How will Brexit affect the patterns of European agricultural and food exports?

Abstract

1. Introduction

The UK’s decision to leave the European Union (EU) after the results of the 2016 British referendum marks a turning point in European history and raises many questions about the future economic relationships between the two parties. The agrifood sector has been caught in the middle of these events. The UK’s large financial contribution to the EU’s common agricultural policy (CAP) was one of the main arguments of the pro-Brexit vote. Still, the fact that the UK is a net importer of agrifood products and an important destination market for EU agrifood exports suggests that a change in trade policy is likely to have important repercussions on this sector. The EU and UK agrifood sectors are strongly interconnected. UK is highly dependent on agrifood imports from the EU, accounting for 69 per cent of the country’s total imports in this sector based on 2015 data (the year before the referendum). More specifically, Netherlands, Ireland and France are the UK’s largest foreign suppliers. Reciprocally, the British market is a major destination of European agrifood exports, absorbing 9 per cent of the EU’s worldwide exports in this sector.

Most of the works in the literature evaluate the effects of Brexit at country level, and only a small number perform sector-level analyses. The present paper investigates the effects on the EU agrifood sector and identifies the products and EU countries most harshly to be affected by Brexit. For that, we estimate the change in trade flows for the entire agrifood sector and 13 groups of products, both at country level and by bilateral trade relationships, under different scenarios. Our analysis provides a higher level of product-level detail than existing works. Moreover, we consider not only potential scenarios but also the UK’s actual trade policy after Brexit.

EU–UK negotiations were marked by strong uncertainty on achieving a trade deal. Negotiations started in March 2017, when the British government notified its intention to leave the EU based on article 50 of the European treaty, and lasted 45 months, largely exceeding the 2 years of transition period stated in the treaty. A trade and cooperation agreement was concluded on 24 December 2020 and ratified by the UK parliament on 20 December 2020 and by the European Parliament on 27 April 2021. The agreement has been provisionally applied since 1 January 2021, although it officially entered into force on 1 May 2021.

The prospects of reaching a deal remained grim until the very last moment, fuelling speculation on a large array of possible post-Brexit trade policy scenarios. In the present paper, we evaluate the effects of Brexit on trade in agricultural and food products by comparing the present EU–UK deal to four alternative policy scenarios. Differently from previous studies, our scenarios refer not only to the outcome of EU–UK trade negotiations but also to the future trade relationships between the UK and its non-EU partners. The EU–UK trade deal introduces zero tariffs and zero quotas on bilateral trade in all products, provided the rules of origin requirements are fulfilled. It also implies simplified customs procedures, although the border-level checks apply to all traded goods. Compliance with EU’s high regulatory standards, such as sanitary and phytosanitary standards and food safety regulation, is required and checked for all exchanged goods, although the agreement introduces some facilitations on certifying the conformity with technical regulations and standards.

First, due to the rules of origin introduced by the present EU–UK agreement, we assume that zero tariffs will apply to most but not all EU–UK trade flows. British imports from countries with which a preferential trade agreement (PTA) was already signed are subject to preferential tariffs, while those from the other non-EU countries are subject to the new UK Global Tariff (UKGT, introduced in May 2020). We consider the rest of bilateral trade relationships to continue under the same terms as before Brexit.

Then, we consider two possible alternatives for the future UK–EU trade relationship: a free trade agreement (FTA) similar to, but deeper than, CETA (EU–Canada Comprehensive Economic and Trade Agreement) and a return to World Trade Organization (WTO) rules implying equal treatment for all foreign suppliers.¹ Then, for each of these two cases, we consider two options for trade between the UK and non-EU countries: the replication of EU trade agreements and PTAs with the UK’s largest non-EU partners. Combining the two policy alternatives for the EU–UK trade relationship and two options for the UK’s trade relationship with third countries, we obtain four symmetric trade policy scenarios. All scenarios are defined by changes in import tariffs and non-tariff measures (NTMs).

We use a structural gravity model and estimate the effects of Brexit following the approach developed by Anderson, Larch and Yotov (2018). This choice is motivated by a number of factors. Being designed to estimate effects at the bilateral level, the model permits to quantify (more accurately than other approaches) trade diversion between the UK and each EU country. Structural gravity is well adapted for estimating effects at the product (or product group) level. It also allows to include a very detailed bilateral trade cost function and to control for both bilateral and unilateral trade determinants without making strong and hard-to-check assumptions. We account for the direct effects of Brexit corresponding to changes in trade costs, as well as the indirect effects induced by changes in multilateral resistance (MR), output and expenditure. Our results reflect long-term general-equilibrium impacts. We base our computations on data over the period 2012–2015, as the baseline for the predictions. Our data panel starts in 2012, after the effects of the 2008–2009 economic crisis were completely absorbed in most countries, and ends in 2015, the year before the Brexit referendum. Unlike most of the previous works on the effects of Brexit, we estimate trade elasticities and use observed data on different types of NTMs.²

In the light of the UK’s new trade agreements and UKGT, the costs of trading with the UK will increase for EU countries and decline for some of its non-EU partners, strengthening the competition on the British market, as well as on all markets to which trade will be diverted. Results indicate only marginal changes in EU imports, exports, prices and real income, with stronger negative effects for Ireland and even more for the UK. We expect trade between the EU and the UK to decline by around 10 per cent and for both parties to register stronger losses for meat and meat products.

The remainder of the paper is structured as follows. The next section summarises the UK–EU negotiation process. Section 3 describes the methodology and data. Results of the five scenarios are presented and discussed in Section 4. Section 5 presents the effects by groups of products and Section 6 resumes our main findings.

