The impact of economic policy uncertainty caused by the Brexit referendum on aggregate and sectoral-level FDI into the UK

(1)

MASTER'S THESIS

THE IMPACT OF ECONOMIC POLICY

UNCERTAINTY CAUSED BY THE BREXIT

REFERENDUM ON AGGREGATE AND

SECTORAL-LEVEL FDI INTO THE UK

Master's Thesis in International Business, Economics

July 2020

Author:

Tran Loc, Vu – S1033141

Supervisor: Dr. A. de Vaal

(2)

Acknowledgments

I would like to take this opportunity to express my gratitude to my supervisor, Prof. Dr. A. de Vaal of Nijmegen School of Management, Radboud University, for his constructive feedback and creative insights during my writing. Without his coaching, this paper would not have been materialized.

Additionally, I want to thank my parent, my friends, and especially my wife, who supports me throughout the entire Master’s course. Their unconditional support motivates me to push through all the difficulties, particularly during this trying times of a global pandemic.

I really enjoy life in the Netherlands. Thank you, Radboud University, for giving me the chance to study here and develop myself to the fullest, while still having fun and exploring new things every day.

Tran Loc Vu

Nijmegen, July 2020 Abstract

This paper performs a time-series and a fixed-effects panel data analysis to investigate the effect of economic policy uncertainty, caused by the Brexit referendum, on aggregate and sectoral-level FDI into the UK for the period 2010-2019. In general, on average, the number of M&A projects decreases by 1,041% after Brexit. A 1% increase of EPU will cause the number of M&A projects to decrease by 0,173% before Brexit; however, after the referendum, an 1% increase of EPU will cause the number of M&A projects to increase by 0,049%. A similar effect is observed on the sectoral model: On average, there are 12,45 less M&A projects per sector after Brexit than before. However, after Brexit, a 1% rise in EPU will result in a rise of 2,697 M&A projects. The positive moderating effect of Brexit can be attributed to firms’ conceptions. The service sector consistently has a higher number of M&A disinvestment compared to the primary and manufacturing sectors. When EPU increases by 1%, the service sector has an average reduction of 0,639 M&A projects more than non-service sector. Banks and Insurance sector is the most affected by the referendum due to potential changes in the regulations regarding the EU market.

(3)

LIST OF TABLES

Table 2.1: The pros and cons of possible Brexit scenarios ... 14

Table 4.1: Descriptive statistics for panel (M&A) dataset ... 32

Table 4.2: Descriptive statistics for the number of M&A projects in each sector ... 33

Table 5.1: Effect of lagged EPU on aggregate FDI ... 42

Table 5.2: Mediating effect of Brexit model ... 44

Table 5.3: Mediating effect of Brexit ... 44

Table 5.4: Fixed-effects model, Number of M&A projects, lag of all variables ... 47

Table 5.5: Fixed effect model, before and after the Brexit referendum, June 2016 ... 50

Table 5.6: Most negatively to least negatively affected industries by EPU, before and after the Brexit referendum ... 51

Table 5.7: Model with Service dummy ... 52

Table 5.8: Individual regression results for each sector ... 53

Table 1: Aggregate model 1.1 and 1.2 ... 61

Table 2: Model 2.1... 62

Table 3: Model 2.2... 63

Table 4: The mediating effect sectoral model ... 64

Table 5: Correlation table ... 65

LIST OF FIGURES

Figure 2.1: The UK’s historical EPU index, 1900-2020 ... 16

Figure 4.1: Number of M&A projects of each sector by time ... 34

(5)

ABBREVIATION TABLE

Abbreviation Full form

BvD Bureau van Dijk

ECSC European Coal and Steel Community

EEA European Economic Area

EEC European Economic Community

EMS European Monetary System

EPU Economic policy uncertainty

EU Europe Union

FDI Foreign direct investment

G7 Group of Seven

GDP Gross Domestic Product

M&A Mergers and acquisitions

MNC Multinational companies

NACE Nomenclature statistique des activités économiques dans la

Communauté européenne - Statistical classification of economic activities in the European Community

OECD Organization for Economic Co-operation and Development

OLS Ordinary Least Squares

ONS Office for National Statistics

OPEC The Organization of the Petroleum Exporting Countries

SD Standard deviation

UK United Kingdom

US United States

(6)

1. CHAPTER 1: INTRODUCTION

On 23rd_{June 2016, 51,9% of British people voted in favor of leaving the}

European Union (EU). By the thinnest margin possible, the legally non-binding referendum result (The Guardian, 2016) paved the way to a new chapter in the UK-EU relationship, cleverly named “Brexit.” Ever since, the uncertainty surrounding Brexit keeps raising questions of how and to what extent the UK economy would change with the new system, and whether British people would be worse or better off. Reviewing the literature, nearly all speculate a dark picture of the UK economy in the upcoming years. One of the reasons is the possible reduction of investment, especially foreign direct investment (FDI) of firms into the UK, as it is uncertain how the Brexit will turn out. Dhingra et al. (2017) mentioned three reasons for this forecast: (1) being in the EU single market “makes the UK an attractive export platform for multinationals as they do not bear potentially large costs from tariff and NTBs when exporting to the rest of the EU” (Dhingra et al., 2017, page 683). Firms invest in the UK not only to serve one of the biggest markets in the world but also to use that as a springboard to conquer the EU market as well. In the Brexit worst-case scenario of no trade agreements between the UK and EU being ratified, trade between them will be governed by the World Trade Organization (WTO) regulations, with most-favored-nation tariff rates and treatment. The subsequently higher tariff barrier will make export much more costly to do. (2) In the current global value chain system, multinational companies (MNCs) tend to fragment their production across borders to minimize cost and take advantage of cheap production factors. They tend to have complex supply chains and complicated relationship between headquarter-subsidiaries. Leaving the common market means MNCs active in the UK now have to adapt to different regulations, further complicating their management scheme. An example would be the free movement of labor. The UK attracts a lot of highly skilled workers from the EU market, which will change if no agreement is made to control for free labor movement between the UK and the EU, causing a disturbance in intra-firm staff transfers of MNCs. (3) The uncertainty that comes with future Brexit

(7)

agreements will most likely slow down FDI inflows since firms will wait until a clear solution is reached.

Many studies are investigating the impact of Brexit on the UK economy. However, most of them are trying to model the impact under possible circumstances in the future, like a total Brexit versus partial Brexit. This is expected, given the uncertainty surrounding Brexit and its far-from-final state. However, those studies seem to forget about the current impact on business, as they disregard the economic policy uncertainty (EPU) foreign firms are facing when they decide to invest in the UK or not. Moreover, Brexit is an ongoing crisis, which means the uncertainty covering the whole situation would not go anywhere soon.

Brexit is a one-of-a-kind event; there are nearly no comparable case studies to apply to the Brexit crisis. Therefore, no one could guess correctly what the outcome of Brexit will be, nor when accurately it would arrive. The process could take many years, evidenced by the slow reaction of the British government now. By the time the situation has settled down, the economy might change a lot; it could render all predictions invalid. Meanwhile, prolonging the uncertain situation will discourage FDI to the UK (Serwicka and Tamberi, 2018), before agreeing on a good deal with the EU. Therefore, it is crucial to assess the impact of EPU by Brexit now, particularly on investment and FDI, because, as Hsieh (2019) showed, investment changes tend to come relatively fast: expecting outward FDI to decrease after two-quarters of an EPU shock in the host country, or even sooner in some industries.

