Influence of the terror attacks of 13 November 2015 in Paris and the terror attacks of 22 March 2016 in Brussels on the European market indices

(1)

Influence of the terror attacks of 13 November 2015 in Paris

and the terror attacks of 22 March 2016 in Brussels on the

European market indices.

Date| 29 June 2016

Bachelor’s Thesis

Name| Nick Kroon

Specialization| Economics and finance

Student ID| 10546642

Study programme| Economics and Business

(2)

Statement of Originality

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

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

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

Abstract

This study examines the effect of the Paris attacks of 13 November 2015 and the Brussels attacks of 22 March 2016 on the European market indices by testing the effect on return and the volatility. This study contributes to the existing literature by the fact that there is less known about the effect of recent terrorism on the volatility in Europe. Besides that this study researches whether there are factors which enhance or weakens the effect on return and volatility. The effect on return is tested by using the event study described by Brown and Warner. The effect on volatility is tested by using a simple F-test. The results shows that it has no significant effect on the return, but the statistics for the Brussels attacks are lower than for Paris. On the other hand the results show that there is a significant effect on the volatility for Paris but an opposite effect for the attacks in Brussel. A possible reason is that investors in Europe are resilient to terrorism attacks but that it brings some risk with it. It is not reliable to draw conclusions from this research for 2 attacks because of the contrary findings.

(3)

1 I

NTRODUCTION

1.1 M

OTIVATION

Last year Europe is plagued by several terror attacks executed by the IS. As we all know it has an impact on people’s normal life. What we wonder is if it has also an effect on the economy. There is much known about the economic effect of the 9/11 attacks. Besides that there are several studies which examines the effect of other terrorism attack on stock markets. But because the Paris- and Brussels attacks are very recent, there are no previous researches available yet. In previous researches about the effect of terrorism in Europe, little attention is being paid to the effect on volatility. The only research that is found that examines the effect of terrorism on volatility in Europe is the research of Arin, Ciferri and Spagnolo (2008). Besides that in this research there is examined if there is an effect of particular factors on the effect on volatility and return. Therefore this research contributes to the existing literature for 2 reasons.

1.2 C

ENTRAL QUESTION

Has the terror attacks in Paris of 13 November 2015 and the terror attacks in Brussels of 22 March 2016 a significant effect on the market indices of the European countries?

1.3 P

REVIOUS LITERATURE

According Arin, Ciferri and Spagnolo (2008) there is a significant effect on the returns and volatility of the stock market. Although the effects varies across countries, the European countries are less affected. Chen and Siems (2004) found that U.S. capital markets are more resilient and thus they recover sooner from terrorist/military attacks than other global capital markets. The results Eldor and Melnick (2004) found shows that the Palestinian attacks had a permanent negative effect on the Israeli stock market. Karolyi and Martell (2006) found an average decrease in market capitalization of 401 million per attack. In the research by Essaddam and Karagianis (2014), they tested the effect of terrorism on volatility taking into account particular country attributes. They found that attacks on firms in wealthier countries are associated with larger volatility.

1.4 S

UB QUESTIONS

To give an answer to the central question it is divided into three sub questions. The first question is whether the attacks has a significant negative effect on the returns of the market indices. The second question is whether the attacks has a significant effect on the volatility of the market indices. The last

(5)

question is whether the factors wealth, threat level and distance has a significant impact on the effect of volatility and the effect on return.

1.5 D

ATA

To examine the effect on return the date 13 November 2015 and 22 March 2016 is considered as the event dates. To examine the effect on the return the daily stock prices are used from November 27 2014 till November 20 2015 for the Paris attack and from April 9 2015 till March 30 2016 for the Brussels attack which are 250 observations. To examine the effect on volatility the daily stock prices are used from October 8 2015 till December 18 2015 for Paris and from February 15 2016 till April 28 2016. As a proxy for the European markets every biggest market index from the 20 countries under examination are used as described in the appendix. To examine which factors does have an effect on the impact of volatility and return, the variables GDP per capita as a proxy for wealth, threat level and the distance from Paris and Brussels per country are used.

1.6 M

ETHOD

The first hypothesis that will be examined is: does the Paris attacks have a negative effect on the return of the European market indices. This will be examined by using the excess return approach by Brown and Warner. The second hypothesis that will be examined is: does the Paris attacks have an effect on the volatility of the European market indices. This will be tested by calculating the standard deviation and use the F-test. The last hypothesis that will be examined is: does the factors wealth, distance and threat level have an effect on the impact on volatility and on the impact on return. To test this a linear regression will be done.

1.7 S

TRUCTURE

This thesis proceeds as follow. In chapter 2 it continues with discussing the results of previous research about the effect of terrorism on stock markets and formulate the expectations. In chapter 3 there will be discussed which data is used from which period. Besides that there will be discussed which method is used to test. After that in chapter 4, the empirical findings will be explained and discussed and finally in chapter 5 there is a conclusion.

(6)

2 L

ITERATURE REVIEW

This study examines the effect of the terror attacks of 13 November 2015 in Paris and of 22 March 2016 in Brussels on the market indices of European countries. To test this effect the central question is divided into 3 sub questions. The following chapter briefly shows the results from previous studies and from this an expectation is derived for the findings. First the result of previous research on the effect of return will be discussed. Thereafter the results on the effect of volatility will be discussed. At last the results about the factors will be discussed.

2.1 T

HE EFFECT OF TERROR ATTACKS ON RETURN

The first sub question that will be examined is if the attacks have an significant negative effect on the returns of the European stock markets. Several previous studies examines the effect of terror attacks on stock markets. Eldor and Melnick (2004) examined the effect of the intensified Palestinian terror attacks after September 27 2000 on the Israeli stock market. For the period of 1990 till 2003 they tested it by using the time-series analysis. They found that these attacks had an permanent negative effect. Furthermore they found that over time the stock market didn’t become desensitized.