2. Negotiations and scenarios

2.1. A long and turbulent road of EU–UK negotiations

Brexit negotiations were led by Michel Barnier for the EU and by British Prime Ministers Theresa May and Boris Johnson for the UK. The very different positions adopted by the UK and the EU and the need to obtain the agreement of each of the 27 EU countries increased the difficulty of negotiations. The tight margin of the Brexit vote and strong differences in results (position pro or against Brexit) across British regions, socio-economic and age groups impeded the country’s main political parties to adopt a uniform position on Brexit. Both the governing (Tory/Conservative) and the opposition (Labour) parties counted a large number of Brexit supporters and Brexit opponents, which complicated even more the negotiations. Surprised by the outcome of the Brexit referendum, the then Prime Minister David Cameron resigned shortly after the vote. Brexit negotiations were launched under Theresa May, who stepped in office in July 2016. On 29 March 2017, the UK notified its intention to withdraw from the EU by activating Article 50 of the EU Treaty. This marked the beginning of a 2-year negotiation process between the UK and the EU on withdrawal conditions and the future bilateral relationship, which was later extended until 31 December 2020.³

From the beginning of the negotiations, the UK has spelled out its intention to negotiate a bold and ambitious FTA with the EU, while respecting four red lines: ending the jurisdiction of the European Court of Justice, controlling immigration from the EU, ending most contributions to the EU budget and being able to strike trade deals with third countries. The EU adopted a firm position that a non-member of the Union cannot enjoy the same rights and benefits as a member country and condemned the UK’s cherry-picking attitude. Matthews (2017) argues that the difficulty of reaching a EU–UK agreement was due to the fact that the UK’s most preferred scenario—a deep and comprehensive economic and FTA similar to the one between EU and Canada, which would avoid NTMs and limit regulatory alignment with the EU—was always rejected by the EU because it dissociated the four pillars of the single market (free movement of goods, services, capital and people).

After several rounds of negotiations, a Withdrawal Agreement was concluded in October 2019 between the EU and the UK government under Boris Johnson and came into force on 1 February 2020. This marks the UK’s de jure exit from the EU and the beginning of negotiations of a future EU–UK trade agreement. The agreement includes a protocol with provisions on preventing the introduction of a hard border between the Republic of Ireland and Northern Ireland. It sets the de jure customs border between the UK and the EU on the island of Ireland but the de facto customs border on the Irish Sea. Also, the UK collects import tariffs on behalf of the EU on goods shipped from Great Britain to Northern Ireland that are ‘at risk’ of being transported into and sold in the EU. The protocol also includes a unilateral exit mechanism for Northern Ireland, which will vote every 4 years on whether to continue these arrangements or not.

To smooth the impacts of the shock on the British economy, in 2019, the UK government announced that, in case of a no deal, it would apply at least temporarily a unilateral trade liberalisation policy by removing tariffs on a large share of British imports. A year later, on 19 May 2020, the British government adopted a new Most Favoured Nation (MFN) tariff regime—the UKGT—that will apply to imports from all trade partners with which the UK has no trade agreement. The UKGT introduces lower import tariffs than the EU’s Common External Tariff and expands tariff-free trade to a wider range of products. Import tariffs are maintained on a number of products backing UK industries, such as agriculture, automotive and fishing (e.g. lamb, beef and poultry). The UKGT applies to imports from all countries, except those with which the UK has negotiated a FTA and those around 70 developing countries that enjoy a preferential access to the British market.⁴

2.2. Quantifying Brexit trade effects, a brief literature review

The economic literature abounds with studies on economic integration, a large number of them focusing on integration within Europe. Episodes of disintegration were analysed by economists mostly from a historical perspective (e.g. Head, Mayer and Ries, 2010, on the deterioration of post-colonial trade; De Ménil and Maurel, 1994, on the breakup of the Austro-Hungarian empire) or from a change of the economic system (e.g. Maurel and Cheikbossian, 1998; Djankov and Freund, 2002; Fidrmuc and Fidrmuc, 2003, on the collapse of the Soviet Union, the dissolution of Yugoslavia and of Czechoslovakia). In this context, Brexit comes as an unprecedented example of economic disintegration and marks a reversal in the recent history of increased integration and globalisation that has characterised the past seven decades.

Another strand of literature highlights that trade patterns change relatively slowly, confirming the presence of hysteresis in trade. For instance, De Sousa and Lamotte (2007) find no evidence that the political disintegration of communist states in the 1990s has led to systematic and severe trade disintegration. Similarly, Buch and Toubal (2009) and Nitsch and Wolf (2013) show, on the case of German re-unification, that borders have a long-lasting effect on trade flows and trade openness. They find a persistent impact of the former East–West border on German domestic and foreign trade flows even 15 years after the fall of the Berlin Wall. Nitsch and Wolf (2013) estimate that it will take 33–40 years to remove the impact of this political border on German trade. These results raise the question of how quickly Brexit will dissolve the EU–UK ties and partnerships developed over the 47 years of joint economic integration, and how rapidly the economic and commercial impediments introduced by Brexit will be reflected in European and British trade patterns.

Over the last years, a number of studies evaluating the economic costs and benefits of Brexit with different methodological approaches have emerged in the literature. Some authors focus more generally on economic disintegration and consider Brexit as a specific illustration of this process (e.g. Sampson, 2017; Larue, 2018). Douch, Edwards and Soegaard (2018) use a synthetic control method to evaluate the effect of Brexit announcement on the UK’s bilateral trade with various partners, considering the Brexit vote as a treatment effect on trade flows involving the UK. Born et al. (2019) employ the same method to estimate the UK’s loss in real gross domestic product (GDP) caused by the Brexit vote, combined with an expectations-augmented vector autoregression model to identify the drivers of this effect.

A large number of works evaluate the effects of Brexit using a computable general equilibrium model (e.g. Kierzenkowski et al., 2016; Bellora et al., 2017; Copenhagen Economics, 2018; Figus et al., 2018; Jafari and Britz, 2020). HM Treasury (2016) and Erken et al. (2018) use a general equilibrium setting, combined with econometric models to measure the costs of Brexit for the UK. Other studies, including van Berkum et al. (2016), Davis et al. (2017), van Berkum et al. (2018) and Choi et al. (2021), use partial equilibrium models. Finally, Dhingra et al. (2017); Oberhofer and Pfaffermayr (2021); Felbermayr, Gröschl and Steininger (2018); Brakman, Garretsen and Kohl (2017); Bruno et al. (2017) and Mayer, Vicard and Zignago (2019) use a structural gravity approach to estimate the effects of Brexit on trade, foreign direct investments and welfare. The latter three studies obtain the effect of Brexit as the opposite of the EU membership effect. More differently, Graziano, Handley and Limão (2020) and Douch, Du and Vanino (2020) focus on the effects of trade policy uncertainty introduced by Brexit. All these studies find a strong negative impact on the British economy and smaller negative effects on the EU, unevenly distributed across member countries. Still, the magnitude of effects differs significantly across studies.

