Influence of the Brexit Announcement on the Firm
Value of Companies Exporting to the UK
Master Thesis Finance
Economics and Business Administration of University of Groningen
Date: 15 February 2018 Student: Berend Zondervan Student number: s2208075
2
Abstract
In this study, we make use of a traditional event-study in an attempt to identify the impact on firm value of exporting companies to the UK following the announcement of the “Brexit”. Therefore we have compared and aggregated the actual and expected stock returns of EU and non-EU firms, which gave us the cumulative abnormal return (CAR).
3
Acknowledgements
This thesis has been written in the context of the master Finance assignment of the Faculty of Economics and Business Administration University of Groningen. I have written this thesis with much pleasure, of course there where difficult times in the long days that I worked on this, however it has been a rather interesting project. I have gained a lot of knowledge in a completely new area of expertise and I learned from the writing process itself. Therefore my thanks goes out to my supervisor Dr. A.S.R de Ridder, for the confidence and motivating discussions that have kept me sharp and critical and his help with the questions I had about various aspects.
Collecting, ordering and comparing the data with the programs DataStream and Excel was probably the most time consuming and difficult part of the project. I did not had much knowledge and working experience with the programs, especially with collecting data from DataStream and converting it to Excel. Therefore, it was quite hard to do the project in the limited time I had given myself, however, with the help of my supervisor, I was able to complete the project. As a closing note, I want to thank my friends and family for their commitment and support.
5
1 Introduction
The UK is one of the longest members of the European Union as it became a member in 1973. However, on 23 June 2016 a referendum in the United Kingdom (UK) was held to decide if the UK should stay or leave the European Union (EU) in 2019. Although opinion polls suggested that the outcome of the referendum was uncertain, it still was a big surprise as the results were to leave by 51.9% to 48.1%. This resulted in the name Brexit, which comes from the words Britain exiting the European Union. Before the Brexit is definite, the UK stays as an active EU member, therefore the companies in the EU who are trading with the UK and the UK companies can trade their goods freely to customers without causing those customers having to pay extra additional taxes to imported goods. As long as UK is a member of the EU it will give consumers and companies in the UK no restrictions on the import from anywhere in the EU, because there are no tariffs on imports. Hence, the name “free trade”. All the EU members have these free trade agreements not only within the EU but also with the non-EU countries: Norway, Switzerland and Turkey.
So why is the UK leaving the EU? One of the most important arguments is that the UK gets her sovereignty back together with her self-esteem. Moving outside the European Union will create more self-control as they can determine their closing trade agreements together with their own legislation. When the UK is no longer in the EU it will have as a non-member, to make new agreements with members of the European Union and the non-EU countries. If the UK cannot agree on those free trade agreements, it may cause tariffs on goods and services that is imported and exported into and from the UK to rise.
6
Therefore, it is interesting to identity the value effects of the Brexit announcement by analyzing the impact of the Brexit announcement on a sample of firms that are exporting to the UK and which have substantial economic relationships with the UK. In this paper, a classical event-study is used in an attempt to identify the impact on firm value following the announcement of the Brexit.
Important key papers in this study are about the general approach of an event study that is used as a benchmark, which are the articles from MacKinley (1997) and Campbell et al. (1997).
Therefore, we obtained data around the date of the referendum from the largest companies that have the highest impact on the import to the UK from both non-EU and EU trade partners of the UK. As an extension to the event study, we run a regression by using different independent control variables from the end of 2015 to test the effects of these variables on securities. We make a distinction between countries that are a member of the EU and non-EU. By doing this we can identify if there are significant different reactions for these countries and their companies following the announcement.
We find a positive and a negative value effect following the announcement for European and non-European firms respectively. The regression results indicate a negative relationship between the accumulative abnormal returns after the announcement and the firm size measured by the firms’ market capitalization and sales. However, we find a positive relationship between the firms’ market capitalization and the abnormal returns at the day after the announcement.
7
2 Literature Review
The methodology that is used throughout this paper is mainly based on the classical method of an event study, which is described by MacKinlay (1997), and Campbell et al. (1997). According to these authors, the advantage of conducting such a study is due to the fact that if a market is efficient, the effect of an event is reflected immediately in the stock prices. In other words, we will see immediately reactions from investors if such an event or news would be made public. Hence, a classical event study allows investigating the effect of an unexpected event on the firm value. The articles are based on the assumption of the semi strong efficient market hypothesis, which was introduced by Fama (1970). He said that an efficient market is a market where the prices always contain and reflect all available information.
In the years after the publication of MacKinlay a number of articles, using his methodology of the event study, where published. One of these was from Lim, Brooks and Kim (2008). They examined the unexpected financial crisis from 1997 by focusing on the Asian stock markets changes for three periods: a) pre-crisis, b) crisis and c) post crisis.
There are already several studies on the influence of economic variables on stock markets in various markets. Mohammad (2011) mapped out the impact of micro and macroeconomic variables on the Dhaka stock Market. Brunie et al. (1972) discovered a positive effect between money supply and stock prices. Boulanger and Philippidis (2015) examined what the estimated impact would be on the budget of both the UK and EU if the UK would leave the EU. The results showed a trade-off between the negative aspects of a smaller market access but in exchange, the UK would benefit from greater freedom. The UK, as a member of the EU has to meet on more regulatory and obligations then a non-member. Overall Boulanger and Philippidis (2015) concluded that UK would be better off as an active member of the EU.
Petersen (2015) expected that if Britain votes for a “Brexit” it would lower the export and
8
also made by Dhingra and Ottaviano (2016) who are saying in their paper that the effects on the economy will, for the majority, be dependent on which trade agreement are made after the Brexit. They have strong believes that due to different trade policies together with the diminishing integration with the EU will result in a negative impact on trades and therefore, the gains from the Brexit will be lower than the cost for the UK.
