements of the ideas of Viner (1950) by Johnson (1962), Kendall and Gaisford (2007) offer clear expositions of CU theory, while Baldwin and Venables (1995) and Panagariya (2000) survey the more recent extensions of the theory. The CU theory describes the welfare effect of a PTA on each member as the sum of the efficiency gains or losses that result from trade creation and diversion and its Terms of Trade (TOT) changes. Non-members are also affected through changes in their TOT. Because the TOT gains of one country are the TOT losses of its trading partner, the net TOT effect is zero at the global level suggesting global welfare to be the sum of gains and losses in regional efficiency.

The impacts of an FTA on the members and non-members of the agreement are mainly transmitted through welfare, macroeconomic and trade effects. As such, the transmission channels of welfare effects include allocative efficiency, endowment effect and TOT effects. For macroeconomic effects, the transmission channels include real gross domestic product (GDP), export volume and trade balance, while those of trade effects include trade creation and trade diversion.

A key question raised by the formation of an FTA is whether it will make member countries better or worse off. In fact, FTAs are expected to be welfare enhancing. This is partially possible through allocative efficiency (i.e. optimal allocation of resources) stemming from factors such as specialisation and comparative advantages, improved market access and competition, dynamic effects on growth and technological progress, consumer benefits and income effects (Leibenstein 1966). In terms of dynamic effects, for instance, through promoting specialisation, innovation and knowledge transfer, free trade can stimulate productivity growth and technological advancement over time (Acemoglu 2008; Romer 1990). These dynamic effects further enhance allocative efficiency and contribute to long-term improvements in welfare. Similarly, the endowment effect can promote efficiency in resource allocation by encouraging individuals to allocate resources towards goods they highly value (Burfisher 2017). Also, improvement in the TOT can enhance welfare by increasing purchasing power, boosting incomes, improving the balance of payments, stimulating investment and growth and reducing relative poverty (Feenstra 2015; Krugman & Obstfeld 2009). Within this context, the following null hypothesis is specified:

Free Trade Agreements can influence the macroeconomic environment of member states through growth in GDP and trade balance. As such, FTAs can stimulate GDP growth by promoting trade, investment and economic efficiency (Hufbauer & Schott 2005). For example, increased exports resulting from improved market access can contribute positively to economic growth. Trade balance can be affected through the alteration of import and export flows. In fact, FTAs can theoretically lead to increased exports because of reduced tariffs and trade barriers, potentially improving the trade balance (Baier & Bergstrand 2007). However, this effect may vary depending on the specific terms of the FTA and the competitiveness of domestic industries. In this setting, the following null hypothesis is specified:

Drawing from Viner (1950), FTAs theoretically affect trade through trade creation and trade diversion, which juxtapose the welfare-enhancing effects within the established or potential trade agreement with the welfare-reducing effects emanating from trade discrimination (Burfisher 2017). According to Viner (1950), trade creation is defined as the shift in the volume of production of traded goods from a high-cost producer in the established agreement to a lower-cost member. The resulting increase in production efficiency unambiguously increases global welfare. Trade diversion, in contrast, is defined as the shift in members’ source of imports from low-cost non-members to high-cost members resulting in deterioration of production efficiency and welfare. Hence, the following null hypothesis is specified:

Insights from the literature relating to the welfare, macroeconomic and trade effects of existing and potential FTAs follow.

The probing of welfare consequences of FTAs is an empirical matter, and estimating welfare implications of potential trade agreements is challenging as it requires knowledge of the counterfactual (Freund & Ornelas 2010). Because this remains unknown, assumptions must be made. It is in this context that we see the prevalence of one strand of extant literature employing gravity models (Caliendo & Parro 2015; Cernat 2001; Qiu, Tian & Wang 2022; Singh 2021; Timsina & Culas 2022), CGE models (Abrego et al. 2019; Mustilli 2015; Quartey 2023; Timini & Viani 2022; Yi 2020) and PE models (Biden, Ker & Duff 2020; Jagdambe & Mouzam 2019; Makochekanwa 2014; Sabaruddin 2022; Wonyra & Bayale 2022) to examine the welfare, macroeconomic and trade impacts of potential trade agreements. In cases where trade agreements are already in existence, we see another strand of literature (Alhassan & Payaslioğlu 2023; Clausing 2001; Díaz-Mora, Esteve-Pérez & Gil-Pareja 2023; Khadan & Hosein 2015; Uzair & Nawaz 2018; Wei, Chen & Rose 2019) using disaggregated data to examine the welfare, macroeconomic and trade impacts of particular trade agreements.