Besides this methodological variety, there are also distinctions according to the geographical reach of analyses. While some studies focus on the impact of Brexit on the British economy, others place the UK’s trade partners in the core of the analysis. For instance, Lawless and Morgenroth (2019) analyse the effects on EU countries (on both British exports to EU countries and EU exports to the UK) by taking into account the great variation in tariffs across products, which fills in a gap in aggregate studies. Their results show an extremely wide range of country-level reductions in trade with the UK. For some EU countries—Estonia, Finland, Latvia and Slovenia—the estimated impact on total trade is minimal, less than 0.5 per cent. Ireland is the most severely affected with an estimated 4 per cent drop in exports. The food sector is one of the hardest hit sectors, with bilateral EU–UK expected to drop by up to 90 per cent in case of a no-deal scenario (especially for meat and confectionary products). Copenhagen Economics (2018) emphasises the effects on the Irish economy. Whatever the scenario analysed, it finds that Brexit will have negative impacts on Irish trade, with adverse knock-on effects on Irish production and ultimately Irish GDP. By 2030, the latter is estimated to be 2.8–7 per cent lower than the non-Brexit baseline GDP level. They also show that the agrifood is one of the five most affected sectors, the predicted fall in trade and production being the largest for processed foods, beef, sheep and other cattle meat, and dairy.

Most analyses are carried on all economic sectors combined. Bellora et al. (2017) and Choi et al. (2021) are some of the few that focus on the effects of Brexit on the agricultural and food sector, for the EU and the UK. Bellora et al. (2017) find that, due to marked dissymmetry, macroeconomic impacts on the UK are significantly larger than on the EU. However, because of the UK’s high integration with EU countries in terms of trade and value chains, authors argue that all EU countries will be negatively affected by Brexit, the magnitude of the impact increasing with economic proximity to the UK, and largest for Ireland, followed at a great distance by Belgium and the Netherlands. Choi et al. (2021) conclude that whatever the future trade relationship between the UK and the EU, there will be disturbances in agrifood markets. They show that, only in the case of a FTA with the EU, the retention of the UK’s CAP contribution permits to increase the country’s net welfare. Net market impacts are expected to be small for the EU, but the loss of the British contribution to CAP leads to a reduction in net welfare in most scenarios. A few studies highlight the effects of Brexit on the British agrifood sector (e.g. van Berkum et al., 2016; Davis et al., 2017; Bradley and Hill, 2019) or on that of separate EU countries (e.g. Donnellan and Hanrahan, 2016, on Ireland; Yu et al., 2017, on Denmark; van Berkum et al., 2018 on the Netherlands). For instance, Davis et al. (2017) show that impacts on prices and production value in the UK agricultural sector vary greatly by products, while van Berkum et al. (2018) find that Dutch exports to the UK and the rest of the world are affected only marginally, either the EU and the UK reach a FTA or not.

Finally, most studies analyse the effects at the country level. Still, a handful of studies illustrate the variation of effects across EU and British sub-national regions (Chen et al., 2018; Figus et al., 2018). Chen et al. (2018) construct an index reflecting the degree to which regions and countries are exposed to negative trade‐related consequences of Brexit, accounting for the geographical fragmentation of the production process. They find that British regions are far more exposed than comparable-size regions in EU countries. The most affected EU region is the Republic of Ireland, with exposure levels similar to UK regions, followed by German, Dutch, Belgian and French regions.

2.3. Actual vs. potential trade arrangements

As the UK leaves the EU, producers and consumers on both sides need to absorb the additional trade costs associated with the border regulations, customs clearance procedures, regulatory, sanitary and phytosanitary checks and food safety controls of traded products, as well as tighter regulation of road transportation. The burden of these changes is particularly strong on agrifood products, which include many perishable goods, are subject to a large number of border controls and often require specialised control and transportation equipment (see Matthews, 2017, for a detailed discussion). Although both the UK and the EU declared their willingness to prevent or limit the increase in bilateral trade costs, Brexit will unavoidably increase the costs of shipping goods across the English Channel. The importance of costs associated with non-tariff barriers is also highlighted by Choi et al. (2021).

The most common approach in the literature on the effects of Brexit is to consider two extreme scenarios, a ‘hard’ and a ‘soft’ Brexit. The ‘hard Brexit’ scenario assumes that the UK will leave the EU and end its current (free or preferential) trade agreements with both the EU and third countries. Furthermore, this scenario assumes that no new FTA between the EU and UK could be established. Therefore, the UK would trade with all countries, including EU members, based on WTO regulations. According to the ‘soft Brexit’ scenario, the UK may stay in the Single Market even after the exit from the EU. It assumes that the UK will pursue all trade agreements with third countries inherited from the EU. It has however some drawbacks for the UK since it would not allow the UK to place restrictions on immigration from the EU and it would require the UK to implement EU economic legislation without having any part in deciding the legislation. Some intermediary scenarios concerning the EU–UK agreement are also considered in the literature (e.g. van Berkum et al., 2016; Dhingra et al., 2017; Davis et al., 2017; Bradley and Hill, 2019).

The present paper questions the effects of long-term trade policy arrangements. Even if negotiations between the EU and the UK were concluded and a trade agreement was already reached, Brexit is not over. There is still some uncertainty on how the current EU–UK trade agreement will be respected, as well as on the evolution of the UK’s future trade policy with third partners (the negotiation of new agreements). The new EU–UK agreement relies on the respect of the Irish/Northern Ireland protocol, the fulfilment of requirements in terms of rules of origin, as well as the compliance of exchanged products with partner’s standards and regulations. Deviations from these provisions may lead to a non-application of the agreement to the concerned flows (e.g. MNF instead of zero tariffs), the application of fines or financial sanctions; or even to a suspension of the treaty (the extreme case). As early as in March 2021, the EU initiated two infringement procedures against the UK for a breach of the Northern Ireland protocol after the British government announced its unilateral intention to extend the entry in force of some provisions of the Irish/Northern Ireland protocol. In late March to early April 2021, the introduction of customs regulations and border controls between Northern Ireland and the Great Britain caused temporary supply shortages in Northern Ireland and fuelled (along with other factors) violent riots across the region. These series of recent events reveal a certain fragility of the EU–UK trade agreement and some of the possible negative effects of Brexit, such as increasing uncertainty for trading firms and putting at risk the accomplishments of the 1998 Good Friday Agreement (that has secured a lasting peace in Northern Ireland).

