The impact of Brexit on financial institutions in the UK and the EU: An event study analysis

The impact of Brexit on financial institutions in the

UK and the EU: An event study analysis

Bachelor Thesis

BSc Economics and Business Economics

Specialization Finance

University of Amsterdam

Date: 30 June 2020

Thesis Supervisor: Ekaterina Seregina

Student Name: Kevin Liu

Statement of Originality

This document is written by Student Kevin Liu who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Abstract

This paper studies the effect of Brexit on stock prices of financial institutions in the EU and the UK. We employ an event study analysis in which we calculate abnormal returns of the financial institutions for each day in our event window, which are later used to calculate cumulative abnormal returns (CAR’s). We observed a greater ratio of significant abnormal results for EU institutions, including banks, than for their counterparts. Our analysis of the abnormal returns showed that 60% of the total were significant results for EU banks, while only 10% were significant for UK banks. Our results from our CAR analysis suggest that UK banks experience more intense negative effects, while EU firms other than banks also are negatively affected by the Brexit outcome. We attribute these negative effects to implications tied to the Capital Market Union, political instability, and a depreciation of the pound.

However, these are only a few of many complex implications Brexit will have on the economies of both parties, while the full extent can only be known when the exit is completed, and we can analysis post-Brexit relations of the two parties.

Table of Contents

SECTION 1: INTRODUCTION ... 5

SECTION 2: LITERATURE REVIEW ... 6

2.1

B

I

... 7

2.2

R

E

S

P

... 9

2.3

H

... 11

SECTION 3: DATA DESCRIPTION & METHODOLOGY... 12

3.1

D

... 12

3.2 METHODOLOGY ... 14

3.2.1

E

S

... 14

3.2.2 Aggregation of abnormal returns... 16

SECTION 4: RESULTS & DISCUSSION ... 17

4.1

A

R

UK

&

EU

B

... 17

4.2

CAR’

&

R

O

... 23

4.2.1 Regression diagnostics ... 25

4.3

D

... 25

4.3.1 Limitations ... 26

SECTION 5: CONCLUSION ... 26

6. REFERENCES ... 28

APPENDIX: ... 30

Section 1: Introduction

Originally meant to take place on the 29th of March 2019, Brexit’s deadline has been

postponed twice due to failed negotiations between members of the Parliament and the now resigned prime minister Theresa May. Brexit refers to the United Kingdom leaving the EU, and is the result of a referendum on the European membership, more specifically the “2016 United Kingdom European Union membership referendum”. On 23 June 2016, a small majority of 51,9% voted to leave the European Union (Welfens, 2018). After 45 years of EU membership, the United Kingdom backs out of the European Union, taking both parties into uncharted waters. However, the UK will continue its activities within the EU single market until 2020, following a transitional period agreed upon in March 2018 (Welfens, 2018).

After this transitional period, there will be two main outcomes, either a hard Brexit or a soft Brexit. A hard Brexit will include leaving the EU single market and the customs union, which will put a strong depreciation pressure on the pound, and a replacement between the current soft border regime between Northern Ireland and the UK by a new hard border regime (Welfens, 2018). In the case of a soft Brexit, the UK will remain in the EU customs union and adapt to EU tariffs (Welfens, 2018). In the case of a no-deal, which means that the two parties failed to agree on either a soft or hard Brexit, the UK will immediately leave the EU and its single and customs union. A no-deal would amplify the negative effects through various channels, which will be discussed later.

The outcome of the referendum resulted in an increasing level of uncertainty for both parties (Schiereck et al., 2016). The direct financial impact have been observed by sharp decreases in share prices of financial institutions. According to Schiereck et al. (2016), share prices of the Royal Bank of Scotland, Barclays, and Deutsche Bank dropped by 18%, -17.7%, and -13.9%, respectively.

In the aftermath of Brexit, one important problem for the remaining EU27 countries will be the allocation of its wholesale banking market. Moreover, since 90% of the EU27’s wholesale banking market is located in London, together with many insurance contracts for the EU27 countries, they will be excluded from the EU after the UK leaves the European Union (Welfens, 2018). The financial headquarters located in London represent a key role in derivatives, which can be strong hedging instruments, foreign exchange trading, and

arranging big loans for EU27 companies (Welfens, 2018).

Consumer protection laws will no longer hold after Brexit. As for the UK, this includes macroprudential institutions such as the Bank of England and institutions

responsible for prudential supervision in banking, insurance and securities markets (Welfens, 2018). For the EU, the European Central Bank is responsible for supervision of around 120 large banks in the Eurozone, together with the European Systemic Risk Board, which covers macroprudential supervision (Welfens, 2018).

Cooperation in the macroprudential supervision and economic policy of both parties is essential, whereas bad cooperation will lead to higher downsides of the Brexit outcomes (Welfens, 2018). According to Welfens (2018), the UK government is likely to take into consideration significant changes in corporate tax structures and banking regulations in order to revive the growth rate of output to a higher level than the reduced level which is observed in the Brexit context.

Furthermore, the European Central Bank’s Composite Indicator of Systemic Stress, in short CISS, depicts an increase in financial market nervousness driven by the UK referendum (Welfens, 2018). Due to the unclearness of the Brexit outcome, higher levels of financial market nervousness can be anticipated in the years prior to the final Brexit deadline. This increase in nervousness can be translated into uncertainty, which is expected to have a

negative impact on the overall UK economy. This will be explained more in detail in the later parts of this paper.

The goal of our paper is to analyze the effects of Brexit on banking and other financial institutions from the two parties, the UK and the EU. We are going to employ an event study on stock prices of banks and other financial institutions around the referendum result date. According to Breinlich et al. (2018), stock prices reflect expectations about the future value of companies, aggregating all available information at any given point in time. This agrees with the Efficient market hypothesis, in which a capital market is considered efficient if it reflects all relevant and available information into security prices (Malkiel, 1989).

We will look into the effects of banks against other financial institutions, and

furthermore compare UK and EU institutions. This way, it is possible for us to see which of the two parties involved were impacted most by the outcome, and in which particular sector the effects were more prominent.

In order to analyze the impact of Brexit on banks and other financial institutions we are going to conduct an event study following the proposed methodology of MacKinlay (1997). An event study is useful given the rationality in the marketplace, because the effects of an event will be reflected immediately in security prices (MacKinlay, 1997). We first define the event window, which includes the event of interest and the period over which we examine the security prices. This window often stretches to multiple days, including the day of the announcement and the day after the announcement, which captures the essential effects of the announcements if the stock market is already closed on the announcement day

(MacKinlay, 1997). The event of our interest will be the UK referendum which took place on 23 June 2016. However, due to the results being announced in the morning of 24 June 2016, this will be our event date (Belke et al., 2016).

In the methodology section of this paper, we are going to present in detail the steps and regressions necessary to carry out our event study. In short, we will first estimate normal performance of the stocks by using the market model, and secondly estimate abnormal returns on the event dates, which will allow us to analyze the impact of our event.

