Have companies in areas within the United Kingdom benefited from their county's Brexit vote?

(1)

1

University of Amsterdam, Amsterdam Business School

MSc Finance, Specialist track: Asset Management

Master Thesis

Have companies in areas within the United Kingdom benefited from their

county's Brexit vote?

Ilkem Ozturk

11686405

Supervisor: Mr. R. Döttling

(2)

2

Statement of Originality

This document is written by Ilkem Ozturk who declares to take full responsibility for the content of this document. I declare that the text and the work presented in this document is 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 solely responsible for the supervision of completion of the work, not for the contents.

This document may not be shared or uploaded without the permission of Ilkem Ozturk.

(3)

3

Acknowledgements

I would like to thank my supervisor Robin Dottling and all other members of staff at the University of Amsterdam that helped contribute towards the completion of this thesis. Without their ongoing support this research would not have been possible.

(4)

4

Abstract

This study will attempt to examine whether individual counties within the United Kingdom have benefitted from their aggregate Brexit vote, by assessing firms in areas which voted to Leave the European Union against those in areas which voted to Remain. The study will then go further by exploring a portfolio made up of these companies. By building upon existing literature which generally inspects either the overall or industry specific impact of the vote outcome, this paper will seek to establish whether individuals have behaved rationally and how they have been impacted as a result of financial changes in relation to their voting choices in the 2016 United Kingdom European Union membership referendum. Similar to other works, we observe that firms in all areas experienced some form of significant negative abnormal returns. Furthermore, the results indicate that firms in Remain areas were more susceptible to this shock. A result, which could indicate that these firms are more likely to suffer from Brexit in the longer term.

(5)

5

Contents

1 Introduction . . . 9

2 Literature review . . . 12

2.1 Rationality . . . 12

2.2 Efficient Market Hypothesis . . . 12

2.3 Voting motivations and characteristics . . . 14

2.4 The effects of Brexit upon the UK economy . . . 17

3 Data . . . 20

4 Methodology . . . 22

4.1 Event Study . . . 23

4.1.1 Structure of study . . . 23

4.2 Formation of Portfolio . . . 26

4.2.1 Definition, structure and contents of portfolio . . . 26

5 Results . . . 28

6 Robustness checks . . . 33

7 Conclusion . . . 37

8 References . . . 38

(6)

6

List of tables and figures:

Figure 1: Visual illustration of event study time sequence.

Table 1: Summary statistics of all data used in initial event studies.

Table 2: Table displaying the abnormal returns for firms in Pro-Brexit & Anti-Brexit areas,

along with a combination of the two in a Total UK group. Event Window (-1, +1).

Table 3: Table displaying the cumulative abnormal returns for firms in Pro-Brexit &

Anti-Brexit areas, along with a combination of the two in a Total UK group. Event Window (-1, +1).

Table 4: Table displaying the abnormal returns for All UK firms, firms in Leave areas and

firms in Remain areas in the (-5, +5) event window.

Table 5: Table displaying the cumulative abnormal returns for All UK firms, firms in Leave

areas and firms in Remain areas in the (-5, +5) event window.

Table 6: Summary statistics of all data used in robustness test event study, with initial Leave

group and Remain group excluding London firms.

Table 7: Table displying the abnormal returns for All UK firms, firms in Leave areas and

firms in Remain area, excluding London in the (-1, +1) event window.

Table 8: Table displaying the cumulative abnormal returns for All UK firms, firms in Leave

areas and firms in Remain areas, excluding London in the (-1, +1) event window. The vales in brackets denote the t-test values.

(7)

7

List of abbreviations: UK - United Kingdom EU - European Union AR - Abnormal Return

AAR - Average Abnormal Return CAR - Cumulative Abnormal Return EMH - Efficient Market Hypothesis FDI - Foreign Direct Investment

(8)

8

“Coming together is a beginning; Keeping together is process; Working together is success.”

(9)

9

1. Introduction:

The United Kingdom (UK) became an official member of the European Economic Community in 1973. Since that time, it has held two public referendums to determine whether or not it should remain a member. Following a 1975 referendum the general voting populous overwhelmingly voted to remain part of the European Community, with a 67.2% majority. Although, skepticism regarding the EU had been present at large in the ensuing decades, the UK did not hold another referendum again until 2015. However, in this case the vote to leave (what has now become the European Union) was victorious with 51.89% of the votes cast.

Ever since the 2015 decision to hold a Brexit referendum was announced, by the then UK Prime Minister, David Cameron, there has been constant debate as to not only the effects upon the UK, but also the entire global market. When viewed in an overall context, extensive research has been undertaken by multiple parties ranging from European Policy advisory groups to giant banking corporations in establishing both the transient and longstanding economic implications, of the UK’s decision to withdraw from the European Union. Beyond this, rather unsurprisingly, there is also an abundance of literature concerning the

characteristics, motivations and aims of proponents at either end of the respective argument. However, to date there appears to be no relevant research interconnecting these two vital ‘components’ of the Brexit decision. This work will seek to address, what could be

considered of not only financial, but also British societal research, by establishing how both firms and individuals have been impacted according to their voting patterns in the June 2016 referendum.

This analysis will initially be performed through an event study, which will sort areas according to how they based their vote in the Brexit referendum. From here, the firms within these areas will be tested to observe any abnormal returns. Furthermore, a portfolio will be created containing firms in these counties to test and see if a significant pattern can be shown between stock returns and counties, depending on how they had voted. A final robustness test will involve only testing firms in areas which demonstrated large majorities. We observe that the main concerns when making a voting decision, were not based primarily upon individual economic wellbeing. However there is a clear characteristic voting pattern linked to financial

(10)

10

prosperity. As a result, by testing the major providers of employment in these areas, this paper hopes to investigate individual area’s prosperity. These county boarders are split according to NUTS 3 data, which has been established within the UK electoral system for a considerable amount of time1. Furthermore, this research will create portfolios of these counties, to culminate in a final rotund conclusion. If we are able to detect a pattern of positive movement for firms in counties with similar voting patterns, then there is cause to believe that individuals within these areas were justified in making their decision. In recent decades, many counties have suffered the consequences of globalization, on much cherished industries, which once employed thousands of local residents. The confidence of the global stock market on whether these counties will now improve on the whole, due to departure from the EU, can be a clear signal for how the future of the UK will develop.

The uncertainty of the event itself, caused a sharp depreciation of the pound to its lowest level since 1985 (Plakandaras, Gupta, & Wohar, 2017) & further, left large employers such as Britvic warning the vote would only create “additional consumer…uncertainty” - Chief executive Simon Litherland. As the Brexit ‘split’ is still yet to occur, no definitive answer can be given as to what the exact consequences of its execution will be. The

implementation of Article 50 is only the framework for the UK’s ‘divorce settlement’ from the EU, and does not outline any future state of affairs (Dougan, 2016). For this reason, the following work will hope to serve as a prelude to the eventual withdrawal of the UK from the EU. Analysing, how the current existence of uncertainty will already be affecting individual companies across the nation. Moreover, this work will endeavor to provide explicit evidence for other countries suffering bouts of Euroscepticism (Taggart 1998) as to the possible future ramifications of leaving this political and economic union.