In order to evaluate the magnitude of Brexit effects, we compare the effects of the newly reached EU–UK agreement and the UK’s new Global Tariff with an array of alternatives in terms of the bilateral EU–UK trade policy and of the UK’s trade policy with third countries. We consider two scenarios for the UK–EU trade arrangements: (i) a FTA similar to, but deeper than, CETA, usually referred in the literature as the ‘optimistic scenario’ and close to the status quo (the UK’s membership of the EU) and (ii) a return to WTO rules, usually referred in the literature as the ‘pessimistic scenario’. For each of these scenarios, we analyse two alternatives for the UK’s arrangements with non-EU trade partners: (i) a replication of EU agreements with third countries or (ii) the negotiation of PTAs with UK’s main extra-EU partners (more ambitious than EU agreements) and a return to WTO rules for the rest of countries (Table 1).

In the above scenarios, a FTA between the UK and EU countries implies zero import tariffs on bilateral trade and an increase in the difference of NTMs to the level observed for the UK’s trade with partners with which it has no preferential trade arrangement. PTAs between the UK and its main non-EU trade partners are modelled as a 50-per cent decrease in import tariffs and pre-shipment measures. We consider the USA, Australia, New Zealand, Switzerland, Chile and Israel the UK’s main extra-EU trade partners with which it has a strong incentive to negotiate PTAs. This list includes the top three developed countries in the UK’s foreign trade that currently do not have any trade arrangement with the UK (the USA, Australia and New Zealand) and the three countries with which the UK has already negotiated a PTA (Switzerland, Chile, and Israel). We exclude the emerging countries that count among the UK’s main trade partners, such as China and India. Despite their large volume of trade with the UK, these countries have a high level of protection of their domestic agricultural markets. Therefore, we find it very unlikely that the UK would easily reach a trade agreement with any of them. The return to WTO rules means the application of MFN import tariffs and setting NTMs between the UK and EU countries equal to the average level observed between the UK and countries with which it has no preferential trade arrangement.

Due to the rules of origin introduced by the treaty, we assume the actual EU–UK agreement (zero tariffs) applies to most, but not all, EU–UK trade flows. Based on recent data on the EU countries’ preference utilisation rates for different product categories (Nilsson and Preillon, 2018), we expect 84 per cent of the EU–UK trade in agrifood products to fulfil the rules of origin requirements and be subject to zero tariffs and the remaining 16 per cent to face MFN tariffs. British imports from countries with which a PTA was already signed are subject to preferential tariffs; British imports from the rest of the countries are subject to import tariffs specified in the UKGT. Since the treaty permits NTMs to diverge between the EU and the UK, we assume that the dissimilarity in NTMs increases to half of the level of those observed between the UK and partners with which it traded under MFN terms before Brexit.

The alternative four scenarios offer some lower and upper bounds for the changes in trade costs induced by Brexit. Comparing the effects of the concluded EU–UK trade agreement with results under different scenarios reveals how exposed are British and EU agrifood exports to the change in the UK’s trade arrangements with the EU and with third countries.

3. Methodology and data

3.1. Structural gravity

We aim to estimate the effects of Brexit in terms of changes in UK–EU trade flows. We focus not only on the direct changes in the trade flows of changes in trade costs after Brexit but also on the indirect impact of these changes on other trade flows via adjustments in price indices, expenditure and output levels.

|${X_{ijt}}$| denotes the value of exports from origin country i to destination country j in year t at destination prices (as paid by consumers in j), |${E_{jt}}$| is the expenditure of consumers in destination country j on products from all origins, |${Y_{it}}$| is the value of output of goods produced in origin country i, |${\tau _{ijt}}$| are the bilateral trade costs between the two countries, |${\Pi _{it}}$| and |${P_{jt}}$| reflect the ease of access to the global market of exporter i and, respectively, importer j, |$\sigma \gt 1$| is the elasticity of substitution of goods from various origins and |${\varepsilon _{ijt}}$| is a zero-mean error term.

Equation (1) is referred in the literature as the gravity model of international trade. Adding the constraints expressed by Equations (2) and (3), one obtains a structural gravity model.

According to Brakman, Garretsen and Kohl (2017), MR terms are at the core of modern formulations of gravity models. They are often interpreted as price indices and are crucial to analyse the effects of a change in trade policy. Without these terms, the simulated effects of a trade policy change would only affect the two countries involved. Including these terms, one accounts as well for effects on countries’ access to the global market, i.e. the way the change in trade policy affects the entire trading system.

3.2. Estimation strategy

We can now turn to computing the effects of a change in trade costs. Each trade policy scenario analysed in the paper corresponds to a hypothetical level of trade costs after Brexit, obtained by changing the values of trade policy variables for trade flows involving the UK. For each scenario, we identify the corresponding matrix of trade flows (as well as the MR, expenditure and output levels) that satisfy the set of Equations (2)–(3) and (5)–(7). To find this new equilibrium, we follow the procedure developed by Anderson, Larch and Yotov (2018). The impact of Brexit on agricultural and food exports for a specific scenario is computed by comparing the value of exports predicted by the model with counterfactual (post-Brexit) trade costs and with observed (before Brexit) trade costs.

More precisely, we use the values of parameters estimated on observed international trade flows and predicted domestic flows to compute the counterfactual trade costs for each scenario. Then, we run the model with counterfactual trade costs and unchanged values of parameters (⁠|${\beta _1}$|–|${\beta _4}$| and |${\delta _m}$|⁠) using an iterative process. Each iteration consists of two steps. First, we keep the baseline values of Π_it and P_jt unchanged and predict the level of trade flows, expenditure, output and domestic (factory gate) prices.⁵ Second, we re-compute MR terms Π_it and P_jt using predicted data from the first step and adjust the value of output, expenditure and factory gate price for each country.⁶ The structure of the model permits to compute countries’ adjustments to the new trade environment. The iterative process stops when the difference between domestic prices in the last two iterations is equal to 1 per cent or less for all countries in the model.

