Introduction

Signed by 54 of the 55 African Union (AU) member states, the African Continental Free Trade Area (AfCFTA) agreement entered into force on 1 January 2021 with the promise to boost economic development and to improve Africa’s position in global trade by leveraging economies of scale and the cost decrease of imported products. This study provides a comprehensive description of the free trade agreement’s impacts by focusing on the food sectors and on food security at multiple geographic scales (continental, regional and national).

The study employed a Global Trade Analysis Project (GTAP)-based economic model, which captured the impacts of the AfCFTA on trade flows at a bilateral level and across a wide range of merchandise and service groupings. The results cover a period up to 2035 and are evaluated against a multiannual socioeconomic development baseline. With the tariff offers still under consideration by the signatory countries, the analysis includes alternative liberalisation scenarios implemented across the continent, which have the objectives of government revenue maximisation on one hand, and the promotion of food security, of intermediate products or of revealed comparative advantage products, on the other.

Another area of uncertainty addressed in this report is the capacity of the trade agreement to reduce the costs related to non-tariff measures (NTMs). In the African context, the existing ad valorem estimations, although highly uncertain, highlight a higher burden of these costs on trade than the burden of tariffs. Therefore, the analysis includes results for both a ‘tariff-only’ reduction under the AfCFTA and a ‘tariff and NTMs’ liberalisation scenario. The latter anticipates that by 2035 the NTM-related costs for all intra-African trade will decrease by 50%, whereas those for imports and exports from and to AfCFTA third countries will decrease by 25%, to reflect wider gains from the harmonisation of standards at the border.

Scenarios

The study examines alternative policy scenarios that consider both tariff cuts alone and combined with NTMs reductions. The modelled tariff cuts take into account the AfCFTA tariff cut modalities, and the study addresses the uncertainty on the final tariff offers by constructing four alternative scenarios. The NTM reductions being even more uncertain, the study resorts to use a plausible level of NTM reductions in the absence of reliable estimates of the actual ones.

Import tariffs

As final tariff liberalisation offers from the AU member states detailing the exact tariff lines and rate cuts were not available at the time of drafting the report, several tariff liberalisation scenarios are constructed for each country separately while respecting regional trade blocs. Specifically, the scenarios are constructed utilising the information on the modalities on liberalisation of trade in goods agreed in the AfCFTA negotiations to date. The scenario construction allocates tariff lines into four lists of products: non-sensitive, non-sensitive G7, sensitive and excluded. The details of the liberalisation commitments for each of the four lists differ by country category – non-LDC, LDC and G7 – and are summarised in Table 1. For the list of excluded products, the AfCFTA adopts a double-qualification approach, limiting the number of tariff lines to 3% (or 156 lines) of all HS6 tariff lines, as well as limiting the value of intra-AU trade covered by selected lines to 10%, measured using the average import value over the period 2014–2016 or 2015–2017. The basis for the tariff liberalisation is the MFN rates applied in 2019, the year the AfCFTA came into force (Lunenborg, 2019).

Member states are tabling liberalisation offers individually unless they belong to an REC that is applying a common external tariff. The EAC, ECOWAS and the Southern African Customs Union (SACU) are customs unions, and the Central African Economic and Monetary Community (CEMAC) is in a late stage of forming one, so these four regional groupings are assumed to each table common offers (United Nations Economic Commission for Africa, 2018; Lunenborg, 2019). Considering some ambiguity about the negotiation and application details of the AfCFTA (Lunenborg, 2019), the following additional assumptions are made in the present analysis: the four customs union RECs are classified as non-LDCs (although they frequently include LDC countries), and each AU member state applies the same set of tariffs to all other AU member states, except if they belong to the same customs union.

Non-tariff measures

The AfCFTA will harmonise rules and regulations in many areas of merchandise and services trade (SPS standards, TBTs, etc.) and thereby reduce the cost of trading not only for trade within the AU but also for trade outside the AU. In the absence of information on how NTMs will be reduced by the AfCFTA, we assume that the scope of the AfCFTA facilitates a reduction of 50% in NTMs on merchandise and services trade for all intra-AU states. The AfCFTA is also assumed to reduce the cost of trading with countries outside of the AU as a result of greater standardisation and harmonisation of rules and regulations and a corresponding reduction in costs of compliance with foreign rules and regulations. NTMs for all imports and exports between AU and non-AU countries are assumed to decrease by 25%. The NTMs are reduced linearly, starting from the year 2020 and completing in 2035.

NTMs on merchandise are distinguished between technical measures (SPS and technical measures) and non-technical measures (contingent trade measures, quantitative restrictions, price controls, finance measures) (see Kee and Nicita, 2016), and are treated differently in the implementation of the simulation scenarios. All NTMs on services are assumed to be technical measures.