However, for a trade agreement to be welfare improving, sufficient conditions must be satisfied. Kemp and Wan (1976) point out that a trade agreement is necessarily welfare improving if external tariffs are adjusted so that its establishment does not discriminate against external trade. According to Freund and Ornelas (2010), the logic for this is facile given that if external tariffs necessitate that external trade is perpetual, it follows that any supplementary intra-trade should be trade creating. This warrants that non-members are not made worse off by the formation of the trade agreement, and this general outcome also extends to FTAs (Panagariya & Krishna 2002).

By using variation in liberalisation across industries to isolate trade creation and diversion, Trefler (2004) finds both trade creation and diversion in the Canada-US FTA but determines positive welfare effects on the average Canadian. Likewise, through the utilisation of CGE analysis, scholars such as Evans (1998) and Lewis, Robinson and Thierfelder (1999) find positive net effects of regional integration initiatives in Southern Africa, while Flôres (1997) reaches comparable inferences concerning MERCOSUR. Cernat (2001) also dismissed the widely held opinions and theoretical expectations that South-South trade agreements are trade diverting and concluded otherwise. However, in the case of the Canada-US FTA, Romalis (2007) finds that the expansion of the Canada-US FTA to Mexico (i.e. NAFTA, now USMCA) has been trade diverting.

Besides the aforementioned seemingly revolutionary and mid-pioneering studies, many other contemporary studies have examined the welfare and trade effects of distinct existing and potential trade agreements. Such studies include Uzair and Nawaz (2018), who find substantial evidence in favour of trade creation in the Pakistan-China FTA; Abrego et al. (2019), who find significant welfare gains from the formation of the African Continental Free Trade Area (AfCFTA), emanating mostly from a reduction in Non-Tariff Barriers (NTBs); Wei et al. (2019), who find positive welfare gains for both participants in the South Korea-US FTA; Laget et al. (2020) who find that the depth of trade agreements contributes to an increase in GVC trade among parties; Oberhofer and Pfaffermayr (2021), who find that the negative welfare effects of Brexit can be partially compensated for by the conclusion of FTAs with other countries, including the European Union (EU); and Mattoo, Mulabdic and Ruta (2022), who find that deep trade agreements lead to more trade creation and less trade diversion than shallow agreements and that some provisions of deep trade agreements have a public good aspect and enhance trade also with non-members.

For the potential SACU-US FTA, in particular, DeRosa and Gilbert (2004) revealed that the economic welfare of SACU would deteriorate as a consequence of an individual FTA with the US. An analysis by Brown et al. (2008) of the expected economic effects from the potential FTA concluded that the welfare benefits of an FTA between SACU and the US were relatively small and that the interests of both parties, along with the global community, could be better served by unilateral liberalisation, particularly in a multilateral setting. In contrast, Kiyota and Stern (2005) showed a significant increase in welfare gains for both SACU and the US as an outcome of the likely formation of the SACU-US FTA. However, whereas Lehloenya (2009) acknowledges that the failed 2003–2007 SACU-US FTA would have brought some benefits such as locking in AGOA benefits, he submits that the finalisation of the trade deal would have probably caused more harm given the extensive intellectual property rights sought by the US, as well as the country’s persistent refusal to reverse its farm subsidy policy.