The remainder of the paper is organized as follows. In section 2, we will present a literature review of important and relevant pieces of literature contributing to our subject. Section 3 consists of a detailed description of the data and methodology. In section 4, we are going to present the main results and findings of this research, followed by a discussion of these results. In the last section we will present a short conclusion of our event study.

Section 2: Literature Review

This paper aims to contribute to existing literature around the subject of the Brexit

conducted on this topic, and we will concentrate on the most relevant papers that target the same direction as our paper.

2.1 Brexit Implications

In his 2016 policy paper, Ständer discusses the potential problems of Brexit regarding the Capital Markets Union. The UK is an important member of the Capital Markets Union, hosting a large part of European capital markets located in London. According to Ständer (2016), the UK is responsible for up to three quarters of total EU activity in some segments of the market, which leads to a huge loss of participation if the UK leaves the Capital

Markets Union. As mentioned before, London’s role as the main wholesale financial centre is essential for other financial markets throughout Europe. Not only does it act as a refinancing, trading and clearing platform provider for institutional investors, but it also hosts Europe’s largest equity and derivative markets (Ständer, 2016). Following the points discussed in his paper, Ständer (2016) concludes that even after leaving the EU, London will remain a major global financial center and an important trading partner for EU27 countries. This requires close cooperation between the two parties after the Brexit. Regarding UK’s departure from the CMU, EU27 member states are determined to compensate this loss with more developed and integrated capital markets in the EU, while building necessary regulatory capacities, allowing the authorities to cope with a potential influx of financial capital (Ständer, 2016).

Belke et al. (2016) describe the Brexit outcome as a “lose-lose” situation for both the UK and Europe, with negative economic effects for both financial markets. Leaving the EU will impact the British economy through trade in goods and services, investment,

immigration, productivity and fiscal costs (Belke et al., 2016). In their paper, Belke et al. (2016) found that the referendum did not cause a long lasting impact, apart from a weakened pound and lowered UK interest rates, the financial markets which have initially struggled during the weeks following the referendum have recovered. Furthermore, they found that consumer spending remained stable, and investment remained constant even though a

significant amount of uncertainty about UK-EU trade relations weakened overall confidence. Markets will remain volatile due to uncertainty about the Brexit outcome for both parties. Previous literature show that especially during crises and political events, financial market volatility tends to increase significantly with spillover effects to other markets (Belke et al., 2016). This can lead to uncertainty about Brexit triggering spillover effects across markets, which has been observed by the authors. The results of their research showed that policy uncertainty which was induced by the Brexit referendum had huge spillover effects to financial markets. This level of uncertainty remained strong, which further weakens the investment decisions and hiring in the UK and the EU (Belke et al., 2016). This leads to their conclusion that the policy uncertainty caused by Brexit has the potential to cause more instability in important financial markets and damage the British real economy.

According to the policy paper prepared by Kierzenkowski et al. and published by the OECD in April 2016, two months before the actual referendum, the outcome of the latter is still uncertain, which leads to financial markets pricing in the risk of Brexit and furthermore an incease in economic uncertainty which is hurting confidence and business investments in the UK, hindering its growth. A prediction made by Kierzenkowski et al. (2016) in case of

Brexit includes several channels, through which the initial shock will be transmitted. The paper discusses both the near term channels and the long term channels. The most relevant channels for our research will be briefly discussed in the following paragraphs.

In the near term, heightened economic uncertainty has a negative impact on

confidence, which would limit spending decisions and lead to tighter financial conditions due to an increased risk premia resulting in an augmented cost of finance and decrease in

accessibility (Kierzenkowski et al., 2016). Furthermore, the loss of its unrestricted access to the Single market would have a significant impact on UK trade and its barriers to trade, including higher tariffs for goods (Kierzenkowski et al., 2016). Another cost to the economy is amplified through reduced economic migration to the UK due to barriers to the free movement of labour from the EU to the UK. Lastly, for the short term exchange rates, an appreciation of other currencies against the sterling would be the result of the financial shock of Brexit (Kierzenkowski et al., 2016). These channels would allow for a 3 percentage points decrease of the UK GDP by 2020, as predicted by the paper.

For the long term, there will be substantial structural changes in the economy of the UK in the process of adapting to the new relationship with the EU including new policies (Kierzenkowski et al., 2016). Relevant to the Single Markets problem, Brexit would decrease foreign direct investment inflows, which would result in lower UK business investments and ultimately a decline in capital stock (Kierzenkowski et al., 2016). Following Brexit, the UK does not have to contribute to EU budget anymore, which results in fiscal savings likely to be 0.3-0.4% of UK GDP per year, according to the predictions made by the paper. However, the lower GDP growth would limit the extent to which these fiscal savings can be used to relax fiscal policy (Kierzenkowski et al., 2016).

Schiereck et al. (2016) analyzed and compared the Brexit market reactions to the Lehman Brothers bankruptcy filing. In their paper, they find that short-run drop in stock prices related to Brexit was more prominent than the stock price decreases related to the Lehman Brothers bankruptcy. In the case of Brexit, as mentioned before, a lot of international financial institutions had their headquarters in the UK, taking advantage of a well-developed UK financial market together with EU passporting rules which lead to more efficient business in other EU member states (Schiereck et al., 2016). However, after the Brexit outcome, uncertainty regarding future regulations for these institutions, whether they will be relocated or else sacrifice beneficial access to EU financial markets increased significantly. A similar spike in uncertainty has also been observed for the Lehman Brothers case due to doubts about further government bailouts in order to avoid more bank failures (Schiereck et al., 2016). Boutchkova et al. (2016) have found that political risk results in higher stock return volatility. Furthermore, Brogaard and Detzel (2015) have found that market returns are negatively affected by higher uncertainty around government policy choices.

Schiereck et al. (2016) have found a strong negative cumulative abnormal returns after the Brexit announcement that amount up to -12% on average during their event window. Furthermore, they have observed a sharper decrease in stock prices for European banks compared to non-European banks. Financial institutions, especially those located in the EU have been affected strongly by the referendum outcome. However, from comparing the CDS spread reaction of the Brexit announcement to the bankruptcy filing of the Lehman Brothers, Schiereck et al. (2016) observed significantly lower CDS spreads for the Brexit case.

Additionally, there was little contagion of increasing CDS spreads for EU banks to non-EU banks, which shows that uncertainty is better managed by financial markets and there is more resilience in handling shocks arising from unexpected policy changes (Schiereck et al., 2016).

Ramiah et al. (2017) conducted a broad analysis of the Brexit announcement on various sectors of the UK economy, in which they find that the banking sector was negatively affected with large negative cumulative abnormal returns. The British banking sector was among the most affected sectors from the referendum outcome. Ramiah et al. (2017) formulate the hypothesis that Brexit is bad news for the banking sector, and will lead to negative abnormal returns and higher risk.