Furthermore, there is a disturbing lack of literature discussing the impact of Brexit now on sector-level FDI into the UK. This needs to be the focus because industries do not react the same. On the whole, Brexit might have an unpleasant influence on aggregate FDI, but some industries might be more severely affected than the other. For instance, new FDI into the banking and financial sector might freeze since firms are not sure whether London can still serve as one of the EU’s biggest financial centers due to unclear commitment of the UK government in maintaining and following the EU’s regulations regarding financial activities.

(8)

Seeing all these problems, I want to do an empirical analysis to answer the question: What is the impact of economic policy uncertainty (EPU) caused by the Brexit referendum, on aggregate and sectoral-level FDI into the UK? First, the impact of EPU caused by Brexit on aggregate FDI inflow to the UK will be quantified. Next, the distinctive characteristics of each sector will be taken into account, when a fixed-effect panel model is regressed to reveal to what extent their FDI inflow react to EPU. The time of research would convey from 2010-2019, starting after the financial crisis and a new booming phase of the UK economy, with the referendum point roughly at the middle, and following years of Brexit preparation. The contribution of this thesis is three-fold. Firstly, it adds to the growing stream of EPU research that tries to explain economic fluctuations using uncertainty. Brexit is certainly an unexpected event that is causing the highest level of uncertainty in the UK economy, even surpassing historical crises like the 2008 global financial crisis (Baker’s EPU index dataset, 2020). While unfortunate for British people, this proves to be a valuable real-world case study for economists who would want to examine how uncertainty can deregulate the economy.

Secondly, the paper adds to the Brexit stream of literature by approaching the problem at a different angle. While most of the papers focus on the Brexit outcomes, this paper focuses on the transition period. FDI is crucial to the UK’s economy; in fact, foreign firms’ FDI activities account for 11% of the UK’s Gross domestic product (GDP), the highest in the G7 group (Dhingra et al., 2016). As a result, minor disruption to FDI flow could drag down the whole economy, in turn causing more firms to postpone their investments. This paper will show how much damage Brexit has done to each sector’s FDI, thus serve as guidance for the government to mitigate the effect effectively for each sector.

Thirdly, the paper adds to the FDI research. Most FDI papers concerned FDI at the national level, using macro data to predict FDI to each country based on their attractiveness, or FDI at firms’ level, analyzing their decisions to invest globally. This research leaves the beaten trails to explore FDI at the sectoral level and how it reacts

(9)

under changes in the economic environment of a single country, specifically, how EPU affects mergers and acquisitions.

The thesis will be presented as follows. Chapter 2 contains some background information about the UK-EU relationship, the Brexit referendum, and possible scenarios. Chapter 3 contains the literature review and hypothesis for the quantitative model. Chapter 4 explains the data collection process and method used. Chapter 5 depicts the empirical results and the explanation of the results, with their economic implications. Finally, chapter 6 summarizes and concludes the thesis, including limitations and recommendations for future research.

(10)

2. CHAPTER 2: BACKGROUND INFORMATION OF BREXIT

2.1. Outline of the UK-EU relationship

The United Kingdom (UK) is a relatively isolated island nation from the rest of Europe throughout its history. Even when European countries formed an alliance between them after World War II, Britain was out of the game. The UK, in turn, refused to join the European Coal and Steel Community (ECSC) in 1951 and then the European Economic Community (EEC) in 1957 (Reuters, 2020).

It was not until several years later that when the French and German economies recovered quickly and formed a strong union, the British authorities gradually changed their view of joining the EEC. The British government applied to join the EEC in 1961 but was rejected twice by French President Charles de Gaulle in 1963 and 1967. It was not until 1973 that Britain officially became a member of the EEC (Reuters, 2020).

However, this relationship quickly met with opposition from the British ruling political parties. During the 1970s, with the collapse of the Bretton Woods system and the steep oil price increase by The Organization of the Petroleum Exporting Countries (OPEC), the ruling Labour party faced enormous pressure to renegotiate terms with EEC, and to hold a referendum deciding member status of the UK in the EEC (Bancroft, 2019). The referendum took place in 1975, with over 67% of the votes supporting the stay.

In the following years, the relationship between the two sides did not seem to improve. In 1990, the UK joined the European Monetary System (EMS) to stabilize the fixed exchange rate in the whole bloc. However, only two years later, Britain announced its withdrawal from the system, after the Pound crisis. In 1995, Britain also refused to join the Schengen Treaty on freedom of movement between member states and not using the European common currency. Following the global financial crisis and public debt in several European countries, the UK refused to sign the 2011 fiscal treaty proposed by the EU in order to overcome some of its financial problems (The Washington Post, 2011).

(11)

Since 2010, polls showed that the British public had had a split in leaving or staying in the European Union (Ipsos Mori, 2016). By January 2013, former Prime Minister David Cameron pledged to hold a referendum if he won the 2015 general election despite opposing Brexit, after pressure from British political parties, including the Conservatives, United Kingdom Independent Party, British National Party, etc. (BBC, 2013). The main arguments were the EU’s economic decline, the over-controlling scheme of EU over the UK legal systems, the lack of UK’s influences, and the EU’s plan for further integration (Vox, 2016).

The referendum took place on 23rd_{June 2016. According to the BBC}

referendum dataset (BBC, 2016), 33,58 million voters cast their votes, making up 72,21% of the total number of voters nationwide. The results showed that 17,4 million people, or 51,89% of British voters, supported the Brexit option. However, there was a clear divide between the age as well as the geographical location of the voters. While people in England and Wales favored leaving the EU, the majority of Scots and Northern Ireland desired to stay, with 62,0% and 55,8% of the "Remain" votes, respectively.

Regarding age, the Brexit support rate increased gradually with the age of the voters. Up to 60% of voters over 65 voted to leave the EU, while the “Leave” rate in 45-54 and 55-64 years old groups are 56% and 57%, respectively. In contrast, only 27% of people aged 18-24 supported “Leave.”

While the first referendum to leave the EU (in 1975) seems to be motivated purely by political aim, the 2016 referendum’s results showed a disapproval attitude of British people towards the EU’s policies and regulations forcing on the UK. Goodwin and Milazzo (2017) studied the vote and share of the 2016 Brexit referendum using data from the 2014-2017 British Election Study Internet panel. They found that “strong public concerns over immigration, and its perceived effects on the country and on communities, were central to explaining the 2016 vote for Brexit” (Goodwin and Milazzo, 2017, page 464). Carl et al. (2019) summarized different surveys on reasons for voting leave EU and found the main reasons were immigration problems and a lack of sovereignty due to EU policies. Leave voters wanted the UK to

(12)

regain control over any role the EU had in UK law-making system, including, and most crucial, control over immigration. They also wanted to cut any payments from the UK to the EU. On a different note, Becker, Fetzer and Novy (2018) found no significant impact of neither measures of EU in immigration and trade exposure, nor the quality of public and fiscal policies, to the variation of the leave vote. Instead, it is the demography, education, and economic structure that caused people to vote leave. That could reflect Brexit as a fundamental problem in the UK itself, with less-educated or older voters being fed up with UK’s own mediocre public services (Becker, Fetzer and Novy, 2018). They thought the EU is to blame, so they voted to leave.

