Brexit : how did the outcome of the British Referendum affect the financial market of UK?

UNIVERSITY OF AMSTERDAM

Faculty of Economics and Business

Bachelor Thesis in Economics & Finance

Specialization: Economics

Brexit: How did the outcome of the British Referendum affect the

financial market of UK?

Abstract:

As a result of the Brexit referendum regarding the membership of UK in the European Union, large amounts of economical and political uncertainty arose in the British economy. This research examines the effects preceding the public announcement of the “leave” vote on the abnormal returns of the financial and non-financial industry of Great Britain. In order to do conduct an empirical research, an event study as well as a time-series regression on abnormal returns is used as the main methodology for this research. The results from both methods indicate the Brexit had a significant negative effect on the returns of financial firms, while no effect was reported for the non-financial industry.

Key Words: Brexit Referendum, Abnormal Returns, Event Study, Time-Series Regression

Author: Nikiforos Kilikidis Student number: 10513876

Thesis supervisor: Ms. Egle Jakucionyte Date: 16/02/2017

1.INTRODUCTION

In January 2013, in a long awaited speech, David Cameron committed to a referendum on whether United Kingdom should remain in the European Union or exit and become an independent nation (Watson, 2013). The main goal of this referendum, which took place on the 23th of June 2016, was to give British people the ability to determine their country’s future. Although large cities such as London, Liverpool and Edinburgh supported the remain vote, the majority of the voting population that accounted to 51.9%, voted in favor of leaving the European Union. Hours after the public announcement of the so-called Brexit, there was a lot of uncertainty in the British stock market. Evidently, London’s main FTSE 100 index plunged by nearly 9%, while FTSE 250 that tracks 250 mid-sized companies, closed that day with a decline up to 7.2% (Sheffield, 2016). As Peter Spence wrote in a recent article on The Telegraph, the money that was pulled out of UK assets as a result of the Brexit fear almost accounted to 65 billion pounds, a magnitude that has not been witnessed since the wake of the financial crisis in 2009 (Reuters, 2016). In addition, the large drop of the sterling to the US dollar, which accounted to more than 10%, created a lot of instability in the financial market. As a result, uncertainty risk and financial instability became an issue of great concern for economists, governments and policymakers, since the UK economy needs large amounts of foreign direct investments to be able pay for its imports. Great Britain’s economy is one of the five largest national markets worldwide and the City of London, one of the biggest trading centers that supply the EU with significant amounts of exports regarding financial and business services (Reuters, 2016). Hence, stepping out of the European Union would imply severe effects on the domestic financial businesses. Therefore, since a similar exit has never happened before, it is very interesting to examine different views on the immediate effects of such an announcement.

As Bouoiyour and Selmi outlined in their study, the effects of the Brexit crash did not only affect the British stock market, but also had significant impacts in other global leading countries such as Germany and France (Bouoiyour & Selmi, 2016). Their study shows that the correlation of the UK stock indices with those of Germany and France are higher than any other country in the union. Therefore, bad news arising in one of these markets would significantly impact each other’s stock indices. Patently, in the light of the Brexit results, the German Stock Market, indicated by the DAX index, resulted in a one thousand point drop just a few hours after the announcement of the “leave” vote. However, Bouoiyour and Selmi highlighted that the consequences of Brexit would be more severe for Great Britain than the rest of Europe.

During the past years, there has been much discussion on the long-term effects of a possible Brexit. However there is limited research regarding the immediate implications of such a decision in the domestic market and since it is a very recent topic, there has been a lot of uncertainty regarding what is going to happen in the preceding months. Therefore, this paper will attempt to identify and analyze the immediate effects of such an announcement by focusing on the financial and non-financial industry of United Kingdom. In other words, we are interested in examining whether the event announcement increased or decreased the abnormal returns of financial and non-financial firms. This paper contributes to the existing literature by providing a comparison of the effects of such an announcement on the financial and non-financial industry of UK. In order to test this, the methodology of an event study is conducted, which attempts to identify the full impact of Brexit by calculating the abnormal returns of the sampled firms for a three-day period surrounding the announcement. Although the referendum took place on the 23rd of June 2016, the event date in this research is taken as the 24th of June, which is the actual date of the public announcement of Brexit.

Another way to estimate the effect on abnormal returns for financial and non-financial firms is to perform two time series regressions. Explanatory variables such as exchange rates, 10-year government bond rates, time-lagged variables and a Brexit dummy variable are used in order to explain variations in abnormal returns. Therefore, a second manner for testing whether the event announcement resulted in negative abnormal returns, is the estimation of the Brexit dummy parameter, for which its sign is expected to be negative. After conducting the research, the results revealed a negative value for the Brexit parameter that equals to -6.3%. In economic terms the result shows that the abnormal returns of financial firms decreased by 6.3% after the Brexit announcement of the 24th of June.

The next section compares and contrasts the existing literature, highlights exemplary studies and fulfills existing gaps in literature around the Brexit impacts on the financial market. Section 3 focuses on an analytical description of the event study as well as the methodology followed for the time series regressions. The last sections provide an analysis and discussion of the empirical results and outline the main conclusions that were observed in this research.

2.LITERATURE REVIEW

After conducting research and having read the existing literature, several studies were found that discuss the implications of new market information on equity prices. According to Brickley (1983), whether the market news is good or bad, investors would always expect that the stock prices reflect all available information in the market (Brickley, 1983). Therefore, any given economical event or announcement is expected to alter stock prices in a very short period of time. Smales (2013) reveals in his research, that on average it takes about 75 to 90 seconds for the Great Britain and about 40 seconds for the U.S to adjust all new information on equity prices (Smales, 2013). Both Brickley and Smales base their arguments on the fundamentals of the Efficient Market Hypothesis, a theory that was originally developed by Eugene Fama. In one of his studies, Fama investigated the effect that a public announcement would have on the behavior of stock prices. Such announcements included stock splits, enterprise performance, and dividends & earnings announcements (Fama, 1991). Fama argued that the market would realize any announcement, whether it is good or bad information, and would re-evaluate the future expected income from the shares of the stocks. As a result, the stock market would only be efficient only when all new information is immediately reflected in equity prices.