In their 2018 paper “The Economic Effects of Brexit – Evidence from the Stock Market” Breinlich et al. (2018) study the reactions of the stock markets towards the Brexit result. More specifically, three events were analysed in their study, the referendum, and two subsequent speeches by Theresa May which are expected to have impacts to tariffs and non-tariffs barriers in the UK-EU trade discussion. According to their research, initial stock price movements can be attributed to fears of a cyclical deterioration and a sterling depreciation caused by the referendum. Stock price changes on the first day after the referendum, namely the 24 June 2016, can be attributed to initial fears of an economic stagnation in the UK, which may be caused by a steep depreciation in the sterling following the decision to leave the EU (Breinlich et al., 2018). The exact impact of the depreciation of the sterling currency ultimately depends on the firms’ participation in international markets (Breinlich et al., 2018). Their conclusion adds evidence to support the hypothesis of an economic deterioration following the referendum. However Breinlich et al. (2018) did not find clear evidence that the decision to leave the EU would end up in an immediate recession, even if UK’s growth slowed down compared to other economies in the years after 2016. An important takeaway from their study is that the full scope of the Brexit decision will ultimately depend on the status of UK-EU relations after Brexit (Breinlich et al., 2018).

Cazan (2017) studied the effects of Brexit on the English banking system, with results that support the negative view of the Brexit outcome. From a panel of 11 financial

institutions that are listed on the London Stock Exchange, the results show sharp negative values for abnormal and cumulative abnormal returns, which have dropped significantly in the first days after the announcement. This implies that stock prices have experienced large drops, with the biggest drops observable for Aldermore, with a 47% drop, and for Virgin Money Holdings with a 33% drop (Cazan, 2017). This decrease in stock prices is a trend observed throughout the study, with cumulative abnormal returns mimicking the stock prices and decreasing as well in the first three days.

2.2 Relevant Event Study Practices

Kothari et al. published their paper on the topic of event studies in 2007, in which they go through existing literatures of the methodology of event studies, and furthermore provide a guideline which one can follow. The methodology of event studies has not changed much over time, the goal still remains in estimating and measuring the mean of sample securities and the cumulative mean abnormal return around the event date, so that we can compare the

two and analyze the impact of the event on the specific securities (Kothari et al., 2007). However, daily security data has become a more common occurrence, which increases the precision in measurement of abnormal returns. Furthermore, Kothari et al. (2007) discussed cross-sectional testing and interpretation in their paper, which plays an important role in interpreting the results of our study. We will make use of the literature of Kothari et al. (2007) in our methodology part, together with MacKinlay’s literature on Event Studies in Economics and Finance (1997).

The use of an event study enables us to measure the effect of a specific event on the values of firms, using financial market data (MacKinlay, 1997). This relies on the rationality in marketplaces, which allows effects to be immediately reflected in security prices. Thus, we can construct and analyse the impact using security prices over a relatively short period of time.

In their paper “Financial sector reform after the crisis: Has anything happened?”, Schäfer et al. (2013) conducted an event study in order to analyze the impact of four regulatory and structural reforms on stock returns and CDS spreads of banks from Europe and the United States after the 2008 financial crisis. Due to the similarities between the financial crisis in 2008 and the Brexit, both in terms of unexpectedness and expected negative impact on the financial sector, we can use their paper as a guideline in terms of conducting a similar event study.

Together with the Khotari et al. (2007) and MacKinlay (1997) papers, this section helps to get a concrete idea of an event study, both in practice and theoretically, in order to correctly apply the methodology to our event study. The following paragraphs are dedicated to briefly deepen our knowledge in terms of conducting an event study, so that we can better understand the methodology behind it.

In their event study, Schäfer et. Al. (2013) investigated four major reforms, namely the Dodd-Frank Act in the United States, the Vickers report in the UK, the Restructuring law and levy in Germany, and the Too big to fail regulation in Switzerland. These are also the four main events, which are used as event dates. The data consists of stock returns and CDS spreads of the largest banks in the United States, the UK, Germany, and Switzerland, all based on their market capitalization. The observations between every bank is determined by the amount of trading days between the first and the last observed regulatory event in each country. An estimation window of 140 trading days prior to the first event is required. In order to test for heterogeneous effects, banks are separated between investment and

commercial banks, which is based on the share of customer deposits in total bank assets and the share of non-interest income in total revenue. Furthermore, banks are split into systemic and non-systemic banks.

The estimation of normal returns is based on the market model, including the Stoxx Global Total Market Index as a benchmark index. This is a broad global index, which is widely diversified, and thus avoids contamination of effects which can be caused by

interdependency of firms or banks within the same country, or in different countries. For the estimation of normal returns, bank equity returns are regressed on a constant, the return of the market index, and dummy variables which are equal to 1 at each event date. Each regulatory event consists of a sequence of events, which are called subevents. For each subevent, there are three dummy variables: a pre-event dummy, equal to 1 the day before the event which

captures anticipatory effects, an event dummy, equal to 1 on the day of the event, and finally a post-event dummy, equal to 1 the day after the event. This dummies out other dates that can be identified as event dates in the estimation period, which avoids mismeasurements of normal returns influenced by other regulatory events. After the estimation of the normal returns, it is possible to estimate abnormal returns, which show whether the event had a positive or a negative impact. Analyzing heterogeneity of effects is done through testing between different types of banks. Furthermore, anticipatory effects are also tested by including abnormal returns on the day prior to the event date. Significance tests for the cumulative abnormal returns over the event date and the day after the event date are ran as their last step of testing.

2.3 Hypothesis

From the previously discussed literature, we can get a clearer overview of what effects have been studied and associated to the Brexit outcome. This allows us to draw up our hypothesis, which will guide our research throughout this paper. From the studies conducted in our literature review, we have observed rather negative impacts of Brexit on financial markets. Firstly, the Brexit outcome shows negative effects for both sides, with spillover effects into different markets (Belke et al., 2016). Secondly, there seem to be a further trend for the UK banking sector, which seems to experience a sharp negative impact (Cazan, 2017). In the case of the EU, further findings have established a same trend, in which EU banks have to deal with more decreasing stock prices (Schiereck et al., 2016).

With our paper, we want to further add evidence to the existing literature.