The contents of this thesis are as follows: Chapter 2 will explore and critically evaluate relevant literature and develop the hypothesis further. Chapter 3 outline the collection and analysis of the data used for the research being conducted. Chapter 4 will highlight and identify different methodologies that are used. Chapter 5 will contain the final results and a detailed explanation of their meaning. Chapter 6 will be an explanation of the robustness tests used, to corroborate the results of the event study. Finally, Chapter 7 will

1

The NUTS classification (Nomenclature des unités territoriales statistiques) is a system developed to divide the territories of the EU for the purposes of socio-economic research in a harmonized fashion by Eurostat.

(11)

11

provide a conclusion summary, while also outline possible future research that could be undertaken as a continuation of this topic.

(12)

12

2. Literature review: 2.1 Rationality

As a brief prelude to this chapter, there will be a short run down of the concept that is rationality. In financial literature, rationality (or bounded rationality) is sometimes referred to as acting to ‘postulate a satisfying strategy’(Simon 1997). Essentially, acting to the best of one’s knowledge, to both maximize utility and achieve ambitions. With this in mind, if in later succeeding sections we are able to establish the motivators of Brexiteers2, by analyzing these through quantifiable variables, we can establish whether individuals had been justified in their choice. It is important to understand, that although many individuals often only expressed one overriding factor for their referendum decision, this was not directly their own financial wellbeing. This said, there is the clear emergence of groups which feel left behind by the forces of globalization (Hobolt, 2016). These individuals who are left with concerns about future British prosperity, were often those with the highest levels of unemployment and lowest levels of income. Therefore, if we are able to substantiate the claim that if there may be a positive movement among these factors, as a result of the Brexit decision, then all those who advocated to leave may have been justified in their choice. Obviously, it can not be claimed beyond all doubt, that all firms who express high levels of returns are guaranteed to provide the local area with vast employment. As stressed repeatedly, this paper will not be able to make any long term conclusions, but only explore these factors in the short time following June 2016, till the present day.

2.2 Efficient market hypothesis

When contemplating our main and initial test, which is the stock market performance of UK firms, it is imperative to make note of this important area of financial literature. A research topic that deals with a combination of the predictability of stock prices and information incorporation. Together, coalescing to the theory that ‘prices reflect most information about fundamental value’ (Fama, 1970).

Although this work serves as the prelude to the UK’s departure from the EU, our initial tests will not accurately measure the effect of the UK’s departure. Instead, they will

2

(13)

13

measure the shock caused by the news of the vote. The central theory which encompasses such research stems from efficient market hypothesis advocated by Fama (1970). According to Fama, if various conditions are met, then there can exist an environment in which prices fully reflect the available information within a market. Or rather, that there exists ‘accurate signals for research allocation’. By performing tests upon the Efficient Market Model, Fama was able to conclude that in the majority of cases, that efficient market hypothesis was supportable with evidence. The consequences of this paper are central to our initial tests. In that, if the results exhibit a clear and significant movement, this is justifiably the reflection of how the market believes the UK economy is acting (or will act in the future). From this, conclusions can be drawn as to how specific counties have benefited or suffered as a consequence.

Further to this, many other papers have explored the relationship between specific or systemic news announcements upon stock returns. One such example Shapiro, Sudhof and Wilson (2017) use predictive models to test the predictive capabilities of sentiment on future economic activity. This very contemporary piece of work uses a proprietary machine learning predictive model. These models come from techniques developed by software company Kanjoya3. Ultimately, in most cases it appears news sentiment is statistically significant at 1%. This capable factor, is frequently able to improve forecasts for important economic variables from industrial production to employment. In addition, sentiment, gaged through financial articles, is able to exhibit log changes in the S&P 500. This work adds to theory surrounding the incorporation of news in general economic activity. The paper goes further to elaborate on how sentiment can be correlated with both future and contemporary

fundamentals. Another factor that could play a large role in Brexit’s future impact on the British economy.

Finally, one other interesting work surrounding a similar field includes,

Sathyanarayana and Gargesha (2016). This paper looks into the effects of the decision to demonetisation of a certain type currency in India on the BSE100. They formally concluded that there is a significant impact, on what claims to be the world’s fastest growing index. This work successfully highlights the large effects such small economic changes can have,

3

(14)

14

begging the question of ‘how sizeable is the Brexit effect?’. Multiple papers exist within the financial literature, from a range of countries, to aid in highlighting how large political decisions can have a detrimental effect on a nation's financial capabilities. More prominent works such as Neiderhoffer (1971), demonstrate how the probability of abnormal movements in returns are increased in periods of large global events. Together, these both display how an event which is so vital domestically and on the global stage, could have huge effects.

Specifically, they demonstrate why, even though there may be long-term benefits from Brexit, in the transitionary period how individuals on both sides of the argument may have suffered. The presence of EMH, could prove to be even more vital in the case of Brexit for various reasons. The largest aggravator of this fact could be the constitutional crisis caused due to Brexit, which saw both Scotland and Northern Ireland vote to remain members of the EU (Hughes & Hayward 2018).

The arguments and evidence for efficient market hypothesis, are still often debated at great length. Given the considerable amount of research on the topic it appears rudimentary to come to the conclusion that news plays some role in effecting price. The key points that need answering for the topic of Brexit, are as to how much this news will affect the economy. 2.3 Voting motivations and characteristics

During what could be described as an esoteric campaign, multiple motives were suggested by high ranking officials, as to why an individual would be inclined to vote either way. Ranging studies have been undertaken in the months following the result, hoping to establish exactly ‘who?’ voted to leave and ‘why?’. Although, no comprehensive answer can be given to exactly why the decision to leave the EU was taken, due to the election outcome being multi-causal and multi-faceted (Becker et al. 2017), extensive research has explored the various dimensions of the voting characteristics. It is these characteristics which provide the motivation for this work.

Prominent major policy makers and commentators have expressed a consensus of opinion as to the root causes for the widely bred Euroscepticism. The overall prevailing argument appears to point towards the question of British parliamentary sovereignty. A constitutional, rather than financial or economic decision, with almost half of all leave votes outlining “the principle that decisions about the UK should be taken in the UK” as their

(15)

15

biggest single reason (Ashcroft, 2016). Some even went to far as to describe this as the ‘most significant constitutional event in Britain since the Restoration in 1660’ (Bogdanor, 2016). Although, data and literary work have confirmed characteristic patterns, as those who have often dealt with low income and high unemployment were the prime voters to leave (Becker et al. 2017), with the question of sovereignty on the line, it may be considered trivial to dissect the short term economic consequences. Which many hope, will eventually be succeeded by long term parliamentary sovereignty.