In addition, there are numerous studies that relate the arrival of new information with market uncertainty. In a paper from Brown, Harlow and Tinic (1988), the concept of Uncertainty Information Hypothesis was introduced. In their study they outline that even when the event conveys good or bad news, the impact on the stock prices would still be uncertain regardless of the direction of the effect (Brown, Harlow, & Tinic, 1988). Therefore, they base their theory Uncertain Information Hypothesis on the assumption that the risk of the securities would increase in a systematic trend. Some indications of validity arise from the study of French, Schwert, and Stambaugh (1987) who argue that risk and expected returns are twice as large in the post-event period compared to a non-event period. Therefore, based on the rationale of risk-averse investors, they claim that investors should anticipate a compensation for the additional risk they bear (French, Schwert, & Stambaugh, 1987). According to the Uncertain Market Hypothesis, the study of Brown, Harlow and Tinic (1988), suggests that post prices preceding an unfavorable event would be much higher than the post prices of a favorable event. However, they outline that price changes would only be significant in the short-run. Another study examined the short-run

results indicate that stock prices were largely affected immediately after the announcement of the result due to the political uncertainty on whether Canada should become an independent country or not. In addition, a similar study of Fernando, May and Megginson (2012) investigated the effect of the Bankruptcy of the Lenham Brothers on the US stock market. Evidently, after the 15th of September 2008 when the Lenham’s stock lost virtually all its value, many large companies that are included in the S&P 500 index, reported an average abnormal return decline of 5% (Fernando, May, & Megginson, 2012). Another firm-specific event occurred in 2010 when BP experienced an oil spill of almost 4.9 million barrels of oil in the sea of Mexico. Due to the large environmental damage and the unethical environmental behavior of the firm, BP reported a stock price plump of 3% a day after the announcement of the accident (Read, 2011).

In a study of Vincent & Bamiro (2013), the authors emphasize the significance of stock price fluctuations on market uncertainty. Whether the event is favorable or not, they claimed that an accurate prediction of short term as well as the long-term prices would not be feasible (Vincent & Bamiro, 2013). They argued that this is caused due to information asymmetry between the handlers of information and the clients of the market, namely the investors. Regarding the price behavior followed by an unfavorable event, Fama (1965) developed a price trend where he found that large immediate price changes would always be followed by large varying market responses. In line with Fama argument, DeBondt and Thaler (1985) found evidence that investors tend to overreact to bad news and as a consequence, they should consistently alter their future expectations about the returns (DeBondt & Thaler, 1985).

Another interesting study conducted by Grossman and Shiller (1980) shows the different factors that determine the stock prices and affect the returns of firms. In their research they also question the extent to which firm specific or a market event is the sole reason for the deviations of the actual and the expected returns (Grossman & Thaler, 1985). Their findings reveal that despite the significance an event might have, there are numerous other factors that create the abnormalities in returns. Supporting evidence to this claim can be found in the research Yurdakul and Akcoraoglu where factors such as company fundamentals, market behavior and other external factors have a significant effect on stock prices and the stock market in general (Yurdakul & Akcoraoglu, 2005).

In the following part of the literature, an attempt to analyze some of the most important factors that influence the stock prices of securities will be made. These factors will be used as a common benchmark of risk to all companies and would help in evaluating how much did the returns of the firms listed in the FTSE 100 dropped due to the outcome Brexit opposed to other additional factors that may have contributed to the fluctuations. These factors include the

domestic interest rate, the exchange rate, the inflation rate and the industrial production of the nation. Studies such as (Liu & Shrestha, 2008) and (Haritha & Uchil, 2016) investigate the effect of these macroeconomic factors on stock prices, and their results indicate a positive relationship with industrial production and negative relationship with exchange rates, inflation and interest rates.

Stock Prices and Interest Rates:

In general, there have been many equity valuation theories that relate the return of a security to other macroeconomic factors. One of the most commonly used models is the Capital Asset Pricing Model (CAPM) developed by Harry Markowitz and is mostly used for the pricing of risky assets (Sharpe, 1964). The distinguishing feature of the CAPM is that the return of the security is directly linked to the risk free rate as well as the risk premium that investors receive for bearing additional risk. In more detail, the security’s return is determined as follows

E (R

) = r

+ β (R

- r

)

Where E (Ri) is the return of the security i Rf is the risk free rate of return

E (Rm) is return of the market portfolio index

β is the asset’s return sensitivity to the market portfolio return

The risk free rate of return is an indication for the time-value of money. It represents the required rate of return of a security with no risk. Thus it is very interesting to examine the relation between the interest rate and the risk free interest rate. According to Cox, Ingersoll and Ross (1985), a given change in the domestic interest rate would directly affect the required return that investors expect from their shares (Cox, Ingersoll, & Ross, 1985). Therefore, as the government increases the interest rate, the risk-free interest rate will rise as well. Controlling for all other variables, a given increase in the risk-free rate, will increase the required rate of return and thus will decrease the stock’s price. Hence, a negative relationship between the interest rate and stock prices was observed (Cox, Ingersoll, & Ross, 1985).

Stock Prices and Inflation Rates:

Chen, Roll and Ross (1986) studied the effect of a set of macroeconomic variables on the stock market return such as inflation rates, industrial production and treasury bills. Regarding the inflation rates, literature showed a negative relationship with stock returns; however not for all the time-series of their research (Chen, Roll, & Ross, 1986). Supporting empirical evidence comes from a study of Fama (1982), were a negative correlation between the stock prices and inflation was found. However, this is not a causal relationship since it is used as a benchmark to establish a positive relationship between stock returns and real activity. Another supporting argument for the negative relation between inflation and stock prices comes from the study of Boucher (2008). He argued that an increase in the rate of inflation would lead to a higher required rate of return and therefore lower stock prices (Boucher, 2008). Hoguet also argued that any considerable increase in the given rate of inflation would create the need for economic policy and as a result, stock prices would further slide down.