Another research by Arin et al. (2008) found evidence of statistically significant causality effects of terrorism in all six countries under examination by using the VAR-GARCH(1,1)-in-mean model. They found that the stock markets of the two West-European countries, Spain and UK, are less effected by the attacks than the other 4 non-European countries.

When examining the effect of 31 attacks on global and local capital markets, Brounrn and Derwall (2010) found an immediate price reaction averages −0.34%, which translates into a negative annual price impact of over 134%. Comparing major events shows that other than the 9/11 events, very few attacks have had a significant price impact that lasted longer than the event day itself.

In the research of Karolyi and Martell (2006) they examined the effect of terrorism on stock markets by testing 75 attacks between 1995 and 2002 in which publicly traded firms are targets. They used the event study to test the effect. An event-study analysis around the day of the attacks uncovers evidence of a statistically significant negative stock price reaction of -0.83%. They estimated an average decrease in the market capitalization of 401 million. Besides that it appears that losses on human capital, such as kidnapping, has an larger negative effect than physical losses like bombings.

Chesney, Reshetar and Karaman (2011) also examined the effect of terrorism on stock markets. They tested 77 terror attacks from the time period of 1994 till 2006 using different methods. According to the event study, 55 out of 77 terrorist attacks had a significant negative impact on the behavior of the stock markets. Using the GARCH–EVT method, 45 out of 77 terror

(7)

attacks had a significant negative effect. Using the non-parametric method, 56 out of 77 attacks had a significant negative effect.

The most result from previous research indicates that terrorism has a negative effect on stock markets. Taking this into account and according to the efficient market hypothesis of Fama (1970) , which states that the stock market prices reflect all the available information, the

expectation is that the ‘Paris attacks’ and the ‘Brussels attacks’ have a negative effect on the stock market prices and therefore have a negative effect on the returns.

2.2 T

HE EFFECT OF TERROR ATTACKS ON VOLATILITY

The second sub question that will be examined is if the ‘Paris terror attacks’ have a significant effect on the volatility of the European market indices. Volatility is the measure of financial market risk and is measured by the standard deviation.

In the research by Arin et al. (2008) they found a significant effect on the volatility on the 6 countries. Like the returns, they also found that the effect on volatility is less for the 2 European countries. This suggest that the European countries are more resilient for events like this.

In the research by Essaddam and Karagianis (2014) they found significant abnormal volatility on the day of the attack and remain significant for at least fifteen days following the day of the attack. Furthermore they found that firms operating in wealthier, or more democratic countries, face greater volatility in stock returns relative to firms operating in developing countries.

Most researches focuses on the effect on returns. That is why little researches are available about the effect of terrorism on the volatility of stock markets. By this reason this research will examine the effect of the ‘Paris attacks’ and the ‘Brussels attacks’ on the volatility of the European markets. The expectation is that it has an effect because according to the efficient market hypothesis the price reflects the beliefs of investors. According to Chen and Siems (2004) when information becomes available about a cataclysmic event, like a terrorist or military attack, investors often flee the market in search of safer financial instruments. This will have effect on the prices of the market index and will affect the volatility.

2.3 T

HE EFFECT OF THE FACTORS ON RETURN AND VOLATILITY

The last sub question that will be examined is if the factors GDP, threat level and distance have a significant effect on the impact of terrorism on the volatility and return of European stock markets. This is comparable with the research by Essaddam and Karagianis in which they examined the effect of terrorism on volatility considering particular country characteristics.

(8)

2.3.1 Wealth

The purpose of testing the effect of GDP on the volatility and return is to test whether it matters how rich a country is. The GDP per capita is used as a proxy for a countries wealth.

Chen and Siems (2004) found that U.S. capital markets recover sooner from terrorist/military attacks than other global capital markets. They also found evidence that suggests the possibility that this increased market resilience can be at least partially explained by a banking/financial. It is

assumed that the efficient functioning of an economy’s banking/financial sector is a key determinant of whether an economy, and hence its capital markets, is able to withstand and quickly absorb exogenous and endogenous shocks. Financial sector development, as measured by the size of activity, and efficiency of banks, nonbank financial intermediaries and equity markets tends to be greater in richer countries (Demirgüç-Kunt & Levine, 1999, p. 5). So there is a link between the GDP of a country and the efficiency of the banking/financial sector.

As was mentioned before, according to Arin et al. (2008) was the effect on the two European countries less than the effect on the other countries. The European countries does have a higher GDP than the other 4 countries. This is not in line with the research by Karolyi and Martell (2006) which have found that attacks on firms from richer and more democratic countries generate a larger negative stock return. According to the research by Essaddam and Karagianis (2014) in which it examines whether the country attributes does effect the abnormal volatility, it found that it is statistically significant when it occurs in wealthier countries. For poorer countries they found no significance. In the researches they did not explained a theory to support their findings.

Because the only theory to support the effect of GDP is the theory of efficient banking and financial sector. Countries with an higher GDP have a more efficient banking and financial sector so they are better to absorb the effects of terrorism. For this reason the expectation is that GDP does have a significant negative effect on volatility and a negative effect on return.

2.3.2 Threat level

About the effect of threat level on the volatility of stock markets is not much known. For this reason this research wants to examine if there is an effect of the threat level on the volatility and return of European market indices.

According to the efficient market hypothesis (Fama, 1970) prices reflects all the available information. When there is a high threat level in a country, investors are already aware of high risk for terrorism and this is already included in the prices. When there is an attack the shock of the attack is already partly absorbed by the high threat level. For that reason the effect is less for a country with a high threat level.