Overall, although the major concerns attributed to Brexit may not be financial, there is ample evidence to support the theory that this is where such Euroscepticism can stem from. The fact that the result led to 3 trillion dollars being lost in global markets is also further testament to how important such an issue is financially. Although, this study is to look specifically at counties and their financial movements, it is testament to the nation-wide socio-economic climate that exists within a polarized United Kingdom. The ‘Left

Behind’(Goodwin, 2016) as Brexit voters have been labelled recently, possessed majorities in the sub-groups for those unemployed and receiving state pensions. Furthermore, there were huge divides between those citing work class backgrounds. One study, even confirmed that the working class index explained 58 per cent of the variation in the Leave vote across districts (Kaufmann, 2016).

While some academics, such as Alexander Betts, claim such polarizations are a sign of the failing distribution of globalization, there are contrasting opinions. Academics such as Kaufmann highly advocate the need for more stringent analyses, pointing towards more ingrained, deep rotted personal traits and values. Opinions such as those relating to party support, apprehension to European integration and even so far as outlooks relating to the death penalty (Kaufmann, 2016). Exploring personal differences, even ones as simple as those who ‘like change’ and those who ‘do not’, provides some fascinating insight into psychological variation that the referendum may have caused. With this said, as much of the existing research analyses ‘visible’ traits, much of the aforementioned literature can only be viewed in an objective manner.

As a relatively new research topic, there exists a minute amount of literature that has begun exploring sentiment (financially and socially) both before and since the UKs

(16)

16

referendum, on whether they should remain a participant of the EU. One such paper that does so is Arnorsson & Zoega (2018). In one of the most in-depth pieces of research currently available apropos of Brexit, this multi-layered piece of work first seeks to explore the pattern of voting. Then, explain its causes, to address the financial ramifications specifically as a result of these motivations. In conjunction with the works above, they address similar topics such as immigration and how the absence of British perceived sovereignty can do little to address this. Following from this, there is an acknowledgement of economic motivations for specific individuals. This telling analysis, again, aids this research in enhancing the rhetoric which describes Leave voters as individuals voting based purely on economic intent. Instead, of those voting purely based on issues surrounded immigration. While it has already been established, that a clear positive contribution has been made by migrants to the UK of £22.1 billion (Dustman & Frattini, 2013).

Despite this, economic motivations are often cohesive with self-interest. In recent decades, as a globalization has sprung hold, areas which had formerly prospered during the industrial revolution through the production of coal, textiles, steel or general manufacturing, have suffered great economic consequences. This is coupled with the ever-growing services and financial sectors in staunch Remain areas, such as London and Northern Ireland.

Furthermore, the authors characterize an important point, which is that the economies in such remain areas have recovered the fastest and most successfully in conjunction with EU

migration, since the 2008 financial crisis. Collectively, these factors help establish a platform for the research being undertaken in this paper which seeks to address who have been the ‘winners’ and ‘losers’ since the referendum.

A final point of interest, sees Arnorsson & Zoega reporting fundamental variations in the UK since the referendum. This is done by assessing the fundamentals of the sterling exchange rate. The authors are only able to establish a small level of causality between the proportion of those expressing a desire to vote leave and sterling movements, prior to the referendum. However, they are able to reach the conclusion that a 1% decrease in the exchange rate, was caused by every 1% increase in the number of individuals wanting to leave. These results are telling, in relation to both EMH and the general future of the UK economy. As the UK becomes a more attractive destination for foreign investment,

(17)

17

news to cause appreciation in the GPB. As no such scenario occurs, this could be construed as a sign that the UK economy will struggle on the global front in the long-term, following the Brexit decision.

2.4 The Effects of Brexit upon the UK economy

This section shall be the pinnacle section of focus for this chapter. Rather

unsurprisingly, there is little research that exists directly examining the county effects caused by the Brexit vote. That said, although the process itself has not fully taken place, there is already an abundance of work ascertaining to the sectoral and industry effects caused by the vote. In a self-explanatory fashion, helping highlight the global importance of such an issue.

The first article worthy of mention is by Bruno, Campos, Estrin and Tian (2016), which shows the increases of FDI inflows, as part of an economic union such as the EU. They draw conclusions that a country with the intention leaving this would ‘face a reduction in FDI inflows of around 22%’. Highlighting the detrimental effect of such a vote, with the use of a ‘gravity’ model. The authors express how this decrease could cause issues, as FDI is often complimentary with trade on an international scale and ‘other elements of financial globalisation’. This work adds more theoretical background and meaning to the undertaken research. Although, the research methods differ, this work also serves as an example of the possible future economic ramifications.

One particularly important piece of work belongs to Belke, Dubova and Osowski (2016), this paper, assesses ‘the impact of Brexit uncertainty on the UK and also on international financial markets’. By estimating the time-varying interactions between UK policy uncertainty and financial market volatilities, the paper successfully highlights the large role which policy uncertainty plays on volatility 'with magnitudes that had never been

observed before'. This paper will also be used as part of the basis for establishing the

economic motivations behind the Brexit vote itself. The paper even goes so far as to looking at Betfair probabilities and data compiled by Bloomberg, as measures of uncertainty.

Comparing these factors, which act as perceived probability, to determine the level stock movement changes as a result. These comparisons will hopefully highlight these event’s large role, beyond just uncertainty and volatility. Finally, this paper’s last important contribution is

(18)

18

its demonstration of 'uncertainty spillovers' which mean that Brexit-caused policy uncertainty could have the potential to damage the UK economy 'in the medium run'.

Recalling, that the majority of works summarise economics strife's within the UK on a short-term basis, the following paper is memorable for examining Brexit in such a way. The work of Raddant (2016) carries out a counterfactual analysis, to determine the responsiveness of the European stock markets to Brexit. Included in this analysis, is also a foremost look at UK stock prices also. The authors reach rather corresponding conclusions to other notable works, when stating that there was a systematic downward movement of stocks in the days following the election. However, the same authors pragmatically suggest this was the result of an 'undifferentiated panic reaction' across Europe and choose to explore the data up to a month following June 2016. These results are telling, in confirming an average decrease in stock prices of 10% across Europe, a fact they argue is unsurprising due to the close connection in behaviour between UK and European stocks. That said, they also establish, with the use of a univariate GARCH model, that in most industries the level of volatility had returned to pre-vote levels within three weeks. Although, this in conjunction with the fact that volatility had already been exceedingly high prior to the vote. Overall, this paper's main contribution is to show that although most industries within Europe had stabilised, that the financial sector (particularly in the UK and Italy) has suffered the most sluggish of these recoveries. Clearly, vital questions remain in this sector as to how the infrastructure of the financial markets will be shaped in the coming years. The prevailing level of uncertainty could prove problematic for areas such as London, which now thrive with financial offices providing services to individuals on a global scale.