Stock Prices and Exchange Rates:

Aside from the interest and the inflation rates, it is very interesting to examine whether a country’s exchange rate has a real effect on stock prices. In general, the extent to which a market performs well is significantly affected by the strength of that country’s currency, which of course is directly related to the strength of foreign currencies. Based on the literature, there are numerous views regarding the relation of exchange rates and stock prices. According to Phylaktis and Ravazzolo (1980), changes in the currency rates have real effects on the international competitiveness and trade balance of the country. However, these effects depend on the level the country is opened up for trade. In their study they mention that any increase in the exchange rate implies a depreciation of the currency since the value of the same good would be more expensive in the foreign country (Phylaktis & Ravazzolo, 1980). The country would therefore realize its cheaper production and increase its exports. As a result, domestic firms would experience increases in their competitiveness as well as their profitability. Increases in profitability results to higher stock prices. Therefore, a positive relationship between exchange rates and stock prices was concluded (Phylaktis & Ravazzolo, 1980).

On the other hand, Ajayi and Mougoue (1996) found an opposing relationship to the one found by Phylaktis & Ravazzolo (1980). In their research they examine several developed countries, such as United States, United Kingdom and The Netherlands, in order to find an

interpolar relationship between stock prices and exchange rates (Ajayi & Mougoue, 1996). The results indicate that there is a negative short-run relationship, because any given increase in the aggregate stock prices would change the inflation expectations and would imply uncertainty risk for investing activities (Ajayi & Mougoue, 1996). From the investor's’ point of view, this would act as a demotivator for investments and would decrease their willingness to hold that currency. Consequently, the currency would depreciate. However, Ajayi and Mougoue claim that this is not a causal relationship since an increase in the domestic stock prices would lead to an appreciation of the currency.

Stock Prices and Industrial Production:

In addition to the inflation rates, the article of Chen, Roll and Ross (1986) also studied the effect of industrial production on stock prices. In general, industrial production refers to the output produced by the industrial sector of an economy. According to Chen, Roll and Ross (1986) industrial output is highly correlated with national production. As a result, in a period of economic growth the industrial production increases, while during a recession it declines. Thus, it reflects similar changes to a country’s GDP. In a growing market, as the Gross Domestic Product increases, the demand for goods rises. As a consequence, the levels of profitability rise and investors find it more attractive to invest in stocks in that country, which in turn increases, the stock prices (Chen, Roll, & Ross, 1986). Thus, a positive relationship between prices and industrial production was found in the paper of Chen, Roll and Ross (1986).

Other studies examine whether this relationship holds for developing countries, too. Both the article of Islam and Habib (2016) and the study of Mustafa and Nishat (2012) show a positive linear relationship between industrial production and stock prices for India and Pakistan accordingly. However, according to Abugri (2008), that was not the case for Mexico and Argentina, since no relation between these variables was found significant (Aburgi, 2008).

3.METHODOLOGY:

This section will attempt to describe the methodology and the data used to conduct this research. The first section provides a description of the type of data and information regarding the sources that it was retrieved. The later sections describe the event study methodology as well as a cross sectional regression that was used in order to analyze the effect of the Brexit announcement on the financial and non-financial sector.

3.1DATA DESCRIPTION

In this research, 40 companies are examined for a total period of 329 days. Since the main focus of the paper lies around the public announcement of the British referendum, which became public on the 24rd of June 2016, the timeline begins on 23rd June 2015 and ends on 23rd September 2016. To examine the effect of such an announcement on the financial sector of UK, the stock prices of 20 financial and 20 non-financial entities was collected. The sample constituting the financial sector include Great Britain’s largest banks, insurance companies and fund management firms, while the sample of nonfinancial firms contains sectors such as retail, mining and consumer goods. That way, a fair comparison between these sectors will be provided.

All aforementioned companies are trading in stock market of the London Stock Exchange, and are listed in the FTSE 100 stock index, which includes 100 of the largest companies listed by market capitalization (Reuters, 2016). Therefore, it is assumed that the index is an accurate representative of the UK market, as it is by the far the most widely index used in Great Britain. As it was mentioned earlier, only 40 out of 100 companies from the FTSE 100 were selected for this research. The reason for excluding the remaining 60 firms is due to the nature of the companies’ industries, which include sectors such as pharmaceuticals, building and building materials and telecommunication services. Having included these sectors in the index of non-financial firms, would probably have led to an ambiguous result since different industries with different behaviors of returns would be aggregated to a single index, and thus results would be confusing. Therefore, it is assumed that consumer goods, retailing and the mining industry, which constitute the non-financial sector for this paper, would have a similar effect after the Brexit announcement.

In the later section, the procedure of conducting an event study will be provided, for which an estimation period is required to predict the normal performance of the sampled firms. In order to do so, a wide market portfolio index is used, which in this case is represented by the FTSE 100 index. The stock prices of all 40 firms as well as the FTSE 100 are retrieved from DataStream,

which contains historical price series and macroeconomic data for more than 170 countries and 60 markets worldwide.

One special manipulation that was performed in this research was the scaling of each sector’s stock prices. In order to aggregate the effect with respect to each sector, the stock prices of financial firms were scaled, by using the absolute cell reference command on Excel. Taking the fraction of each daily price over the first referenced price of each security did this. Therefore, the referenced value of the first sampled price would act as a constant denominator for all fractions. Subsequently, by calculating the average of these scales for each day as well as the underlying returns, a new index was created that represents the average stocks of a financial company in FTSE 100 index. The same procedure was followed for the non-financial firms, too.