(9)

On the other hand, when a country has a low threat level, investors are not aware of risk for terrorism. The prices reflects a low risk for terrorism. When there is an attack, it is unexpected and for this reason the effect on prices will be higher.

For this reason the expectation is that the threat level has a negative effect on the effect on volatility. Besides that it is also expected that threat level has a negative effect on the effect on return. In short, a higher threat level softens the shock of a terror attack.

2.3.3 Distance

There are no studies found in which they examines the effect of distance on the volatility of stock markets. For this reason this research wants to examine the effect of distance on the volatility and return of European market indices. The factor distance is measured by using the absolute distance from Paris and Brussel to the city in which the market index is located.

Because global capital markets are closely and tightly inter-related, policymakers and regulators around the world must always be aware of what is going in other parts of the world. Today’s real-time information economy means that news spreads rapidly (Chen & Siems, 2004). Financial markets are becoming increasingly integrated internationally. (Heathcote & Perri, 2004). This is called financial globalization. Because of the globalization and the current communication tools distance does not matter anymore. For this reason it is expected that it does not have a significant effect on volatility and return.

Figure 1:overview of relationships

Terror attack

Volatility of European market indices Return on European market indices Threat level Distance from paris Wealth - +/- - - + - +/- -

(10)

Table 1: Summary previous researches

Authors (year)

Study Period Method Event window

Results return Results volatility Chen and

Siems (2004)

Test effect of Kuwait invasion and 9/11 on global capital markets

1990 and 9/11 Abnormal excess returns (t = 0) (t = 0,5) (t = 0,10) Significant at 0.01, 0.05 and 0.1 respectively:

9/11: AR: 31, 0 and 2 out of 33 CAR 6: 10, 4 and 8

CAR11: 2, 3 and 4

Kuwait: AR:9, 4 and 2 out of 18 CAR6: 6, 2 and 2 CAR11: 6, 1 and 2 Arin, Ciferri and Spagnolo (2008)

Effect of terror on the stock markets of Indonesia, Israel, Spain, Thailand, Turkey and UK 2001-2007 VAR- GARCH(1,1)-in-mean model. 1/1/2002– 31/12/2006, for a total of 1368 observations.

A significant negative effect for Indonesia, Israel, Spain, Thailand, Turkey and UK:

0.0026 (2.16), 0.0011 (2.2), 0.0018 (2.57), 0.0014 (2), -0.0022 (-2), -0.0003 (-3) respectively

-evidence of statistically significant for Indonesia, Israel, Spain, Thailand, Turkey and UK: 0.0037 (2.06). 0.0027 (3.85), -0.0016 (-4), 0.0015 (1.88), -0.0018 (-2), -0.0002 (-1) respectively Karolyi and Martell (2006) Test 75 attacks between 1995 and 2002 on publicly-traded companies 1995-2002 Abnormal excess returns (t = 0) (t = -10,10)

-Negative return on day of attack: With 9/11: 2.2%.

Without 9/11: -0.83% (-10,10)

no other abnormal returns found -average reduction in market capitalization of 401 million Essaddam

and Karagianis (2014)

examines the impact of 44 terrorist attacks against publicly traded American firms using volatility factors 1995-2010 GARCH(1,1)-model (t= 0,i) where i=1,2,5,10,15 (t = -i, -1) where i = 2, 5, 10, 15

Significant cumulative abnormal volatility at 1% for all event windows after the attack(t = 0,i). No significance for event windows prior to attack. Chesney, Reshetar and Karaman (2011)

Effect 77 attacks on global, European, U.S., and Swiss capital markets 1994-2006 abnormal excess returns (t = 0) (t = 0, 5)

-global, , U.S, European and Swiss capital markets:

AR:

30, 19, 29 and 28 out of 77 attacks respectively. CAR6: 25, 12, 13 and 25 out of 77

attacks respectively Eldor and

Melnick (2004)

Testing the effect of Palestinian attacks on the Israeli stock market distinguishes location, type of attack and target, number of casualties, and the number of attacks per day for 639 terror attacks.

1990-2003

Time serie analysis

13 year -a permanent negative effect on the stock market

-no evidence that markets became desensitized to terror attack over time.

Brouren and Derwall (2010)

Tested effect of 31 attacks on global and local capital markets. 1990-2005 Abnormal excess returns (t = 0) (t = 0,2) (t = 0,5) (t = 0,10)

All attacks and 9/11 excluding respectively:

AR global: -0.34% and -0.22% AR local: -0.92% and -0.72% CAR3, CAR6 and CAR11 are not tested at significance

(11)

3 M

ETHODOLOGY

3.1 E

FFECT OF TERROR ATTACKS ON RETURN

To examine whether the attack has an significant effect the event study methodology described by Brown and Warner (1985) is used. The event-study methodology is an approach that focuses on identifying abnormal returns to firms or markets from a specific event. If investors react favourably to an event, a positive abnormal stock returns around the event date is expected. Alternatively, if investors react unfavourably to an event, a negative abnormal stock returns is expected.

The event-study methodology is based on the efficient markets hypothesis. This hypothesis generally states that as new information becomes available, it is fully taken into consideration by investors assessing its current and future impact. Investors immediately reassess individual firms and their ability to withstand potential economic, environmental, political, societal, and demographic changes resulting from the event. (Chen & Siems, 2004, p. 3)

To test this a maximum of 250 daily return observations is used for the period around its respective event, starting at day - 244 and ending at day + 5 relative to the event (Brown & Warner, 1985). In this study the event dates are 13 November 2015 and 22 March 2016. From day -244 till -5 equals L1 and this is considered as the estimation window . From -5 till + 5 equals L2 and is

considered as the event window. The data from 27 November 2014 till 20 November 2015 and from April 9 2015 till March 28 2016 are used, which are 250 observations. In figure 1 the time line is illustrated.