Furthermore, for our hypotheses, we will analyze whether Brexit has a significant negative effect on financial institutions in the EU and the UK. For this, we will draw up the following hypotheses:

Our first hypothesis states that Brexit has no effect on UK Banks:

𝐻₀: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑈𝐾 𝑏𝑎𝑛𝑘𝑠 = 0 𝐻₁: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑈𝐾 𝑏𝑎𝑛𝑘𝑠 ≠ 0 Our second hypothesis states that Brexit has no effect on EU banks:

𝐻₀: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝐸𝑈 𝑏𝑎𝑛𝑘𝑠 = 0 𝐻₁: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝐸𝑈 𝑏𝑎𝑛𝑘𝑠 ≠ 0 Our third hypothesis states that Brexit has no effect on UK firms other than banks: 𝐻₀: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑈𝐾 𝑓𝑖𝑟𝑚𝑠 = 0 𝐻₁: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑈𝐾 𝑓𝑖𝑟𝑚𝑠 ≠ 0 Our last hypothesis states that Brexit has no effect on EU firms other than banks: 𝐻₀: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝐸𝑈 𝑓𝑖𝑟𝑚𝑠 = 0 𝐻₁: 𝐵𝑟𝑒𝑥𝑖𝑡 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝐸𝑈 𝑓𝑖𝑟𝑚𝑠 ≠ 0

When we speak of firms here, the banks are excluded, however later in the paper, we will use firms when we speak of all the institutions.

Section 3: Data description & Methodology

In this section we are going to describe the data and the methodology used in our event study. 3.1 Data description

Our study examines the effect of Brexit announcement on the financial institutions in the EU and UK. The financial institutions covered in the study is disaggregated into banking and non-banking institutions in the research area. Based on the fact that it is impossible to cover the whole banking and non-banking institutions, we take into consideration the use of

financial entities listed on the MSCI index which is used as the benchmark for the market. All indexes listed in the MSCI is classified within the financials according to the Global Industry Classification Standard (GCIS). To further work with reasonable sample size, we use a benchmark of top 20 banking institutions in the EU and top 10 banking institution in the UK as well as 20 non-banking institutions respectively using the total of their balance sheet which is a crucial indicator of the soundness of the financial standing as criteria for the selection.

The reason for selecting more banks from the EU lies in her size which outweighs that of UK and more so some of the banks in the UK are only incorporated in the UK but have their headquarters outside. The selected top 10 and 20 in both banking and non-banking institution of the UK and the EU are based on the availability of data. This means that we concentrate on financial institutions for which data is readily available and retrievable. Moreover, with the perception that the banking and non-banking institutions employed in the analysis are the top institutions in the industry, the results obtained can be used as a

parameter to evaluate the whole financial system in EU and UK (Cazan, 2017). The stock prices of the selected institutions as shown in Table 1 and Table 2 below are retrieved from Yahoo Finance.

Table 1:

Banking Institution

UK EU

Name SYMBOL Total

Asset Billion (€)

Name SYMBOL Total

Asset Billion

(€) 1. HSBC Holdings HSBC 2,100.13 BNP Parisbas BNPQY 1,963.43

2. Barclays BCS 1,275.62 Credit Agricole

Group

CRARY 1,763.17 3. Lloyds Banking

Group

Source: Statista.com and Rlist.io

Note: Total asset figure is the financial standing as of December 31, 2019

Table 2:

Non-Banking Institution

UK EU

Name SYMBOL Total

Asset Billion ($)

Name SYMBOL Total

Asset Billion

($)

1. Prudential PUK 631.32 Allianz Group ALV.DE 1,060.82

2. Legal & General Group

LGGNY 618.98 AXA S.A. AXA.BE 1,035.07

3. Aviva AVVIY 531.35 Assicurazioni

Generali

ARZGF 585.41 4. Phoenix Group PHNX.L 282.10 CNP Assurances CNPAY 450.62 5. RSA Insurance

Group

RSAIF 22.87 Aegon AEG 449.62

6. AIG AIG 20.27 Zurich Insurance

Group

ZFIN.SG 374.58 7. Direct Line DIISF 10.377 Swiss Life Holding SLHN.SW 216.53

8. Chesnara CSN.L 10.196 Swiss Re SREN.SW 200.75

9. AON AON 2.94 Talanx TLX.BE 186.43

10. Ecclesiastical ELLA.L 2.8 Chubb CB 167.77

11. Ageas AGS.BR 98.56

12. Unipol Gruppo UNI.MI 87.81

4. Royal Bank of Scotland Group

RBS 930.78 Banco Santander SAN 1,446.15

5. Standard Chartered SCBFF 552.56 Societe Generale SGE.SG 1,275.13

6. Abbey ABBY.L 435.95 ING ING 846.22

7. Nationwide Building Society

NBS.L 306.96 Intesa Sanpaolo IES.DE 801.01

8. Bank of Ireland BKRIF 146.80 UBS Group UBS 782.45

9. AIB Group AIBRF 11.86 Credit Suisse Group CS 680.46

10. Arbuthnot ARBB.L 5.49 Banco Bilbao

Vizcaya BBVA 671.02 11. Nordea NRDBY 581.61 12. Danske DNKEY 475.39 13. Commerzbank CBK.SG 452.49 14. KBC Group KBCSY 350.18 15. Svenska Handelsbanken SVNLY 336.69 16. Skandinaviska SKVKY 311.45 17. Landesbank LLBN.SW 285.09 18. Banco BPM BAMI.MI 193.10

19. Banca Monte dei BMDPY 166.68

13. Baloise Holding BLON.SG 81.65

14. Hannover Re HNR1.F 73.78

15. Mapfre S A MPFRF 70.23

16. Storebrand ASA SREDF 66.14

17. Helvetia Holding AG HELN.SW 59.23

18. Vienna Insurance VNRFY 58.33

19. Sampo Oyj SAXPF 56.12

20. Cattolica di

Assicurazione

XCW.F 37.88

Source: ADV Ratings 2019

Note: Total asset figure is the financial standing as of December 31, 2018

Using the data of the financial institutions listed above and following the event study methodology, we can ascertain whether the confirmation of UK exit from the EU has an effect on the market stocks of the financial institution in the UK more than European Union member states or otherwise. More so, we will highlight whether the banking institution or non-banking is affected by the changes as well as which banking and non-banking institution are affected most.

3.2 Methodology

In this section, we will discuss the methodology used in this paper. We are using an event study methodology based on the papers of Kothari et al. (2007) and MacKinlay (1997).

We begin by defining our event window and our estimation period. The event

window contains our event of interest. In our case, it is the day of the 2016 referendum result, which is the 23 June 2016. However, the results became public the morning after, so we are going to use the 24 June 2016 as our event date. It is important that we choose the exact date at which the information hit the public market. MacKinlay (1997) defines the event window as “the period over which the security prices of the involved firms will be examined”, which is commonly expanded to multiple days. This includes the day of the announcement and the day after the announcement. Including the post-event day into our event window allows us to capture any effects which may have occurred when the stock market was closed during the event date (MacKinlay, 1997). Thus how event window will have three days, namely one day before our main event date, our main event date, and one day after the event date.

The estimation period represents the returns that are not affected by the event, and usually ends several days prior to the start of the event window (De Jong, 2007). This estimation period is used to estimate the normal returns using the market model, which we will discuss more in detail later. Our estimation period begins 200 trading days prior to the event date, ending one day before the start of our event window. It is important that the event window and the estimation period do not overlap, since this would lead to the event

influencing normal measures, which would make our estimation inaccurate (MacKinlay, 1997). Figure 1 represents a timeline of our event study.