After the referendum, former Prime Minister David Cameron resigned. The successor, Theresa May, triggered Article 50 - the only legal mechanism for a member state of the EU to leave - in March 2017 (Time, 2017). The UK and EU would have a period of two years to build a vision for the future of the two sides, including a new trade agreement, labor movement, and barriers of trade. After numerous withdrawal agreements being rejected by the UK parliament and subsequent extension from the EU, the new government of Boris Johnson finally got the agreement approved and

became law on 23rd_{January 2020 (BBC, 2020). On 29}th_{January, The European}

Parliament gave its consent to ratification of the withdrawal, and two days after, the UK officially ended their 47-year membership of the EU (BBC, 2020).

In the meantime, until 31st_{December 2020, the UK and EU enter a transition}

period, which means free movement of goods and services and labors remain unchanged. The UK will be able to negotiate and ratify trade deals with the EU and other partners without EU consent (BBC, 2020). However, unless a “soft Brexit” is reached, these conditions would not hold after that, and the UK will have to deal with various difficulties in regards to EU trade and business relationship. The “soft” and “hard” Brexit scenarios will be discussed in the next part.

2.2. Possible Brexit scenarios

One of the hardest things when coming to negotiations after Brexit is that there are no antecedent cases of a country leaving the European Union to base on. However,

(13)

there are examples when a country, while not a member of the EU, still has special treatment treaty with the EU that could be used as a guideline for the UK in upcoming negotiations. Dhingra et al. (2017) listed three possible scenarios for Brexit: Soft Brexit “Norway option,” Soft Brexit “Switzerland option,” and Hard Brexit.

Norway is not an EU member, but it is a member of the European Economic Area (EEA), together with other EU nations, Iceland and Liechtenstein. EEA gives Norway access to the European common market, which means free movement of goods, services, and labor; however, they only have to abide by the economic policies of the EU only. It is, unfortunately, a double-edged sword. Staying in the single market could avoid the UK from a considerable trade cost increase with the EU, and free the UK from binding monetary and common foreign policies (Dhingra et al., 2017) – the freedom they fought for long. Yet, the UK will have to obey the EEA economic policies without any saying, while still have to contribute to the EU’s budget like an annual fee. Furthermore, there could be new tariff and non-tariff barriers being added when importing goods from EEA members, due to rules of origin requirement and anti-dumping measures from the EU, with the Norwegian salmon being an example (Dhingra et al., 2017).

The second possible scenario would be in the form of an agreement. At the lowest level, a free trade agreement would maintain the current tariffs on goods trade; however, the freedom of service and movement is out of the question, while non-tariff barriers would likely be higher. The UK could follow the Switzerland model and have multiple treaties with the EU, ranging from the movement of people, insurance, and fraud prevention (Dhingra et al., 2017). The UK could freely negotiate trade deals with other countries separately from the EU and could choose which programs they like to partner with the EU. However, like EEA countries, Switzerland has to make some financial contributions to the EU without the ability to influence the EU programs they decide to join. Finally, the UK would have much less bargaining power when negotiating with the EU since the UK’s market is nowhere near the size of the EU (Dhingra et al., 2017).

(14)

If both parties cannot come to a mutual agreement, the Hard Brexit is imminent. WTO’s most-favored-nation term would govern trade between the UK and EU. As a result, tariff and non-tariff barriers on UK’s export to EU will be higher, as well as much more limited trading in services and free movement of labor (Dhingra et al., 2017). In return, they will be able to establish economic policies and foreign relations freely from any preferences from the EU, getting rid of EU’s brought-in problems like immigration. Of course, with the increase in trade costs, this could be the costliest solution for the UK economy, especially the financial sector; but it is inevitable if the UK-EU could not agree with a common solution after the transition period ended.

Table 2.1: The pros and cons of possible Brexit scenarios

Source: Dhingra and Sampson (2016, CEP document, page 8)

2.3. Observations for the current situation regarding Brexit and FDI inflow Based on the bittersweet historical relationship between the UK and EU, and the possible future between them, some observations can be made about the situation regarding Brexit and FDI inflow into the UK:

- It is undeniable that the uncertainty caused by Brexit is the highest ever in the UK, even surpassing historical events of World Wars, financial crises, or any other political distress (Figure 2.1). However, one might say that Brexit is

(15)

expected because the UK did not have the same level of commitment to the EU as other nations since the beginning. The UK refused to sign the EU fiscal treaty in 2011, and the following public opinion studies gave a very early hint of what was going to happen. So, firms knew fairly clearly the situation when they decided to commit to the UK market, way before the actual referendum took place. As a result, it might mitigate some of the adverse effects of EPU on FDI, simply because firms already adapted or chose to ignore the UK well before Brexit was a reality.

- On the other hand, almost no one believed in a full separation of the UK from the EU. Even former Prime Minister Cameroon, who risked his campaign on the referendum result. Because, although frequently disagreeing with each other, the UK-EU relationship was still healthy. Both benefitted from each other, especially the UK, from FDI inflow. So much, in fact, all papers about the influence of EU membership on FDI inflows predicted a significant positive effect, ranging from 25% to 48%, even 62% in some cases (Straathof et al., 2008; Campos & Coricelli, 2015; Fournier et al., 2015; Bruno et al.,

2016;Welfens & Baier, 2018). Everybody expected the 2016 referendum to be

similar to the first referendum held in 1975 and used its results to push government renegotiating terms with the EU. A total discharge, therefore, is shocking and is not predicted by firms. Foreign companies will have to change strategy regarding the UK market with the mindset that the UK market is not part of the EU anymore, with possibly stricter rules and higher costs. Disinvestment might occur as a result.

- The vote results could mean trouble for foreign firms. UK citizens are separated, especially about the sovereignty and independence of the UK from outsiders. Some are becoming more and more annoyed by foreign entities controlling their government decisions and the wave of refugees adding up to already overloaded public infrastructure. That could affect their view of foreign brands and firms. The boycotting of foreign-branded products is the worst-case scenario that firms must consider, and it adds up to the uncertainty firms faced when FDI into the UK.

(16)

- Each possible Brexit scenario requires different actions from foreign firms. The Soft Brexit would be much easier for firms, and would likely be less destructive to FDI inflow, but the probability for it to be true is getting slimmer.