For the event study the data was collected from DataStream. First the return of the 20 market indices under examination are calculated using the following formula: 𝑅 = ln(_𝑃𝑃𝑡

𝑡−1). After that the excess

returns for the event window is calculated as described in the appendix. Because the statistics described by Brown and Warner considers multiple securities while in this study only one security is used, the will be made use of the statistics described by Brooks (2014, p. 635). These test statistics will be asymptotically normally distributed, so 𝐴𝑅𝑖,𝑡~ 𝑁(0, 𝜎2(𝐴𝑅𝑖,𝑡)).

Thereafter the statistics for the event date and the estimation window from -5 till 5 are calculated. After that the standardized abnormal returns are calculated, all further explained in the

(12)

appendix. These statistics are needed to test the null hypothesis for significance at an alpha of 1%, 5% and 10%. The hypothesis for the event date and the cumulative excess returns are as following:

𝐻0 : 𝐴𝑅𝑖,0 = 0

𝐻1: 𝐴𝑅𝑖,0 < 0

𝐻0: 𝐶𝐴𝑅−5,5= 0

𝐻1: 𝐶𝐴𝑅−5,5< 0

3.2 E

FFECT OF TERROR ATTACKS ON VOLATILITY

To test whether the Paris attack has a significant effect on the volatility the data from DataStream from 8 October 2015 till 18 December 2015 is used. To test whether the Brussels attack has a significant effect on volatility the data from February 15 2016 till 28 April 2016. This obtains 52 observations start from 26 days prior to the event date till 25 days after the event date. As a measure of a stock’s volatility the standard deviation is calculated.

First the standard deviation from -26 till -1 is calculated and after that from 0 till 25. To test whether there is a significance difference in standard deviation the variances will be calculated. The test statistics to compare the variances is the F-test:

𝐹 = 𝜎

2 0−49

𝜎2 −50−−1

To test the null hypothesis on significant the F-distribution is used and an alpha of 1&, 5% and 10%. The hypothesis for the volatility are as follow:

𝐻0: 𝜎0−49= 𝜎−50−−1

𝐻1: 𝜎0−49 > 𝜎−50−−1

3.3 E

FFECT OF THE FACTORS ON RETURN AND VOLATILITY

To examine whether the factors distance, wealth and threat level has a significant effect on the effect of terrorism on return and volatility a linear regression is done. The data for GDP per capita denominated in US dollars from the World Bank Development Indictors database are used as a proxy for the factor wealth. The GDP data used are from 2014 because 2015 is not available yet. The data for the threat level is derived from the traveling advice from the UK and Dutch government and from the telegraph.co.uk which mapped all the threat levels.

The dependent variables in the regressions are the difference in the volatility and the abnormal excess returns. The independent variables in the regression are distance, wealth,

threatlevel2, threatlevel 3 and threatlevel 4. The variables threat level are dummy variables and are categorical predictors. For example, when the threat level in a country is the highest, variable

(13)

threatlevel 4 equals 1 and the other threatlevel variables equals 0. The sample size of the regression is the number of market indices under examination and equals 20. According to Schwert and Seguin, heteroscedasticity in stock returns is a pervasive phenomenon (1990, p. 1152). Because daily returns are used, this data will exhibit heteroscedasticity. For this reason there will be a robust regression done. By doing the robust regression it does not assume homoscedasticity. The regression equations will look like the following:

𝜎𝑎𝑓𝑡𝑒𝑟− 𝜎𝑏𝑒𝑓𝑜𝑟𝑒 = 𝛼 + 𝛽1 𝐷𝑖𝑠𝑡 + 𝛽2𝐺𝐷𝑃 + 𝛽3𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙2 + 𝛽4𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙 3 + 𝛽5𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙 4

𝐴𝑅𝑖,0= 𝛼 + 𝛽1 𝐷𝑖𝑠𝑡 + 𝛽2𝐺𝐷𝑃 + 𝛽3𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙2 + 𝛽4𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙 3 + 𝛽5𝑇ℎ𝑟𝑒𝑎𝑡𝑙𝑒𝑣𝑒𝑙 4

To test the hypothesis on significance for the factors distance and GDP a t- test will be done. To test the hypothesis on significance for the threat level factors a F-test will be done. The hypothesis for both regressionsare as follow:

𝐻0: 𝛽1= 0 𝐻1: 𝐵1≠ 0

𝐻0: 𝛽2= 0 𝐻1: 𝐵2≠ 0

(14)

4 R

ESULTS

In this chapter the results are shown and interpreted for each sub question. The found results will be discussed for each sub question and concludes whether the null hypothesis defined in chapter 3 should be rejected or not.

4.1 E

FFECT OF TERROR ATTACKS ON RETURN

In table 2 the abnormal returns and statistical significance levels for the 1- and 11-day event windows are shown for the Paris attacks for the 20 European stock markets under examination. Also included in the table is a column showing the number of trading days before each stock market is returned to its pre-attack level.