Figue 1:

After having defined our estimation period, we can now use the returns of this period to estimate normal returns. An event study analyzes stock return behavior of sample firms following a common event, which impacts the sample firms. In their paper, Kothari et al. (2007) define the model as following:

𝑅_𝑖𝑡 = 𝐾_𝑖𝑡+ 𝑒_𝑖𝑡

where 𝑅_𝑖𝑡 is the return of the security for sample firm i at time t=0, which represents the time of the event. 𝐾_𝑖𝑡 is the expected normal return, and 𝑒_𝑖𝑡 is the abnormal component which is unexpected. Thus, the abnormal return 𝑒_𝑖𝑡 is the difference between the estimated return and the normal return of the security:

𝑒_𝑖𝑡= 𝑅_𝑖𝑡− 𝐾_𝑖𝑡

Moreover, the abnormal return can be interpreted as the difference between the return which is conditional on the event and the expected return which is unconditional on the event (MacKinlay, 1997). The abnormal return directly measures the change in return of the

security following the event. In order to estimate abnormal returns, a model for normal returns has to be established first.

For this, we will use the market model, which assumes a stable linear relation between the market return and the stock return (MacKinlay, 1997). The market model is a statistical model in which the return of the stock is related to the return of the benchmark market portfolio. Following the market model illustrated in MacKinlay’s paper, for any stock i the market model is:

𝑅_𝑖𝑡 = 𝛼_𝑖 + 𝛽_𝑖𝑅_𝑚𝑡+ 𝜀_𝑖𝑡 𝐸(𝜀_𝑖𝑡) = 0 𝑣𝑎𝑟(𝜀_𝑖𝑡) = 𝜎_𝜀2

Where 𝑅_𝑖𝑡 is the returns of stock i in period t, and 𝑅_𝑚𝑡 is the return of the market portfolio. 𝜀_𝑖𝑡 is the zero mean disturbance term and 𝛼_𝑖, 𝛽_𝑖, and 𝜎_𝜀2are the parameters of the model. In our estimation process, we will use the MSCI Index as a benchmark for the market portfolio.

The main advantage of using the market model as mentioned above is that by

reduce the variance of the abnormal return. This leads to a better detection of the event effects (MacKinlay, 1997). The 𝑅2 of our market model regression will ultimately reflect the benefits of using a market model, if we have a high 𝑅2 we have successfully reduced a great amount of variance of the abnormal return, and furthermore achieved a greater gain

(MacKinlay, 1997).

As mentioned above, abnormal returns are used to measure the effect of an event on the event date. We compare the actual returns, derived from actual stock return data, with our estimated expected returns. Using the abnormal returns, we are aiming to capture the event effects. The abnormal return for firm i in period t is:

𝐴𝑅_𝑖𝑡 = 𝑅_𝑖𝑡− 𝐸(𝑅_𝑖𝑡|𝑋_𝑡)

Where 𝑅_𝑖𝑡 is the actual return of the firm, and 𝐸(𝑅_𝑖𝑡|𝑋_𝑡) is the expected return conditioned on information X in period t, which is unrelated to the event. Moreover, 𝐸(𝑅𝑖𝑡|𝑋𝑡) is the expected return that we estimate with our model market. Substracting the

actual returns by the expected returns on the event date gives us the abnormal return on the event date.

3.2.2 Aggregation of abnormal returns

Moreover, a cross-sectional test can help us attribute the event-related effect to firm specific attributes (Kothari et al., 2007). This means that a cross-section of firms will allow us to compare the abnormal returns regressed against specific firm characteristics, such as the firm being a bank or a non-bank, or the firm being in the EU or the UK. From this, we receive more evidence which we can use to discuss further economic hypotheses (Kothari et al., 2007). The results from our cross sectional test are relevant even for mean zero effects, or for insignificant results, which lead to the acceptance of the null hypothesis (Kothari et al., 2007). Furthermore, we will be able to analyze how the effects vary through firms, since abnormal returns differ cross-sectionally due to several reasons. First of all, the economic effect of the event is not the same for every firm with regard to its extent and intensity (Kothari et al., 2007). The degree to which an event is anticipated by each firm also changes the extent of the effect with respect to the firm (Kothari et al., 2007).

We calculate the cumulative abnormal returns for each firm during our event window which is (-1;1). Hence, we will have three abnormal returns per firm in our event window, which we will sum together and receive cumulative abnormal returns for each firm. The CAR calculation will look like the following:

𝐶𝐴𝑅(𝑡₁, 𝑡₂) = ∑ 𝐴𝑅𝑡 𝑡2

𝑡=𝑡₁

Where the horizon length 𝐿 = 𝑡₂ − 𝑡₁ + 1. In order to test for significance of our CAR, we will use a student t-test. This will allow us to infer whether the mean abnormal performance is different from 0. The t-test will look like this:

𝑡_{𝐶𝐴𝑅(𝑡1,𝑡2)} = 𝐶𝐴𝑅(𝑡1, 𝑡2) 𝜎_𝐶𝐴𝑅 /√𝑁

With the calculated CAR’s, we will run an OLS regression, in order to be able to analyze our results. Using a multivariate analysis, we can regress our CAR on dummy variables, which will allow us to disaggregate effects between banks and non-banks, and the UK and EU. The OLS regression will have the following dependent and independent variables:

𝐶𝐴𝑅_𝑖𝑡 = 𝛽₀ + 𝛽₁𝑈𝐾 + 𝛽₂𝐵𝑎𝑛𝑘 + 𝛽₃𝑈𝐾 ∗ 𝐵𝑎𝑛𝑘 + 𝜀_𝑖

- 𝐶𝐴𝑅_𝑖𝑡 is our dependent variable

- 𝑈𝐾 is a dummy variable which is equal to 1 if the firm is in the UK, and 0 if it is in EU

- 𝐵𝑎𝑛𝑘 is a dummy variable which is equal to 1 if the firm is a bank, and 0 if it is a non-banking institution

- 𝑈𝐾 ∗ 𝐵𝑎𝑛𝑘 is an interaction term between the two dummy variables

The betas in our regression will ultimately show us the relationship between the dummy variable and our CAR’s. This allows us to draw final conclusions on the effects of the Brexit announcement on our sample, hoping that our findings align with expected findings which are partly described by our results of abnormal returns.

Section 4: Results & Discussion

In this section, we will go over our results of our event study. We will first go over abnormal returns of each individual firm, and compare them to one another, which will provide a deeper understanding of our CAR’s. Furthermore, we cannot draw inferences in this part of the results, since this will be done in the later parts of this section when we analyze our CAR’s and our regression output.

4.1 Abnormal Returns of UK & EU Banks

Our main goal is to highlight the market reaction in the UK and EU relative to the Brexit referendum outcome. Furthermore, we analyze its effects on the financial system of both parties. Tables 3-6 show the predicted returns, the abnormal returns, the variation of the abnormal returns relative to the predicted returns, and finally the T-values, which tell us whether our observations are significant or not.