Figure 2.1: The UK’s historical EPU index, 1900-2020

Source: Baker’s Economic Policy Uncertainty, https://www.policyuncertainty.com/

0 100 200 300 400 500 600 1 90 0 1 90 2 1 90 4 1 90 6 1 90 8 1 91 0 1 91 2 1 91 4 1 91 6 1 91 8 1 92 0 1 92 2 1 92 5 1 92 7 1 92 9 1 93 1 1 93 3 1 93 5 1 93 7 1 93 9 1 94 1 1 94 3 1 94 5 1 94 7 1 95 0 1 95 2 1 95 4 1 95 6 1 95 8 1 96 0 1 96 2 1 96 4 1 96 6 1 96 8 1 97 0 1 97 2 1 97 5 1 97 7 1 97 9 1 98 1 1 98 3 1 98 5 1 98 7 1 98 9 1 99 1 1 99 3 1 99 5 1 99 7 2 00 0 2 00 2 2 00 4 2 00 6 2 00 8 2 01 0 2 01 2 2 01 4 2 01 6 2 01 8

(17)

3. CHAPTER 3: LITERATURE REVIEW AND HYPOTHESES

3.1. Literature review of the impact of Brexit on FDI inflow of the UK

Studies are evaluating the impact of Brexit on FDI into the UK. Among them, two angles are prominent. First, researchers look at Brexit as a (coming soon) shock in international trade relations. Forecasting models are used to predict the impact of loss of free access to the EU market and labor force on future FDI. The second trend is calculating the impact of EPU, caused by Brexit, on the UK economy. Despite the diverse approaches, the literature is somewhat united in concluding that Brexit would likely cause FDI inflow to decrease.

3.1.1. Brexit as a shock in international trade relations

Centre for Economic Performance (2016) calculated the UK FDI stock to be over £1 trillion, with half of it coming from the EU. The paper was based on Bruno et al. (2016) model of FDI between 34 Organisation for Economic Co-operation and Development (OECD) countries (1985-2013), which, in turn, is a type of gravity model regressing bilateral FDI flow and stocks on GDP (home and host), EU membership, and share of export/import and manufacturing in the economy. EU membership had a significant positive effect on FDI inflow to the UK by an average of 28% increasing in FDI, with the financial services sector being the largest recipient. With the counterfactual model in place, if a country was to leave the EU, a 22% fall in FDI in the next ten years would be the least damaging consequence, with a corresponding fall of 3,4% GDP per capita. The impact was much more severe when comparing to income losses due to lower trade, at 1,3% to 2,6%. The financial sector would be highly impacted as well, due to potential restrictions in the free labor movement and the inability to have a voice making regulations at the EU court.

The result matched other papers of Campos and Coricelli (2015), who compared the UK FDI growth with a comparison group of matched countries. They found that the UK benefited significantly from the EU in trade, FDI, and finance. EU membership was estimated to boost FDI by 25-30%. Similarly, Welfens and Baier (2018), with the use of a similar gravity equation, looked at the period 1985-2012 for 34 OECD countries and found (1) being in the EU increased inward FDI by 62%, so

(18)

leaving it will reduce FDI dramatically, (2) The effect of Brexit to FDI could be mitigated if foreign ownership of UK capital stock increases, which was a possibility due to the Pound’s depreciation causing UK’s stock to be relatively cheaper, (3) The UK government could mitigate some of the Brexit impacts by reducing taxes on firms; however, the UK corporate tax rate was already meager, so the reduction might not be possible (Welfens and Baier, 2018).

McGrattan and Waddle (2018) analyzed some post-Brexit scenarios using neoclassical growth simulations. They extended the multi-country dynamic general equilibrium model of McGrattan and Prescott (2009) by adding trade frictions and bilateral cost of FDI, with the main feature being technology capital, playing a vital role for FDI (McGrattan and Waddle, 2018). The framework allowed to calculate welfare loss from barriers of trade and investment due to the blockage of foreign innovation and the substitution of costly domestic technology investment. The model was parameterized using data from all nations having major investors in the UK and EU, before the 2016 referendum, then it was used to forecast the different Brexit scenarios. If the UK tightens regulations on trade and FDI alone, their firms and citizens would be significantly worse off, and the EU would be better off from investments from the UK. If the EU also tightens regulations, UK firms would likely to disinvest into EU and increase international lending, leading to welfare losses for both the UK and EU. Finally, if the UK reduces restrictions to nations outside the EU, significant welfare gains would be expected.

3.1.2. Brexit as EPU

Tripier (2019) assessed the cost of uncertainty in the UK economy. Focusing on the period after the referendum (2016-2019), UK’s economy lost around £16 billion per year of GDP. There is an entangled effect of Brexit and of the economy itself when producing uncertainty: Brexit created uncertainty, and the uncertainty was exaggerated by market reaction waiting for the economic policy in response. Thus, the author contended to use uncertainty shocks as a proxy to EPU, since it was identifiable by econometric method (in this paper, Structural Vector Auto-Regressive model) based on the fact it affected volatile market variables first like exchange rate

(19)

or stock prices, but had a lagged effect on other variables like production and investment behaviors. Using the uncertainty shock that way, it would be strictly independent of the economic situation, thus truly reflect the effect of Brexit as EPU. EPU shocks reduced the UK economic growth rate from 2,3% (counterfactual model) to 1,9% in real life. While it is not enough to put the UK into a recession, it is undoubtedly a massive problem for the government to deal with.

Serwicka and Tamberi (2018) explored the FDI inflows after the Brexit referendum and found correlations between aggregate FDI inflow and uncertainty

caused by Brexit. After the referendum date of 23rd_{June 2016, investments into the}

UK had the most prolonged continuous decline since 2003, with UK FDI shares of EU28 declining sharply from 25% to 18%, indicating some relocation activities of investing firms to the EU. Using the EPU index by Baker (2015), the author found the UK uncertainty skyrocketed, and the number of FDI projects and capital investments plummeted right after the referendum point. Comparing with the counterfactual “synthetic UK” created by the weighted average of biggest UK trade partners, if the UK had not voted to leave the EU, an approximately 24% more FDI projects would have been started in the UK, compared to reality. Again, there was a decline in investment in the services sector, mostly banking, software, and management. Based on this work, Tamberi (2020) developed a new partial equilibrium framework for the impact of Brexit uncertainty on export-platform FDI in the UK. Brexit uncertainty resulted in a reduction of 13,5% export-platform FDI projects, and evidence of firms relocating to the EU was found.

In general, while some Brexit papers tried to connect EPU and the UK’s inward FDI around the referendum point, they either (1) made descriptive comparison between EPU index and aggregate FDI (Serwicka and Tamberi, 2018); or (2) concluded that uncertainty would affect financial variables in the short run but probably not production and investment behavior (Tripier, 2019). The use of counterfactual models brings graphical results, but there is no way to make sure the “artificial” UK would be the same UK if the referendum did not take place. The framework proposed by Tamberi (2020) is too new and will need further evaluation

(20)

and expansion out of export-platform FDI. What is lacking in the field now is straightforward empirical research to show a simple, direct impact of current EPU situation on FDI into the UK, and whether or not that EPU situation will encourage firms to divest out of the UK as a way to avoid uncertainty.

3.2. Literature review of the impact of EPU on FDI level 3.2.1. The impact of uncertainty on firm’s investment in general

From the 1980s, economists pointed their focus on the causal relationship between country uncertainty and the level of investment of firms. All in all, a similar theme emerged from those early studies. EPU was generally a result of changes in the macroeconomic environment, either from monetary, fiscal, tax, and other regulatory policies. The typical reaction of firms was holding back and delay or reduce investment. While the discussion was mostly surrounding the theoretical perspective, the theories developed from the papers certainly help future research when isolating the effect of a specific EPU event on MNC’s investing decisions.