Finally, a paper which employs some similarities in the research methods to those used in this work is conducted by Ramiah, Pham & Moosa (2016). This literary work makes up the majority of the previous literature, most commonly associated with the event study research being undertaken. Although, it greatly differs in approach, the work comparably seeks to address the issue of what financial changes will befall the UK. This paper explores sectoral changes, as a direct result of the referendum, by similarly contemplating CAR, for the period of June to July of 2016. The authors weigh out the positive and negative factors that could have repercussions for sectors such as banking, travel & even tobacco4. Eventually,

4

(19)

19

the authors conclude that Brexit effects the stock returns of almost all sectors tested (all but 6). The most important of which, banking, exhibited a cumulative AR of -15.37%. Even in industries they deemed to have received ‘extensive media coverage…in term of outcome’, the majority of cases again showed negative ARs. However, it should be noted that they cite many other works, such as a study by Woodford Investment Management (2016) and Booth et al. (2015), which all depict varying effects on a sectoral and national basis5. Their claim that all studies by and large ‘contain an element of ideological bias’, is testament to the lack of certainty as to the future bilateral agreements that will be reached (or not reached). These agreements, on migration, law and importantly trade which will determine the future

prosperity of the post-referendum British establishment.

5

Woodford Investment Management claims that due to the presence of fewer tariffs with non-EU members the UK has a higher probability of a successful Brexit - as a result of the EU’s diminishing gravity on the global economic playing field – than most predictions consider. Booth et al. conclude the UK’s GDP ‘realistically’ stands to either be 0.6% better or 0.8% worse off by 2030 depending on its policy stance.

(20)

20

3. Data

In this section, the data used for this research is described. For the purposes of the study being most accurate and informative, data was collected concerning UK firms from Compustat IQ. Data has been collected, encapsulating all current UK registered firms in the build-up to and aftermath of the June vote. More precisely, data concerning their daily returns along with the FTSE 250 index have been obtained to determine the excess returns. For various reasons some firms were omitted from the study. The primary reasons, were often due to lack of data concerning either location or missing returns. Others were excluded due to their location being within the Channel Islands or the Republic of Ireland, which are areas which ultimately had no say in the referendum vote. Firms which were located in Gibraltar were included in the study. This is due to the fact, that the area is legally a British Overseas Territory and residents received a vote in the referendum. One final issue which resulted in exclusion, was that many firms became publicly listed after the beginning of this research. Firms were then divided according to their location, depending upon how the area had voted. Areas were divided into districts, more commonly known as Local Government or Authority districts. This data was collected in conjunction with Eurostat and YouGov, on areas more precisely split into Nomenclature of Territorial Units for Statistics Level 3

(NUTS3). From here, 150 firms were selected at random from either subgroup, to give a total of 300 firms. This was done due to an imbalance, as a large number of firms preside in London. This will be addressed further with in the robustness tests. For the purposes of gaining clarity on the performance of both sets of firms, the two groups will be combined to provide a Total UK group, to accurately gain an overall representation. This combined group should also give more clarity to the overall effect upon the UK.

The FTSE 250 was selected as the index from which excess returns would be calculated. The reasoning for this stems from a series of counterfactual outcomes and facts. Mainly, that the FTSE 100 may fail to represent the possible downfalls of Brexit due to the movement and uncertainty surrounding the pound currency. As an overall conclusion, it appears that the FTSE 250 would represent most accurately represent the investment

sentiment surrounding the UK but also the potential effects of the Brexit decision, directly on individuals. With ‘two-thirds’ (Financial Times) of the FTSE 100’s revenue earned overseas

(21)

21

it is clear the FTSE250 would fare better in this position. Data for the index has been collected via Thomson Reuters. The availability of daily data has greatly aided in the coherence of this work.

All data collected within the tests is from the 1st of January 2015 till the 1st of January 2017. For context, the announcement that a referendum would take place was released in February 2016. By collecting data prior to this period, we hope to offer more insight into the entire process that has taken place. A data summary for the firms in the pro and anti-EU areas are displayed below. Also included is a summary of a combination of the two groups to provide a total UK group.

Table 1: Summary statistics of all return data used in initial event studies. The table displays

the mean, standard deviation and median of returns calculated for firms in areas which voted to leave and remain. A combination of the two groups is created to give a total UK group.

Group N Mean Std. Dev. Median

Remain 150 0.207% 9.844% 0.000%

Leave 150 -0.016% 2.159% 0.000%

(22)

22

4. Methodology

As previously outlined, this research shall perform tests advocated as event studies, in an attempt to capture the movement of abnormal returns across areas. An event study is the method most commonly used to examine the effect of a specific event, often measuring the impact on stock returns. The outcome of the referendum would clearly have had two very different reactions upon the stock returns of UK firms. Had the public voted to remain a member state of the EU, in most likelihood stock markets would have experienced little change. Instead, this research will measure the degree to which the Brexit vote impacted upon the stability and the stock market within the economy. For the application of the following event studies, prominent works of literature have been used to provide an adequate and elaborate framework. MacKinlay (1997), will be used to provide the main structure of the subsequent test. Simultaneously a combination of Thompson (1985) & Bowman (1983) will be used as references to outline an event study free of error.

With a combination of Fama's efficient market hypothesis and MacKinlay (1997), the effect of the referendum should be fully incorporated into the stock returns, as information discovery is immediately assessed. In defining the event, the day after the referendum will be used as the event of interest. This is due to the fact, that the results were not fully confirmed until the early hours of the 24th of June. Furthermore, many early exit polls predicted a victory for the 'Remain' campaign, a fact that would most like have diluted the data, making most results inconclusive. Overall, markets would have most clearly demonstrated the effects on the 24th of June, the day following the vote.

Beyond this, the research will hope to address whether a portfolio could be formed, which would long stocks from firms which preside in areas that voted to remain and shorts firms in areas which voted to leave. The main purpose is to highlight the variation in returns. This, along with the result from the event study it will hopefully provide an answer to the matter at hand. Some similarities and overlaps can be draw from previous pieces of work striving to ascertain conclusions on Brexit. Despite this, the main hypothesis for this work, is unique and should have a coherent and unique place within the financial literature:

H0: Companies within counties that voted in favour Brexit have experienced greater negative abnormal returns than those in counties which voted against.

(23)

23 4.1 Event Study

In this section an outline will be given of the event study methodology. Primarily, the structure of methodology will be expressed.