This prediction is also given by Kierzenkowski et al. (2016) were they found significant results about a large negative shock that would hit the UK economy in the first years after the Brexit, which will be transferred to other European countries. By the assumption of fixed interest rate of zero, the UK GDP growth would decrease by 0.5 percent in both 2017 and 2018 and GDP growth decrease of 1.5 percent in 2019. As in 2020, the GDP would be 3 percent lower than it would be without a Brexit. By 2023, as the economy is adjusting, the real GDP of the UK will increase with 0.5 percent, but still be 2.5 % lower than it would be otherwise.
Armstrong (2017) addresses one of the most important issues that the UK and EU faces after the Brexit, which is the negotiation on new trade deals. Both the EU and the UK are members of the World Trade Organization (WTO). Even after the Brexit is definite it still remains of a member of the World Trade Organization and therefore the UK has obligations towards the WTO and it faces consequences if the trade agreements between members are adjusted. Armstrong points out that there are some common agreements when being a member of the EU. The tariff rate quotas and tariff schedules that the UK has with other countries from outside the EU are common for all members of the EU. However if UK leaves the European Union it will have to implement its own tariff rate quotas and schedules between them and all members of the WTO. If the UK leaves the European Union without these agreements with the EU it will have negative consequences for both the EU and UK.
9
However, the literature has also some published articles who reject the predictions about the negative impact of the Brexit. Keen (2017) believes that economical predictions are too narrowly based on “the reversal of the gains from specialization”. In other words, within the European Union we have a free trade policy and due to this policy countries will be able to increase their specialization and therefore increase their output due to higher efficiency. Therefore decreasing the free trade, the counterpart should be true, which will happen if the UK leaves the EU. Steve Keen examined this hypothesis by using measurements of ubiquity and diversity, which according to the theory from “the reversal of the gains from specialization”, high-developed countries should score low on both measurements. However, Keen found that this assumption does not hold for a large sample of developed countries.
Beck (2016) and Simionescu (2016) addresses other potential issues for the UK if it leaves the EU. They both examine the potential impact of the Brexit on the Foreign Direct Investments (FDI). Were they asses the gain from inward investments and the relationship of the FDI in the UK with the European Union. The FDI in the UK, of which 80 percent from FDI sales belongs to EU accounts, is an important tool for the economy as it decreases interest rates, increases wages and productivity. Beck estimates a worst prediction of a decrease of 7 percent in the UK’s stock of FDI, which would have potentially effect of ½ percent decrease in Gross Domestic Products. Simionescu (2016) used panel data from 2012 until 2015 for different regions across the globe. Where she found out that Brexit has a significant negative impact on the unemployment rate created through FDI.
10
3 Methodology
I. Research Problem
The first focus of this thesis will be to examine which companies has the highest import influence on the UK. For time limitation reasons we will use data from a sample of 10 countries from Europe containing 60 companies in total that are trading with to UK. The data contains stock data from 7 months before and 5 months after the announcement of the Brexit. The second step will be to identify which companies of each of those countries has the largest share on the UK import. We extent our study by collecting economic control variables that can have an influence on firm value.
In the next phase, we identify if there is a value effect of the announcement by implementing the methodology from an event study for all the firms of the EU and non-EU countries over a one-year timeframe. The last step will be to combine the control variables with the value effect. By doing so we are able to identify the effects of implemented economic variables on the firm value around the Brexit announcement.
With this information, we formulate the following research question:
“What is the Influence of the Brexit announcement on the firm value of companies that are exporting to the UK”.
In addition to the main research question, the project will be extended with the following sub- questions: “What is the Influence of the Brexit announcement on EU firms and non-EU firms?” and “What is the Influence of the Brexit announcement on small firms and large firms?”
11
II. Project Description
The major focus of the project is concentrated on investigating the effects of the Influence of the “Brexit” announcement on the UK import market. Therefore, we apply a classical event-study in an attempt to identify the impact on firm value following the announcement of the Brexit. We do this by using stock data over a 12-month window.
The stock data are collected from companies that are exporting to the UK. We make a distinction between countries from the EU and non-EU. By doing this we can identify if there are significant different reactions for these countries and their companies following the announcement. We use multiple periods around the announcement to study if there is indeed a value effect. The announcement is considered as day 0, and by using only trading days this gives us an time period from [-159…-1,0,1…109]. Using the estimation window from [-159,-23] we are able to calculate the expected returns for the event window [-22, 22] in absence of the event. The timeline sequence is illustrated in Figure 1. The period before the announcement (7 months) is used to identify if the market, in some extent, is already gradually taking the announcement in considering and therefore reacting to the outcome of the event. While the time window of 109 trading days (5 months) after the announcement is used to examine the market stability. The time window of one month before and after, -22 days and +22 days respectively, are especially taken to calculate the effect of the announcement on the stock returns and therefore on the firm value.
Figure 1
Timeline of the event window
12
III. Model Specification
In this study, we make use of an event-study in an attempt to identify the impact on firm value following the announcement of the Brexit. Based on the findings of Fama (1970), the advantage of conducting such a study is that in an efficient market, revelation of an event or news is reflected directly in the stock prices. Meaning, we will see direct reactions from investors if such an event or news would be made public. The methodology behind event studies is that the impact of an event is captured by the abnormal returns. Hence, an event study allows us to investigate and identify the effect of an unexpected event on the firm value. Therefore, we have to find and compare the actual and expected returns. In addition, we extend our event study by making use of regression analysis to identify the influence of economic control variables on firm value. We conduct our research under the assumption of normal distribution.
To apply the event study we need to find the average abnormal returns (A͞R). This could be done by obtaining the actual stock and market returns as well as calculating the expected returns. The expected returns can be calculated with various models such as the Capital Asset Pricing Model (CAPM) described by Fama (1970) or the market model. The market model is most used by conducting an event study, because it is a better fit than the constant mean return model, as it deletes the variation in the market return. Hence, for simplicity reasons we apply the market model in order to calculate and find the average abnormal returns.