Emerging from the literature, the real options theory successfully explained why and how firms react to uncertainty. It appeared in one of the earliest works of Bernanke (1980), where he used cost-benefit analysis to show that investments would be slowed down and recruiting activities be delayed if firms could sense high uncertainty when investments and labor were expensive. Bernanke suggested that firms’ investment decision-making progress is comparing the cost of deferring the project and the potential value of information gained during the waiting time. “Uncertainty, because it increases the value of waiting for new information, retards the current rate of investment” (Bernanke, 1980, abstract page). The cost-comparing process was realized in Dixit and Pindyck’s paper as “real options theory” (Dixit and Pindyck, 1994). The options approach’s most important assumption is the irreversibility and possibility of delay of investments. Firms with the opportunity to invest hold an option to buy assets at a certain time in the future, like a financial call option. When the investment is approved, the option is lost; firms give up their rights to wait for new information that can affect the viability and timing of the investment. That is why the lost option must be included as an opportunity cost of investment.

(21)

This opportunity cost is sensitive to uncertainty (Dixit and Pindyck, 1994), explaining why uncertainty can significantly disturb investment decisions. Using the theory, Bloom et al. (2007) showed changes in the regulatory environment “increases real option values making firms more cautious when investing or disinvesting.” The “cautionary effect” of uncertainty could offset companies’ responsiveness to a stimulus by the government in hard times because they would restrict from expanding and retracting and keep doing what they do until the situation is better.

Other papers focused on the government-firm relationship and how uncertainty could backfire government intervention. Rodrik (1991) looked into the developing economies of Africa, Asia, and South Africa and their economical-political reforms. When doing policy reforms, the massive changes in structure and macroeconomic environment actually disfavored the investment climate for private firms, due to uncertainty regarding future policies acting as a tax on investment. Therefore, macroeconomic measures with stabilization and sustainability in mind would have greater payoffs in terms of investment when compared to measures focusing on structural reforms and liberalization. Hassett and Metcalf (1999) found a significant negative impact of tax policy uncertainty on firm-level investments when tax policy in time series had a random walk. However, if tax policy had a discrete and stationary jump process that followed historical experiences, higher uncertainty could speed up time to invest, because firms would speed up investment to avoid future tax. In both cases, increases in uncertainty increased loss of tax revenues to government, so if the government could not come up with a fixed tax for investment, uncertainty could be the implicit subsidy to investment (Hassett and Metcalf, 1999). The extent of uncertainty in comparison with the value of waiting determined investment incentives. 3.2.2. Proxy for EPU – a new look at uncertainty in business study

Ghosal and Loungani (1996) were one of the first to measure uncertainty, as a standard deviation of the residual from the forecasting equation firms used to forecast their product price. They investigated the uncertainty-investment relationship in manufacturing industries of the United States 1958-1989, particularly the difference between high and low concentrated industries. Each industry would have a certain

(22)

uncertainty measure, and Ordinary Least Squares (OLS) fixed effects were employed for the regression. Uncertainty did not have a significant impact when the industries were not categorized. However, for low concentrated industries, which had low seller competition and intense competition in the market, price uncertainty had a significant negative effect on investment. The higher the concentration of the industries, the less effect of price uncertainty. The highest seller concentration industries suffered nearly no impact from price uncertainty. The standard deviation method was applied in various uncertainty studies with similar results, such as Huizinga (1993, cited in Ghosal and Loungani, 1996) using inflation uncertainty measure.

By the 2010s, studies looked for new ways to describe policy uncertainty by using proxies. Proxies are better because they can point out the exact cause of uncertainty, not just generalizing on changes in regulations. Proxies are great for event studies, with each event having the corresponding proxy capturing the specific uncertainty caused by that event and not anything else. An example is from the paper of Julio and Yook (2012). They used election years in 48 countries and found, on average, corporate investment rates dropped by 4.8% before an election, comparing to the same time in nonelection year. Julio and Yook fitted this effect with political uncertainty, holding businesses back from their investing decisions until the uncertainty surrounding the election was resolved. The effect was stronger in less stable government and politically sensitive industries.

Another example of the new approach is using indexes. The index is perfect if economists want to assess its impact on other economic variables empirically. Solomon and Ruiz (2012) used political risk index and exchange rate uncertainty as proxies for EPU. In 28 developing countries from 1985-2004, expectedly, both variables showed a significant negative effect on FDI, especially in Africa, due to overly risky perception of businesses to the black continent. In a similar study, Huang et al. (2015) used international crisis events (Gulf War, 11th September, Iran Nuclear crisis, etc.) to investigate their effects on the cooperating payout. Past payers tended to cancel their dividend payments, and non-payers tended not to initiate dividend payments. MNCs were affected the most.

(23)

However, each method had its advantage and disadvantage. For example, election years or international crisis events cannot capture the heterogeneity in EPU in the period between the events. Exchange rate uncertainty cannot distinguish uncertainty coming from own country or international events. Recognizing the problems, Baker et al. (2016) introduced a new paper-based index to measure EPU, aiming to capture “who will make economic policy decisions, what economic policy actions will be undertaken and when, and the economic effects of policy actions” (Baker et al., 2016, page 5). The EPU index is a weighted average of four components: news-based uncertainty, tax expiration, CPI disagreement, and state/local government purchase disagreement. The index was constructed using keyword search results from the most prominent newspaper of a country. The keywords are in 3 sets, Economy(ic), Uncertain(ty), and “congress,” “legislation,” “regulation,” “deficit,” among others. All articles containing three sets of keywords would be accounted for and divided by the total number of articles in that month. This index was found to have similar patterns to other uncertainty indexes, like the VIX index for future stock market return uncertainty (Baker et al., 2016). Baker et al. (2016), as well as many other economists, tested the index to be a reliable and easy to use EPU indicator.

There were paper-based studies of uncertainty before, but Baker and colleagues were the first to standardize the process and published their entire index online, of more than 20 countries, free of charge, with monthly updates. The method of Baker is better in the sense that it captures EPU in time series – no gaps between the event. Crisis events will be presented as uncertainty shocks, and their prolonged consequences will be shown in the EPU index of subsequent periods. The EPU index captures every event that involves large enough concerns to the economic policies, not just binding on a typical kind of occasions like a stock market crash or election (Baker et al., 2016). Also, the EPU index takes both fiscal and monetary policies into account (Baker et al., 2016).

3.2.3. The impact of EPU on FDI

The new EPU index stimulated a whole string of literature of EPU and investments of firms in modern setting. Most of the studies found EPU and

(24)

investments to have a negative correlation. For example, Kang et al. (2014) used the error correction model of capital stock adjustment and found EPU affected firms via “economic policy shock” times “stock price volatility” interaction channel. The effect of EPU was greater with firms having higher uncertainty properties; yet, it did not affect the top 20% biggest firms. Gulen and Ion (2013) indicated that policy uncertainty significantly negatively related to firm and industry-level investment. Approximately 2/3 of the drop in investments in 2008 depression was attributed to policy uncertainty.