4.1.1 Structure of study Event window:

For the purposes of an event study, an event window must be selected. This is to include other dates which may have experienced abnormalities in investment activity and therefore stock returns, as a result of the observed event. As we are unaware, of exactly when the new information is completely incorporated the selection of multiple windows allows us to determine in which period the largest effect is observed. In conjunction with MacKinlay (1997) the first event window is selected for this study. The first is set 1 trading day before and after the event date (+1, -1, a total of 2 trading days), a total of 3 trading days. The second will be in conjunction with Ahern (2008) and will feature a window set 5 days before and after the event (+5, -5, a total of 10 trading days), a total of 11 trading days. As

mentioned, not only was there an unclear narrative presented by the political and media discourse in the hours following the vote, but this also existed to varying extents in the days and weeks leading up the 23rd of June. For this reason, it may be that each of the selected event windows has differing results, depending upon the financial mood created as a consequence of daily fluctuating polls.

Estimation window:

The estimation window, is a period of time selected prior to the event of study, to determine the expected return that would be have been realised. An estimation window must be long enough that it provides a clear and accurate representation of how the economy had developed in the time preceding the event itself. The event and estimation window will not overlap at any point (as shown in the time line represented below). Thus, as a result of this, the length of both can be calculated as shown:

L1 = T1 – T0

(24)

24

For this study, 250 days will be selected as the estimation window, again in

conjunction with MacKinley (1997) who suggests using this amount. The reason for this is to include a long enough period, as the referendum is likely to have been affecting the UK economy from the very moment it was announced that there would be a vote. For the purpose of clarity, this will mean that the estimation window will would be 249 days exactly for the event window of 1 day prior to the event. The length of the event window itself would be precisely 2 days, 1 day before and 1 day after the event.

Figure 1: Visual illustration of event study time sequence

Measuring normal & abnormal returns:

In conjunction with the estimation window, which will track the regular pattern of returns prior to an event, abnormal returns must then be calculated to determine how much this caused a fluctuation differing from the status quo. This abnormal return is the actual realised return minus the normal expected return. There exist multiple models and methods which are used in order to calculate normal returns. The most prominent of these methods are often split into either economic or statistical models, depending on the conditions required for each. For this study, an economic approach will be used, expanding upon the market model (MM).

The MM is essentially the use of a ‘single index market model as part of the

equilibrium returns generating process’ (Coutts, Mills & Roberts 1994). The MM links the return of any given security or stock to the return of the overall market portfolio. Although, in the majority of empirical research this method outperforms the CAPM model, there is still much dispute as to its effectiveness. For example, in cases of long-horizon event studies there

(25)

25

can be clear deviation from either the CAPM or MM, in the form of size or momentum effects. In such cases, it is more advisable to use Fama & French’s (1996) three factor model as an accurate benchmark. However, in this case is appears more agreeable to denote

abnormal returns is as residuals of the market model as part of the study.

Abnormal returns will be defined as follows, in order to calculate normal returns: ARit = Rit – NRit (1)

Where actual return is defined as:

Rit =αi +βiRmt + εit (2)

From here, abnormal return can be calculated as the error term of normal return model. This, in the case of the MM, is essentially the equation below. The unknown parameters of α & β are then used in order to calculate the final abnormal returns for each firm.

𝐴𝑅it = Rit - 𝛼̂i - 𝛽̂iRmt (3) Where: 𝛽̂ = ∑𝑇1 (𝑅𝑖𝜏 − 𝜇̂𝑖)(𝑅𝑚𝜏 − 𝜇̂𝑚) 𝜏=𝑇0+1 / ∑𝑇1𝜏=𝑇0+1(𝑅𝑚𝜏 − 𝜇𝑚) (4) And: 𝛼̂𝑖 = 𝜇̂𝑖 − 𝛽̂𝑖𝜇̂𝑚 (5)

Aggregation to gain AAR & CAR:

As the penultimate part of this analysis, aggregation must be performed to arrive at the final cumulative abnormal return (CAR). Aggregation will first involve calculating the average abnormal return (AAR) for each period of the event. This is calculated using the formula below. This is simply, the sum of abnormal returns, divided by the total number of firms. From this formula, it can be determined that any large differences between an AAR from zero should signal abnormal performance. Any and all abnormalities as a result of other causes, as

(26)

26

a pose to the referendum, should be cancelled out, as the study is entirely based around one specific event. AARt = 1 N ∑ ARit N i=1 (6) 𝐶𝐴𝐴𝑅 = ∑𝑡 𝐴𝐴𝑅𝑡 𝑡=𝑡1 (7) Testing significance:

Finally, to confirm the validity of the AR and CAR results obtained, their significance must be tested. The most common form of test, to observe if the null hypothesis of no abnormal returns, is a t-test (Bowman, 1983). As part of this parametric test, certain assumptions are made. The most specific being that, we assume all AR’s are independently and identically distributed.

As stated, the null hypothesis (for this specifically) is that all AAR are equal to zero, meaning that the Brexit would not have had an effect on prices or returns. For this, the standard

deviations of CAR and AAR are calculated (SAARt & SCAARt) and tested as below:

TAARt = √𝑁 𝐴𝐴𝑅𝑡 𝑆𝐴𝐴𝑅𝑡 (8) TCAAR = √𝑁 𝐶𝐴𝐴𝑅 𝑆𝐶𝐴𝐴𝑅 (9) 4.2 Formation of Portfolio

As an extra inclusion to this study, a portfolio will be formed of all firms to gain full financial retrospect. In a similar fashion to Fama & French (1993, 1996) a BREXIT portfolio will be created. This portfolio will entail the average return of firms in counties which voted with a majority to remain within the EU minus the average return of firms from counties which voted to leave to the EU.

4.2.1 Definition, structure and contents of portfolio

Fama & French (1993,1996) is often regarding as the pioneering research when considering the predictability of stock returns. In their works they advocate factors, which

(27)

27

build upon the CAPM model to further analyse and better measure stock performance. Their work has seen them reach conclusions that various characteristics such as value stocks and those with a small-cap were are likely to perform better (Fama & French 1996). For the final part of this research, similar methods will be used, to those advocated in these notable pieces of work.

Although, the aim of this research is not to create or test factor predictability, it will employ similar techniques to form a portfolio. The purpose of this section is to provide an overall representation of how firms in Leave areas performed compared to those in Remain areas. In the past, Fama and French have created factors such as SMB (small-minus-big) and HML (high-minus-low). This is done by subtracting the ‘average return on the three big portfolios’ from the ‘average return on the three small portfolios’ for the HML factor. The same technique will be employed here by subtracting the average AR of firms in Leave areas from the average AR of firms in Remain areas. Although, they choose to value weight their portfolios, for this study all firms will be equally weighted, as they are chosen at random.

As stated, the purpose of this section is to provide some overall financial retrospect. If we are able to observe a positive a value for this factor, it would aid the conclusion that firms in areas that voted to Remain were impacted upon less as a result of the Brexit vote. That said, if a negative value is found this would aid the opposite conclusion that firms in areas that voted to Leave were impacted upon less.