Table 3 represents the abnormal returns of the top 10 UK banking institutions. Before the announcement, the abnormal returns witnessed in stock is influenced by the devaluation of the exchange rate as well as the expectation of economic slowdown or recession (Breinlich et al., 2018). Out of the 10 banks, only one bank, The Royal Bank of Scotland, has a

significant abnormal return value. 8 of the 10 banks examined in this study have stock returns that are negatively affected by the referendum outcome. In our analysis of the top 10 UK banks, there is a clear negative effect visible.

For the only significant outcome The Royal Bank of Scotland, the abnormal returns were -0.1876652, with a -17% variation from the predicted returns. Furthermore, The Bank of Ireland has the highest variation with -65%, and Abbey has the lowest variation with close to 0%.

Moreover, among the 5 independent banks in the UK, HSBC Holding has the smallest variation. It shows that the impact the announcement had on their activity is not as intense as in other cases. As the biggest bank in the UK with almost 2,000 billion EUR in total asset, the HSBC provides extensive services in many different branches which allows them to manage the negative effects more effectively (Cazan, 2017). Additionally, the diversity of their financial operations and the segmentation of their clients also affects the time frame and degree of its recovery. This observation is in agreement with the findings of Cazan (2017) who maintained that the negative impact of the referendum is smaller for HSBC Holdings.

Table 3: Abnormal Returns of the top 10 UK Banking Institution Predicted Returns Abnormal Returns Variation t-value AIB Group .0454683 -.0639407 -11% -.0602535 Abbey -.0029593 .000759 0% .0015181 Arbuthnot -.023245 -.0944756 -7% -.4100501 Bank of Ireland .2384142 -.406965 -65% -.5801448 Barclays .0754268 -.2363654 -31% -.4938474 HSBC Holdings -.0198992 -.0282411 -1% -.6724905 Lloyds .023938 -.0053428 -3% -.1443738 Nationwide Building Society -.0383665 .0383665 8% .1661932 Royal Bank of Scotland -.017327 -.1876652 -17% -13.00729

Standard Chartered .1784029 -.2607785 -44% -.5507096

Note: The t-value is based on a 5% level of significance

Table 4 shows the abnormal returns for the top 20 EU banking institutions. We found that out of the 20 banks analyzed, 12 showed significant results. For our EU analysis, ING has the biggest variation, with around 184% deviation from its predicted return. Furthermore, it represents a positive effect regarding the Brexit announcement. The lowest variation can be found at around 0% with Landesbank. Out of the 20 EU banks, 5 banks were positively affected, while 15 banks experienced a negative effect. This can partially be explained through the linked banking activities such as lending and deposit-taking between the EU and the UK. In an article posted on the European Central Bank webpage, Bergbauer et al. (2020) describe the reliance of the EU on the UK with respect to derivatives as an important source of interdependency. As documented by the article, in 2019, UK CCPs serves as the channel for clearing almost 80 per cent of clearing members’ OTC derivatives positions.

Predicted Returns Abnormal Returns Variation t-value BNP Parisbas -.0277984 -.2927315 -26% -9.094932

Banca Monte dei .0887822 -.0816804 -17% -.2639444

Banco BPM -.0403123 -.0157069 2% -.0796295

Banco Bilbao Vizcaya -.0187924 -.1026661 -8% -6.297039

Banco Santander -.0147316 -.1632659 -15% -4.527944

Commerzbank -.0196716 -.0725137 -5% -5.328961

Credit Agricole Group -.0300732 -.1520009 -12% -3.99374 Credit Suisse Group -.0686167 -.7339991 -67% -10.17996

Danske -.0131643 -.3956471 -38% -.4250838 Deutsche Bank -.0320773 -.1514069 -12% -6.588132 ING .200855 2.038976 184% 3.07161 Intesa Sanpaolo -.3933574 -.8664355 -47% -1.701843 KBC Group -.0184302 -.1738009 -16% -8.490634 Landesbank -.0011077 -.0006608 0% -.0323946 Nordea -.0137305 -.0665646 -5% -7.47476 Raiffeisen -.0157741 .0157741 3% .5336741 Skandinaviska -.0136074 .0148667 3% .9312124 Societe Generale -.0248715 -.3127046 -29% -9.463634 Svenska Handelsbanken -.0134315 -.0418391 -3% -3.063458 UBS Group -.0104336 -.5952556 -58% -1.647781

Note: The t-value is based on a 5% level of significance

From our initial analysis on banking institutions between the EU and the UK, we can observe that there are more significant abnormal returns for EU banks than for UK banks. The ratio shows that abnormal returns were significant for 10% of the UK sample while for the EU sample it was 60%. UK banks may have been able to manage the negative impact of the announcement better than their counterparts in the EU, and were able to cushion its effect.

Figure 2: Abnormal Return values of the top 10 Banks in the UK

Figure 3: Abnormal returns value of the top 20 Banks in the EU

We will now move on to our observations of non-banking financial institutions and the Brexit announcement. This expands the scope of our research with the aim of having a

-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3

UK Banking Institution Stock Returns

predicted Returns abnormal Returns

-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

EU Banking Institution Stock Returns

more descriptive and detailed analysis of the whole Brexit announcement on the financial sector of the EU and the UK.

Table 5 shows the abnormal returns of the top 10 UK non-banking financial institutions. Out of the 10 financial institutions studied, only 2 show significant results, namely Legal & General Insurance and Prudential. The highest negative variation of -339% is also observed for Legal & General Insurance. The lowest variation of 1% and -1% can be observed for RSA Insurance and Direct line respectively. A total of 4 financial institutions experienced positive variation in their stock return, while the rest of the 10 experienced negative variation in their stock return.

Table 5: Abnormal Returns of top 10 UK Non-Banking Institution

Predicted Returns Abnormal Returns Variation t-value AIG .0010401 .017236 2% 1.181231 AON -.0368038 .1109117 15% .2675612 Aviva .1914381 -.2927248 -48% -.2016389 Chesnara .0157043 -.0113946 -3% -.080851 Direct Line .0057284 -.0057284 -1% -.2354222 Ecclesiastical -.0632372 .0604322 12% .1490718

Legal & General -.1440592 -3.537652 -339% -31.99791

Phoenix -.0161817 -.0918653 -8% -1.945541

Prudential -.035402 -.1568389 -12% -6.695431

RSA Insurance -.0046437 .0046437 1% .2075644

Note: The t-value is based on 5% level of significance

Table 6 shows the abnormal returns of the top 20 EU non-banking financial institutions. For the EU, 14 out of the 20 studied entities show significant results. Unipol Gruppo experienced the largest negative variation of 14% with abnormal returns of 0.1702897, while Vienna Insurance has a variation of around 0% with abnormal returns of -0.0005341. In total there are 5 non-negative variations in returns observed for Assicurazioni Generali, Chubb, Hannover Re, Mapfre S.A, and Vienna Insurance.