Four noteworthy studies can be applied to the UK Brexit situation. Quang Nguyen et al. (2018) used Baker's EPU index to find the correlation between firm-level FDI, EPU, and firm hedging behavior. They ran a regression model including 881 non-financial firms FDI on relative EPU between home/host countries, firm-specific variables (sizes, leverage, etc.) and country control variables (GDP, unemployment, natural resource, exchange rate, distance, etc.). They found EPU differences between home/host to have a significant relationship with firm-level FDI. MNCs would increase FDI in a country that had a lower level of EPU.

Cebreros et al. (2018) built up a trade policy uncertainty index based on Baker’s index, using Google trends. Using that index, they found a negative correlation between Mexico’s trade policy uncertainty and FDI inflow into this country, especially in export-oriented states. Similarly, Noria and Fernandez (2018) studied the effect of EPU on FDI flows into the Mexican manufacturing sector 2007-2015. Using a panel dataset of sector level, a reduction of 1% of uncertainty (proxied by firm’s economic situation in 12 months) for every quarter from 2010-2015 would lead to an additional of 4,6% average FDI inflow per year. Idiosyncratic uncertainty for each manufacturing subsector was a better explanatory for FDI inflow than aggregate uncertainty. However, both EPU measures by Baker and the global risk aversion index by UBS still showed a significant adverse effect on FDI inflow into the Mexican manufacturing sector. What makes these studies valuable is that Mexico has been facing increasing trade protectionism from the US, Mexico’s leading trading partner, causing the trade policy uncertainty to jump because of the possibility of the

(25)

US renegotiating the North American Free Trade Agreement terms, or even a complete withdrawal. This situation is somewhat similar to Brexit, with growing concern about future trading policies with the EU causing the UK’s EPU to rise higher.

Bonaime et al. (2018) studied the association between political and regulatory uncertainty and merger and acquisition activity at the aggregate and firm-level. With a panel dataset of 151,925 merger and acquisition deals ranging from 1985-2014, the authors ran a vector autoregression model with M&A, EPU, and controls for mispricing, liquidity, and stock market volatility. A one standard deviation increase in EPU would lead to 6,6% and 3,9% decrease in deal value and number of deals in next year, with no evidence of future uptick indicating a postponed reaction of firms. On firm-level, a similar negative effect of EPU on acquisition likelihood was observed, one standard deviation increase in EPU would lead to 11,74% decline in the probability of a firm to perform a merger/acquisition in the next year. Again, there was no evidence that firms simply delayed their M&A until the EPU was back to normal. Even with controls for future economic conditions, industry-related shocks, or depressed valuation waves, the results remained the same. Four channels were suspected to link EPU and M&A activities (Bonaime et al., 2018): (1) real options theory, (2) macroeconomic uncertainties causing target values to drop before merger deal is completed, therefore discouraging merger, (3) higher uncertainty level might increase merger activities because it distracts investor attention, allowing the managers to quickly build up without immediate consequences, (4) mergers can be used as a risk management tool in high EPU. Between the four channels, Bonaime et al. (2018) found significant evidence for the first channel only, with channel (2) and (4) partially supported, and channel (3) not supported.

Altogether, the growing literature about EPU has shown connections between EPU and changes in the economic activities of businesses. However, these studies tend to be on the general side, circulating familiar sources of EPU (political fluctuations, policy, and regulatory changes, etc.). Later studies tried to go to the microeconomic level and succeed in linking EPU with business investments

(26)

domestically or globally. They will serve as an excellent source of models and explanations for further research into a particular EPU event like Brexit.

3.3. Hypotheses

Building upon the literature shown I propose some hypotheses. To begin with, almost all the studies revealed a negative relationship between EPU and investment level (Bernanke, 1980; Dixit and Pindyck, 1994), between EPU and FDI level, either into a country in question or out of the uncertainty-riddled country (Quang Nguyen et al., 2018, Cebreros et al., 2018, Sarkar, 2019). Predicting models of UK economists also pointed out the high likelihood that FDI into the UK would suffer due to Brexit (Centre for Economic Performance, 2016; Serwicka and Tamberi, 2018; Tamberi, 2020). For these reasons, I expect the level of aggregate FDI to the UK to go down when the EPU of the UK goes up. The negative effect of EPU on FDI will be much stronger after the referendum point (June 2016) due to the additional force created by Brexit uncertainty.

Hypothesis 1: UK's economic policy uncertainty index is negatively related to the inflow of FDI into the UK.

Hypothesis 2: The effect of EPU on FDI inward is stronger after the Brexit referendum point (June 2016).

Like any other shocks, firms need some time to adjust their strategy after witnessing changes in EPU in their host country. The real options theory explains that firms would hold back and “wait and see” when they encounter uncertainty in their business environment, either coming from policy changes, stock market uncertainty, or even from firms themselves (Dixit and Pindyck, 1994, cited in Hsieh et al., 2019). For FDI decisions, it is even more true because most FDI projects have high costs to start and withdraw (Hsieh et al., 2019). Also, FDI decisions show a whole new level of commitment of the firm in the host country and tend to be in the long-term perspective (Hsieh et al., 2019). Therefore, it is expected for foreign firms investing in the UK to have a lagged response to EPU (caused by Brexit), because of the unique nature of this event. The UK is still one of the biggest markets in the world; it is irrational just to give up. On the other hand, firms still have to act soon and probably

(27)

divert some of their FDI activities to other countries, in order not to be caught by surprise in case an unfavorable Brexit agreement is settled.

Hypothesis 3: Inward FDI shows a lagged response to EPU.

EPU will not affect industries equally. The manufacturing industry has (1) extremely high capital investment in machinery and equipment, (2) very high substantive investments to maintain operation, and (3) very complex network of supply (Serwicka and Tamberi, 2018). All those conditions make it hard for companies active in the manufacturing industry to divert their FDI across the border. On the other hand, the service sector has mainly offices and computers, which means firms can move their business to other locations much more manageable. The service sector also relies on the EU's single market regulation to provide services to the EU market, as well as relying on the free movement of labor law for high-quality workers. Therefore, hypothesis 4 follows:

Hypothesis 4: The increase in EPU caused by Brexit will severely affect the FDI in the service sector the most.

