Will the Future EU-UK Free Trade Agreement Affect Foreign Direct Investment?

This article aims to provide new insight on how Brexit will affect foreign direct investment (FDI) into the UK. By estimating an augmented gravity equation which accounts for the depth of free trade agreements (FTAs) as well as for EU and euro area membership, the article gauges the potential impact that different EU-UK trade scenarios might have on FDI flows and stocks. Results show that under a no-deal scenario, FDI flows from the EU into the UK would plunge by 25.9% – 40.6%, and inward FDI stocks would decrease by 49.2% – 53.9%. However, the depth of the future FTA can mitigate this negative outcome. More generally, the article shows that the FDI costs of leaving the EU would be significantly higher for the euro area countries.

Extensive research has highlighted the benefi ts from the European Union (EU) and the Economic and Monetary Union in terms of trade and foreign direct investment (FDI). The acquis communautaire, common currency and the free movement of goods, services, people and capital appear to have reduced transaction costs and fostered the development of economic activities by multinational enterprises (e.g. Carril-Caccia and Pavlova, 2018;Coeurdacier et al., 2009;De Sousa and Lochard, 2011;Martínez-San Román et al., 2016;Umber et al., 2014).
There has been a great deal of Euroscepticism in the last few years despite the fact that the EU project seems to have delivered relevant economic benefi ts to its member states. In the UK, this scepticism resulted in the Brexit vote on 23 June 2016. More than three years later, after signifi cant political convulsion and uncertainty, the UK has left the EU. This year, the EU and the UK will negotiate the future trade deal that will condition the future relations between both parties. The economic consequences of Brexit will depend on this deal (Dhingra et al., 2017). While the British government seeks a trade deal similar to the one that the EU signed with Canada (CETA), the EU is after a more comprehensive one (Adler, 2020, March 2).
Most of the available analysis agrees on the negative consequences that Brexit could have on trade, FDI and welfare (e.g. Bailey et al., 2019;Bruno et al., 2016;Dhingra et al., 2017;Driffi eld and Karoglou, 2019;Greenaway and Milner, 2019;Mulabdic et al., 2017;Simionescu, 2018). In fact, the uncertainty brought about by the Brexit announcement appears to have already negatively impacted the UK's fi nancial market stability and trade (e.g. Belke et al., 2018;Douch et al., 2018;Korus and Celebi, 2019). Dhingra et al. (2017) point out that in the hard Brexit scenario, in which no free trade agreement (FTA) is signed and the UK and the EU trade under the World Trade Organization (WTO) rules, British income per capita would plunge by 9.4%. Dhingra et al. (2017) highlight that not only trade but also the potential drop in FDI might be responsible for this outcome. In this case, research highlights the positive effect of FDI on economic growth or productivity in developed countries (e.g. Alfaro et al., 2004;Ashraf et al., 2015).
Theoretically, there are four main channels through which trade may affect inward FDI. First, with horizontal FDI, FDI and bilateral trade are substitutes. Multinational enterprises (MNEs) follow this strategy to serve the foreign market and to avoid trade costs (Horstmann and Markusen, 1987). If this MNEs investment strategy is predominant in the UK, the future new trade barriers between the EU and the UK will foster bilateral FDI.
Second, vertical FDI is positively associated with bilateral trade liberalisation. In this type of investment, MNEs establish production networks across borders that are linked through trade (Hanson, 2005).