(28)

28

5. Results

The results of the initial event study are presented first. The abnormal results for both sets of firms are given on the three corresponding dates chosen as the event window. Also included are the results of the overall Total UK group, to demonstrate the entire nation’s performance.

Event window (-1, +1):

Table 2: The table shows the abnormal returns for All UK firms, firms in Leave areas and

firms in Remain areas in the (-1, +1) event window. The vales in brackets denote the t-test values.

Denotations of significance: 1% significance is denoted as ***, with 5% denoted as ** and 10% as *.

As we can observe, the results are significant at 1% for AR (+1) in both Leave and Remain areas and the Total UK group. This is however not the case for AR (0) and AR (-1) in all areas, where only two of the AR’s are significant at 1%. This appears to aid the conclusion that the Brexit decision’s impact was sizable enough to be observed, but to a varying extent. We are also able to observe that all three AR are negative for firms in Leave areas, whereas this is not the case in Remain area firms. It appears that these firms suffered most as a direct consequence of the Brexit referendum. This said, our results also indicate

Date Total UK Remain area firms Leave area firms

AR (+1) -4.121% (-4.642)*** -6.37% (-5.92794)*** -0.582% (-3.860617)*** AR (0) -4.821% (-2.853)*** -5.61% ( -2.84089)** -1.699% (-3.98393)*** AR (-1) -0.0045% (-1.7028)** 2.718% (-2.10025)*** -0.736% (-0.943325)

(29)

29

they recovered the most following the event of interest. The reason for this could stem from the fact that many financial institutions that preside in London did not expect the result, and as a result the uncertainty and shock was much more detrimental to them. This coupled with the fact that firms in areas which voted to Leave may expected the result to a much greater degree, and were therefore less unaffected. Another reason could be that the global market genuinely predicts that financial institutions and services (which were primarily based in London and other Remain areas) will suffer the most as a result of the vote, and this is now being incorporated and reflected in market returns.

For firms in Remain areas the AAR was -3.087%. For those in Leave areas this was -1.006% and for all firms this was -2.982%. The fact that the largest AAR were not found in all groups on same day leads us so suggest, that certain firms in different areas may have begun to incorporate the information at different speeds. The largest negative AAR was observed by firms in Remain areas on the 23rd of June. In forming a portfolio of the two groups we find a final value of -2.081%, again suggesting firms in Remain areas were the one to suffer most.

Next, the CAR is calculated to fully represent the effect of the vote. For this, CAR (-1, +1) and CAR (0, +1) are both calculated. The table below provides a full representation of these results for all groups.

Table 3: The table below shows the cumulative abnormal returns for All UK firms, firms in

Leave areas and firms in Remain areas in the (-1, +1) event window. The vales in brackets denote the t-test values.

CAR (-1, +1) -7.807% (-6.40973)*** -9.656% (-7.4701)*** -2.763% (-2.305502)*** CAR (0, +1) -2.446% (-1.6982)* -2.891% (-1.86791)** -2.151% (-1.94864)**

From the table above certain conclusions can be draw. We notice the CAR is negative and significant to some extent for firms in both the Remain and Leave areas. The average

(30)

30

CAR for the Total UK group is -5.1265%, for the Remain group this was -6.2735% and for the Leave group -2.457%. Our results clearly indicate one again that the effect of Brexit was highest on Remain firms.

Event window (-5, +5):

firms in Remain areas in the (-5, +5) event window. The vales in brackets denote the t-test values.

Date Total UK Remain area firms

Leave area firms

AR (+5) -1.21% (-1.7586)* -0.674% (-1.5763)* -1.839% (-1.8938) AR (+4) -0.784% (-2.7931)*** 0.0556% (0.8836) -1.308% (-1.7457)** AR (+3) 0.0752% (2.4823)** 1.015% (1.9789)** -0.924% (-1.7356)* AR (+2) 0.791% (1.90326)** 0.842% (1.0593) 0.446% (2.1355)** AR (+1) -4.898% (-4.1014)*** -6.578% (-6.184)*** -0.6209% (-3.4748)*** AR (0) -5.162% (-3.043)*** -5.781% (-3.5498)** -1.818% (-3.84854)*** AR (-1) -0.239% (-1.8847)** 2.881% (1.994)*** -0.784% (-0.89745) AR (-2) 0.528% (2.5538)** 0.638% (3.437)*** 0.429% (1.8208)* AR (-3) 0.185% (0.9374) 0.605% (1.3639)* -0.407% (-0.9574) AR (-4) -0.401% (-1.6718) -0.0278% (-0.8092) -1.332% (-2.195)* AR (-5) 0.116% (0.8372)* 1.155% (1.882)** -0.879% (-1.514)

(31)

31

From the table above we can observe that all AR are significant for the Total UK group and the firms in the Remains areas from -2 until +3. While the Leave areas still poses high degrees of significance at certain points, this is not sustained continually. The average AR for the Total UK group in this period was -1.04%, for the Remain group it was -0.534% and for the Leave group -0.822%. When forming a portfolio we reach a final value of 0.288%. Although, all these values are negative, they appear to contradict the ones obtained above. However, many reasons can be given as to why this may be the case. Namely, it may be that firms in Remain areas where labelled with greater confidence in the run-up to the vote. This could also be in conjunction with the fact that these firms may have had much larger recoveries following the results. Another, factor may be that fact that multiple large financial firms operate in London which would regularly experience greater returns compared to smaller firms in industrial areas. Finally, this may be due to the fact that an event window that is too large has been selected for an event such as Brexit, meaning the results are

inconclusive.

Leave areas and firms in Remain areas in the (-5, +5) event window. The vales in brackets denote the t-test values.

CAR (-5, +5) -7.905% (-6.4538)*** -5.328% (-4.75098)*** -8.763% (-6.58782)*** CAR (-3, +3) -5.881% (-1.7218)* -6.197% (-5.4571)* -3.251% (-2.170852)***

Unfortunately, from the results above we are unable to gain many conclusive insights to help address the hypothesis at hand. We are able to observe negative CARs for all three categories with varying levels of significance, which again solidifies the conclusion that the total UK suffered as a result of Brexit. A fact, which has been confirmed within other noteworthy pieces of literature. The average CAR for the Total UK group was -9.393%, for

(32)

32

the Remain group this was -5.763% and for the Leave group this was -6.007%. Leading us to the conclusion Leave area firms suffered more, but only to a minor extent.

At this stage of the research our results begin to express telling conclusions. Notably, in the first event window we have observed larger negative CAR for firms in Remain areas, in most periods. Furthermore, by observing the portfolios created for this window we again reach the conclusion that these firms have suffered to a greater extent. However, we can see that there are some differences in the two event windows, which is due to the change in number of days. It appears the second event window chosen has not been conclusive in understanding the hypothesis set forth earlier within the research. Due to their existence, at this stage we cannot, beyond reasonable doubt, substantiate the hypothesis that Remain area firms suffered more as a consequence of Brexit.