Table 2: Abnormal returns (Paris)

As shown in table 2 only the WIG experienced significant negative effect at the event day at a 0.10 level. This is the opposite of the expectations formulated earlier which stated that it should have a negative effect. At the event window from -5 till 5 only the OMX IPI out of 20 stock markets has a significant negative effect at the 0.10 level. Looking at the number of trading days before this market

Stock market Event day

AR

t statistic CAR (5,5) t statistic Days to rebound CAC40 -0,01061 (-0,73907) -0,01985 (-0,41681) 3 AEX -0,01418 (-1,05695) 0,000252 (0,005664) 3 DAX -0,00727 (-0,49327) 0,017062 (0,348836) 3 FTSE MIB -0,00118 (-0,07185) -0,00839 (-0,15345) 3 ATX 0,002632 (0,209486) 0,022554 (0,541241) 2 IBEX -0,0031 (-0,22215) -0,01231 (-0,26583) 3 FTSE in £ -0,00959 (-0,87197) -0,00225 (-0,06166) 3 ISEQ -0,00388 (-0,31786) 0,016038 (0,396514) 3 SMI in CHF -0,00758 (-0,57513) 0,00355 (0,081263) 3 BEL20 -0,00212 (-0,17982) 0,023256 (0,596046) 2 OMX Copenhagen -0,01248 (-0,96034) 0,007503 (0,174089) 3 OMX Helsinki -0,01097 (-0,83228) 0,011764 (0,269206) 7 OMX Stockholm -0,00838 (-0,67916) -0,00268 (-0,0654) 3 Athex -0,01012 (-0,29229) -0,02608 (-0,2272) 4 OMX IPI 0,009453 (1,406296) -0,03107 (-1,3937)* 1 PSI 20 -0,00341 (-0,22454) -0,03942 (-0,78283) 3 LuxX -0,01452 (-1,13384) 0,001272 (0,029933) 3 OSEAX -0,00389 (-0,33896) -0,01014 (-0,26616) 1 WIG -0,01322 (-1,48462)* -0,01096 (-0,37129) 2 PX 0,002787 (0,289327) 0,01975 (0,618273) 4

(15)

is returned to its previous level and to the fact that in de days before the attack it had a negative return it can be concluded that this significance is not caused by the Paris attacks.

According to the research by Chen and Siems (2004) they suggest that most stock markets recover from a terrorist attack between 6 and 11 trading days. Table 2 shows that 17 out of 20 stock markets are within 3 trading days returned at their pre-attack level. The outlier in trading days is OMX Helsinki with 7 trading days before rebound. It cannot be explained why OMX Helsinki

recovered the least. For the event day all the null hypothesis are not rejected except for the WIG. For the event window from -5 till 5 all the null hypothesis are not rejected except for the OMX IPI.

In table 3 the abnormal returns and statistical significance levels for the 1- and 11-day event windows are shown for the Brussels attacks for the 20 European stock markets under examination. Also included in the table is a column showing the number of trading days before each stock market is returned to its pre-attack level.

Table 3: Abnormal returns (Brussels)

As shown in table 3, none of the stock markets experienced a negative significant effect on the stock market. This is not in line with the expectation formulated earlier. For the event window from -5 till 5 none of the stock markets experienced a negative significant effect. This is also not in line with the expectations. Comparing the days to rebound, there is a wide range in days. This cannot be explained by theory.

Stock market Event day

AR

t statistic CAR (-5,5) t statistic Days to rebound CAC40 0,001547 (0,098313) -0,02067 (-0,39598) 2 AEX -0,00109 (-0,0724) -0,00334 (-0,067) 1 DAX 0,004973 (0,305945) 0,00659 (0,122232) 1 FTSE MIB 0,001049 (0,055754) -0,03622 (-0,58056) 20 ATX -0,00233 (-0,15826) -0,01769 (-0,36274) 14 IBEX -0,00218 (-0,13784) -0,03549 (-0,67798) 19 FTSE in £ 0,001843 (0,148884) 0,005924 (0,14429) 1 ISEQ -0,00041 (-0,03096) 3,38E-05 (0,000768) 1 SMI in CHF 0,001108 (0,08914) -0,01932 (-0,46865) 1 BEL20 0,002125 (0,161979) -0,01686 (-0,38754) 14 OMX Copenhagen 0,003514 (0,226047) -0,04078 (-0,79095) 2 OMX Helsinki -0,00365 (-0,24364) -0,01399 (-0,28166) 4 OMX Stockholm 0,00427 (0,28965) -0,03386 (-0,69253) 14 Athex 0,005298 (0,174522) 0,052621 (0,522615) 3 OMX IPI -0,00237 (-0,29347) -0,00148 (-0,05515) 2 PSI 20 0,001639 (0,103642) -0,00185 (-0,03528) 15 LuxX -0,00766 (-0,50728) 0,004458 (0,089005) 9 OSEAX 0,001703 (0,125685) -0,01041 (-0,2317) 11 WIG 0,00303 (0,295216) 0,04267 (1,25363) 2 PX -0,00747 (-0,68631) -0,00894 (-0,24751) 7

(16)

Comparing the t statistics from the Paris attack with the t statistics from the Brussels attack, the significance for the Brussels attack is less than Paris. A possible reason for that is that capital markets become more resilient and more desensitized for terror attacks.

The findings of this research are not in line with the expectations that are stated in chapter 2. Because none of the stock markets are significant, the null hypothesis will not be rejected.

4.2 E

FFECT OF TERROR ATTACKS ON VOLATILITY

Table 4 shows the volatility before the attacks, the volatility after the attacks and results from the F-test for the Paris attacks. The statistical significance is F-tested at 0.10, 0.05 and 0.01 level.

As Table 4 shows that 11 out of 20 stock markets experienced a significant positive effect on the volatility. 2 out of 20 experienced a positive significant effect at 0.01 level. Of the 20 stock markets, 5 had a significant positive effect at the 0.05 level. 4 out of 20 stock markets experienced a significant positive effect at the 0.10 level. The stock market with the highest significance is the OMX Helsinki and WIG. The reason why these stock markets have the highest significance cannot be

Table 4: Volatility before and after with the corresponding F-values (Paris)