Therefore we first need to obtain the returns of the stock market which is calculated by the following model:
𝑅𝑖,𝑡= ln( 𝑝 𝑝𝑖,𝑡
𝑖,𝑡−1), (1)
where 𝑅i,t is the period-t return of security 𝑖, pi,t is the price of security 𝑖 on period 𝑡, and pi,t-1
is the index value of security 𝑖 on period 𝑡 − 1.
13
market model can be explained by the following model:
𝑅it = 𝛼̂i + β̂i𝑅mt + 𝜀it , (2)
where, 𝑅it is the actual return for security i in period t, 𝑅mt is the period t returns on the
market portfolio, 𝛼̂i and β̂i are the parameters of the given market model, and, finally, εit is
the zero-mean disturbance term.
Given the parameters estimates of 𝛼i and 𝛽i,, one can calculate the abnormal returns with the
following model:
𝐴𝑅it = 𝑅it - 𝛼̂i – β̂i 𝑅mt , (3)
where ARit is the abnormal return of security i, Rit is the actual return of security i in period t, 𝛼̂i, β̂i are the parameters of the market model, where 𝛼̂i represents the intercept of the
market index i and β̂i the slope of the market index i and Rmt is the market return.
After the abnormal returns from model 3 are obtained, we need to change the abnormal return (AR) to the average abnormal returns (A͞R) in order to test our null-hypothesis. This is calculated by dividing sum of the abnormal returns by the number of events N, which is represented with the following model:
A͞Rt = 𝑁1∑𝑁𝑖=1𝐴𝑅𝑖,𝑡 , (4)
where A͞Rt is the average abnormal return on date 𝑡, N is the sample size and 𝐴𝑅i,t is the
abnormal return for security 𝑖 on period 𝑡.
14
𝐶𝐴𝑅𝑖(𝑡1,𝑡2) = ∑𝑡2𝑡=𝑡1𝐴𝑅𝑖𝑡 , (5)
Where 𝐶𝐴𝑅(𝑡1,𝑡2) is the sum of the included abnormal returns for security i from period 𝑡1 to
t2 and 𝐴𝑅𝑖,𝑡 is the abnormal return of security I at period t.
The average abnormal returns from (4) can be aggregated over the event window using the same approach as in (5) for each security i. For any interval in the event window
C͞A͞R(𝑡1,𝑡2) = ∑𝑡2 A͞R𝑡
𝑡=𝑡1 , (6)
where C͞A͞R(t1,t2) is the cumulative average abnormal return on period 𝑡1 to t2 and A͞Rt is the
average abnormal return on period t.
After the (cumulative) average abnormal returns are obtained, a parametric t-test can be implemented to test our hypothesis. Which can be calculated with the following expression:
Θ1
=
C͞A͞R(𝑡1,𝑡2)
σC͞A͞Rt/√n
,
(7)where C͞A͞R(t1,t2) is the cumulative average abnormal return on period 𝑡1 to t2, 𝜎𝐶𝐴𝑅𝑡 is the
standard deviation of the abnormal returns at period t and n is the sample size.
15
By combining the eight independent control variables with the calculated cumulative average abnormal return (C͞A͞R) of model 6, we are able to identify the effects of the independent variables on the firm value around the Brexit announcement. Hence, our baseline regression is as follows:
CAR i,t = 𝛼 + β1 MCAP + β2 M/B + β3 ROA + β4 D/E + β5 C/A + β6 ASSETS + β7 SALES + β8 (Dummy
variable) + Ɛit, (8)
where CARi,t is the cumulative average abnormal return, β1 till β8 are the beta’s used for the
control variables, β8 (D) represents the dummy variable which takes the value of 1 in case it is
a European Union firm and takes the value 0 otherwise and Ɛit is the zero mean disturbance
term. We use (8) to formulate three different models to find a relationship between firm size and the cumulative abnormal returns by using 3 different measurements for firm size: market capitalization (9), assets (10) and sales (11) for security i in period t. Hence, our regression models are as follows:
model 1
CAR i,t = 𝛼 + β1 MCAP + β2 M/B + β3 ROA + β4 D/E + β5 C/A + β8(D) + Ɛit, (9)
model 2
CAR i,t = 𝛼 + β2 M/B + β3 ROA + β4 D/E + β5 C/A + β6 ASSETS+ β8(D) + Ɛit, (10)
model 3
16
IV. Hypothesis
As long as UK is a member of the European Union it will have no restrictions on trades from anywhere in the EU, because there are no tariffs on the trades. This is called “free trade”. All the EU members have these free trade agreements not only within the EU but also outside the European Union.
When the UK is no longer in the European Union it will have, as a non-member, to make new agreements within the EU. If the UK cannot agree on those free trade agreements, it will impose tariffs on goods and services that are imported and exported into and from the UK respectively.
With this information, we obtain the following prediction:
UK trades with both EU as non-EU-countries will decrease due to a negative impact of the new trade agreements. This prediction is based on previous research from authors like Beck (2016), Kierzenkowski (2016), Simionescu (2016) and Armstrong (2016). Firms that are exporting to the UK will see a decline in firm value due the fact that the market reacts to the foreseeing decline in export to the UK due this announcement. However, we expect that the larger firms will feel the effect of the Brexit announcement less than the smaller firms because smaller firms are probably less global than the larger firms and therefore losing a higher percentage of the index value. With this knowledge, we expect a decrease in abnormal returns for larger firms and a larger decrease in abnormal returns for smaller firms.
To test our prediction by using an event study we have considered that the abnormal returns will be jointly normally distributed with a corresponding zero conditional mean. Therefore we obtain the following hypothesis for the main research question:
H0: The announcement of the Brexit has no effect on the security returns.
H1: The announcement of the Brexit has an effect on the security returns.
Together with the following hypotheses for the sub questions:
H0: There is no difference between the abnormal returns of EU and non-EU countries.
H1: There is a difference between the abnormal returns of EU and non-EU companies.
H0: There is no relationship between the abnormal return and firm size.