3.2EVENT STUDY

The primary objective of this research is to identify and measure the effect preceding the announcement of Brexit. More specifically, we are interested in checking whether the event announcement decreased abnormal returns for the shareholders of public companies. In order to investigate this effect, the method of event studies is employed. According to Fama, Fisher, Jensen and Roll (1969), event studies are usually applied to capture the stock market reaction of a major announcement by the public or a publicly traded firm. Here, the main idea is to isolate the effect of the Brexit announcement from other general market fluctuations and subsequently calculate the change in value of the sampled securities. Appraisal of changes in equity values requires the measurement of abnormal returns, which according to MacKinlay (1997) represents the difference of the ex post actual return of a security during the event window (-1, +1) and the normal return over the same period. A three-day event window (-1, +1) was used in order to capture the complete effect as well as check if there were prior adjustments before the event announcement.

Hence, according to MacKinlay (1997), abnormal returns are calculated as follows:

AR

= R

– E (R

/X

)

Where, Rit and E (Rit / Xt) indicate the actual and predicted return of security i during the event

window (-1, +1). The variable Xt is discussed below. According to the central limit theorem, if the sample size is large enough (n>30), which is true in this case, the distribution is assumed normally distributed (Tone, 2012). Thus, under the null hypothesis that Brexit had no effect on the financial sector, the abnormal return of any given observation is:

AR

- N (0, σ

(ε

))

MacKinlay (1997) outlined that there are two common models for measuring a security's abnormal performance. The Constant Mean Return Model, where Xτ is a constant and implies that the mean return of a firm is constant through time, and the Market Return Model, which assumes that there is a linear relationship between a stock’s returns and market returns. Thus, Xτ represents the market returns. For this research, the Market Return Model is employed as it relates the returns of any security to the market returns. According to Fama (1991), based on the assumption of rational expectations and the efficient market hypothesis, the market model predicts that each firm’s return is proportional to the market return. Therefore,

R

= α

+ β

R

+ ε

E (ε

ε

Rmt represents the return of the FTSE 100 portfolio index. E (εit) and Var (εit) represent the zero

mean and the variance of disturbance term, respectively. The parameters αi and βi are estimated

through an OLS regression for T=260 trading days, which represents the estimation window. According to Fama et. al. (1969), usually 100 to 300 days are enough to predict the normal performance of firms. So that the full effect of Brexit is captured, the event window (-1, +1) includes one day before and one day after the public announcement of Brexit. Appendix 2 shows the start and end dates of the estimation period as well as the event window. By calculating the parameter estimates for the normal performance, the abnormal returns for each sector are calculated. However, since one day’s abnormal return does not provide any significant information, the cumulative abnormal returns of each sector are calculated for the three-day event

window. That way the entire effect on abnormal returns over three days is shown. Thus, the cumulative abnormal returns are calculated using the formula below:

In order to check whether the abnormal returns of financial companies were statistically different than zero, the following t-statistic is calculated:

Where σΑR is the standard deviation of the abnormal returns for each sector (financial and non

financial) and n represents the number of days in the event window (-1, +1). Thus, N=3. Given

that the t-statistic is a two-tailed test, the table below depicts the rejection regions at 1%, 5%, 10% and at a 20% level of significance.

Table 1: Rejection Regions for different alpha values

Level of Significance α (%) T-statistic (2-tailed) α=1% ±2.576 α=5% ±1.960 α=10% ±1.645 α=20% ±1.282

3.3TIME SERIES REGRESSION

Another way of identifying the impact of the Brexit announcement on both the financial and the non-financial sector is by estimating a time series regression, where Brexit takes the form of a dummy variable. In line with the event study method, the main variable that is examined here is the abnormal returns of each industry. Abnormal returns are therefore used as the dependent variable of the regression. Explanatory variables such as exchange rates, government bond rates, Brexit and seven time lagged variables are depicted in the function below:

AR

= β

+ β

AR

+ β

AR

+ … + β

AR

+ β

Brexit + β

ER + β

G.Bond

+ε

Where t = 1,2… 7

Table 2: Definition of Variables

Symbol

Variable

Definition

AR

Abnormal Returns

Difference between actual and

predicted returns

AR (t-N)

Time Lagged Variables of

AR

N = 1,2...7 time lagged variables

ER

Effective Exchange Rate

British sterling (£) against US dollar

($)

G.Bond

Government Bond

Rates of 10 year UK-Government

Bond

3.3.1DESCRIPTION OF THE VARIABLES

As it is depicted in the table 2, seven time-lagged variables are selected as explanatory terms for the time series regressions. According to Blumash and Trivoli (1991), by including lagged terms of the dependent variable, the model can predict both the current and the time-lagged values of the predictor variable. Therefore, by selecting them as explanatory variables is expected to explain some variations in abnormal returns. In this regression, the amount of time-lagged variables was chosen based on the Adjusted R-squared, which increases only when a new explanatory term improves the model. The main difference between squared and Adjusted R-squared is that the first one always improves when a variable is added even when the model is not improved (Stock & Watson, 2012). Consequently, by looking at table 3 and the Adjusted R-squared in all columns, it could be argued that choosing seven time-lagged variables would explain a greater variation of abnormal return fluctuations. The values below represent the estimated coefficients of the explanatory variables. Those marked with a star characterize the ones that are statistically significant.

Table 3: Checks for time-lagged variables based on Adjusted R-squared.

Moving onwards, exchange rates represent the daily effective spot rates of the sterling against the US dollar quoted by the Bank of England. As in was earlier mentioned in the research,

abnormal returns and exchange rates. They argued that an increase in exchange rates would increase a given firm’s competitiveness, thus its profitability and as a result its stock price. Since prices are negatively related to returns, a negative relationship was concluded. Therefore, including it as an explanatory variable is expected to capture part of the variations in abnormal returns. Although there are some opposing views from Ajayi and Mougoue (1996) regarding the sign of the correlation, it is expected that β9, the estimated coefficient of Exchange Rates, would have a positive value.