Table 6: Abnormal Returns of top 20 EU Non-Banking Institution Predicted Returns Abnormal Returns Variation t-value AXA S.A. -.0137749 -.1054407 -9% -5.34329 Aegon -.0269338 -.1201991 -9% -5.124716 Ageas -.0176518 -.0534578 -4% -4.158309 Allianz Group -.0199066 -.0897821 -7% -6.216332 Assicurazioni Generali -.0056536 .0056536 1% .240827 Baloise Holding -.010707 -.0337946 -2% -2.913022 CNP Assurances -.0042915 -.0562698 -5% -3.122136 Cattolica di Assicurazione .0015026 -.0926415 -9% -4.472492 Chubb -.018593 -.0186272 0% -1.978569

Hannover Re -.0219057 -.006537 2% -.4618993

Helvetia Holding AG -.0128605 -.035401 -2% -2.247949

Mapfre S A -.0043743 .0043743 1% .1528911

Sampo Oyj -.0168995 -.0749036 -6% -5.715446

Storebrand ASA .0045294 -.0045294 -1% -.2793445

Swiss Life Holding -.0157932 -.0395974 -2% -2.933648

Swiss Re -.0164152 -.0348473 -2% -2.93715

Talanx -.0042028 -.0933951 -9% -5.458076

Unipol Gruppo -.0352007 -.1702897 -14% -6.962881

Vienna Insurance .0005341 -.0005341 0% -.0224759

Zurich Insurance -.01559 -.0301995 -1% -1.677981

Note: The t-value is based on 5% level of significance

From figures 4 and 5, which show the abnormal returns of the top 10 UK non-banking financial institutions and the abnormal returns of the top 20 EU non-banking financial

institutions respectively, we can observe that non-banking financial institutions in the EU are more affected than those in the UK. However, this is merely an observation and is based on abnormal returns, thus we cannot make any overall inferences yet.

Figure 3: Abnormal returns value of the top 10 Non-Banking Institution in UK

-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5

UK Non-Banking Institution Stock Returns

Figure 5: Abnormal returns value of the top 20 Non-Banking Institution in EU

4.2 CAR’s & Regression Output

With the above mentioned abnormal returns, we computed our CAR’s of our event window (-1;1), which we then used to test for significance. The following table provides a brief

overview of the t-test conducted for our CAR’s:

Table 7

Event Window Observations Mean Std. Error Std. Dev. T-test

CAR[-1,+1] 60 0.0755065 0.0987587 0.07649814 0.7646

From this we can see that the t-test is 0.7646, which is insignificant. Thus we do not have enough statistical evidence to infer that at a 5% significance level, differences of our CAR are significantly different from 0. We will nonetheless continue with our regression, and analyze the effects of our two dummy variables together with their interaction term.

We will now move on to our OLS regression results, from which we can draw up our final inferences about the Brexit announcement on our CAR’s.

Table 8 represents our regression output, in which the coefficients for our dummy variables can be observed. As mentioned in the methodology part, our regression used the dummy variables “UK”, which is 1 for any UK firm, and 0 for any EU firm, and “Bank”, which is 1 for any banks, and 0 for any other type of financial institution. From this table, we can see that the dummy variables UK and Bank both resulted in insignificant results. UK has a t-test value of 1.46, while Bank has a slightly higher value of 1.59, however both are insignificant, and we can’t draw inferences from these values. Our constant, which is

significant with a t-test value of -6.45, shows us the effect if our two dummy variables are at their base case, which is 0. So if both dummy variables are 0, that means we have a non-banking institution from the EU. For a non-non-banking institution from the EU, we interpret a coefficient of -0.0936, meaning that it decreases our CARs by -0.0936. However, it is worth noting that interpreting the constant may be problematic, since the constant acts as a bias absorber for the residuals, it makes sure that our residuals have a mean of zero.

-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05

EU Non-banking Institution Stock Returns

For our interaction term between UK and Bank, we can observe a significant negative coefficient of -1.094. The interaction term of UK and Bank shows us that even though solely being in the UK or being a bank does not provide significant results, an interaction between the two of them shows a significant outcome, and we can interpret these results. This can be attributed to crossover effects of the two dummy variables, in which the interaction of the two will result in increases of the marginal effect of either dummy variable, which means that the marginal effect of UK, will now be the coefficient of UK and the coefficient of our

interaction term together with the dummy variable Bank, instead of just the coefficient of UK in the case without the interaction term. From our regression, we can infer that UK Banks performed worse than the rest, since the CAR’s for UK Banks will be -1.094 lower than for other CAR’s. This leads us to reject our first hypothesis, which stated that Brexit has no impact on UK banks. For our second hypothesis, regarding Brexit effect on EU banks, we do not have enough statistical evidence to draw conclusions on it. The same is valid for our third hypothesis, in which we stated that Brexit has no effect on UK firms other than banks.

However, from our interpretation of our constant, which is indeed significant, we can infer that Brexit has an effect on EU firms other than banks, which leads us to reject our last hypothesis.

Table 8

Regression output of CAR’s

(1) CAR UK 0.533 (1.46) Bank 0.348 (1.59) UK_Bank -1.094* (-2.55) Constant -0.0936*** (-6.45) Observations 60 R-squared 0.1158

4.2.1 Regression diagnostics

In order to run our OLS regression, several assumptions exist which we have to take into consideration and test, so that we can make sure that our results are unbiased estimates. From the Central Limit Theory, which states that a sample size from a population of over 30 observations results in the sample mean being normally distributed, we can test our sample for normality. Our sample consists of 60 observations, which means that normality can be assumed. Additionally, we have conducted a Sharpio-Wilk test, to back up the Central Limit Theory, and we have found that our sample is indeed normally distributed. We test our model for multicollinearity by using the VIF test. For multicollinearity to exist, we have to observe values over 10. For our regression, we observe values less than 5, meaning that we can assume that there is no multicollinearity. Lastly, we use a Cameron-Trivedi test to check for heteroskedasticity. The test shows us a low p-value, and a high chi2 value,which shows us that there is in fact heteroskedasticity. However, homoskedasticity is rarely observed for financial data, we can use robust standard errors for our estimations. The resulting tables of our tests can be found in the appendix.

4.3 Discussion of results

Our findings are agree for the most part with the findings of our literature review in which we discussed the results of previous studies concerning the Brexit outcome. Moreover, to a large extent, Brexit is expected to have negative effects and implications on the economy of both parties. This is suggested by the paper of Belke et al. (2016). From our abnormal returns observation, we can observe mainly negative abnormal returns, for both the EU and the UK firms. However, we fail to reject the null hypothesis of our CAR t-test, which implies that our CAR’s are not statistically significant and different from zero for our entire sample. In other words, we have no clear evidence of Brexit implications for our full sample of UK and EU firms. However, from our regression of our CAR’s, we can infer that UK banks are

performing worse than any other firm from EU or UK. This is however only observable through the inclusion of our interaction term, in which the relationship between UK and Banks is shown.