(28)

4. CHAPTER 4: DATA AND METHODS

4.1. Data

This thesis will investigate the effect of EPU on UK’s FDI in aggregate and sectoral levels. The mergers and acquisitions (M&A) dataset is used as a proxy for FDI. The M&A activities dataset is downloaded from Zephyr, from 2010-2019. The choice of this proxy is due to three reasons. (1) There is a lack of firm-level FDI dataset. Due to the COVID-19 situation, I could not contact and acquire FDI databases from well-known sources like FDI markets. (2) M&A data allows the segregation into sectors, based on the target and acquirer's primary sector. (3) M&A data indicates new FDI. New FDI will be affected much more than concurring FDI under the uncertainty caused by Brexit – as explained above, with higher level of cost and commitment when doing a brand new project; new MNAs will be much more cautious, especially now they cannot be guaranteed that the EU treatment would be the same. On the other hand, firms that already had established FDI into the UK would probably continue to support the subsidiary, because the UK is still a huge, developed market with much potential on its own. Therefore, the impact of Brexit uncertainty can be best seen in the perspective of new FDI projects, making M&A activities a suitable proxy for FDI. In total, there are 8.503 unique M&A instances from foreign-owned firms with targeting UK firms. One of the challenges in the Zephyr database is the occasional missing of data, particularly in the acquirer's country code and deal amount. Manual mapping is done in Excel to map the acquirer's ultimate ownership, global ultimate ownership, and controlling shareholder country code to each acquirer. The deal completion date is used to determine which quarter the deal is in. The sector is determined by targeting firm’s primary sector code assigned by Bureau van Dijk (BvD). In the case this is not available, the target primary NACE Rev.2 code is used instead. Both of those are available in the dataset.

At first, all observations are grouped into 19 main BvD sectors. They are

distributed into a panel dataset of 40 quarters, from the 1st_{quarter of 2010 to the 4}th

quarter of 2019. However, due to lack of observations, the “public administration and defense” sector is removed. Moreover, five sectors with relatively low M&A projects:

(29)

Banks; Gas, Water, Electricity; Post and telecommunications; textile and leather; and Wood, paper are combined with other sectors, forming the final 15 sectors as follow:

- Chemicals, rubber, plastics, non-metallic products - Construction

- Education, Health

- Food, beverages, tobacco - Hotels & restaurants

- Machinery, equipment, furniture, recycling - Metals & metal products

- Other services

- Primary Sector (agriculture, mining, etc.) - Publishing, printing

- Transport

- Wholesale & retail trade

- Banks + Insurance Companies (Finance)

- Textiles, wearing apparel, leather + Wood, cork, paper (TLWP) - Post and telecommunications + Gas, Water, Electricity (Utilities)

In total, the panel dataset is formed by pooling the time series of 15 sectors’ M&A deals, quarterly, from 2010-2019, resulting in 600 individual observations.

For the EPU data, this study uses Baker’s EPU index of the UK, from

1998-2019. The data is available for free on https://www.policyuncertainty.com/. Due to the

availability of the index (only at daily or monthly level), monthly data is transferred into quarterly data by the averaging process. Details on how the EPU index is constructed can be found in the literature review.

Next, a dummy variable is added, equalling 1 when the time of the observation

is after the Brexit referendum date of 23rd_{June 2016 (from the 3}rd_{quarter of 2016}

onwards). This dummy is included to test if the Brexit phase has a significant impact on FDI into the UK.

A set of control variables is included. The first control variable is real GDP growth rate, defined as the growth rate of real GDP of the UK

(30)

quarter-on-quarter-last-year (%). GDP growth is expected to have a positive effect on FDI in the new theory of trade and the Melitz trade model (Assuncao et al., 2011). GDP growth rate implies the economic development prospects of a country, with increasing rates suggest a booming economy, boosting investments. Mohamed and Sidiropoulos (2010) found the same effect of real GDP growth rate with his panel dataset of 36 Middle East, North America, and developing countries. In this study, the effect of GDP growth rate can be different, considering only one country is researched. The data for GDP growth rate is obtained via the Office for National Statistics (ONS) of UK, series: Gross Domestic Product: quarter-on-quarter-last-year growth (current price, %, seasonally adjusted). However, MNCs can also look at other countries’ GDP growth to compare which economy has greater prospects for investment. To control for that, the average GDP growth rate of OECD countries will be used in addition. OECD average is chosen due to its economic similarity to the UK since the OECD consists of most developed European countries. The data for OECD GDP is from the OECD database.

The second control variable is exchange rate. The exchange rate of home to host country has a positive and significant effect on FDI inflow (Alba et al., 2009; Cleeve, 2008). A higher value of the host country currency will increase the FDI MNCs profit since firms FDI into host country will earn in host currency, which is appreciated over the home currency. However, a lower home/host exchange rate does not necessarily mean a lower FDI. According to Baldwin (1989), firms enter a market bear a sunk cost that cannot be recovered if exiting. The exchange rate would have to fall under a certain level to make earnings of foreign firms, after exchanging into home currency, lower than the entry-triggering level. Data on UK’s real effective exchange rates quarterly is from Federal Reserve Economic Data. Real effective exchange rate is used, which is the weighted average UK’s Pound in relation to an index or basket of other major currencies (Hayes, 2020, Investopedia).

The third control variable is interest rates. It is added to see the financial environment attraction of the UK. Interest rates are the return of investment; a higher interest rate in host country compare to home means the marginal product of capital for firms FDI into host country will be higher than into home; thus, they can make

(31)

more profit. So higher interest rates will help channel investors from low-interest to high-interest rate countries (Anna, 2012). The UK’s real long-term interest rate is collected from EconData.uk, which, in turn, is calculated from OECD data on long-term nominal interest rates and the GDP deflator (EconData, 2020). Similar to GDP growth rate, to accommodate interest rates from other countries, OECD’s average interest rate is used. The data is acquired from the OECD database.

The fourth control variable is trade openness. Following Cleeve (2008), Bruno et al. (2016), export and import percentage shares of GDP is used to calculate the openness of the UK in terms of trading. The ability to trade with more countries gives rise to firms opening subsidiaries and expanding activities to participate in the global value chain. Data for trade openness is obtained from Federal Reserve Economic Data. Gross Domestic Product, Imports, and Exports of Goods and Services into the UK (Millions of Pounds, Quarterly, Seasonally Adjusted) are collected separately. Trade openness is calculated as: (Export+Import)/GDP, and showed as % of GDP.

The fifth control variable is GDP of sector. Other papers used a variety of variables to control for growth in each sector, mostly based on Tobin’s Q, which equals (market value of an additional unit of capital)/(replacement cost) (Noria and Fernandez, 2018). For industry level, cash flow is usually used as a proxy for profitability (Noria and Fernandez, 2018). It is manufacturing sales minus wages of a sector. However, the data for wages at sectoral level is not available for the UK. Thus, the GDP of each sector is used as a proxy for the growth of the sector. Similar to the aggregate GDP growth rate, I expect this variable to have a positive effect on FDI inflows. Data for GDP for sector is acquired from ONS (current price, million Pound).

Table 4.1 provides the list and descriptive statistics of the variables. The general trend of controlling variables are followed. The exchange rate fluctuates around 83-90, but surges to 101,3 in Q4 2015, until drops straight back to 84,3 in Q4 2016. UK’s real interest rate has a decreasing trend, peaks at 3,29% in Q1 2011 and ends at -0,37%. Since Q3 2015, the real interest rate of the UK is negative. The GDP growth rate of the UK inconsistently moves around the mean of 3,65% with no apparent trend to capture. The trade openness increases from Q1 2010 to Q1 2012

(32)

(56,3% to 62,4%), then slumps to 55,2 in Q3 2015. Trade openness recovers quickly to an all-time high of 65,7% in Q1 2019. The descriptive table shows evidence for scaling discrepancy between some variables (GDP of sector, trade openness, and exchange rate). They will be taken care of by the transformation of variables in the following empirical chapter.