Stock Market σbefore σafter F-value

CAC 40 0,010056 0,014281 2,016925** AEX 0,009321 0,014184 2,315663** DAX 0,010714 0,016073 2,250423** FTSE MIB 0,010808 0,01454 1,809881* ATX 0,012504 0,012218 0,954726 IBEX 0,009773 0,012849 1,728463* FTSE 0,007508 0,011108 2,188759** ISEQ 0,008064 0,009698 1,446414 SMI 0,007367 0,010971 2,217762** BEL 20 0,007852 0,010772 1,881947* OMX Copenhagen 0,012394 0,011371 0,841823 OMX helsinki 0,008611 0,013942 2,621231*** OMX stockholm 0,009891 0,0129 1,701151* Athex 0,017023 0,021056 1,53004 OMX IPI 0,006611 0,007604 1,322761 PSI 0,013776 0,011163 0,656633 Luxx 0,013373 0,012795 0,915347 OSEAX 0,009178 0,01086 1,400108 WIG 0,0059 0,011371 3,714874*** PX 0,007099 0,007062 0,98947

* Statistically significant at 0.10 level ** Statistically significant at 0.05 level *** Statistically significant at 0.01 level

(17)

explained by a theory. Also the volatility from the cumulative returns were calculated. A significant positive effect at a 0.10 level was found for the cumulative volatility with a corresponding F-value of 1.937

It was expected that the attacks should have a positive significant effect on the volatility. Because 11 out of 20 (55%) stock markets experienced a significant effect are the findings in line with the expectations. For these 11 stock markets the null hypothesis is rejected. For the remaining 9 stock markets the null hypothesis is not rejected.

Table 5 shows the volatility before the attacks, the volatility after the attacks and results from the F-test for the Brussels attacks. The statistical significance is tested at 0.10, 0.05 and 0.01 level.

Table 5 shows that 12 out of 20 stock markets experienced a significant negative effect on the volatility. This means that the volatility decreased after the terror attack. This is not in line with the expectations which states that volatility should increase. The cumulative volatility also

experienced a significant negative effect at a 0.05 level with a corresponding F-value of 2.427.

Stock Market σbefore σafter F-value

CAC 40 0,014058 0,01185 1,40739 AEX 0,013481 0,009483 2,020938** DAX 0,015678 0,011766 1,775491* FTSE MIB 0,017979 0,017102 1,105257 ATX 0,01492 0,010386 2,063532** IBEX 0,016215 0,012506 1,681137 FTSE 0,011966 0,008174 2,143127** ISEQ 0,011944 0,007402 2,603934** SMI 0,011002 0,008129 1,831622* BEL 20 0,011994 0,00843 2,024502** OMX Copenhagen 0,018114 0,007865 5,303957*** OMX Helsinki 0,014626 0,011816 1,532008 OMX Stockholm 0,015688 0,011605 1,827382* Athex 0,031552 0,021297 2,194784** OMX IPI 0,008439 0,005485 2,36709** PSI 0,013807 0,011368 1,475038 Luxx 0,018078 0,014371 1,582341 OSEAX 0,015013 0,012393 1,467483 WIG 0,009774 0,008258 1,400922 PX 0,013327 0,007365 3,274826***

Table 5: volatility before and after with the corresponding F-values

* Statistically significant at 0.10 level ** Statistically significant at 0.05 level *** Statistically significant at 0.01 level

(18)

The alternative hypothesis states that the volatility after the attack is higher than the volatility before the attack. Because in this case the volatility is higher before the attack, the null hypothesis will not be rejected.

In conclusion, the volatility increased after the Paris attacks but decreased after the Brussels attacks. For this reason it is not reliable to conclude that the effect on volatility is due to the terror attacks.

4.3 E

FFECT OF THE FACTORS ON RETURN AND VOLATILITY

First a linear regression is done with the abnormal returns as the dependent variable. Table 6 shows the result of the regression for the Paris attack.

As shown in table 6 none of the variables has an significant effect on the abnormal returns. The variable with a low t statistics is distance, which is in line with the expectations that distance does not has a significant effect. For the threat level variables and the factor GDP, the findings are not in line with the expectations that it should have a negative effect. For all the variables the null hypothesis will not be rejected.

Thereafter a linear regression is done with the difference in volatility as the dependent variable. Table 7 shows the result of the regression for the Paris attack.

As shown in table 7 none of the variables has a significant effect on the difference in volatility. The variable with the lowest t statistics is distance and GDP. This is in line with the

expectations that distance does not have a significant effect but not in line with the expectation that GDP does have a negative effect. The variables TL3 and TL4 are contrary to the expectations that threatlevel should have a negative effect on the volatility. The variable TL2 has a negative effect on

Table 6: Linear regression AR Distance GDP TL2 TL3 TL4

AR

Coef. t P>|t|

Dist 2.51e-06 3,5E-06 0.486

GDP -3.64e-08 7,23E-08 0.623 TL2 -.0024541 0,00529 0.650 TL3 -.0067877 0,005409 0.230 TL4 -.0023599 0,005235 0.659 σafter – σbefore _Coef. _t _P>|t| Dist 3.62e-07 0,35 0.734 GDP 1.67e-09 0,06 0.951 TL2 -.0016448 -0,86 0.403 TL3 .0004449 0,26 0.798 TL4 .001623 0,95 0.360

(19)

the volatility, but is not significant enough. For all the variables the null hypothesis will not be rejected.

Subsequently, the same linear regressions have been done for the Brussels attacks. Starting with the regression on the abnormal returns. Table 8 shows the result of the regression for the Brussels attack.

As shown in table 8, none of the variables have a significant effect on the abnormal returns. The variables TL3 and TL4 are contrary to the expectations that threatlevel should have a negative effect on the volatility. Variable TL2 is in line with the expectation that it should have a negative effect on the abnormal returns, but there is no significance enough. The findings for the factor distance is in line with the expectations that there is no significant effect. For the factor GDP it should have a negative effect, so it is not in line with the expectation. For all the variables the null

hypothesis will not be rejected.

At last the linear regression have been done for the difference in volatility for the Brussels attacks. Table 9 shows the results of the regression.