To address variation in these types of NTMs and their effects, we assumed that the technical measures are cost generating and represented them as iceberg costs in the model, and that the non-technical measures are rent generating and represented them as ad valorem tariff equivalents. The NTM reductions in the scenarios are distributed accordingly.

MAGNET enables policy experiments to be conducted, in which a reference scenario or baseline is first simulated over a future period and then, after changing one or more underlying assumptions (e.g. about policy settings or about exogenous macroeconomic developments, weather trends, etc.), a new scenario incorporating these changes is run over the same period. A comparison of the new scenario with the reference scenario at a given point in the simulation period, usually in terms of percentage differences, establishes the direction and relative magnitude of the impacts on all the endogenous variables of the change that is depicted in the hypothetical scenario at that point in time. In other words, the model is intended to compare, for the same moment in time (i.e. holding time constant), the outcomes prevailing in two or more different hypothetical ‘states of the world’ that might prevail at that point in time.

Although MAGNET could be used to project individual values of all variables, it must be stressed that it is not a forecasting model and users should be aware that the values projected may be unreliable as to what will happen in that year. Although this type of model is calibrated to fit a given year closely, its solutions become less reliable the further into the future it is used to simulate outcomes. Given the substantial number of assumptions, estimated or calibrated parameters, and stylised specification features that these models assemble, each of which is ‘correct’ only up to an (unknown) probability, it is impossible to establish confidence intervals or margins of error around individual projected numbers.

A further limitation relates to the coverage and the disaggregation of the countries and of the agricultural products in the database used to calibrate the model. The CGE model MAGNET has a comprehensive coverage of the world and of the economy, and thus of the agri-food sector and beyond. However, some of the most important cash crops and processed agricultural products that fall in another food category cannot be included in this analysis owing to data limitations. These products, which include coffee, cocoa, tea, and other cash crops, are typical flagship export products for various African countries. This limitation leads to underestimating the trade gains for the African agri-food sector in a broad sense. In addition, not all the African countries are included as individual countries in the database and some of them only appear aggregated in composite regions.

MAGNET also suffers from a typical caveat that applies to all CGE models employing the constant elasticity of substitution assumption: the so-called small share problem. The constant elasticity of substitution treatment underestimates trade creation opportunities when the import flow in the benchmark data is ‘small’. When the GTAP benchmark data import share is ‘small’, significant tariff reduction might induce a fall in the price but, even when coupled with a large trade elasticity, changes to bilateral imports will still be negligible (Philippidis et al., 2014). This is the case in many trade flows among African countries within the GTAP database. This caveat also refers to trade in new products that might appear on the market in the subsequent years. When the initial trade in these products is zero, it will remain zero in the model. This could be particularly relevant in the context of a rapid industrialisation process in African countries and their potential to enter new industries and produce goods they are not producing now.

Another notable caveat relates to the adjustment of the database to reflect the most up-to-date tariff rates between countries. As mentioned earlier, we apply the Altertax method by Malcolm (1998) for this purpose. In addition, the same adjustment is required for NTMs when they are modelled as AVEs rather than as iceberg costs. Typically, the required Altertax adjustment is much larger for NTMs than for tariffs. Therefore, the database deviates more from its original form in the case of the NTMs. The consequent changes in trade and income flows are a major weakness of the AVE approach. In addition, the Altertax method assigns all the rent generated by the NTMs as government income, which is hardly realistic and can lead to misleading changes in tariff revenues.

Database and aggregations

This study employs a fully consistent and academically recognised global database based on contributions from members of the GTAP network and constructed by the GTAP team at Purdue University, United States (Aguiar et al., 2019). Version 10 of the GTAP database contains a complete record of all economic activity (i.e. production, trade, primary factor usage, final and input demands, taxes, trade tariffs, and transport margins) for 65 activities and 141 regions for the year 2014. Our analysis employs an aggregation of the database that catches the most salient features of agri-food industries in various African countries. The aggregated database includes 40 tradable commodities and 36 regions, of which 29 are in Africa (see Figure 1 for details of regional aggregation). We make the results available in the REC level also. However, this poses a challenge since some of the countries that belong to an aggregated GTAP region belong to several RECs. In those few cases, we needed to deviate from the actual REC composition by assigning the aggregated region to its REC that was economically the largest. However, the deviation is small, and we believe that our REC aggregation gives an adequate view of the REC-level results. See the section on regional aggregations for more details.