In light of the aforementioned literature appraisals, it is apparent that there are mixed conclusions about the welfare, macroeconomic and trade implications of the failed SACU-US FTA. These divergences may be partially attributable to variations in the models applied (i.e. CGE, PE or gravity model) and whether the assessment was carried out ex-ante or ex-post. Also, in studies where CGE modelling has been applied using the GTAP model, the simulated scenarios, the model version and the regional and sectoral aggregation schemes adopted often differ. Despite this, it must be acknowledged that the global political and domestic economic conditions within both SACU and the US have significantly transformed since then. In this regard, AGOA, which is set to expire in 2025, might not be renewed. Hence, considering the lingering uncertainty surrounding the expiry of AGOA and the significance of the US in SACU members’ trade and economic development endeavours, SACU has to be proactive. This entails locking in AGOA benefits through re-engaging the US in negotiations of an FTA that builds on AGOA by strengthening trade and investment relations while addressing AGOA drawbacks and taking reciprocity into account. It is here where the present article joins the conversation by assessing the welfare, macroeconomic and trade effects of this prospective SACU-US FTA while simultaneously reviving the scholarly discourse on the flouted 2003–2007 SACU-US FTA negotiations and contributing to the knowledge base of the extant trade negotiations and agreements literature.

The GTAP model, initially formulated by Hertel (1997), is the most extensively applied CGE model in trade policy analysis (Plummer, Cheong & Hamanaka 2011).The model’s theory is based on the multisectoral model of the Australian economy, the ORANI model, developed by Dixon et al. (1982). Hence, the standard GTAP model is a multi-region, multisectoral CGE model with perfect competition and constant returns to scale (Burfisher 2017).

In the GTAP model, bilateral trade is handled via the Armington assumption, which states that products that are manufactured in distinct regions are imperfect substitutes in demand (Armington 1969; Burfisher 2017). In other words, products are differentiated not exclusively by their nature but also by their place of origin. For example, SACU chemicals, US chemicals, SACU machinery and US machinery might be four different products distinguished in the GTAP model.

The GTAP-CGE model, tried and tested by many scholars including Siriwardana (2006), Stenberg and Siriwardana (2015), Hussain and Ali Shah (2017), Itakura (2020), Kitetu, Ko and Sone (2023), is appropriate for analysing the effect of crucial policy transformations, such as the formation of the potential SACU-US FTA (Zeng 2013). It can compute the probable welfare, macroeconomic and trade implications of the prospective SACU-US FTA ex-ante by means of mathematical simulation (Rosyadi & Widodo 2018). This simulation method is suitable for the analysis in this article as there is currently no ex-post data generated from the likely full liberalisation resulting from the implementation of the potential SACU-US FTA.

This article, therefore, utilises the static version of the standard GTAP model (Version 10), regarded as the latest version following numerous revisions of the standard GTAP model owing to various modelling concerns (Itakura & Hertel 2001). The GTAP 10 database features four reference years (i.e. 2004, 2007, 2011 and 2014). The latest reference year (i.e. 2014) with 141 GTAP regions for all 65 GTAP sectors and eight factors were selected for aggregation (Aguiar et al. 2019). To test the hypothesis developed and specified in this article, the welfare, macroeconomic and trade of the potential SACU-US FTA are simulated under the following ambitious or full trade liberalisation policy reform scenario: SACU-US full trade liberalisation policy reform in which they eliminate all their import taxes and export subsidies to their bilateral trade. In this regard, the import tariffs and export subsidies applied by SACU and the US on their bilateral trade are eliminated.

A synopsis of the simulation process is visually illustrated in Figure 2. It begins by first aggregating the GTAP regions, sectors and factors as well as creating the aggregation model (SACU_US_13 × 10 model) in the GTAP Aggregation (GTAP-Agg) database. For the model experiment simulated in this article, the 141 GTAP regions are aggregated into 13 new regions, while the 65 GTAP sectors are aggregated into 10 new sectors (see Table 1-A1 in Appendix 1). The eight GTAP factors are aggregated into four new factors, with labour and capital considered mobile. In contrast, land (ETRAE value = -1.000) and natural resources (ETRAE value = -0.001) are considered sluggish (see Table 2-A1 in Appendix. ETRAE is a transformation elasticity for mobile factors. It indicates how easy or difficult it is to transfer a sluggish factor (e.g., capital) from one sector to another.

Besides isolating SACU members and the US as the trading partners of interest in this article, firstly, the regional aggregations were based on global regional classifications. China was isolated because of its vast influence in global trade, which could be of interest, along with being one of the largest trading partners of both SACU countries and the US. The sectoral aggregations follow the default aggregations in the GTAP model database because the foci of the assessments in this article were placed on aggregate welfare, macroeconomic and trade effects of the prospective SACU-US FTA. This also applies to the factor aggregations where all labour skill levels were aggregated into labour.