We estimate the model on observed 2012–2015 bilateral trade flows using the PPML approach developed by Santos Silva and Tenreyro (2006, 2011), which permits to account for zero flows and correct for heteroscedasticity in the data. We rely on a very convenient feature of the PPML estimator demonstrated by Fally (2015) that estimated fixed effects are exactly equal to the MRs that satisfy the structural gravity model.

Table 2 summarises our five Brexit scenarios based on the trade cost specification expressed by Equation (4). Each scenario implies a change in import tariffs and some NTMs between the UK and some other partners. The evolution of trade flows under each scenario enables to evaluate the magnitude of the effects of Brexit on international trade in agrifood products.

3.3. Data

We use a trade cost structure combining tariffs, NTMs, geographic distance and standard variables on bilateral linkages between trading countries used in the trade literature (common land border, common language and common colonial past). Accordingly, our final data panel is obtained by combining several sources of data.

Trade flows at the Harmonized System (HS) six-digit level come from the BACI database. Geographical distances and bilateral links between countries (common land border, common language and common colonial past) are obtained from the CEPII. Data on import tariffs and NTMs at the HS six-digit (HS6) level come from the UNCTAD’s TRAINS database. We use the applied import tariffs computed by UNCTAD from data on reported ad-valorem duties and specific duties converted in ad-valorem equivalents. Most NTMs vary only across importing countries and products. One cannot separate the effect of this variable from other importer-specific terms in the model. To overcome this limit, we compute NTM regulatory distances following the approach introduced by Cadot et al. (2015) using original UNCTAD data on the NTMs applied at the HS six-digit product level. For each NTM type m, the regulatory distance is the normalised number of narrowly defined categories of NTMs (within type m) imposed on a given product, both in the exporting country and the importing country. It takes values between 0 and 1. A value equal to 1 corresponds to the case when the two countries apply totally different types of NTMs (total dissimilarity), while a value equal to 0 reflects the situation when countries apply exactly the same narrowly defined categories of NTMs.⁷ We compute regulatory distances for each type of NTMs within the six most commonly used broad categories of NTMs, which account for 78.5 per cent of the data on NTMs (see Table 3). NTM data are not well documented before 2012. Therefore, in estimations, we use data starting from 2012.

The regulatory distances for SPS and TBT measures (types A and B) are highly correlated with each other. Therefore, we take the average of the two measures for each country pair. The regulatory distances for contingent trade-protective NTMs (type D) are very close to zero for all country pairs. The regulatory distances for price-control NTMs (type F) are highly correlated with regulatory distances for other types of NTMs. For these reasons, we drop these two regulatory distances.

Due to data limitation on tariffs and NTMs, as well as for computational convenience, we consider only trade between the world’s largest exporters and importers of agrifood products. Solving the structural gravity requires a full trade matrix and associated trade costs. To keep in the analysis some large countries that report few data on import duties, missing tariff data were filled in by extrapolating the existing data. Still, we cannot overcome the absence of NTM data for some important exporters (e.g. Norway and Egypt) without making arbitrary or unrealistic assumptions. Therefore, we excluded from our analysis countries that do not report any NTM data. Our final data panel contains the full trade matrix of 57 exporting and importing countries (Table A1). The data panel accounts for 76–77 per cent of the global trade in the agrifood sector from 2012 to 2015. It covers 84–85 per cent of the selected countries’ annual exports and 91–92 per cent of their annual imports of agricultural and food products to/from all partner countries.

We run the analysis on the entire agrifood sector and on 13 constituent groups of products. To obtain import tariffs and NTM regulatory distances at this level of aggregation, we take the average of values at the HS6 level, weighted by the global trade in each HS6 product. Table 4 provides descriptive statistics on import tariffs and regulatory distances used in aggregate-level estimations, for different types of trade flows. The data confirm that the EU market is highly protected in terms of both tariffs and NTMs. Our data show as well that SPS measures are the NTMs that vary the most across countries. SPS measures and TBTs reflect norms and standards that are defined not only for a trade perspective and take time to be changed. Quantity-specific NTMs complement the country’s trade policy reflected by tariffs. Differently, pre-shipment measures capture most of the procedures and border controls introduced by Brexit on the EU–UK trade. For these reasons, our scenarios include only changes on the regulatory distance of pre-shipment NTMs. The regulatory distance under the MFN regime is the weighted average of the regulatory distances between the UK and its main MFN trade partners at the HS6 level, using as weights the share of each HS6 product in the UK’s trade with these countries.⁸ Under all scenarios, we assume that the UK maintains the same regulatory standards as the EU.

The product groups and their representation in our data panel are listed in Table 5.

4. Results at the aggregate level

Table 6 presents the PPML estimates of Equation (5) for aggregate agrifood trade between the 57 countries in the panel over 2012–2015. For our estimation strategy, we choose the USA as the reference country. Accordingly, the inward multilateral resistance (IMR) term of this country is normalised to 1, P_USA= 1, for all years and product groups. This choice stems from the fact that the USA is an important trade partner for all countries in our data panel, including the UK, but not an EU member (hence not involved in Brexit negotiations). Another aspect that supports our choice is the fact the USA has the most balanced trade in agrifood products in our panel of 57 countries.

We obtain domestic trade (⁠|${X_{iit}}$|⁠) by plugging into Equation (7) the estimated value of parameters and importer and exporter fixed effects obtained from regressing Equation (5) on observed international trade data. The results of estimating Equation (5) on international trade flows (with omitted fixed effects for the USA) are reported in column (1) of Table 6. Running the estimation on the full trade matrix with both domestic and international flows, displayed in column (2), yields very small adjustments in the value of parameters. The effect of traditional gravity variables is in line with the empirical trade literature. Bilateral trade decrease with distance and import tariffs and is stronger for countries sharing a common land border, language or colonial past. Different pre-shipment and quantity-control NTMs in the exporting and importing countries (large regulatory distance for type C and type E NTMs) also deter trade. The estimated effect of discrepancies in countries’ sanitary and phytosanitary standards and technical barriers is not significantly different from zero.