17
4 Data
I. Data Resources
To understand and identify the effects and the influence of the “Brexit” announcement we need to obtain data from the largest companies that have highest impact on the import to the UK from both largest non-EU and EU import countries to the UK. The data, collected from DataStream, contain the Total Return Index (TRI) of sixty firms and market indices. The countries total return index (TRI) is used as a benchmark to calculate the cumulative abnormal returns, where the TRI itself accounts for dividends payments or stock splits. This gives a better representation of the value securities and therefore the firm value rather than using only the stock price.
II. Data Collection
Figure 2 illustrates which 12 industries have the most impact on the total import of the United Kingdom over the year 2015, where electrical and mechanical machinery together with car import has the largest share on the import value of the UK.
Figure 2
Import in the UK by sectors Source: Statista Portal
Note: This figure shows the 12 largest industries that have the most impact on the import value of the United Kingdom over the year 2015. (In million GBP)
18
Furthermore, in this paper we are looking at the EU and non-EU countries that have the highest influence on the total import value of the UK. In Table 1, we report the largest importers to the UK, over the year 2015. It describes the values for the EU countries and for the non-EU countries. We use a total sample of 10 countries, which is chosen from the highest export countries to the UK.
Table 1: Largest exporters to the UK, 2016
source: (International Trade Centre, Trade Map)
Country Exported value to the UK (%)
1 Germany 79.6 (28.4 %) 2 Netherlands 42.8 (15.3 %) 3 France 32.5 (11.6 %) 4 Belgium 28.5 (10.2 %) 5 Switzerland 24.0 (8.6 %) 6 Italy 21.8 (8.0 %) 7 Spain 19.2 (6.9 %) 8 Norway 15.8 (5.6%) 9 Turkey 11.0 (3.9 %) 10 Denmark 4.8 (1.7 %) Total 280.1 (100 %) EU countries Germany 79.6 (33.8 %) Netherlands 42.8 (18.2 %) France 32.5 (13.8 %) Belgium 28.5 (12.1 %) Italy 21.8 (9.3 %) Spain 19.2 (8.2 %) Denmark 11.0 (4.7 %) Total 235.4 (100 %) Non-EU countries Switzerland 24.0 (53.8 %) Norway 15.8 (35.4 %) Turkey 4.8 (10.8 %) Total 44.6 (100 %)
19
Based on Table 1 and the annual reports 2015 we can determine the sample size of companies for each country. In Table 2, we are using a decreasing sample size of the number of firms for every country, ranked from the highest to the lowest influential countries. This will give a total sample of 60 companies that is used in this study.
Table 2: List of countries and their number of firms
Country Number of Firms (%)
Germany 10 (16.67 %) Netherlands 9 (15 %) France 9 (15%) Belgium 7 (11.67%) Switzerland 7 (11.67%) Italy 6 (10%) Spain 4 (6.67%) Norway 3 (5%) Denmark 3 (5%) Turkey 2 (3.33%) Total 60 (100%)
Note: List of countries, ranked on export influence to the UK and total impact on sample size
The different companies that are used in this paper are based on the data of Figure 2, which reports the industries who have the highest export influence to the UK. Based on these industries we selected the largest companies that have an export trade with the UK over the year 2015. In Table 9, listed in the appendix, all the firms are collected and ranked according to the company value. This table contains the full list of companies that are used in this paper together with the corresponding countries. The ranking of these companies is based on the market capitalization (MCAP) for each country, which is defined as total number of outstanding shares multiplied by the stock price of the annual report at the end of 2015.
5 Results
20 Table 3: Market model
Event Day European Companies Non-European Companies
A͞R
C͞A͞R A͞R C͞A͞R
-22 -0.573 -0.573 1.997 1.997 -21 -0.150 -0.723 0.015 2.012 -20 0.306 -0.417 0.571 2.584 -19 -0.315 -0.732 -0.116 2.468 -18 -0.026 -0.758 1,066 3.534 -17 0.262 -0.496 -2.353 1.181 -16 0.206 -0.290 -0.047 1.134 -15 0.289 -0.001 -0.321 0.813 -14 0.152 0.151 0.945 1.758 -13 0.137 0.287 0.362 2.120 -12 0.193 0.480 0.707 2.827 -11 -0.084 0.396 -0.636 2.192 -10 -0.208 0.188 -1.397 0.795 9 0.034 0.222 -0.110 0.685 -8 -0.030 0.192 2.519 3.204 -7 0.135 0.328 -2.140 1.064 -6 -0.174 0.154 0.836 1.901 -5 -0.156 -0.003 -0.187 1.713 -4 -0.100 -0.103 1.221 2.935 -3 0.122 0.019 -0.665 2.269 -2 -0.255 -0.236 -0.851 1.418 -1 -0.004 -0.240 0.514 1.932 0 0.161 -0.079 0.465 2.397 1 0.937 0.857 -1.113 1.284 2 -0.197 0.660 -0.139 1.145 3 -0.078 0.583 1.427 2.572 4 -0.090 0.492 -0.154 2.418 5 0.291 0.783 0.740 3.158 6 0.507 1.290 -0.945 2.213 7 -0.046 1.244 -0.426 1.787 8 -0.365 0.879 0.165 1.952 9 0.055 0.935 0.282 2.233 10 -0.160 0.775 -0.399 1.834 11 -0.064 0.711 1.393 3.227 12 0.168 0.879 0.021 3.248 13 -0.167 0.712 0.293 3.541 14 -0.013 0.699 -0.101 3.440 15 -0.055 0.644 0.534 3.974 16 0.105 0.748 -1.081 2.893 17 0.173 0.921 0.626 3.520 18 -0.160 0.761 -0.035 3.485 19 0.008 0.768 -0.387 3.097 20 -0.116 0.652 -0.160 2.937 21 0.146 0.798 -2.655 0.283 22 0.341 1.139 -0.492 -0.209
21 Figure.3
Plot of the average cumulative abnormal return for EU and non-EU companies
Note: Plot of the average cumulative abnormal returns (C͞A͞R) of companies from both the EU nd and non-EU countries from event day -22 to event day 22.