From a food security perspective, the trade agreement is expected to lead to a rise in food availability and food consumption. However, agricultural production is predicted to become more concentrated in a few regions. For primary agricultural commodities, the output is predicted to increase most in Southern Africa but to decrease in central areas and many sparsely populated countries. A similar outcome can be seen for processed foods. There are a few exceptions to this pattern on a more detailed commodity level. Cereal output is predicted to decrease almost everywhere, the exceptions being increased wheat output in Eastern Africa and increased paddy rice output in Southern Africa. Oilseed output is predicted to increase, especially in Eastern Africa. Dairy production is predicted to grow notably in Nigeria and Côte d’Ivoire. The expansion of beef cattle production is predicted to be concentrated in Southern Africa, and the output of other meats is expected to increase, mostly in Southern and Eastern Africa. The output of fish is predicted to increase throughout Africa. The supply of services and food services is also predicted to increase throughout Africa, reflecting the change in income levels. The outcomes for food prices are mixed, with predicted decreases in prices in some regions and increases in others. The predicted increase in prices for some commodities and countries marks the need to address the issue of food affordability for low-income groups.

Both production and price changes in agri-food sectors reflect a reallocation of production means following the AfCFTA implementation as countries and regions steer trade towards products and services in which they have a comparative advantage. As a result, agricultural production at a continental level is predicted to grow less than GDP, indicating that, on average, non-agricultural sectors will become more important in the national economies and thus the share of value added in total output increases.

Overall, with tariff and NTMs liberalization intra-African trade is predicted to be 22% above the baseline level by 2035, with all regions and countries experiencing an increase in both exports from and imports to other AfCFTA members. On the exports side, the Economic Community of West African States (ECOWAS) countries are expected to have the highest growth rate, with a more uniform expansion across the other RECs. For imports, the largest growth in relation to the baseline is expected in Ghana (73%), Côte d’Ivoire (58%), Nigeria and South-Central Africa (both 50%), whereas, at the other end of the spectrum, many Southern African Development Community (SADC) countries and some Arab Maghreb Union (AMU) countries are predicted to have the lowest expansion.

At REC level, the ECOWAS is predicted to see the most significant increase in general exports (intracontinental and extracontinental), and the highest increase in export diversification away from agri-food trade. This will enable the largest increase in income per capita across the RECs (2.7% above the baseline in 2035). At the same time, the GDP gains are expected to be in line with the AU average (0.4%) and the corresponding positive welfare effect is predicted to be slightly reduced by an increase in food prices (1.0% above baseline values); nevertheless, there are differences between member countries, with Ghana and Nigeria having negligible GDP gains, whereas the other countries obtain a much greater increase (0.7–1.6% by 2035).

In the Common Market for Eastern and Southern Africa (COMESA), the predicted income expansion by 1.7% is in line with the AU average, with GDP increasing by 0.3%. The region is also predicted to have the second highest agri-food exports expansion (5.9% above baseline values), facilitated by an increase in trade flows within the continent, with intra-African exports increasing the most in Rwanda (33.5 %), Uganda (29.5 %) and Ethiopia (27.0 %).

Partially overlapping COMESA, the East African Community (EAC) and the SADC have different expected outcomes resulting from the AfCFTA implementation. The EAC is predicted to benefit from above the average GDP gains, resulting from an increase in total exports. At the same time, the EAC is predicted to have the highest increase in food imports (5.7% above the baseline) and food prices (1.4%), whereas the food consumption growth (0.6%) is predicted to be below the AU average.

The SADC is expected to have the second-highest increase in GDP facilitated by the rise in agri-food exports (6.2% above baseline values in 2035), with Rwanda and Botswana expanding the most. However, the food security gains in the region are moderate; the food consumption expansion is below the AU average (0.65% above the baseline) despite the second-highest decrease in food prices at a regional level (0.25% below the baseline), although not uniform across member states.

The Economic Community of Central African States (ECCAS) is expected to have the largest GDP, but a moderate food consumption expansion compared to the other RECs. The region has the lowest expansion of total intra-African exports (11.4% above the baseline in 2035), whereas the agri-food trade with the rest of the continent is predicted to expand above the AU average for both exports (20.3 %) and imports (28.0 %).

AMU is the REC in which the AfCFTA is predicted to have some of the lowest impacts on the economic and food security metrics. GDP, income, and trade are also estimated to be below the AU averages. The changes in trade diversification indices are lower than in the other RECs, although the countries in this region are already more integrated within the global trade networks. Nevertheless, by 2035, food exports expand at the highest rate of the RECs (5.6% above the baseline), while food consumption increases by 1%.

With very few exceptions, the tariff liberalisation results in a decrease in government tariff revenues relative to baseline levels, despite an increase in total trade volumes at the continental scale and across countries. Nevertheless, the AfCFTA stimulates general economic activity and thus indirectly generates additional fiscal revenue from other tax classes. When the effects of NTM liberalization are also included, this increase in non-tariff revenue is enough to compensate for the tariff liberalisation and to result in an overall positive impact on public finances in most countries.