Secondly, the created aggregation file was opened in RunGTAP, and the model was prepared for the running of the model experiment by checking the closure and shocks. In this regard, the ‘standard general equilibrium closure’ of the GTAP model was adopted for the simulation performed in this article. According to Hertel (1997), under the standard general equilibrium closure, price elasticity parameters can react to shocks from both the demand and supply sides. Furthermore, unemployment closure was applied in the case of SACU countries. This is sensible because all SACU countries suffer from high unemployment. The shock list was cleared to ensure that there are no shocks lingering in the experiment file other than the ones introduced in the model experiment of this article.

Thirdly, the base import and export tax rates applied by SACU and the US on their bilateral trade were checked (see Table 3-A1 in Appendix 1) before defining the model experiment. The shocks were defined by setting the tms (import tariff) and txs (export tax) variables to a zero-target rate for all the traded products originating from the US imported by the SACU members (Botswana, Namibia, South Africa and the Rest of SACU) and vice versa. For conciseness’ sake, the closure and shock definitions are available upon request.

Fourthly, the model experiment was then saved and solved using Gragg’s 2-4-6 steps solution method with the ‘automatic accuracy’ option permitted to ensure maximum accuracy of the results obtained. The method splits the shock with interpolation into minor increments and iterates the calculation numerous times (Rosyadi & Widodo 2018). The final solutions are obtained from the average values of each iteration’s solution. After solving the experiment, the updated GTAP import and export tax rates applied on the bilateral trade between the US and SACU were checked if they were now zero (see Table 3-A1 in Appendix 1).

Illustratively, when running a CGE model experiment such as the one in this article, the modeller is likened to a billiard player who hits the ball initiating a chain of reactions and interactions among all of the balls on the table and who must wait to see where all the balls come to rest (Burfisher 2017). In the same manner, the entire CGE model equations should be resolved to obtain new solution values for the endogenous variables in the model. The new values represent a new equilibrium at which the demand and supply quantities are equal again at some set of prices. The GTAP-CGE model experiment simulated in this article does not show the adjustment process. In other words, we did not watch as the billiard balls knocked against each other as they traverse the table.

This article does not contain any studies involving human participants performed by any of the authors.

The equivalent variation (EV) was used to measure the net welfare gains originating from the potential SACU-US FTA, simulated under an ambitious or full trade liberalisation policy reform scenario. Equivalent variation is a money-metric measure of the value to the consumer of the price changes because of a shock (Burfisher 2017). The decomposition of the EV of the potential reciprocal trade deal (see Table 1) reveals positive net welfare gains for both SACU (US$316.47 million) and the US (US$678.01 million). This points to an expansion in economic welfare because of the trade creation resulting from the potential reciprocal bilateral trade agreement (BTA).

As expected in the case of SACU, South Africa achieved more significant welfare gains of US$283.74 million compared to the combined welfare gains of US$35.86 million achieved by Lesotho, Namibia and Eswatini. Botswana experienced a welfare loss of US$3.14 million emanating primarily from its TOT and endowment effect. However, the endowment effect is the primary source of welfare gains for SACU, contributing US$311.27 million to its net welfare gains originating from the formation of the potential SACU-US FTA. The endowment effect arises from changes in the availability of primary factors, such as the expansions in the stock of machinery and buildings (Hanslow 2000).

The magnitude of the welfare gains emanating from the potential SACU-US FTA is much larger for the US relative to SACU. Unlike SACU, the endowment effect contributes nothing to the net welfare of the US. However, the TOT effect is the primary source of the US’s welfare gains. In this regard, the TOT effect contributes US$536.62 million towards the net welfare gains of the country resulting from the formation of the potential SACU-US FTA.

At a global level, the formation of the potential SACU-US reciprocal trade agreement improves global welfare by US$314.40 million. This is despite only SACU and the US achieving positive welfare gains. At the same time, all the other regions are expected to attain negative welfare gains because of the formation of the potential reciprocal trade agreement. The most affected regions are the EU and China with a net welfare loss of US$216.89 million and US$165.53 million, respectively. This might be partially attributable to the diversion of trade from the EU and China, which are also important longstanding trading partners of SACU and the US.