Brexit
Third, export supporting FDI is also positively moderated by bilateral trade liberalisation. It refers to the investments that seek to enhance the market penetration from exports in a host country (Krautheim, 2013).
Fourth, MNEs may also create export platform subsidiaries in a country with the objective of serving third countries' markets through exports (Ekholm et al., 2007).
Through these FDI strategies, MNEs set subsidiaries to perform specifi c economic activities and link through trade, and in doing so, confi gure their global value chains (e.g. Amendolagine et al., 2017;Beugelsdijk et al., 2009;Krugman et al., 1995). If these MNEs' investment strategies are predominant, new barriers to trade brought by Brexit will hamper inward FDI into the UK. In this regard, Bailey et al. (2019) describe the potential negative consequences that Brexit might have on the automotive sector. More generally, Carril-Caccia and Pavlova (2020) show how global value chains participation and the capacity of exporting to (and importing from) a wider number of countries has a positive impact on the capacity of attracting cross-border mergers and acquisitions.
This article presents a study on how the depth of FTAs affects bilateral FDI, a topic rarely considered by previous literature. Indeed, FTAs have been extensively included as a determinant of FDI (e.g. Carril-Caccia and Pavlova, 2020;Garret, 2016;Hyun and Kim, 2010;Jang, 2011;Paniagua and Sapena, 2014), but FTAs are often assumed to be homogeneous by only considering whether a pair of countries have signed one or not. Nevertheless, countries sign FTAs with varying numbers of provisions that cover different policy areas like anti-dumping, competition, tariffs, intellectual property rights or data protection. In this way, the depth of FTAs is determined by their coverage in terms of policy areas.
Based on the relationship between FDI and the depth of bilateral trade integration, this work also extends the literature, which gauges the consequences of Brexit on FDI (e.g. Bruno et al., 2016). In particular, it sheds new light on the potential consequences that Brexit might have on FDI in different EU-UK trade agreements scenarios, and more generally disentangles the effect of leaving the EU for those countries that belong to the euro area (EA) and those that do not. Three different scenarios are considered: WTO rules (no FTA), an agreement similar to that between the EU and the European Free Trade Association (EFTA) and an agreement like the EU-South Korea FTA. 1 The results are in line with the literature that indicates the complementarity of FDI and trade and adds to the depth of the FTA as a relevant dimension when considering the FDI-FTA relationship. Furthermore, this article illustrates how the future FTA between the EU and the UK may affect bilateral FDI. If no FTA is signed, the UK could face a 25.9% -40.6% plunge in terms of FDI fl ows from the remaining EU members. Alternatively, if an FTA like the EU's with South Korea is signed, the drop would be between 13.3% and 22%. In case an agreement is signed like the one between the EU and the EFTA, FDI into the UK would fall by between 4.2% and 7.2%. Estimates for FDI stocks corroborate these fi ndings, showing that in the best-case scenario (i.e. an EFTA-like agreement is signed), EU FDI stock in the UK would decline by 9.2%-10.5%.

Data and methodology
The data sources are the OECD bilateral FDI fl ow and stock BMD3 and BMD4 data, World Bank's FTA database, World Bank's Development Indicators, and UNC-TAD's International Investment Agreements database. The World Bank's FTA database covers 279 agreements signed by 189 countries between 1958 and 2015. The database includes not only the date of FTA signature but also their content; it provides each agreement coverage in terms of provisions that target distinct policy areas (see Hofmann et al., 2017).
The analysis covers the 68 host and source countries 2 during the period 1985-2015. 3 As proposed by Silva and Tenreyro (2006), we use the Poisson pseudo-maximum likelihood to estimate the following gravity equation: in which FDI is the investment fl ow from country i to country j in year t. GDPsum ijt is the combined economic size of the source and host country, diffGDPpc ijt is the difference in GDP per capita between both countries, and SIMI ijt represents the economic size similarity between country i and country j in t. 4 BIT ijt is a dummy variable that takes the value 1 whenever a pair of countries have signed a bilateral investment treaty (BIT). Country pair fi xed effects (λ ij ) are included in order to avoid the potential endogeneity between bilateral investment and the likelihood of signing an FTA or BIT (Baier et al., 2008;Bergstrand and Egger, 2013). Furthermore, country pair fi xed effects account for the multilateral resistance, that is to say, the relative attractiveness of countries to FDI (Anderson and van Wincoop, 2003). The year fi xed effects (λ t ) control for the world's macroeconomic trends.
Finally, as in Mulabdic et al. (2017), depthFTA ijt is the number of provisions included in the FTA signed between country i and j in year t divided by the maximum number of provisions of the deepest FTA in our sample. depthFTA ijt is interacted with a dummy variable that takes the value 1 when a country becomes an EU member (EU jt ) and by a dummy that represents those countries that become EU members but do not belong to the EA yet (EUnvEA jt ). The objective of these interactions is respectively to quantify how the depth of the FTA affected the EU member states inward FDI and see whether the impact is different for those that did not adopt the euro. The signifi cance of the sum of coeffi cients is tested with: ( r + k ) t = ; r, k = 4, 5, 6. σ 2 r + σ 2 k -2 x Cov ( r , k )  Table 1 shows the regression result. Columns 1-3 refer to bilateral FDI fl ows and columns 4-6 refer to stocks. In line with previous research, FDI fl ows and stocks are positively moderated by the combined economic size from the home and host country (GDPsum ijt ) and go from capitalabundant to labour-abundant countries. In addition, signing a BIT appears to boost bilateral FDI fl ows, but is insignifi cant in the case of stocks. 5 Furthermore, in both cases, economic size similarity (SIMI ijt ) is insignifi cant.