Across the four liberalisation scenarios considered in this report, the one promoting trade in agri-food products results in the lowest food prices. This scenario also results in the largest increases in per capita food consumption. However, a trade-off comes from a slight reduction in the income gains. From a GDP perspective, the differences in impact across the scenarios, although small, are country and region specific and are dependent on the structure of each economy– at the REC level, the agri-food promotion liberalisation scenario leads to the highest GDP gains in the ECOWAS and the SADC, whereas a scenario promoting industrialisation through trade in intermediate products leads to better outcomes for COMESA, EAC and AMU.

An important aspect of food security is the affordability of food for the most vulnerable groups. We find that this indicator changes to positive on average, but in some regions basic food prices increase faster than the wages of low-income groups. This result indicates that due to the AfCFTA some regions could experience increase in food security risks in vulnerable groups. We also find that the negative effects can be mitigated if the regions uniformly choose to promote industrialization. Thus, our results suggest that promoting industrialization rather than low food prices is a better strategy for mitigating food security risks as it leads to a more positive income development in vulnerable groups. However, this result is not uniform across regions, and therefore non-uniform strategies, which take into consideration regional conditions, should be further investigated.

Policy recommendations

The study shows that, although food consumption increases on average throughout Africa, changes in food prices could make low-income earners more vulnerable to food security risks in some regions. There are regions where the prices of basic food items increase more than the wages on low-income groups. Therefore, governments in these regions are advised to take measures to protect the vulnerable groups. In particular, establishing food safety nets, and providing adequate income transfers, should be high on these governments’ agendas.

The study has also shown that the decisions on tariff offers can already mitigate the worst food security outcomes as the uniform adoption of promoting industrialization yields both the highest overall improvement of food security in vulnerable groups, it also avoids the worst outcomes in regions that will encounter negative consequences. Promoting low food prices in particular is not as beneficial as regards food security, since it does not increase the incomes of the African households as much as aiming for a more rapid industrialization and improvement of general value added of production do. However, as the results vary by region, the recommendation should be considered within the local context.

Secondly, given the structure of the database underlying the MAGNET model, the impacts of the agreement cannot be evaluated in detail either for all countries (some of them are not available within the latest GTAP version as individual countries) or at sectoral level (the commodity disaggregation is set by the structure of the GTAP database). To overcome these limitations, a linkage between the MAGNET model, a single-country CGE model and a farm-level model developed and maintained in house by the JRC is under development. These linkages should enable the results produced by the global CGE MAGNET model to be used in country-level models to produce a refined analysis, including more detailed sectoral results and impacts on a wider range of agents (e.g. urban and rural households in various regions of the countries, and farmers).

Future research should also benefit from an improved modelling approach in allocating the NTM reductions between import and export costs. Despite all these uncertainties around the analysis of the NTM reduction, the results highlight that there is a potentially large gain for the AU from an ambitious commitment to lower NTMs.

An additional stream of research should focus on the impact of a higher level of integration among African countries, including free mobility of people and/or capital.

Finally, further model enhancements should be developed to include a more refined analysis of social (nutrition, employment, migration) and environmental (emissions, resource use, footprints) indicators affected by the trade agreement and the possible interactions and impacts on food security between Africa integration and climate change issues.

Abbreviations and definitions

AfCFTA - African Continental Free Trade Area

AMU - Arab Maghreb Union

AU - African Union

AVE - ad valorem - equivalent

CEMAC - Central African Economic and Monetary Community

CGE - computable general equilibrium

COMESA - Common Market for Eastern and Southern Africa

EAC - East African Community

ECCAS - Economic Community of Central African States

ECI - export concentration index

ECOWAS - Economic Community of West African States

EDI - export diversification index

EU - European Union

G7 - Group of seven particularly vulnerable AU Member States

GDP - gross domestic product

GTAP - Global Trade Analysis Project

HS - Harmonized System (of tariff nomenclature)

HS4 - Harmonized System (of tariff nomenclature; four-digit level)

HS6 - Harmonized System (of tariff nomenclature; six-digit level)

IMF - International Monetary Fund

JRC - Joint Research Centre

LDC - least-developed country

MAcMap - Market Access Map

MAGNET - Modular Applied GeNeral Equilibrium Tool

MFN - most-favoured nation

NTM - non-tariff measure

REC - Regional Economic Community

SACU - Southern African Customs Union

SADC - Southern African Development Community

SPS - sanitary and phytosanitary

SSP - Shared Socioeconomic Pathway

TBT - technical barrier to trade