As shown in table 9 none of the factors experienced a significant effect on the volatility. For the factor distance this is in line with the expectations that stated that there is no significant effect. For the factor GDP and the threatlevel factors, the negative effect on the volatility is in line with the expectations but the t value is too low to be significant. For all the variables the null hypothesis will not be rejected.

Table 8: Linear regression AR Distance GDP TL2 TL3 TL4

AR Coef. t P>|t| Dist 1.42e-06 0,84 0.417 GDP 1.20e-09 0,02 0.981 TL2 -.000284 -0,11 0.911 TL3 .0042077 1,62 0.128 TL4 .0040592 1,12 0.281

Table 9: Linear regression σafter – σbefore Distance GDP TL2 TL3 TL4

σafter – σbefore Coef. t P>|t| Dist -6.52e-07 -0,56 0.586 GDP -3.86e-09 -0,16 0.874 TL2 -.0000745 -0,07 0.946 TL3 -.0026348 -1,20 0.250 TL4 -.0005543 -0,35 0.730

(20)

5 C

ONCLUSION

This study examines the effect of the terror attacks in Paris of 13 November 2015 and the effect of the terror attacks in Brussels of 22 March 2016 on 20 European stock market indices. To examine this the effect of the attacks on the return and volatility of the stock markets are tested. This study also sought to explore whether there are factors that affect the impact on return and volatility. This study contributes to the existing literature on three fields. First because it is relatively recent, there is no research done yet to this attack. Second not much previous researches does give attention to the effect of terrorism on volatility in Europe. At last there are no previous studies which examines the effect of factors on the return and volatility.

The first research question of this study examines the effect on the return of the market indices by calculating the abnormal returns (AR). The obtained result shows that there is no

significant effect on the market indices. The markets are more resilient for terrorism than expected. The second research question examines the effect on the volatility. 11 out of 20 market indices experienced a significant positive effect for the Paris attacks. For the Brussels attacks 12 out of 20 markets experienced a significant negative effect. The last research question examines the effect of the factors wealth, distance and threat level on the effect of return and volatility. For both Paris and Brussels there was no significant effect found for any of the factors.

The findings on returns of this study is opposite to the findings of previous studies. Chen and Siems (2004) found for the attacks they examine for all the stock markets under examination a significant effect. This is not in line with the findings of this study. Comparing the t values of Paris and Brussels, the values found for Brussels are relatively lower. A possible reason for these findings are that investors become more desensitized for terrorism. The findings on volatility for Paris are in line with the findings of previous studies. Both Arin et al. (2008) and Essaddam and Karagianis (2014) found a significant effect on the volatility. The findings for Brussels are the opposite of what was found in previous studies.

The theory which support this study is the efficient market hypothesis. The results of this study are not in line with the theory. According to the efficient market hypothesis prices reflects all the available information. Considering this, the expectation is that it would have a significant effect. The effect on volatility for Paris in line with the efficient market hypothesis but is not in line for Brussels.

This study has been examined 20 market indices for 2 terrorism attack. As a consequence using this methodology, this study does have a number of limitations which need to be considered. The first limitation is that because only two attacks are examined, the power to draw conclusions on the effect of terrorism on the European market indices is low. Because there is no more information

(21)

available yet about other terror attacks in Europe, it may be included in later researches. The second limitation is that because only 20 market indices are used, the statistical power of the linear

regression is low. By researching more market indices the effect of the factor may become clearer. The method used in this event study is the excess mean return method. By using the market model instead, the variance of the abnormal return is reduced by removing the portion of the return that is related to variation in the market's return (MacKinlay, 1997). Using the market model, a global market index should be used as the market proxy. In short, there are suggestions for three area’s in which more research could be done to improve the quality of this research.

The answer on the main question whether there is an effect of the Paris attacks of 13 November and of the Brussels attacks on 22 March 2016 on the European market indices is

contradictory. The attacks does not have a significant effect on the return. However, the Paris attacks does have a significant effect on the volatility but the Brussels attacks has the opposite effect. Based on these 2 attacks it is not possible to draw reliable conclusions about the effect on volatility.

(22)

6 B

IBLIOGRAFIE

Arin, K. P., Ciferri, D., & Spagnolo, N. (2008). The price of terror: The effects of terrorism on stock market returns and volatility. Economics Letters, 164-167.

Brooks, C. (2014). Introductory economics for finance (3 ed.). Cambridge : Cambridge University Press.

Brounrn, D., & Dewrall, J. (2010). The Impact of Terrorist Attacks on International Stock Markets.

European Financial Management(16), 585-598.

Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: the case of event studies. Journal of

Financial Economics(14), 3-31.

Chen, A. H., & Siems, T. F. (2004). The effects of Terrorism on Global Capital Markets. European

Journal of Political Economy(20), 349-366.

Chesney, M., Reshetar, G., & Karaman, M. (2011). The impact of terrorism on financial markets: An empirical study. Journal of Banking & Finance (35), 253-267.

Demirgüç-Kunt, A., & Levine, R. (1999). Bank-based and Market-based Financial Systems:

Cross-country Comparisons. Washington: World Bank Publications.

Eldor, R., & Melnick, R. (2004). Financial Markets and Terrorism. European Journal of Political

Economy(20), 367-386.

Essaddam, N., & Karagianis, J. M. (2014). Terrorism, country attributes, and the volatility of stock returns. Research in International Business and Finance(31), 87-100.

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

Finance(25), 383-417.

Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. (1969). The Adjustment of Stock Prices to New Information. International Economic Review(10), 1-21.

Heathcote, J., & Perri, F. (2004). Financial globalization andreal regionalization. Journal of Economic

Theory, 207-243.

Karolyi, A. G., & Martell, R. (2006). Terrorism and the Stock Market. ssrn.com.