The analysis aggregating the average abnormal returns is primarily based on the assumption that the event window of the included firms do not overlap over time. This research is based on this assumption and therefore it allows us to calculate the variance of the C͞A͞R without concerning about possible covariances between securities as this assumption assumes they are zero. If overlap exist between the event windows, the distributional results for the aggregated abnormal returns do not longer hold, as the covariances between the securities will not be longer zero.
22
23
The evidence of Table 3 and 4 strongly supports the hypotheses that there is a relation between the abnormal return and both European and non-European firms. However, this hypothesis does not hold for every country and period. The sample average abnormal return (A͞R), at the day after the announcement (day 1) for European firms is 0.937 percent. Given the standard error of the one-day average abnormal return, we identify a strong positive relationship between the average abnormal return and EU-firms. We also identify strong relationships between the cumulative average abnormal returns from the periods 5, +5], [-3, +3] and [0, +4]. Hence, the null hypothesis that there is not a relationship between average abnormal return and European firms is rejected.
In addition, the sample average abnormal return (A͞R), at the day after the announcement (day 1) for non-European firms is -1.113 percent. Given the standard error of the one-day average abnormal return, we identify a strong negative relationship between the average abnormal return and EU-firms. In addition, we also identify strong relationships between the cumulative average abnormal returns for the period [-5, +5]. Hence, the null hypothesis that there is not a relationship between average abnormal return and European firms can be rejected for the period [-5, +5] and for the day after the announcement. ……….
The results indicating the negative value effect for non-European firms are quite consistent with the predictions from existing literature as well as our own hypothesis. However, the positive value effect for European firms is quite surprising, as the predictions would indeed indicate a less negative effect on European firms than on non-European firms. However, the positive and significant average (cumulative) abnormal returns over the periods [-5, +5], [-3, +3], [0, +4] and [0, +1] were not expected. Therefore, we can conclude there is a positive and a negative value effect on EU and non-EU firms respectively.………..
24
Table 5: Results of the C͞A͞R based on firm size for different periods
C͞A͞R (%) Event Day
Region Firm size (-10, 10) (-5, 5) (-3, 3) (0, 4) (0, 1)
EU Large 0.058 (0.098) -0.019 (-0.031) -0.423 (-0.708) -0.592 (-1.362) -0.365 (-0.612) EU Small -0.220 (-0.368) 0.913** (2.100) 0.346 (0.580 -0.273 (-0.629) 2.225*** (5.122) Non-EU Large 0.126 (0.290) -0.130 (-0.217) -0.488 (-1.123) 0.250 (0.419) 0.287 (0.481) Non-EU Small 0.187 (0.431) 0.483 (0.809) 0.568 (1.308) 0.024 (0.040) -0.310 (-0.520) EU + non-EU Large 0,092 (0.155) -0.074 (-0.171) -0.455 (-0.763) -0.171 (-0.286) -0.039 (-0.066) EU + non-EU Small -0,016 (-0.027) 0.698. (1.606) 0.457 (0.766) -0.125 (-0.209) 0.958** (2.204)
Note: Table 5 describes the C͞A͞R for either large or small companies both from the EU and non-EU countries, which are ranked according the MCAP of these companies. T-statistics are reported in parenthesis. Statistical significance levels at 10%, 5% and 1% are indicated by *, ** and ***
respectively. ……….
Figure 4………
Plot of the average cumulative abnormal return for small and large companies.………
Note: Plot of the average cumulate average abnormal returns (C͞A͞R) for all small and large companies that belong to the EU and non-EU from event day -22 to 22.---
25
Table 6: Average cumulative abnormal returns through time an across firms for all months.
Panel A Period (-7,-6) (-6,-5) (-5,-4) (-4,-3) (-3,-2) (-2,-1) (-1:0) (0, 1) (1, 2) (2, 3) (3, 4) (4, 5) All Firms -0.297 (-0.344) -0.801 (-0.69) 1.793 (-1.369) 1.081 (-1.169) -0.154 (-0.049) -1.845* (-1.765) 0.184 (-0.869) 0.403 (-1.239) 1.632** (-2.369) -1.105 (-1.039) -0.100 (-0.049) -0.551 (-0.369) EU Firms -0.210 (-0.175) -1.626 (-1.355) 1.765 (1.471) 1.163 (0.969 -0.160 (-0.133) -1.215 (-1.513) -0.790 (-0.966) 0.878** (1.981) 1.181 (1.485) -0.722 (-0.602) -0.154 (-0.128) -0.208 (-0.174) Non-EU Firms -1.497 (-1.013) 3.373* (1.809) 5.577*** (2.720) 2.675* (1.667) -0.414 (-0.821) -9.94*** (7.651) 2.397*** (4.069) -2.114** (-2.466) 8.046*** (7.039) -5.993*** (-4.749) 0,081 (0.039) -4,049*** (-3.069) Panel B Period (-7,-6) (-7,-5 (-7,-4) (-7,-3) (-7,-2) (-7,-1) (-7, 0) (-7, 1) (-7, 2) (-7, 3) (-7, 4) (-7, 5) All Firms -0.297 (-0.027) -1.098 (-1.269) 0.695 (0.569) 1.777 (1.331) 1.623 (1.267) -0.221 (-0.099) -0.037 (-0.594) 0.366 (-0.767) 1.997 (1.552) 0.893 (0.869) 0.793 (0.711) 0.242 (0.047) EU Firms -0.210 (-0.063) -1.836 (-1.419) -0.071 (-0.062) 1.091 (1.240) 0.931 (0.767) -0.284 (-0.765) -0.363 (-0.861) 0.514 (1.169) 1.696* (1.669) 0.974 (1.139) 0.820 (1.125) 0.612 (0.838) Non-EU Firms -1.497 (-1.133) 1.876 (1.461) 7.453*** (4.059) 9.128*** (9.038) 9.714*** (8.069) -0.327 (-1.065) 2.169** (2.469) 0.055 (0.059) 8.101*** (6.564) 2.108** (2.011) 2.188 (1.569) -1.861 (-1.364)
26
Table 6 represents the average cumulative abnormal returns through time an across firms for all months. Panel A describes the C͞A͞R of the companies of each separate month around the announcement and panel B describes the aggregated results of panel A. Both panels contain the returns for all companies, EU companies and non-EU companies. There we identify high fluctuations in time. Whereas the most C͞A͞R of each month lay in a range of -/+ 2 percentage we notice especially larger outliers for non-EU companies. In addition, we identify a significant negative effect for non-EU companies and a significant positive effect for the EU-companies in the month after the announcement. However, we see a positive value effect for non EU-companies in the month towards the announcement. The plot of Table 6.A is illustrated in Figure 5.