Results: FDI and FTA depth
Column 1 shows that the depth of FTAs has an overall positive impact on bilateral FDI fl ows. According to the estimate of β 4 , the deepest signed FTA in our sample (the EU with 42 provisions) would have increased bilateral FDI fl ows by 38%. 6 Alternatively, a shallower FTA like the one between the EU and South Korea, which has 22 provisions, results in only a 20% higher FDI.
5 The resulting lack of signifi cance of BIT is not surprising. The literature highlights that the sign and signifi cance associated with BIT is prone to depends on the sector of investment, the level of development of the signing countries or the existing intensity of bilateral FDI (e.g. Berger et al., 2011;Colen et al., 2011;Paniagua et al., 2015). 6 The effect from the depth of FTA are calculated by

Brexit
In column 2, the positive signifi cance of β 6 suggests that in comparison with the rest of the world, the depth of FTAs is particularly relevant in explaining FDI into the EU. According to this estimate, joining the EU has increased FDI fl ows from its member states by 68% (e β 5 + β 6 -1).
In line with the literature that highlights the benefi ts from the euro (Coeurdacier et al., 2009;De Sousa and Lochard, 2011), β 7 from depthFTA ijt x EUnvEA jt is signifi cant and negative. This indicates that the gains on FDI from the depth of the FTA are lower for those countries that have not adopted the euro yet. Accordingly, euro area member countries have seen an increase in FDI fl ows from other member states of 141% (e β 5 + β 6 -1), while the fi gure for those not in the euro area is 35% (e β 5 + β 6 + β 7 -1).
Estimates of the depth of FTAs on FDI stocks are in line with the fi ndings for FDI fl ows. The FDI stock estimates represent the long-term effects of FTAs. Overall, the EU membership increased the FDI stock among its members by 116.8%. The increase amounts to 156.8% for countries that adopted the euro and 97% for those that did not.
The consequences of leaving the EU Based on the estimates from columns 2 and 3, for FDI fl ows, Table 2 approximates the consequences of leaving the EU under different FTA scenarios. On average, under a no-FTA scenario, a country that leaves the EU could expect a 40% drop in FDI fl ows from the remaining members, the fall being higher if it adopted the euro (58%) than if it did not (26%). 7 It is likely that the impact of Brexit on FDI into the UK will be closer to the average expected impact of those countries that never adopted the euro. Nonetheless, the UK has been an EU member for 40 years and it is one of the world's main FDI stockholders. Thus, the average impact of leaving the EU is considered to be the upper bound.
The negative impact of Brexit on FDI fl ows could be signifi cantly lower if an FTA is signed. If an agreement is reached like the one between EFTA and the EU, the drop would be of only 4% -7%. Instead, if a shallower FTA is signed, like the one between the EU and South Korea, the decrease in FDI would be between 13% and 22%. Table 3 reports the different scenarios for the case of FDI stocks. Leaving the EU will have sizable effects on the stock of FDI from the remaining member states in the UK. Even if the future EU-UK trade agreement is EFTA-like, the investment stock from the remaining states is likely to fall by between 9.2% and 10.5%. If the WTO scenario is reached, the FDI stock in the UK from the remaining EU members is likely to be reduced by half.

Conclusions
Today, the future of the EU-UK relations is uncertain. The EU hopes to sign a more comprehensive FTA than the one that the British government seeks, and a no-deal scenario, in which both parties trade under WTO rules, is not fully unlikely. This article quantifi es the consequences that different EU-UK FTA agreements could have on bilateral FDI.
Results show that there is a complementary relationship between bilateral FDI and trade liberalisation. Consequently, fi ndings suggest that the new trade barriers be-