Table 4.1: Descriptive statistics for panel (M&A) dataset

Variables N Mean SD Min Max Unit

No of M&A projects 600 14,15 23,75 0,00 139,00 Projects

EPU 600 155,72 59,01 76,69 359,74 Points

Exchange Rate 600 88,61 4,93 83,45 101,62 Pound/Basket

Interest Rate 600 0,33 1,12 -1,40 3,29 %

GDP Growth Rate 600 3,65 0,76 1,70 5,20 %

Trade Openness 600 60,55 2,58 55,18 65,70 %

GDP of Sector 600 27085,27 30279,55 3706,00 150149 Million Pound

OECD GDP Growth 600 2,15 0,60 0,83 3,50 %

OECD Interest Rate 600 2,84 1,27 0,84 5,04 %

Notes: The time period is 2010-2019. Data is in quarterly level.

Table 4.2 and Figure 4.1 present the number of M&A projects in each sector. Almost all industries observe a reduction in average M&A projects after the referendum, with the exception of Construction; Food, beverages, tobacco; Other services; and Publishing, printing. The most affected industries include Education, Health; Hotels & restaurants; Metals; Primary sector; Banks and Insurance. Every industry has a dip in M&A projects at or some quarters after Q2 2016 (the red line on the figure), the point of Brexit referendum, with some near bottoming out (Banks + Insurance, Construction, Chemicals, Education Health, or Primary sector to name a few). While most sectors show a slow recovery, Publishing and printing sector sees a surge in new M&A projects comparing to the other. One last thing to note is that the

(33)

deviations differ largely between sectors. Some sectors have exceptionally low deviations between quarters. This will be picked up by the empirical models.

Table 4.2: Descriptive statistics for the number of M&A projects in each sector

Sector N Mean SD Min Max All Before Brexit After Brexit

Chemicals, rubber, plastics, non-metallic products

40 11,03 11,27 10,57 4,42 5 24

Construction 40 4,10 3,73 4,79 2,45 0 11

Education, Health 40 3,73 4,00 3,21 2,05 0 9

Food, beverages, tobacco 40 6,43 6,42 6,43 2,88 0 11 Hotels & restaurants 40 4,03 4,50 3,14 2,27 0 10 Machinery, equipment,

furniture, recycling

40 21,85 21,96 21,64 4,10 14 31

Metals & metal products 40 6,75 7,46 5,43 3,14 1 17 Other services 40 98,63 97,23 101,21 13,06 70 139 Primary Sector (agriculture,

mining, etc.)

40 3,40 3,80 2,64 1,77 0 7

Publishing, printing 40 10,30 8,35 13,93 6,30 0 26

Transport 40 3,90 4,19 3,36 2,51 1 12

Wholesale & retail trade 40 19,85 19,92 19,71 5,04 10 28 Banks + Insurance

Companies (Finance)

40 10,28 11,73 7,57 3,90 4 19

Textiles, wearing apparel, leather + Wood, cork, paper

40 3,23 3,42 2,86 1,87 0 7

Post and telecommunications + Gas, Water, Electricity

(34)

Figure 4.1: Number of M&A projects of each sector by time

Looking at Figure 4.2, the slight negative effect of the Brexit referendum is expected to M&A projects, with the per-quarter average of M&A projects before the referendum is 213,3; and after is 211. From Q1 2010 to Q2 2016, the number of M&A goes up gradually with the EPU index going the other direction, as expected. In Q3 2016, EPU skyrockets, peaks at 359,7. The number of M&A responses two quarters after by a dip to the lowest level since Q3 2010, at 169. In the coming quarters, M&A projects recover a bit, fluctuating around the 200 mark; however, it never reaches the high point of 269 in Q3 2014. The EPU index goes down as well when the effect of Brexit settled down from Q3, 2016 to Q2, 2018, but quickly goes back to 239,9 in Q1 2019, following failed negotiations between the UK and EU. Then it settles down again only to boosts up back to 236,7 in Q4 2019. Thus, it seems that although before the Brexit referendum, EPU and M&A projects have the opposite trends as expected;

after 26th_{June 2016 they share the same trend. Both drastically decrease, then start to}

(35)

Figure 4.2: Time chart of Number of M&A projects and EPU index

4.2. The empirical approach

4.2.1. Aggregate FDI and EPU: A time series approach

For the aggregate method, I aim to test hypotheses 1, 2, and 3. Mainly, the effect of EPU on the number of M&A projects will be assessed, with the Brexit event acting as either moderating or mediating effect of the relationship. But why are both needed? Theoretically speaking, a moderator variable influences the strength of a relationship between two other variables, and a mediator variable explains the relationship between the two other variables. In this case, it is unsure if Brexit makes the effect of EPU on M&A stronger, or if Brexit causes a spike in EPU, which in turn affects M&A projects. 2 model groups will be regressed, to test the mediating and moderating separately. The first model concerns EPU, Brexit dummy, and an interaction term between EPU and Brexit dummy. If they both show significant results, Brexit has a moderating effect on EPU and the number of M&A projects. The

(36)

second model concerns EPU and Brexit dummy only. A special mediating regression and test will be performed to test the mediating effect of EPU and Brexit. More clearly, individual pair regressions between EPU-Brexit dummy-Number of M&A projects determine the level of causality between pairs, and the full model with bootstrap to find out the direct effect of Brexit on M&A and the indirect effect of Brexit that passes through EPU on number of M&A.

The main variables will be FDI inflow into the UK (proxied by the number of M&A projects), and the EPU index, with the exogenous variable of exchange and interest rate, GDP growth rate, trade openness, and the OECD GDP growth rate and interest rate.

MAAt = B0 + B1 EPUt + B2 ZBR + B3 ZBR*EPUt + B4 Ct +  (1.1)

MAAt = B0 + B1 EPUt-2 + B2 ZBR + B3 ZBR*EPUt-2 + B4 Ct +  (1.2)

MAAt = B0 + B1 EPUt-2 + B2 ZBR + B3 ZBR*EPUt-2 + B4 Ct-2 +  (1.3)

Where:

 t: Sub-indexes for time.

 MAAt is the natural logarithm of the aggregate number of M&A projects in

quarter t.

 EPUt is the natural logarithm of the economic policy uncertainty index of the

current quarter. EPUt-2 is the natural logarithm of the economic policy

uncertainty index of 2 quarters ago. The second lags are chosen due to the high explanatory capability comparing to other lags. Also, it fits with other studies of EPU and FDI, for instance, Hsieh (2019) showed investment changes tend to come relatively fast: expecting outward FDI to decrease after two-quarters of an EPU shock in the host country.

 ZBR is the Brexit dummy. ZBR*EPU is the interaction term for the moderating

effect of Brexit.

 Ct is the vector of control variable in quarter t, including GDP growth rate, real

effective exchange rate, real interest rate, trade openness, and the OECD GDP

The impact of economic policy uncertainty caused by the Brexit referendum on aggregate and sectoral-level FDI into the UK

MASTER'S THESIS