Figure 5
Plot of the average abnormal return for each month for EU, non-EU and all companies
Note: Plot of the average abnormal returns (A͞R) of the companies for every month around the announcement. From event month -7 to event month 5. Containing the average CAR returns for all companies, EU companies and non EU-companies.
27
companies. However we identify a decreasing C͞A͞R in the period [-3,-1] for both EU and non-EU companies. Where the CAR of non-EU firms becomes more stable in the period [-1, +1], the C͞A͞R of non-EU firm shifts up in the month before the announcement but decreases directly following the month after the announcement. Therefore, in some extent the market is learning if the day of the announcement is coming closer.
Figure 6
Plot of the average cumulative abnormal return for EU, non-EU and all companies.
Note: Plot of the average cumulative abnormal returns (C͞A͞R) for all, EU and non-EU companies around the announcement. From event month -7 to event month 5.
To examine the influence of the Brexit announcement on the security values we are using control variables of all companies to check the relationship between the variables and the cumulative abnormal return (CAR). We apply 3 different firm size measurements to identify the effects of firm size on the cumulative abnormal returns. This is tested by using (9), (10) and (11) together with the cumulative abnormal returns of the three periods in the event window: [0, +1], [-3, +3] and [0, +4] as the dependent variable. Which gives us the following hypothesis;
H0: The control variables do not explain the abnormal return day.
H1: The control variables do explain the abnormal return.
The regressions results are summarized in Table 7. Where the regressions includes different measurements for the firm size in combination with the various control variables. Where model 1, 2 and 3 represents the regressions using the variables: market capitalization, assets
-2 -1 0 1 2 3 4 5 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 C A R EVENT MONTHS
28
and sales respectively. In addition, panel A, B and C represents the cumulative abnormal returns of the three periods in the event window: [-3, +3], [0, +1] and [0, +4] respectively.
Table 7.A: Regression of the Cumulative Abnormal Return over the Period [0, +1]
Panel A: [0, +1] Variables: Model 1 Coefficient Model 2 Model 3 C 13.070** (0.0356) 7.667 (1.241) 8.647 (1.439) MCAP -0.714** (-2.205) - - Assets - -0.414 (-1.273) - Sales - - -0.480 (-1.477) M/B 0.466* (1.690) 0.366 (1.316) 0.394 (1.411) ROA -12.390 (-0.927) -10.755 (-0.773) -11.680 (-0.841) D/E 0.144 (0.303) 0.222 (0.443) 0.109 (0.223) C/A -12.808* (-1.799) -11.040 (-1.49) -11.099* (-1.888) Dummy variable 0.029 (-0.026) 0.376 (0.335) 0.446 (0.409) R-Squared 0.133 0.082 0.091 Adjusted R-Squared -0.035 -0.022 -0.012 F-statistic 1.355 0.784 0.883
29
Table 7.B: Regression of the Cumulative Abnormal Return over the Period [-3, +3]
Panel A: [-3, +3] Variables: Model 1 Coefficient Model 2 Model 3 C -0.174 (-0.026) -2.288 (-0.301) 2.699 (0.364) MCAP 0.252 (0.116) - - Assets - 0.122 (0.306) - Sales - - -0.145 (-0.373) M/B 0.327 (0.937) 0.305 (0.894) 0.359 (1.042) ROA -2.904 (-0.172) -1.556 (-0.091) -4.623 (-0.270) D/E 0.104 (0.172) 0.063 (0.102) 0.112 (0.186) C/A -15.871* (-1.761) -14.887 (-1.635) -16.930* (-1.888) Dummy variable 0.706 (0.504) 0.829 (0.602) 0.596 (0.660) R-Squared 0.085 0.087 0.087 Adjusted R-Squared -0.018 -0.016 -0.015 F-statistic 0.746 0.843 0.852
30
Table 7.C: Regression of Cumulative Abnormal Return over the Period [0, +4]
Panel A: [0, +4] Variables: Model 1 Coefficient Model 2 Model 3 C 20.747** (2.347) 14.899* (1.662) 20.781** (2.438) MCAP -1.101** (-2.33) - - Assets - -0.771* (1.651) - Sales - - -1.118** (-2.425) M/B 1.074*** (2.670) 0.943** (2.337) 1.043*** (2.634) ROA -29.235 (-1.499) -28.180 (-1.395) -32.472* (-1.648) D/E -0.134 (-0.193) 0.0314 (0.043) 0.168 (-0.243) C/A -31.46*** (-3.030) -29.809*** (-2.771) -31.485*** (-3.054) Dummy variable -0.769 (-0.477) -0.281 (-0.173) -0.312 (-0.201) R-Squared 0.226 0.187 0.232 Adjusted R-Squared 0.138 0.095 0.145 F-statistic 2.576 2.036 2.663
Note: Results of regression for all control variables using OLS regression method. Dependent variable: abnormal return at day one for all firms. T-statistics are reported in parenthesis. Statistical significance levels at 10%, 5% and 1% are indicated by *, ** and *** respectively.