In another study, Cox, Ingersoll and Ross (1985) outlined that interest rates is a crucial determinant of stock return fluctuations and based on the Capital Asset Pricing Model, they inferred that there is a strong negative relation between interest rates and returns. However, since there was no available data for daily interest rates, the daily rates of 10-year UK-government bonds were used as a benchmark for assessing market uncertainty and risk. In general, interest rate risk affects bonds more directly than stocks as it represents the amount of risk that bondholders are bearing. Thus by including it in the model, it is expected to explain some of the variations in returns. For both the exchange rates and the government bond rates, daily data was retrieved from the official website of the Bank of England for a time period of 329 days starting on 23rd June 2015 and ending on 23rd September 2016.

The most important variable in this regression is the BrexitDummy, which is a binary variable and is used in order to assess the implications of Brexit in the financial and the non-financial sector. According to book written by Stock & Watson (2012) binary or dummy are the variables that can only take two values, 0 and 1. Since the effect of the Brexit announcement is captured throughout the three-day event window (-1, +1), the dummy variable reported values of 1 only during these three dates. Therefore, for the period before the announcement as well as the period after the event window, the BrexitDummy reported values of 0. Since such an event has never occurred before, there was no information regarding what is going to follow. As a result, due to the uncertainty, the market reacted negatively. Hence, the estimated coefficient of the BrexitDummy variable is expected to have a negative effect on abnormal returns.

4.EMPIRICAL RESULTS:

This section analyzes the main findings of this paper. The first section provides an elaborate representation of the event study results, which focus on the effect of Brexit during the event window (-1, +1). The later section provides an interpretation of the estimated coefficients of the

time series regressions in an attempt to analyze the impact of the Brexit announcement for both the financial and the non-financial companies.

4.1EVENT STUDY RESULTS

According to the main hypothesis of this paper, the Brexit announcement on the 24th of June 2016 created abnormal returns for financial firms. In order to testify this fact, an event study was conducted which focused on the three-day event window surrounding Brexit. Thus, the 23rd, the 24th and the 25th of June are the focus dates of this research. According to MacKinlay (1997), the main financial tool for analyzing effects of event announcements on stock prices is the calculation of abnormal returns. Thus, it is crucial to present the two constitutes of abnormal returns, namely the actual and predicted returns of the financial and non-financial sectors during the event window. Table 4 depicts that during the event day, the 24th of June, as well as the day after, actual returns for the financial and the non-financial firms were found to be negative.

Table 4: Average Actual Returns of Financial/Non-Financial Institutions (-1, +1) event window

Sector 06/23/2016 06/24/2016 06/25/2016

Financial Institutions 1.56% -10.64% -7.95%

Non-Financial Institutions 1.33% -3.55% -2.79%

Evidently, returns for the financial sector accounted to -10.64% and -7.95% respectively for each date, while the non-financial sector experienced negative actual returns accounting for -3.55% during the day of Brexit and -2.79% the day after. This indicates that the actual performance of both sectors was negatively affected by the event announcement, especially the one of financial intermediaries. However, the positive returns one day before the event show that there were no significant adjustments of stock prices prior to the event and thus no leakage of information regarding the outcome of the referendum seems to have occurred. As it was previously mentioned, conducting an event study requires an estimation period for measuring the normal performance of the sampled firms. Table 5 shows the predicted returns for the financial and the non-financial industry.

Having data for both the actual and the predicted performance, abnormal returns were calculated during the event window. As it is depicted in table 6, the financial sector reported negative values for all three dates, with the 24th of June resulting in the highest abnormal return

equal to -7.27%. The negative value of -5.22% one day after the Brexit announcement shows the market instability resulting after the “leave” vote of Great Britain. The results from Table 6 indicate that the negative abnormal returns of financial firms during the three day event window were found statistically significant at a 5%, 10% and 20% level, but not at a 1% level. Therefore, there is enough statistical evidence to infer that the Brexit announcement created negative abnormal returns for firms constituting the financial sector, a result that is in line with the hypothesis of this paper.

Table 6: Average Abnormal Returns of Financial/Non-Financial Institutions, (-1, +1) event window Sector 06/23/2016 06/24/2016 06/25/2016 Financial Institutions -0.24% -7.27% -5.22% (0.028) (0.028) (0.028) Non-Financial Institutions 0.14% -0.40% -0.23% (0.003) (0.003) (0.003)

*Values in parentheses represent the standard errors of the estimates

On the other hand, that was not the case for the non-financial firms. Although negative abnormal returns were reported for the day of the event and the day after, their magnitude seem to be significantly lower than that of the financials. Evidently, during the event date the non-financial sector reported values of -0.40%, while during the same day, non-financial firms resulted in a negative 7.27%. As to the test outcome, Table 7 shows that there is not enough evidence to conclude that the abnormal returns of the nonfinancial sector were significantly affected by the Brexit announcement. The absolute t-value of 1.038 was therefore not adequate to infer conclusions at any significance level. Due to the nature of the sample, which includes industries such as consumer goods, mining and retailers, it was initially expected that Brexit would have a negative effect on their stock prices. According to Bouoiyour & Selmi (2016), a possible Brexit would be very harmful for retailers and consumer related firms, as leaving the EU would lower trade between the UK and lose the tariff-free advantages of its membership. As a result cost of EU raw materials would increase and as a result prices would increase, too. A possible pitfall in accurately measuring the abnormal returns of the non-financial sector might be the inclusion of the mining industry together next to the consumer-goods industry. Tyler Broda, director of global mining research at RBC Capital Markets outlined that the weak value of sterling has helped the mining industry since the majority of their revenues are in US dollars. As a result mining reported growth after the Brexit announcement (Tyler, 2016). Since a negative performance was expected

from sectors such as consumer goods and retail, the positive returns from the mining industry led to an insignificant test statistic and no conclusions could be inferred. .