Koen, V., Lenain, P., & Bonturi, M. (2002). The Economic Consequences of Terrorism. OECD

Economics Department Working Papers(334). doi:http://dx.doi.org/10.1787/511778841283

MacKinlay, C. A. (1997). Event studies in finance. Journal of Economic Literature(35), 13-39. Seguin, P. J., & Schwert, W. G. (1990). Heteroskedasticity in Stock Returns. The Journal of

(23)

A

PPENDIX

Starting with calculating the abnormal returns for the event window from -5 till 5 using the following formulas: 𝐴𝑅𝑖,𝑡= 𝑅𝑖,𝑡− 𝑅̅ 𝑖 𝑅̅ = 1 239∑ 𝑅𝑖,𝑡 −6 −244

After that calculating the standardized abnormal returns for the event day using the following formulas: 𝜎²(𝐴𝑅𝑖,𝑡) = 1 239∑(𝑅𝑖− 𝑅̅) 2 𝑆𝐴𝑅𝑖,𝑡 = 𝐴𝑅𝑖,𝑡 √𝜎²(𝐴𝑅𝑖,𝑡) ~ 𝑁(0,1)

After this calculate the test statistics for the event window from -5 till 5 also using the method described by Brooks. First summing up the abnormal returns for the event window:

𝐶𝐴𝑅𝑖(−5,5) = ∑ 𝐴𝑅𝑖,𝑡 5 𝑡=−5

To calculate the variance for the CAR(-5,5) use the variance from the event date and calculate the standardized cumulative abnormal returns:

𝜎2_{(𝐶𝐴𝑅}

𝑖(−5,5)) = 11 ∗ 𝜎²(𝐴𝑅𝑖,𝑡) 𝑆𝐶𝐴𝑅𝑖(−5,5) =

𝐶𝐴𝑅_𝑖(−5,5)

(24)

Table 11: Regression AR Dist GDP TL2 TL3 TL4 (Paris)

country AR diff distance

(km) GDP (in US$) TL 2 TL 3 TL 4 CAC France -0,01061 _0.004225 0 42725,7 0 0 1 AEX Amsterdam -0,01418 _0.004863 430 52138,7 0 1 0 DAX Germany -0,00727 0.005359 479 47773,9 0 0 1

FTSE MIB Italy -0,00118 0.003732 640 35222,8 0 1 0

ATX Austria 0,002632 -0.00029 1035 51122,4 1 0 0 IBEX Spain -0,0031 _0.003076 1054 29721,6 0 0 1 FTSE U.K. -0,00959 0.0036 344 46297 0 0 1 ISEQ Ireland -0,00388 0.001634 782 54339,3 1 0 0 SMI Switzerland -0,00758 _0.003604 489 85616,6 0 0 0 BEL20 Belgium -0,00212 _0.00292 264 47327,6 0 0 1

OMX cop Denmark -0,01248 -0.00102 1028 60718,4 0 1 0

OMX hel Finland -0,01097 0.005331 1911 49842,7 1 0 0

OMX sto Sweden -0,00838 _0.00301 1546 58898,9 0 1 0

ATHEX Greece _0.004034 2099 21672,7 0 1 0

OMX IPI Iceland 0,009453 0.000992 2234 52036,7 0 0 0

PSI 20 Portugal -0.003409 -0.00261 1453 22,124.4 1 0 0 LuxX Luxembourg -0.014524 -0.00058 294 116,612.9 1 0 0 OSEAX Norway -0.003892 _0.001682 1341 97,299.6 1 0 0 WIG Poland -0.013218 0.005471 1366 14,336.8 0 0 0 PX Czech Republic 0.0027865 -3.7E-05 882 19,502.4 0 0 0

(25)

Table 12: σafter – σbefore Dist GDP TL2 TL3 TL4 (Paris)

country AR diff distance

(km) GDP (in US$) TL 2 TL 3 TL 4 CAC France 0.001547 -0.00221 264 42725.7 0 0 1 AEX Amsterdam -0.00109 -0.004 174 52138.7 0 1 0 DAX Germany 0.004973 -0.00391 317 47773.9 0 0 1

FTSE MIB Italy 0.001049 -0.00088 697 35222.8 0 1 0

ATX Austria -0.00233 -0.00453 914 51122.4 1 0 0 IBEX Spain -0.00218 -0.00371 1316 29721.6 0 0 1 FTSE U.K. 0.001843 -0.00379 321 46297 0 0 1 ISEQ Ireland -0.00041 -0.00454 776 54339.3 1 0 0 SMI Switzerland 0.001108 -0.00287 492 85616.6 0 0 0 BEL20 Belgium 0.002125 -0.00356 0 47327.6 0 0 1

OMX cop Denmark 0.003514 -0.01025 765 60718.4 0 1 0

OMX hel Finland -0.00365 -0.00281 1648 49842.7 1 0 0

OMX sto Sweden 0.00427 -0.00408 1280 58898.9 0 1 0

Athex Greece 0.005298 -0.01025 2089 21672.7 0 1 0

OMX IPI Iceland -0.00237 -0.00295 2126 52036.7 0 0 0

PSI 20 Portugal 0.001639 -0.00244 1714 22124.4 1 0 0 Luxx Luxembourg -0.00766 -0.00371 187 116612.9 1 0 0 OSEAX Norway 0.001703 -0.00262 1085 97299.6 1 0 0 WIG Poland 0.00303 -0.00152 1160 14336.8 0 0 0 PX Czech Republic -0.00747 -0.00596 717 19502.4 0 0 0

(26)

Table 14 AR Dist GDP TL2 TL3 TL4 (Brussels)

Influence of the terror attacks of 13 November 2015 in Paris and the terror attacks of 22 March 2016 in Brussels on the European market indices