Specifically looking at the Total UK groups we are able to make various comparisons to other pieces of literature which have explored similar hypotheses. Unlike papers such as Ramiah, Pham and Moosa (2016) we are unable to observe a negative return above 10% for the overall UK economy, with the paper finding a result as high as -15.37% for the banking sector. This said, we are able to aid in the confirmation that Brexit clearly had a negative and significant impact on the UK as a whole. Even finding a figure similar to Raddant (2016) of around -10%. The difference in results with Ramiah, Pham and Moosa (2016) is most likely explained by use of different sample datasets and the fact they specifically look at different sectors. As a pose to, the entire UK or different geographical locations.

With regard to or main hypothesis, which was to test which firms had specifically performed worse as a direct result of Brexit, again from the initial event study fairly clear conclusions can be outlined. As highlighted above these results indicate how Remain area firms may have suffered most from the shock of the vote. The reasons for this are also outlined above. Due to these reasons, which are at a glance, are primarily linked to the vast amount of financial firms, our robustness test will be vital. If in the following section we are able to establish similar AR and CAR for both Leave and Remain area firms, while excluding London, we may be able to reach the a conclusion similar to that in Ramiah, Pham and Moosa (2016), that financial firms were the prime sufferers of the Brexit vote shock.

(33)

33

6. Robustness tests

For the purpose of this study being as detailed and thorough as possible this chapter will investigate an alternative method of testing to ensure our previous results are credible. This section will consist of similar techniques to those above, while now excluding all firms in London.

6.1 Excluding London

The first part of this section involves excluding all firms from London. This is linked to the fact that over 30% of the firms in the initial part of the study were based in the capital. Furthermore, a large number of these firms are more likely to have been giant multinational banking corporations. A fact, which could have distorted the results. London is viewed to be the main area in the UK to have benefitted from European diversification, by excluding this area the study can explore the effects on other areas. Areas which do not have the same growing financial services industry, to provide economic growth. Which are instead reliant on other industries, which may have suffered as a result of UK membership to the EU. All London firms have been removed from the initial study and replaced at random by others in different areas of the country. For the purpose of keeping the results clear and coherent the same firms from the early sections have been selected as the firms in the Leave areas. The tests are performed in a similar fashion to those outlined in Section 4. The results are outlined below, again in a corresponding manner to previous results.

As displayed by the table below, we can observe a sharp decrease in the mean level of returns for firms in the Remain category now that there has been an exclusion of firms within London. From here the AR is calculated once again in a similar fashion, for both subgroups and the Total UK group.

(34)

34

Table 6: Summary statistics of all return data used in second event study. The table displays

the mean, standard deviation and median of returns calculated for firms in areas which voted to Leave and Remain, excluding London. A combination of the two groups is created to give a total UK group.

From here the AR is calculating using the same methods shown in Section 4. For this test, only the (+1, -1) event window is used.

firms in Remain area, excluding London in the (-1, +1) event window. The vales in brackets denote the t-test values.

AR (+1) -1.742% (-2.015)** -2.804% (-3.899)*** -0.582% (-3.86062)***

AR (0) -4.0247% (-5.345)** -5.095% (-6.487)** -1.699% (-3.98393)***

AR (-1) -0.258% (-1.846)** 0.319% (1.0947) -0.736% (-0.9433)

Group N Mean Std. Dev. Median

Remain 150 0,004% 1.589% 0.000%

Leave 150 -0.016% 2.159% 0.000%

(35)

35

From the table above, the results are again significant at 1% for AR (+1) in both Leave and Remain areas and the Total UK group. Again similar to the results above, this is however not the case for AR (0) and AR (-1) in all areas. The average AR for Remain firms was -2.527%, for Leave firms this was once again -1.006% and for the Total UK group this was 2.0082%. In forming a portfolio of the two main subgroups we find a final value of -1.521%. This section of analysis aids in the conclusion that Remain area firms suffered the most as result of the Brexit vote. Although the exclusion of London has played some role in the mitigation of these figures, the overall conclusion remains similar. Possible reasons for this could be linked to the fact that the majority of major cities within the UK voted to Remain members of the EU (e.g. Edinburgh & Liverpool). As the likelihood is that firms in these areas are most accustomed to the benefits of EU trade and migration, such a shock decision would impact upon them to a much larger extent. Other factors could be that these firms in large areas which expressed such a strong desire to remain part of the EU, simply did not expect the result.

Lastly, the CAR is calculated for the (+1,-1) window, while excluding all London based firms.

Leave areas and firms in Remain areas, excluding London in the (-1, +1) event window. The vales in brackets denote the t-test values.

CAR (-1, +1) -3.437% (-3.801)*** -4.5797% (-6.4781)*** -2.763% (-2.305502)***

CAR (0, +1) -3.263% (-3.473)* -4.775% (-6.671)* -2.151% (-1.94864)**

Finally, by observing the results in the table above, we come to our closing conclusion. We can see that the majority of observations were once again significant, demonstrating the Brexit vote’s negative impact. The average CAR for the Total UK group

(36)

36

was -3.35%, for the Remain area group without London this was -4.677% and again for the Leave group this was -2.457%. Overall, this leads us to confirm hat to some extent firms in Remain areas across the country suffered most as a direct consequence of the Brexit vote.

(37)

37

7. Conclusion

Overall the results of this research have aided in confirming the hypothesis that in the short term, the Brexit decision has been detrimental to the UK. Confirmation has been made of the fact that firms in pro and anti-Brexit counties have suffered from the widespread uncertainty and shock of the decision. Further evidence has been provided potentially displaying Brexit’s negative upon the UKs financial sector. This is due to the fact that a larger average CAR is found for Remain area firms excluding and including the capital, of -4.6773% and -6.2735% respectively. The hypothesis regarding whether, on the whole, Remain area firms suffered more than those in Leave areas could be proven to some extent. This is due to our finding that for the initial tests performed in this study, when creating a portfolio of the two sets of firms, a negative value of -2.081%. In our robustness test, when excluding the capital we find a portfolio figure -1.521%.

Overall our results suggest that London was the area to suffer most as a direct consequence of the 2016 European Union referendum. Further to this, that on the whole, Remain area firms suffered more to some extent, compared to those in areas that voted for the UK’s departure from the EU. Future studies will no doubt be undertaken, specifically

tracking the outcome for the overall UK economy as a direct result of Brexit. What this research appears to have highlighted (in a similar fashion to works detailed above) is that at present global financial firms are most susceptible to the uncertainty created by this political decision.

It has repeatedly been stressed that the final word on Brexit will need to be given at a later date once, the outline has been specified as to how the UK will operate. Again, as the election was often labelled as a point of sovereign and parliamentary independence, rather than an issue of economic gain, the financial changes in the short-term are likely to be trivialised by those championing these causes. This said, with the Governor of the Bank of England recently announcing that UK households were each on average ‘£900 worse off’ since the vote (The Guardian) it appears all individuals have suffered irregardless of geographical location.