This in line with the findings of Cazan (2017), who has observed sharp negative effects for UK Banks after the Brexit announcement. Ramiah et al. (2017) also found that UK banks are significantly worse off than other sectors in the UK economy. Even though we do not have significant evidence for UK and Bank dummy variables, we can still draw up inferences on the impact of the Brexit announcement to UK Banks. The observed negative impact on the CAR’s of our sample from UK Banks can be attributed to the interdependency of banking institutions and the Capital Market Union, as explained by Ständer (2016). The uncertainty of the future of the Capital Market Union, together with rising political instability as explained by Welfens (2018) may have pushed UK banks into bad performance.

Furthermore, this may have been amplified by near term channels discussed by

Kierzenkowski et al. (2016), where the uncertainty negatively impacts confidence, which then again can limit spending decisions and ultimately lead to a deterioration of the economy. The depreciation of the pound also plays a role for UK banks, as it will push inflation above

preferred levels, with inflationary effects which will make it difficult to cushion shocks from the Brexit situation by cutting interest rates (Strauss, 2019).

4.3.1 Limitations

Regarding our empirical analysis, several limitations are to be noted. First of all, the sample size is relatively small compared to most studies of this type. In our study, we worked with a sample size of 60 different firms, resulting in 60 cumulative abnormal returns. From this, our statistical inference has been limited as well. We find several insignificant results, which can be attributed to many issues, and the sample size can be one of them. Furthermore, our CAR t-test for the whole sample turned out to be insignificant, which hindered us from observing statistically significant effects. Additionally, our regression of the CAR’s is fairly simple and easily structured, meaning we have not included control variables, which could increase important statistical values such as the R-squared. Furthermore, due to the Brexit event being fairly recent, there are relatively less accessible literatures than for other topics.

Further research can be done taking into consideration the limitations presented in our study, which will ultimately lead to more significant results and a bigger scope of explanatory power for event related effects. This can be done through increasing the sample size for both EU and UK firms, and including control variables which can increase efficiency and validity factors for the regression outputs. Also, as this is a fairly recent event which is still not completed, there will be more literatures and studies available for this topic in the future, which is beneficial for future research.

Section 5: Conclusion

This paper studies the reaction of stock returns of EU and UK firms to the result of the referendum on the EU membership, which took place in June 2016. The referendum ended in a decision in which UK voted to leave the EU. The key analysis is done with the help of abnormal returns, which when cumulated into cumulative abnormal returns, show the effects Brexit has on the firms from both parties during the respective event window.

Our findings have shown that the Brexit decision will cause important structural change of the financial landscape in the EU and the UK, as negative effects after the announcement can be observed for both sides. In our regression of CAR’s regressed on the dummy variables UK and Bank, together with an interaction term between the two, we find that the interaction term is the only significant evidence of the Brexit effect. From this, we find that UK banks performance were negatively affected by the announcement, since the CAR’s for these banks are -1.094 lower than for any other firm in our sample. Furthermore, we find that UK firms and all banking institutions have increased CAR’s compared to the other options. However, these findings are based on low t-test values, so these are not statistically significant and cannot be used as evidence to draw inferences.

Out of our four hypotheses, we have found enough statistically significant evidence to reject two hypotheses , the first being that Brexit has no effects for UK banks, and the second that Brexit has no effects for EU firms other than banks. The rejection of the first hypothesis

is justified by the significant result of our interaction term, which we mentioned above, represents a statistically significant negative effect on UK banks. Moreover, the second hypothesis is rejected by emphasizing the constant in our regression, which show that EU non-banking experience significantly lower CAR’s, which rejects that Brexit the effect on EU firms is not equal to zero.

While our findings can be used to draw inferences on the overall effects of Brexit on our sample, it is difficult to pinpoint the extent of the impact Brexit will have on the UK and the EU27 countries, as this is largely depending on the relationship between the two parties after the exit is completed. The future relationship is still unclear, while there are views that consider the UK joining the EEA after the exit (de Vries et al., 2017).

6. References

Ansgar Belke, A. B., Dubova, I., & Osowski, T. (2016). Policy Uncertainty and International Financial Markets: The case of Brexit. CEPS Working Document No. 429, December 2016.

Boutchkova, M., Doshi, H., Durnev, A., & Molchanov, A. (2012). Precarious politics and return volatility. The Review of Financial Studies, 25(4), 1111-1154.

Breinlich, H., Leromain, E., Novy, D., Sampson, T., & Usman, A. (2018). The economic effects of Brexit: evidence from the stock market. Fiscal Studies, 39(4), 581-623. Brogaard, J., & Detzel, A. (2015). The asset-pricing implications of government economic

policy uncertainty. Management Science, 61(1), 3-18.

Cazan, S. A. (2017). Brexit Implications over the England Banking System–An Event Study Approach. Journal of Public Administration, Finance and Law, (11), 81-92.

De Vries, C., Bracco Gartner, P., Haentjens, M., Korteweg, J., Markakis, M., Repasi, R., & Tegelaar, J. (2017). Implications of Brexit on EU Financial Services.

De Jong, F., & de Goeij, P. (2007). Event studies methodology. Lecture Notes.

Kierzenkowski, R., Pain, N., Rusticelli, E., & Zwart, S. (2016). The economic consequences of Brexit: a taxing decision.

Kothari, S. P., & Warner, J. B. (2007). Econometrics of event studies. In Handbook of

empirical corporate finance (pp. 3-36). Elsevier.

MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of economic

literature, 35(1), 13-39.

Malkiel, B. G. (1989). Efficient market hypothesis. In Finance (pp. 127-134). Palgrave Macmillan, London.

Ramiah, V., Pham, H. N., & Moosa, I. (2017). The sectoral effects of Brexit on the British economy: early evidence from the reaction of the stock market. Applied

Economics, 49(26), 2508-2514.

Schäfer, A., Schnabel, I., & Weder di Mauro, B. (2013). Financial sector reform after the crisis: Has anything happened?.

Schiereck, D., Kiesel, F., & Kolaric, S. (2016). Brexit:(Not) another Lehman moment for banks?. Finance Research Letters, 19, 291-297.

Ständer, P. (2016). What Will Happen with the Capital Markets Union After Brexit?.

Strauss, D., (2019). What Is The Effect Of The Falling Pound On Brexit Britain?. Retrieved from: https://www.ft.com/content/0ee55f40-b2c9-11e9-8cb2-799a3a8cf37b

Welfens, P.J.J. (2018). Macroprudential Risk Management Problems in Brexit. Retrieved fromhttps://www.intereconomics.eu/contents/year/2018/number/5/article/macroprude ntial-risk-management-problems-in-brexit.html

Appendix:

Table A: CAR Shapiro-Wilk W test for normal data

Table B: VIF test for multicollinearity