4.2TIME-SERIES REGRESSION RESULTS

Followed by the event study, the results of the time-series regression are presented in Tables 9, 10, 11 and 12. As it was mentioned in the previous section, the aim of this regression was to estimate the parameter of the Brexit dummy variable in order to identify and analyze the effect of the event announcement on each sector. Since abnormal returns, which represent the dependent variable, have different values for the financial and the non-financial sector, two time-series regressions were performed.

The summary statistics from the ordinarily least squares regressions are depicted in tables 9 and 10. However, the seven time-lagged explanatory variables are not included for convenience, since their estimates are not of our interest. Table 9 shows that the minimum abnormal return for financial firms was -7.27% which most likely occurred during the day of Brexit and thus it value is in line with estimated results of the event study that reported the same exact value. However, that was not the case for the abnormal performance of non-financial firms since the minimum estimated value was -1.68%. By comparing this value to the abnormal return from the event study on the day of Brexit (-0.40%), it could be argued that the lowest value of abnormal returns was not reported during Brexit, but on some other trading day within the examined period. Moving onwards, tables 11 and 12 present the estimated coefficients of the OLS regressions. The only difference between the two models is that the first one includes the abnormal returns of financial firms as dependent variable, while the second one the abnormal returns of the non-financial sector. By looking at the Adjusted R-squared for both regressions, the results indicate that the model for the financial firms explains a greater variation in abnormal returns (44.38%) when compared to the financial sector (0.99%). This is because the abnormal returns of non-financial firms were not significantly affected by the Brexit announcement, as indicated by the event study, and thus little variations occurred compared to the financial sector.

Table 11: Time Series Regression Estimates for Abnormal Returns of Financial Firms

Regarding the Brexit dummy variable, the results from table 11 show a statistical significant relationship between Brexit and abnormal returns, with the estimated parameter accounting for -0.043. Therefore, the model indicates that given a Brexit announcement, abnormal returns are expected to decline by 6.32%, a value that is somewhat close to the estimated result of the event study (7.27%). In the model for the financial firms, Brexit was the only variable that was statistically significant at 5%, while the variable of the 10-year government bond rates was only significant at 10% and 20% level. However, the negative sign of the estimated parameters was

found in line with the literature of Harry Markowitz who provided evidence of a negative relationship between interest rates and returns. The reason for the non-significance of the exchange rates could be blamed on the inappropriate selection of the effective exchange rates. Selecting the exchange rate of the sterling against the euro would probably have created significant results.

Table 12 shows the OLS estimates for the model of the non-financial firms. As it depicted in the table, none of the explanatory variables of interest are statistically significant. This is because little variations occurred in the abnormal returns of the non-financial sectors and therefore, the chosen explanatory variables failed to explain the model accurately. The Adjusted R-squared that amounted to less than one percent also indicates the poor performance of the model.

5.1ROBUSTNESS CHECK:EVENT WINDOW (-2,+2)

This section attempts to provide a robustness check to the event study, by analyzing the impact of the Brexit announcement for a larger event window. Therefore, we are interested in comparing the results of a three-day event window (-1, +1) to a five-day event window (-2, +2).

Joyce et. al (2011) investigated the effect of Quantitative Easing on stock prices of UK banks by conducting an event study. In their research they suggested that widening the event window would result in lower abnormal returns (Joyce et. al, 2011). Therefore, by increasing the event window to five days we expect that abnormal returns resulting from Brexit would become less significant. Table 8 shows the test outcome of abnormal returns with (-2, +2), while table 6 presents the results of the event study with (-1, +1). Originally, by performing the event study with (-1, +1), we found that abnormal returns were negatively affected by the Brexit announcement at a 5%, 10% and 20% significance level. However, after extending the event window to (-2, +2), the effect on abnormal returns was insignificant at all levels. As to the non-financial firms, increasing the event window resulted in an even lower test statistic, and the effect on abnormal returns was still insignificant. Therefore, the overall effect of increasing the event window is a decline in the estimated abnormal returns, a result that is in line with the proposition of Joyce et. al (2011).

5.2ROBUST TIME-SERIES REGRESSION:

Followed by the robustness check for the event study, this section provides evidence of structural validity of the time-series regression for financial firms. As Stock and Watson outline in their textbook (2012), an ordinarily least squares regression assumes that the variance of the disturbance term should be a constant value and that it should not be correlated to any of the explanatory variables (Stock & Watson, 2012). However, that is not always the case. In order to check whether this assumption is violated, the robust standard errors are used. According to Lu and White (2014), using a robust time-series regression helps in examining how the main explanatory variables behave when the model modification is changed (Lu & White, 2014). They argue that even if the homogeneity of the variance were violated, the robust method would produce unbiased and consistent with the population estimates. Table 13 shows the estimated of the original regression without the robustness check as well as the regression with the robust standard errors.

Table 13: Robustness check for time-series regression

By comparing the pre-robust regression with the post-robust regression estimates, one can see that the estimated parameters are exactly the same in both cases. The only difference reported is that standard errors under the robustness check are higher than the non-robust standard errors, which according to Lu and White (2014) was expected to happen. Therefore, inferring that the estimated parameters share exactly the same values and that their significance

is not affected, it could be argued that the structural specifications of this model are valid and results are unbiased and consistent with the population parameters.

6.1CONCLUSION

The aim of this paper is to identify and analyze the implications of the Brexit announcement on the financial and the non-financial sector of Great Britain. In order to investigate this effect, two different methods were employed that examined 20 financial and 20 non-financial companies for a total period of 329 trading days.