31
is positive and negative for the period [0, +1] and [0, +4] respectively. Firm size measured by the number of sales only shows a negative relationship between the accumulative abnormal returns from period [0, +4] and the firm size. In comparison, we cannot find enough evidence to conclude a relationship between the firm size measured by the number of assets over any period. Based on the results from Table 7.C, we identify significant more relationships between the independent variables and the cumulative abnormal returns. Hence, the relationship between the firm variables and the value effect increases if the abnormal returns are measured over a longer window after the announcement of the Brexit.
6 Discussion
Some implications about this study are: The countries total return index (TRI) is used to calculate the cumulative abnormal returns, where the TRI itself accounts for all corporate actions such as changes to the stocks by dividends payments. This gives a better representation of the value securities and therefore the firm value rather than using only the stock price. In other words, it improves the stock price after dividends or stock splits. However, DataStream does not contain the TRI for Belgium and Turkey and therefore the price index (PI) is used which is less accurate then the TRI. To acquire results that are more accurate it might be necessary to adjust these countries PI to TRI by using other additional data sources. Using such a large time window to test the cumulative abnormal returns before and after the the event window can lead to biased results as it can capture more effects than only the announcements of the Brexit. However, it is very difficult to assign these fluctuations over time. As addressed in MacKinley the power of the test is substantial higher if the sampling interval is reduced. Another limitation is that the sample size of companies is measured according to the country’s export value to the UK, causing the sample size for lower influential countries to be smaller. Hence, making adjustments for outliers not possibly due the small sample size. To improve the results of this paper implementing a higher number of firms would be useful to create the possibility to adjust for outliers.
32
extension of this paper, it also may be useful to include the impact of the announcement on the firm risk by using the beta of the market model. This beta can be used to examine the stability of the market risk by altering the beta over time. Hence, for further research we could implement the influence of microeconomic risk and how to reduce this risk by using diversification strategies.
7 Conclusion
By applying a short period before and after the announcement, we were able to identify that there are significant different reactions for European Union and non-European Union companies. By using longer periods before the announcement we were able, in some extent, identify pre-market movements for both EU and non-EU companies.
The results indicating the negative value effect for non-European firms are quite consistent with the predictions from existing literature as well as our own hypothesis. However, the positive value effect for European firms is quite surprising, as the predictions would indicate a less negative effect on European firms in comparison with non-European firms. However, we find positive average (cumulative) abnormal returns before and after the announcement. In summary, we can conclude there is a positive and a negative value effect for EU and non-EU firms respectively.
The findings of this paper in general support the hypothesis that announcement of the Brexit has a positive and negative effect on the firm value of EU and non-EU firms respectively. Hence, we can conclude that there is a relationship between the sample (cumulative) average abnormal return and EU and non-EU firms and therefore the null hypothesis that the announcement has no impact is rejected.
33
References
Armstrong, K. A. (2017), “Brexit time: leaving the EU – Why, How and When?” Cambridge
University Press, 151-164.
Baker, S. and Krawiec, K. D. (2004), “The penalty default canon.” George Washington Law
Review, 663-723
Beck, M. (2016), “Brexit and FDI.” Economic Outlook, 40(2), 26-30.
Boulanger, P., Philippidis G. (2015), “The End of a Romance? A Note on the Quantitative Impacts of a ‘Brexit’ from the EU.” Journal of Agricultural Economics, 66(3), 832-842 Brooks, C. (2002), “Introductory Econometrics for Finance.” Cambridge: The University of
Cambridge.
Brown, S.J., Warner, J.B. (1985). "Using daily stock returns. The Case of Event Studies." Journal of Financial Economics 14, 3-31.
Brunie, C. H., Hamburger, M. J., and Kochin, L. A. (1972), “Money and stock prices: The channels of influence.” The Journal of Finance, 27(2) 232-249.
Campbell, J. Y., Adrew, W. L., and MacKinlay, A. C. (1997), "The econometrics of Financial
Markets." Princeton University Press 1997.
Dhingra, S., Ottaviano, G. (2016) “The consequences of Brexit for UK trade and living standards." London School of Economics and Political Science.
Fama, E. (1970), “Efficient Capital Markets: A Review of Theory and Empirical Work.”
Journal of Finance, 25(2), 383-417
Hauswald, R. B. H. (2003),”Event studies: Assessing the Market impact or Corporate Policy.” Kogod School of Business, American University, Washington.
Keen, S. (2017), “Trade and the Gains from Diversity: Why economists failed to Predict the consequences of Brexit.” Globalizations, 14(6), 803-809.
Kierzenkowski, R., Pain, N., Rusticelli, E., Zwart, S. (2016), “The economic consequences of Brexit.” OECD Economic Policy Paper (16)
Lim, K. P., Brooks, R. D., Kim, J. H. (2008), “Financial crisis and stock market efficiency: Empirical evidence from Asian countries”, International Review of Financial Analysis, 17(3),
571-591
MacKinlay, A. C. (1997), “Event studies in economics and finance”, Journal of Economic
34
Masuduzzaman, M. (2012) “Impact of the Macroeconomic Variables on the Stock Market Returns: The Case of Germany and the United Kingdom”, Global Journal of Management
and Business Research, 12(16)
Mohammad, B. A. (2011) “Impact of Micro and Macroeconomic Variables on Emerging Stock Market Return: A Case on Dhaka Stock Exchange (DSE)”, Interdisciplinary Journal of
Research in Business, 1(5), 8-16
Petersen, T. (2015), “Brexit potential economic consequences if the UK exits the EU”,
Shaping Sustainable economies.
Simionescu. M. (2016), The impact of Brexit on the Foreign Direct Investment in the United kingdom”, Bulgarian Economic Papers, 2-17.
Appendix
Table 8: Control variables and their corresponding mean, median, 25 and 75 percentile.