Using estimation results from Table 6 and the methodology outlined in Section 3.2, we estimate the direct and indirect effects of a change in trade costs. Table 7 reports the estimated impacts in a general equilibrium framework for the aggregate agrifood sector for an elasticity of substitution |$\sigma $| equal to 4.⁹ This value is at midway between the 1.6 estimate of Raimondi and Olper (2011) for aggregate trade in food products and the 8.2 estimate of Imbs and Mejean (2017) for processed foods. Estimated impacts for each of the five scenarios described in Tables 1 and 2 and discussed in Section 2.3 are reported below. Among our sample of 57 trade partners, we focus on the impacts for the UK and its main EU partners (Netherlands, Ireland, France, Germany, Spain, Belgium-Luxembourg and Italy),¹⁰ as well as for its main non-EU trade partners. The latter comprise six countries with which the UK is assumed to conclude PTAs in scenarios S3 and S4 (USA, Australia, New Zealand, Chile, Israel and Switzerland) and other six countries with which the UK trades the most (China, Brazil, Thailand, India, Canada and Argentina). We also report average impacts for the EU.¹¹ For each country, we display the percentage change in imports and exports at the global level and bilateral with the UK, as well as the change in the factory gate price of domestic producers and in the price-adjusted revenue. To understand the mechanisms behind some of these results, we also list changes in the average price of products sold in the country (IMR) and in the average price of domestic products sold worldwide (OMR).¹²

A change in trade costs for a given bilateral relationship affects the average price faced by consumers in the importing country, with a potential impact on the competitiveness of products sold in this market. Thus, if the UK’s post-Brexit trade policy leads to an increase (decrease) in the average price of products purchased by UK consumers, goods for which the price remains unchanged appear as more (less) competitive since they are now compared to a higher (lower) reference value. These changes in relative prices will lead to a reallocation of consumers’ demand across products of different origins. Similarly, changes in global trade patterns affect producers’ revenues in exporting countries. Thus, if non-EU countries’ overall exports (to all destinations) increase after Brexit, the revenues of these countries will also increase, with a larger amount to be spent on all products, including on imports from EU countries. The general equilibrium framework of the structural gravity model permits to account for direct effects (changes in trade costs of concerned flows), as well as for indirect effects (changes in trade costs between other partners and worldwide adjustments in terms of price indices, supply and demand) on global trade patterns. We report in Tables 7 and 8 the results for the UK and EU countries and those for non-EU countries in Table A2.¹³

The UK’s actual trade policy

The UK’s actual trade policy, reflected by scenario S5, produces intermediate size effects compared to the other four scenarios. We find that it will generate a decline of nearly 6 per cent in British exports and of 4 per cent in imports. This is the net effect of a relatively strong contraction of the UK’s trade with EU countries (−10 to −11 per cent) and of a strong increase in its trade with third partners, especially of imports from Chile (+31 per cent), Switzerland (+53 per cent) and Canada (+61 per cent). The intensification of trade with Chile and Switzerland is explained by the PTAs signed with these countries; we expect the effect for Canada to diminish with the increase in EU–Canada trade due to CETA. Besides, results suggest a 1.76-per cent increase in the average price of agricultural and food products for British consumers and a 0.33-per cent fall of real income. Contrary to the UK, effects are much smaller for the EU, with a marginal decline in overall imports and exports (−0.32 per cent and −0.25 per cent, respectively), and real income (−0.05 per cent). Results in Table 8 show that EU exports to the UK will be partly redirected to both intra-EU and extra-EU markets. Within the EU, Ireland is the most severely affected country. Still, the negative effects of Brexit on the Irish economy are much smaller than on the UK. The model predicts a slight decrease in the average price of products sold on the EU market (captured by the IMR), except for Ireland where it generates the opposite effect.

Effects on the UK

Although imperfect, the UK’s actual trade policy permits to avoid the higher losses in terms of UK’s overall trade in agrifood products and real income (−0.86 to −0.95 per cent), as well as the strong increase in consumer prices that would have been observed in case of an exit with no deal (scenarios S2 and S4). The latter effect comes from the higher price of EU products (charged with high MFN tariff rates), as well as from the switch to more distant alternative source countries (subject to larger transport costs, some also subject to import tariffs). The present deal also limits the contraction of the EU–UK trade, which would have lost 25–30 per cent of its value under these scenarios according to estimations in Table 8. Results suggest that scenario S3 is the less damaging for the British economy. In this case, PTAs with main third partners compensate for the drop in exports and imports induced by larger trade costs with the EU. The UK has already made a first step in this direction by negotiating agreements with half of these countries (Chile, Switzerland and Israel). Liberalising trade with the other three (USA, Australia and New Zealand) would greatly attenuate the negative effects of Brexit.

Effects on the EU

For the EU, the implemented trade agreement leads to a stronger disturbance of trade patterns (larger amounts of trade reoriented from the UK to alternative markets), a more pronounced decrease in supply prices and a higher relative drop in real income compared to the outcomes of scenarios S1 and S3 (which assume the status quo for the relationship with the UK). The real income of EU consumers decreases under all scenarios, the strongest effect being with scenario S4 (−0.13 per cent). Similarly, we observe a decrease in overall exports and imports of all EU countries in all five scenarios, the effect being the largest (2.3–2.6 per cent) under scenario S2. In this case, the increase in EU–UK trade costs deteriorates the access of European goods to the global market: their average price index to consumers in the destination country (OMR) increases by 0.9 per cent. The average price of goods sold on the EU market (IMR) decreases by 0.6 per cent. While scenarios with no EU–UK trade deal (S2 and S4) predict a relatively small drop in EU’s overall exports and imports, the effect on bilateral EU–UK trade may get quite large.

Effects on third countries

The effects of different scenarios on the UK’s main non-EU trade partners are displayed in Table A2. Results are highly differentiated across countries. All these countries increase their trade with the UK under all scenarios. In some cases, effects are very large: up to a 2-fold increase in trade flows under scenarios S3 and S4. As in the case of the EU, results in terms of real income are very small in magnitude. Among developed countries, the effect is the largest for Switzerland (+0.03 per cent under S5; +0.08 per cent under S2 and S3; +0.09 per cent under S4). Similarly, effects on supply prices and consumer prices are the strongest for Switzerland, but they remain small (below 1 per cent).