The event study attempted to measure the effect of such an announcement on the abnormal returns of the sampled firms. The results from the event study provide evidence that the Brexit announcement on the 24th of June 2016 had a significant negative effect on the abnormal returns of financial firms. However, that was not the case for the non-financial sector since an insignificant test-statistic value was reported. Failing to identify a significant negative effect on the returns of the non-financial sector most likely resulted by including the mining industry in the sample, which according to Tyler Broda, experienced large increases in revenues during the days surrounding Brexit. However, as Bouoiyour & Selmi (2016) outlined in a recent article, the industry of consumer goods and retail would be severely affected by Brexit, due to the higher costs of trading (Bouoiyour & Selmi, 2016). Therefore, by including various opposing effects in the same index, the results failed to infer any significant conclusion.

The second method attempted to interpret the effect of Brexit dummy variable on returns through a time series regression. By selecting abnormal returns as the dependent variable and including explanatory terms such as the 10 year UK Government Bond rates and the effective exchange rates, the parameter of the Brexit dummy was estimated through two ordinary least square regressions, one for each sector. For the financial industry, the results show that both Brexit and the 10-year UK government bond rates had a significant negative effect on abnormal returns, however no relationship with exchange rates was reported. On the other hand, the results for the non-financial industry do not indicate any significant relationships, which again could be blamed on the inclusion of the mining industry in the sample.

By looking at the results of both methods one can observe that the value of the Brexit effect on abnormal returns was very similar in the event study and the time series regression. Evidently, a reduction of 7.27% was reported by the event study, while a negative 6 % was estimated through the OLS regression. Therefore, after employing two different methods to

led to approximately a 7% decline in the abnormal returns of financial firms. Therefore, due to the nature of the announcement, the uncertainty regarding what is going to happen after Brexit led to abnormal return losses in the financial sector. Negative abnormal performance implied that financial stocks performed worse than they were expected and thus the inability to accurately predict future earnings brought significant uncertainty in the stock market. As a result, investing in riskier assets would require higher risk premiums for investors to bear additional risk according to Fama (1982).

6.2LIMITATIONS AND DISCUSSION

In general this essay attempted to provide a clear analysis of the implications of Brexit on the financial and non-financial industries in the UK. However, some shortcomings were realized after conducting the research. Firstly, as it was earlier mentioned, despite the overall negative performance of the market, the non-financial sector appeared to have a non-significant effect after the event announcement. One reason for this unexpected outcome is the inclusion of the mining industry in the sample. It seems that due to the positive performance of this industry during the Brexit period, overall abnormal returns of the non-financial sector did not fluctuate much and thus insignificant t-statistic was reported. Therefore, by leaving mining companies out of the nonfinancial sample of firms, or including them in a separate industry, the results would most likely meet the original expectations. Another possible reason for not getting similar effects in both sectors is that the Brexit announcement had more severe implications on the banking sector. As it was earlier mentioned, the financial industry of the UK is of the largest business exporters of the European Union. Leaving the union would therefore imply that these banks could no longer run their operations in Europe. Consequently, the uncertainty was higher and as a result greater effects were reported for the financial industry.

Another shortcoming of the research is the selection of explanatory variables that determine abnormal returns. Despite the evidence arising from the literature, exchange rates did not report any significant effect on the abnormal performance. One possible reason could be the nature of the data. As it was earlier mentioned, daily effective exchange rates of sterling against the dollar were quoted. Therefore, it might be the case that the exchange rates variable would explain a greater variation of abnormal returns if the rates of sterling against the euro were quoted instead of the US dollar. That way it would be more relevant for the existing research, since Great Britain (sterling) and the European Union (euro) are the main counterparts of this discussion and their currencies experienced large fluctuations during the examined period.

7.BIBLIOGRAPHY

Abugri, B. A. (2008). Empirical relationship between macroeconomic volatility and stock returns: Evidence from Latin American markets. International Review of Financial Analysis, 17(2),

396-410. doi:10.1016/j.irfa.2006.09.002

Ajayi, R. A., & Mougouė, M. (1996). On The Dynamic Relation Between Stock Prices And Exchange Rates. Journal of Financial Research, 19(2), 193-207. doi:10.1111/j.1475-6803.1996.tb00593.x Beaulieu, M., Essaddam, N., & Cosset, J. (2006). Political Uncertainty and Stock Market Returns:

Evidence from the 1995 Quebec Referendum. SSRN Electronic Journal. doi:10.2139/ssrn.841344 Bondt, W. F., & Thaler, R. (1985). Does the Stock Market Overreact? The Journal of Finance, 40(3),

793. doi:10.2307/2327804

Boucher, C. (2008). Stock Prices, Inflation and Stock Returns Predictability. SSRN Electronic Journal. doi:10.2139/ssrn.650171

Bouoiyour, J., & Selmi, R. (2016). Is uncertainty over Brexit damaging the UK and European equities?

Working Paper,CATT, University of Pau.

Brickley, J. A. (1983). Shareholder wealth, information signaling and the specially designated dividend. Journal of Financial Economics, 12(2), 187-209. doi:10.1016/0304-405x(83)90035-1 Brown, K. C., Harlow, W., & Tinic, S. M. (1988). Risk aversion, uncertain information, and market

efficiency. Journal of Financial Economics, 22(2), 355-385. doi:10.1016/0304-405x(88)90075-x Bulmash, S. B., & Trivoli, G. W. (1991). Time–lagged interactions between stocks prices and selected

economic variables. The Journal of Portfolio Management, 17(4), 61-67. doi:10.3905/jpm.1991.409351

Chen, N., Roll, R., & Ross, S. A. (1986). Economic Forces and the Stock Market. The Journal of

Business,59(3), 383. doi:10.1086/296344

Cox, J. C., Ingersoll, J. E., & Ross, S. A. (1985). A Theory of the Term Structure of Interest