The positive welfare gains revealed in the assessments of this article receive empirical support from the findings of Kiyota and Stern (2005) who showed a significant increase in welfare gains for both SACU and the US as an outcome of the likely formation of an FTA between SACU and the US. Makochekanwa (2014) also estimated a positive net welfare gain of US$205.5 million in the Common Market for Eastern and Southern Africa (COMESA)-East Africa Community (EAC)-Southern African Development Community (SADC) tripartite FTA, while Abrego et al. (2019) find significant welfare gains from the formation of the AfCFTA, and Wei et al. (2019) achieve similar findings in the case of the South Korea-US FTA.

The GTAP simulation results of the macroeconomic effects of the potential fully liberalised SACU-US FTA are shown in Table 2. The formation of the FTA is expected to improve the real GDP of SACU by 0.10%, while that of the US will increase by 0.02%. Comparably, a study by Zhang et al. (2015) that evaluated the macroeconomic effects of the Canada-US FTA showed a positive and permanent effect of 1.86% on Canada’s real GDP growth. However, the scale of real GDP growth achieved in the present article is lower. Within SACU, the combined real GDP of Lesotho and Eswatini increases by 0.52%, while that of Namibia and South Africa increases by 0.19% and 0.02%, respectively. The real GDP of Botswana, however, declines by 0.03% and that of other regions declines by very small magnitudes. For Botswana, the decline in the GDP may be because of sectoral shifts expected from the potential SACU-US FTA and the economic structure of the country that may be vulnerable to competitive pressure from the US and South Africa, SACU’s largest economy.

In terms of export volume, the potential trade deal is expected to improve the export volumes of SACU, the US and China by 0.61%, 0.02% and 0.01%, respectively. The export volumes of all the other regions are expected to remain unchanged, while their import volumes (including that of China) deteriorate by smaller magnitudes. The import volumes of SACU and the US improved correspondingly by 0.90% and 0.07%. Within SACU, South Africa’s export and import volumes increased by 0.68% and 1.05%, while Botswana saw a decline in its export and import volumes by 0.01% and 0.04%, correspondingly. For both SACU and the US, the improvement in their import volumes is greater than the improvement in their export volumes. This is partially the reason for the expected deterioration of the SACU (except for Botswana, whose trade balance improves by US$0.33 million) and the US trade balances by US$544.54 million and US$728.63 million, respectively. In contrast, the trade balances of all the other regions are expected to improve.

Only the TOT for the US improves by 0.02% because of the formation of the potential SACU-US FTA. However, the TOT for SACU deteriorates by 0.08%, while that of America deteriorates by 0.01% and that of all the other regions remains unaffected. The potential trade deal also improves the regional household income of SACU and the US by 0.04% and 0.02%, respectively. Nevertheless, the household income of all the other regions is expected to deteriorate by very small magnitudes.

The real import (quantity) changes, which indicate trade creation and trade diversion effects of the potential SACU-US FTA, are presented in Table 3. The potential trade deal is expected to be significantly net trade creating in all the sectors for both SACU and the US. This is because the increase in the quantities of their bilateral imports in those sectors exceeds the diversion of their imports of those sectors from other regions. For instance, the increase in SACU’s light manufacturing imports from the US by US$1.47 billion diverts only US$793.89 million worth of light manufacturing imports from all the other regions.

Similarly, except for the mining and extraction sector where SACU members enjoy considerable export supply sustainability, with a net trade diversion of US$2.34 million, the increase in the US imports from SACU does not divert any trade from other regions in all the sectors. Instead, US imports from other regions increase most noticeably in the heavy and light manufacturing sectors, where the real imports of the US increase by US$329.40 million and US$215.21 million, respectively. Overall, the formation of the potential SACU-US FTA is welfare improving as it creates a net trade of US$3.06 billion. This is in line with the findings of studies such as Cernat (2001), Trefler (2004) and Uzair and Nawaz (2018) that revealed evidence in favour of trade creation in the respective trade agreements assessed.