5. Effects by groups of products

For the product-level analysis, we let trade cost parameters in Equation (5), including the elasticity of substitution |$\sigma $|⁠, to vary across product groups.¹⁴ Product-level estimations, reported in the Appendix in supplementary data at ERAE online, display a great variation in the statistical significance and the magnitude of estimated effects across product groups. We obtain counter-intuitive significant effects mainly for products subject to excise duties (‘tobacco’, ‘wines’ and ‘other alcoholic beverages’).¹⁵ Cross-country differences in the level of excise duties or in other types of regulations aiming to reduce the consumption of these products often outweigh the trade cost advantage induced by geographic proximity, historical ties and/or similar levels of NTM protection. Unsurprisingly, a few effects opposite to the ones usually found in the literature are also obtained for the highly heterogeneous group ‘other products’.¹⁶ Using a |$\sigma $| different from its estimated value only marginally affects the magnitude and the statistical significance of most estimated parameters (see the Appendix in supplementary data at ERAE online). The most sizable changes are the increase in the magnitude of effects of regulatory distances for ‘meat & meat products’, ‘oilseeds & vegetable oils’, ‘wines’, and ‘other alcoholic beverages’, with some of these effects becoming statistically significant. Accordingly, the simulation results for these product groups should be interpreted with some reservation, privileging the sign and the relative evolution of effects rather than actual figures.

The regulatory distances for the three types of NTMs included in the model have mainly a negative effect on trade flows. Still, our estimates show that an increase in the SPS-and-TBT regulatory distance leads to an increase in the volume of trade in ‘meat & meat products’ and in ‘coffee, spices, cocoa, sugar’. Although different SPS measures and TBTs in two countries do not necessarily reflect a difference in the stringency of their norms, this should be the case at least for some (if not most) country pairs and products. Therefore, we can interpret the above-mentioned result as evidence of the fact that for concerned product groups, SPS measures and TBTs, unlike other NTMs, are perceived by foreign competitors as informative measures that provide details on market entry conditions, rather than as barriers.

The UK’s actual trade policy

Under the UK’s actual trade policy, the European ‘meat & meat products’ sector will be most severely affected. As can be seen from Figure 1 and Table A4, this product group will suffer the strongest decline in EU global exports (4.14 per cent), followed at a great distance by ‘vegetables’ (−0.50 per cent) and ‘coffee, spices, cocoa, sugar’ (−0.32 per cent). The drop in EU exports to the UK are much higher: −16.9 per cent for ‘meat & meat products’, −8.2 per cent for ‘coffee, spices, cocoa, sugar’, −5.9 per cent for ‘fish & sea products’ and −5.4 per cent for ‘vegetables’. For all products, quantities that can no longer be exported to the UK are redirected mainly to the extra-EU market. The drop in EU exports to the UK is accompanied by a stronger increase in exports to extra-EU partners and a small or negative change in exports to EU partners (see Table A4). Unlike other product groups, European exports of meat products and vegetables will be redirected exclusively to non-EU markets. Adjustments in terms of prices are also strongest for meat products, with a 1.45 per cent drop in supply (factory-gate) prices of EU products and a 1.77 per cent drop in the average price of products sold on the EU market.

Fig. 1.

Effects of the UK’s actual trade policy on the EU (scenario S5).

Open in new tabDownload slide

Similarly to results at the aggregate level, negative effects are considerably larger for the UK than for EU countries. Detailed simulation results by groups of products and countries are reported in the Appendix in supplementary data at ERAE online. Results suggest that the present trade arrangement will generate a significant drop in British exports of ‘meat & meat products’ (−14.5 per cent) and a small drop in exports of ‘dairy’ (−3.1 per cent) and ‘fish & sea products’ (−2.0 per cent), effects being stronger on exports to EU partners. British imports of these products will also register an important switch to extra-EU suppliers, with the aim to adjust to the new market conditions. Despite this effect, consumer prices (the IMR index) for these products will increase by 1–2 per cent, leading to a contraction of British imports of 1.9–8.5 per cent. The UK’s trade in other products will be less affected due to newly negotiated trade agreements and the improved access to the British market for non-EU partners provided by the country’s new Global Tariff.

Effects on the EU: comparison across scenarios

The European meat sector would have also suffered the most if no EU–UK trade deal was reached (the group with the largest negative effects under scenarios S2 and S4). However, results for this group of products and five more are obtained with a higher value of |$\sigma $| than the estimated one¹⁷ and therefore need to be used with caution.¹⁸ Rather than focus on actual figures, the interpretation of results for these groups of products should rely on comparing effects across countries and scenarios (and less with other groups of products). Ireland is particularly vulnerable in the meat sector: it registers the strongest decrease in global trade within the EU, comparable to that of the UK, and the lowest decrease in exports to the UK.¹⁹ This result is induced by the strong interconnection of Irish producers with British value chains and their increased difficulty to access the EU and extra-EU markets.²⁰ The absence of a post-Brexit agreement would have had a strong impact also on EU exports of ‘fish & sea products’, ‘vegetables’ and ‘fruit’. For these groups of products, a no-deal Brexit would have generated a 1.2–1.6 per cent drop in EU exports worldwide and a 19–32 per cent contraction of EU exports to the British market. More generally, the absence of a deal would have led to a significant drop in the EU–UK trade in all products. This effect is largely attenuated under the UK’s current trade policy.

Introducing regulatory checks on the EU–UK border (scenarios S1 and S3) yields very small variations in the volume of EU trade. For example, under scenario S1, a change is observed only for products subject to a large number of sanitary and food safety controls and NTMs, namely ‘meat & meat products’, ‘dairy’, ‘fish & sea products’ and ‘tobacco’. EU products that can no longer be sold to British consumers are redirected to both EU and extra-EU markets, as shown in Table A4. Changes in trade patterns generate negligible price adjustments on the EU market and on products exported by EU countries.