Rates. Econometrica, 53(2), 385. doi:10.2307/1911242

Fama, E. F. (1965). The Behavior of Stock-Market Prices. The Journal of Business, 38(1), 34. doi:10.1086/294743

Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. W. (1969). The Adjustment of Stock Prices to New Information. SSRN Electronic Journal. doi:10.2139/ssrn.321524

Fama, E.F.(1982). Inflation, Output, and Money. The Journal of Business, 55(2), 201. doi:10.1086/296161

Fama, E.F.(1991). Efficient Capital Markets: II. The Journal of Finance, 46(5), 1575. doi:10.2307/2328565

Fernando, C. S., May, A. D., & Megginson, W. L. (2012). The Value of Investment Banking Relationships: Evidence from the Collapse of Lehman Brothers. SSRN Electronic Journal. doi:10.2139/ssrn.1572780

French, K. R., Schwert, G., & Stambaugh, R. F. (1987). Expected stock returns and volatility. Journal

of Financial Economics, 19(1), 3-29. doi:10.1016/0304-405x(87)90026-2

Grossman, S., & Shiller, R. (1980). The Determinants of the Variability of Stock Market Prices. doi:10.3386/w0564

Haritha, P., & Uchil, R. (2016). Conceptual Framework on Market Factors Affecting Investor’s Sentiments and the Effect of Behavioral Pitfalls on Investment Decision Making. IOSR Journal of

Economics and Finance, 29-34. http://www.iosrjournals.org/iosr-jef/papers/SIFICO/Version-1/5.

29-34.pdf

Islam, K. U., & Habib, M. (2016). Do Macroeconomic Variables Impact the Indian Stock Market? Journal of Commerce and Accounting Research, 5(3). doi:10.21863/jcar/2016.5.3.031 Joyce, M., Lasaosa, A., Stevens, I., & Tong, M. (2011). The financial market impact of quantitative

easing in the United Kingdom. International Journal of Central Banking, 7(3),113-161.

Liu, M., & Shrestha, K. M. (2008). Analysis of the long‐ term relationship between macro‐ economic variables and the Chinese stock market using heteroscedastic cointegration. Managerial

Finance, 34(11), 744-755. doi:10.1108/03074350810900479

Mustafa, K., & Nishat, M. (2012). The arrival of public information and asset price behavior in emerging markets: evidence from stock market in Pakistan. Savings and Development, 36(1), 1. doi:10.2307/savideve.36.1.1

Phylaktis, K., & Ravazzolo, F. (1980). Stock Prices and Exchange Rate Dynamics. SSRN Electronic

Journal. doi:10.2139/ssrn.251296

Read, C. (2011). The Market Response. BP and the Macondo Spill, 202-206.

Reuters, P. S. (2016). Investors pulled £65bn cash out of the UK in March and April as Brexit fears

grow. Retrieved January 15, 2017, from

http://www.telegraph.co.uk/business/2016/06/07/investors-pull-cash-out-of-uk-assets-at-fastest-pace-since-finan/

Sharpe, W. F. (1964). Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. The Journal of Finance, 19(3), 425. doi:10.2307/2977928

Sheffield, H. (2016). FTSE 100 sees £120bn wiped off its value in worst day of losses since financial crisis. Retrieved January 15, 2017, from http://www.independent.co.uk/news/business/news/ftse-100-eu-referendum-brexit-sterling-pound-bank-housebuilders-a7099951.html

Smales, L. A. (2013). Impact of Macroeconomic Announcements on Interest Rate Futures: High-Frequency Evidence from Australia. SSRN Electronic Journal. doi:10.2139/ssrn.1975205

Stock, J. H., & Watson, M. W. (2012). Introduction to econometrics: global edition. Boston: Pearson. Tone, C. (2012). A functional central limit theorem for empirical processes under a strong mixing

condition. Statistical Inference for Stochastic Processes, 15(2), 177-192. doi:10.1007/s11203-012-9069-3

Tyler, B. (2016, August 16). Mining sector outlook improves despite losses and Brexit. Retrieved February 15, 2017, from http://www.bbc.com/news/business-37095581

Vincent, O. & Bamiro K. (2013). Fluctuations in Stock Market Prices: What went wrong, its Implications to Nigerian Economy? International Journal of Computer Applications, 63(19), 13-20. doi:10.5120/10573-5612

Watson, I. (2013, January 23). David Cameron promises in/out referendum on EU. Retrieved January 15, 2017, from http://www.bbc.com/news/uk-politics-21148282

Yurdakul, F., & Akcoraoglu, A. (2005). Stock Returns, Macroeconomic Factors And Structural Breaks: An Application For The Turkish Stock Market. Ekonomik Yaklasim, 16(55), 17. doi:10.5455/ey.10518

8.APPENDICES

Appendix A - List of Financial and Non-Financial Institutions

Table 4: Average Actual Returns of Financial/Non-Financial Institutions (-1, +1) event window

Sector 06/23/2016 06/24/2016 06/25/2016

Financial Institutions 1.56% -10.64% -7.95%

Non-Financial Institutions 1.33% -3.55% -2.79%

Table 5: Average Predicted Returns of Financial/Non-Financial Institutions (-1, +1) event window Sector 06/23/2016 06/24/2016 06/25/2016

Financial Institutions 1.32% -3.37% -2.73%

Non-Financial Institutions 1.19% -3.15% -2.56%

Table 6: Average Abnormal Returns of Financial/Non-Financial Institutions, (-1, +1) event window Sector 06/23/2016 06/24/2016 06/25/2016 Financial Institutions -0.24% -7.27% -5.22% (0.028) (0.028) (0.028) Non-Financial Institutions 0.14% -0.40% -0.23% (0.003) (0.003) (0.003)

Table 8: Hypothesis testing for Abnormal Returns during Event Window (-2, +2)

Table 9: Summary Statistics for Abnormal Returns of Financial Firms including indicators