(38)

38

References

Ahern, R., (2008) Sample Selection and Event Study Estimation, Journal of Empirical Finance, Vol. 16, No. 3, June 2009.

Arnorsson, A., Zoega, G., (2018) On the causes of Brexit, European Journal of Political Economy, ISSN 0176-2680, https://doi.org/10.1016/j.ejpoleco.2018.02.001.

Ashcroft, L., (2016), How the United Kingdom voted on Thursday… and why, Lord Ashcroft Polls,

www.LordAshcroftPolls.com

Becker, S., Fetzer, T., Novy, D., (2017), Who voted for Brexit? A comprehensive district-level analysis, Economic Policy, Volume 32, Issue 92, 1 October 2017, Pages 601–650.

Belke, A., Dubova, I., & Osowski, T., (2016) Policy Uncertainty and International Financial Markets: The Case of Brexit. CEPS Working Document, No. 429/November 2016, ISBN 978-94-6138-560-4 . Available at SSRN: https://ssrn.com/abstract=2965985

Bogdanor, V., (2016) Brexit, the Constitution and the Alternatives, King's Law Journal, 27:3, 314-322, DOI: 10.1080/09615768.2016.1253220

Booth, S., C. Howarth, M. Persson, R. Ruparel, and P. Swidlicki (2015) “What if. . .? The Consequences, Challenges and Opportunities Facing Britain outside EU.”

http://openeurope.org.uk/intelligence/britain-andthe-eu/what-if-there-were-a-brexit/

Bowman, R. G. (1983), Understanding and Conducting Event Studies. Journal of Business Finance &

Accounting, 10: 561-584.

Bruno, R., Campos, N., Estrin, S., Tian,M., (2016), Gravitating Towards Europe: An Econometric Analysis of the FDI Effects of EU Membership, BREXIT 2016 Policy analysis from the Centre for

Economic Performance, London School of Economics and Political Science pp. 134-146.

Coutts, J., & Mills, T., & Roberts, J.,(1994). The market model and the event study method: A synthesis of the econometric criticisms. International Review of Financial Analysis. 3. 149-171. 10.1016/1057-5219(94)90023-X.

Dougan, M., (2016), ‘Transcript: Professor Michael Dougan on the EU Referendum’ – News –

University of Liverpool, https://news.liverpool.ac.uk/2016/06/22/transcript-professor-michael-dougan-eu-referendum/

Dustmann, C. and Frattini, T. (2014), The Fiscal Effects of Immigration to the UK. Econ J, 124: F593-F643. doi:10.1111/ecoj.12181

Fama, E. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of

Finance, 25(2), 383-417.

Frazzini, A., Pedersen, L., (2013), Betting against beta, Journal of Financial Economics, Volume 111, Issue 1, 2014, Pages 1-25, ISSN 0304-405X.

(39)

39

Goodwin, M., & Heath, O., (2016) The 2016 Referendum, Brexit and the Left Behind: An

Aggregate-level Analysis of the Result.The Political Quarterly, 87 (3). pp. 323-332. ISSN 0032-3179.

Hobolt, S., (2016) The Brexit vote: a divided nation, a divided continent, Journal of European Public Policy,23:9, 1259-1277, DOI: 10.1080/13501763.2016.1225785.

Hughes, K., Hayward, K., (2018) Brexit, Northern Ireland and Scotland: Comparing Political Dynamics and Prospects in the Two ‘Remain’ Areas, Scottish Centre for European Relations, Policy Paper No. 6, SC559785

Kaufmann, E., (2016) It’s NOT the economy, stupid: Brexit as a story of personal values. LSE

European Politics and Policy (EUROPP) Blog (09 Jul 2016). Blog Entry.

MacKinley, A. S. (1997) Event Studies in Economics and Finance, Journal of Economic Literature, vol.35 pp. 13-39

Niederhoffer, V., (1971), The analysis of world events and stock prices, The Journal of Business, Vol. 44, No. 2, pp. 193-219, The University of Chicago Press.

Peterson, P. (1989). Event Studies: A Review of Issues and Methodology. Quarterly Journal of

Business and Economics,28(3), 36-66.

Plakandaras, V., Gupta, R., Wohar, M., (2017), The depreciation of the pound post-Brexit: Could it have been predicted?, Finance Research Letter, Vol.21, pp.206-213, 1544-6123.

Raddant, Matthias (2016) : The response of European stock markets to the Brexit, Kiel Policy Brief, No. 100

Ramiah, V., Pham, H. N. A., & 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, DOI: 10.1080/00036846.2016.1240352

Rieth, M.; Michelsen, C.; Piffer, M., (2016) : Uncertainty shock from the Brexit vote decreases investment and GDP in the euro area and Germany, DIW Economic Bulletin, ISSN 2192-7219, Vol. 6, Iss. 32/33, pp. 575-582.

Sathyanarayana, S.; Gargesha, S. (2016), The Impact of Policy Announcement on Stock Market Volatility: Evidence from Currency Demonetisation in India, IOSR Journal of Business and

Management (IOSR-JBM) e-ISSN: 2278-487X, p-ISSN: 2319-7668. Volume 19, Issue 1. Ver. VII

(Jan. 2017), PP 47-63.

Shapiro, A., Sudhof, M., & Wilson, D., (2017). Measuring News Sentiment. Federal Reserve Bank of San Francisco, Working Paper Series. 01-A2. 10.24148/wp2017-01.

Simon, H. A. (1997), Models of bounded rationality. Vol. 3, Vol. 3. Cambridge, Mass, MIT Press. Taggart, P., (1998) ‘A Touchstone of Dissent: Euroscepticism in Contemporary Western European Party Systems’, European Journal of Political Research 33: 363–88.

(40)

40

Thompson, R. (1985). Conditioning the Return-Generating Process on Firm-Specific Events: A Discussion of Event Study Methods. The Journal of Financial and Quantitative Analysis,20(2), 151-168.

Woodford Investment Management (2016) “The economic impact of ‘Brexit’.” http://woodfordfunds.com/economicimpact-brexit-report/

Online articles:

Hunter, M., (2018), FTSE 100 hits record high as pound falls to 2018 low, Financial Times, article available at: https://www.ft.com/content/7d5eb2ca-5cc9-11e8-9334-2218e7146b04

Partington, R. (2018) Brexit vote has cost each UK household £900, says MarkCarney, The Guardian, article available at: https://www.theguardian.com/politics/2018/may/22/brexit-vote-cost-uk-mark-carney-bank-of-england

(41)

41

Appendix:

Figure: Map displaying regional voting in the 2016 UK Brexit Referendum. Image Source