Effects on the UK: comparison across scenarios

Effects on the British agrifood sector also differ greatly across groups of products. We find that a no-deal Brexit would have produced the largest drop in British exports in product groups ‘vegetables’ (−46 to −49 per cent), ‘meat & meat products’ (−43 to −49 per cent), ‘dairy’ (−29 to −32 per cent) and ‘fruit’ (−27 to −29 per cent). These effects were greatly reduced under the present trade arrangement and even annihilated for vegetables and fruits. The UK’s new trade policy also permitted to considerably limit the increase in consumer prices for these products. In case of a no-deal Brexit (scenarios S2 and S4), the latter may have increased by 37 per cent for dairy, 24 per cent for meat and 10 per cent for vegetables. Concluding PTAs with main trade partners (scenario S3) would further reduce these negative effects on the UK’s agricultural and food sectors and even reverse them for a number of products (e.g. +7 per cent for British exports of alcoholic beverages other than wines; +5 per cent for tobacco).

6. Conclusion and further work

Brexit marks a turning point in the history of economic integration across Europe and raises many questions on the future trade relationship between the UK and the EU. The EU agrifood sector is strongly internationalised, with the UK accounting for a high share of European trade. Therefore, it is important to evaluate the challenges and potential risks faced by producers and consumers of agrifood products. The current paper measures the potential effects induced by Brexit in the agricultural and food sector, both at the country level and for specific bilateral trade relationships. We consider five scenarios for the UK’s trade policy with EU and third countries after its exit from the EU. These include a EU–UK deep FTA or an exit with no EU–UK trade deal, the continuation of the UK’s trade commitments towards third partners made when the UK was still a member of the EU or the negotiation of new PTAs with main non-EU partners, and the actually implemented trade policy.

We use a structural gravity model to estimate the effects of Brexit on EU countries under each trade policy scenario. We use data on annual bilateral agrifood trade flows over the 4 years preceding the Brexit vote (2012–2015). This permits to eliminate any bias induced by the global economic slowdown during the 2008 financial crisis and the subsequent recovery, by the dramatic worldwide increases in food prices in the late 2007, or by the announcement of Brexit vote results in mid-2016. Our data panel is composed of the world’s largest 57 exporters and importers of agrifood products and covers more than 75 per cent of global trade in these goods. We use a trade costs structure combining tariffs, NTMs, geographic distance and standard variables on bilateral linkages between trading countries used in the literature. Each scenario is defined as a change in import tariffs and NTMs on the EU–UK bilateral trade and on British imports from third (non-EU) countries. We estimate the effects of each Brexit scenario for the entire agrifood sector and separately for 13 groups of products. Our model accounts not only for direct effects of changes in trade costs on trade between concerned countries but also for the indirect impact of these changes on other partners and shifts in prices, supply and demand.

Our results show that under the current trade arrangements, the EU as a group suffers only a marginal decline in imports, exports and real income, losses being stronger for Ireland. EU products that can no longer be sold in the UK will be redirected to intra-EU and extra-EU markets alike. Changes in trade costs and trade patterns will require EU producers to sell their products at slightly lower prices (−0.62 per cent). We also expect a small decrease in the average price of agrifood goods sold on the EU market (−0.21 per cent), except for Ireland (+0.15 per cent). The increase in prices paid by Irish consumers is largely compensated by the gains of consumers in other EU countries. Hence, setting up a mutual European fund to smooth the effects of Brexit across EU countries appears as a viable strategy. The new EU–UK trade deal, the UK’s new trade agreements with third countries and new Global Tariff permit to avoid the strong decrease in British imports and exports, the nearly 1-per cent drop in real income and the imported increase in consumer prices that would have been observed in case of an exit with no deal. The present deal also limits the contraction of the EU–UK trade to only 10–11 per cent. For the EU, the magnitude of effects would have been smaller if the statusquo was conserved in the relationship with the UK and/or if the UK continued to apply the EU PTAs with third countries. For the UK, a strong liberalisation of trade with other major trade partners (in particular USA, Australia and New Zealand) would annihilate most of the negative effects of Brexit. Still, under all scenarios, British consumers will pay higher average prices for agrifood products. The losses incurred by the UK are considerably stronger than that of the EU as a group or of individual member states.

At the product level, meat and meat products appear as the most severely affected, within the EU. The model suggests that European exports of these products will drop by 4 per cent. EU meat producers will redirect the quantities that they will no longer be able to sell to British consumers exclusively to extra-EU markets and will cut prices by 1.5 per cent. Despite the strong switch to non-EU meat suppliers, the price of these products on the British market will increase by 2 per cent, and the UK’s overall imports will decrease.

We model each scenario by a change in import tariffs and the pre-shipment regulatory distance. It may be argued that some of the increase in EU–UK trade costs are not captured by these variables. A number of border requirements do not fall under any NTM category, while some qualify as SPS measures or TBTs. However, integrating changes in the regulatory distance for SPS/TBTs in our scenarios would not affect the results, since we do not find a statistically significant effect of this variable on trade flows. In addition, changes in the pre-shipment regulatory distance pick up not only the increase in direct pre-shipment costs but also indirectly some of other border-specific trade costs correlated with this variable (e.g. physical inspections and time delays for customs clearance). In the event that in the future the UK will adopt different standards (SPS and TBTs) than the EU, we expect the volume of trade with the EU to decrease. Therefore, our results can be considered as a lower limit of the expected effects of Brexit.

To conclude, let us remind that Brexit coincided with the Covid-19 pandemic, which makes it difficult to disentangle the effects of these two shocks on supply and demand and consequently on international trade flows. Accordingly, the effects obtained by the present and other ex-ante analyses remain relevant even once post-Brexit data become available.

Acknowledgements

We thank Alan Matthews, Carl Gaigné and Guillaume Gaulier for fruitful discussions and comments on earlier versions of this work. We are grateful to the editor and three anonymous referees for valuable remarks and suggestions on improving the manuscript. Any remaining errors are our own. Lucile Henry was a PhD student at UMR SMART-LERECO, Institut Agro-AGROCAMPUS OUEST in Rennes, France. She acknowledges financial support from Region Bretagne and INRAE.

Supplementary data

Supplementary data are available at ERAE online.

Footnotes

References