Germany
variables MCAP M/B ROA D/E cash/assets assets sales size mean 54.8 2.079 0.036 1.355 0.050 116.5 74.9 18.226 median 64.3 1.897 0.035 1.220 0.038 69.1 58.4 18.052 25% 33.2 1.514 0.025 0.741 0.026 36.4 25.5 17.411 75% 7.1 2.770 0.058 1.939 0.063 181.3 106.4 19.010 Netherlands
variables MCAP M/B ROA D/E cash/assets assets sales size mean 45.0 3.021 0.049 0.784 0.057 54.6 45.8 17.134 median 21.6 2.836 0.051 0.507 0.052 16.7 20.5 16.633 25% 15.1 1.636 0.014 0.379 0.026 14.1 7.4 16.459 75% 79.4 3.983 0.080 0.930 0.063 44.4 45.7 17.593 France
variables MCAP M/B ROA D/E cash/assets assets sales size mean 43.2 1.527 0.033 0.712 0.061 71.2 44.1 17.881 median 30.9 1.192 0.031 0.646 0.028 48.5 35.1 17.698 25% 24.7 1.076 0.026 0.449 0.015 35.9 24.5 17.391 75% 67.2 1.909 0.041 0.917 0.103 93.7 50.0 18.354 Belgium
35
Switserland
variables MCAP M/B ROA D/E cash/assets assets sales size mean 112.0 4.652 0.089 0.434 0.102 56.9 35.8 17.401 median 39.4 3.594 0.085 0.347 0.063 41.0 34.3 17.529 25% 19.0 2.621 0.071 0.284 0.040 12.9 9.1 16.371 75% 230.0 7.618 0.121 0.514 0.111 122.4 48.6 18.622 Italy
variables MCAP M/B ROA D/E cash/assets assets sales size mean 23.8 1.596 0.012 1.683 0.104 83.7 49.5 17.648 median 22.4 1.183 0.004 1.689 0.101 86.0 45.8 18.253 25% 7.2 0.871 0.018 0.536 0.048 33.4 15.3 16.761 75% 39.9 2.069 0.041 2.391 0.150 136.3 82.5 18.728
Spain
variables MCAP M/B ROA D/E cash/asset assets sales size mean 22.8 1.741 0.027 2.486 0.039 55.0 25.2 17.612 median 14.6 2.063 0.027 2.326 0.033 41.6 24.7 17.455 25% 12.8 0.828 0.010 0.887 0.007 24.3 5.6 17.006 75% 40.9 2.333 0.064 4.244 0.078 99.0 45.4 18.375 Norway
variables MCAP M/B ROA D/E cash/asset assets sales size mean 240.2 2.007 0.016 0.771 0.059 422.1 233.8 19.454 median 222.7 1.386 0.017 0.802 0.069 201.5 128.2 19.122 25% 105.0 1.107 0.040 0.204 0.028 115.9 108.0 18.568 75% 393.1 3.527 0.070 1.307 0.080 948.9 465.3 20.671 Denmark
variables MCAP M/B ROA D/E cash/asset assets sales size mean 440.6 8.289 0.133 0.405 0.097 210.3 148.6 18.917 median 189.5 1.978 0.013 0.363 0.065 123.2 107.9 18.629 25% 93.4 0.772 0.024 0.023 0.025 85.0 65.4 18.258 75% 1038.9 22.11 0.410 0.828 0.199 422.7 272.6 19.862 Turkey
variables MCAP M/B ROA D/E cash/asset assets sales size mean 14.9 0.415 0.020 1.001 0.041 111.5 27.0 12.485 median 22.3 0.623 0.030 1.502 0.061 167.2 40.6 18.727
25% * * * * * * * *
75% * * * * * * * *
36 N o te : This tabl e d esc rib es th e fu ll sa m p le list o f co m p an ies that are u sed in thi s p ap er. Based o n F ig u re 1 an d T ab le 1 the co m p an ies w er e sel ect ed. The n u m b er o f co m p an ies diff ers per co u n tr y bec au se o f t h e co u n tr y inf lu ence o n t h e im p o rt o f t h e U K, d es crib ed in t ab le 3 . Th e ran kin g o f t h e co m p an ie s is b ased o n th e M arke t cap itali zatio n (M CA P ) p er co u n try , which is d efin ed as to tal n u m b er o f o u tst an d in g sha res m u ltip lied b y the st o ck p rice o f the an n u al report at th e end o f 2 01 6 . 10 9 8 7 6 5 4 3 2 1 ran k Table 9 : L is t o f Co m p an ie s Thysse n Krup p Gro u p Lin d e Fr ese n iu s M erc k BM W Si em ens BA SF Vo lks wagen D aiml er AG Baye r A G Germ an y C ou n tr ies * D SM ko n in klijk e KP N Akzo No b el Ah o ld P h ilip s AS M L H ein ek en Un ilev er Ro yal D u tch Sh ell Neth erlan d s * Peug eo t Sa in t-Go b ain Renau lt P ern o d Ricard Sch n eid er Ch ristia n D io r D AN ON E Sa n o fi TO TAL Fr an ce * * * Bro EVS ad cas t new Barco AG FA Um ic o re So lva y UCB An h euse r-Bu sch Belg iu m * * * Sch lin d er Gro u p Swatch Gro u p Syng enta AB B Gro u p N o varti s N est lé Ro che Switserlan d * * * * Fr eni Brem b o CN H in d u strial Fi at Ch rysler Telec o m
Enel Eni Italy
37 Wilcoxon signed-rank test
When conducting a T-test, it is assumed that abnormal returns are normally distributed, which is not always the case. That is why, it would be wise to conduct an additional non-parametric test to test the significance. Non-parametric test is sometimes called ‘’distribution-free’’ test, since it assumes that the abnormal returns are not distributed normally. A chosen non-parametric test for this study is the Wilcoxon signed-rank test. It is not based on the assumption of normal distribution and considers both the sign and the magnitude of the abnormal return. Test can be done through the following formula:
Z = W−
N(N+1) 4
√N(N+1)(2N+1)24 (12)
