How terrorist attacks influence the stock markets in France : an event study from different industrial perspectives

(1)

How terrorist attacks influence the stock markets in France — an event study from different industrial perspectives

Author: Said Khaled Schakib Student Number: 10841288 Supervisor: Jens Martin

University of Amsterdam Msc. Business Economics – Track Finance Thesis credits: 15 ECTS Date: June 30th, 2017

(2)

Statement of Originality

This document is written by Student Said Khaled Schakib 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

(3)

Abstract

This thesis aims to investigate the effect of domestic terrorism on the stock market in France. After an extensive analysis of considerable literature six potential industries were identified, which are believed to show significant response towards terrorist events. Based on the event study analysis, evidence is provided that terror attacks with high number of fatalities mostly have a short-term effect on excess stock returns. French stocks associated with the airline industry and leisure & tourism industry show the most negative decline in excess stock returns, while the results suggest that stocks associated with the defence industry respond with positive returns towards terrorist events

(4)

Table of Content

1. Introduction ... 1

2. Literature Review ... 3

3. Hypotheses ... 8

4. Methodology ... 11

4.1 The Event Study Approach ... 11

4.2 The Market Model ... 14

5. Data ... 16

6. Results ... 19

6.1 Tati store bombing ... 19

6.2 Air France Flight 8969 Hijack ... 20

6.3 Paris Métro and RER bombings ... 21

6.4 Paris Métro bombing 1996 ... 22

6.5 Toulouse and Montauban shootings ... 23

6.6 Charlie Hebdo attack ... 24

6.7 Paris November attacks ... 26

6.8 Truck attack in Nice ... 28

6.9 Additional test analyses ... 29

7. Conclusion ... 30

8. Appendix ... 32

(5)

List of Tables

Table 1: Major terror events in France from 1986-2016 ... 16

Table 2: Descriptive statistics ... 17

Table 3: Event study results of event 1 ... 19

Table 11: Results for event 1 of entire industries with additional tests ... 32

Table 12:Results for event 2 of entire industries with additional tests ... 33

(6)

1

1. Introduction

Terrorism has arrived into the heartland of Europe, as the major attacks in Paris 2015 show. With the rise of terrorist groups such as the “Islamic State of Iraq and the Levant” (ISIL) and Europe's incapability of external border control, European countries are facing an unprecedented threat of terror. Especially France seems to move into the attention of terrorist groups, as recent incidents show. Several empirical studies suggest that major terrorist attacks have a significant negative impact on the economy. The 9/11 attacks for instance caused a loss of output of approximately $75 billion dollars according to the International Monetary Fund in the United States. This corresponds to a 0.7 percentage points decrease of the total US GDP. Against this background the question arises to what extent the increased threat of terrorism affects the French stock market. While literature examined its effects on stock markets in regions with a pronounced history of terrorism such as Israel, Turkey, the Middle East, Spain, UK and the US, the question arises whether France’s Stock markets react in comparable manners.

The aim of this research is therefore to answer the question if significant abnormal returns do exist around terrorist events in France. If so, it would be interesting to know to what extent the reaction of stock returns can be observed for each industry. The final objective is to investigate whether the French stock market is able to absorb these terrorist events and if not, how long the different industries´ stocks need to rebound. Since media coverage exaggerates the negative impact of terrorism among economic and financial figures (Pirog, 2005), it is important to give the reader a more scientific view of this topic. Especially for investors it will be beneficial to know in detail how the past terror events affected French stocks in terms of portfolio diversification and risk management.

In this paper I analyse eight major terror events in France between 1986-2016. I will identify the events with significant abnormal returns for six different industries that are partly found to be affected by terrorist events in previous literature. The six examined industries in this thesis are the following - Airline, leisure & tourism, insurances, oil & gas, media and defence. For the analysis I use the event study methodology, which is the most conventional approach for measuring the effects of abnormal returns after terrorist events.

The results of this thesis are that primarily terror events with a relatively large number of casualties have a significant effect on particular French stocks, indicates the finding of previous literature. Out of the six analysed industries, stocks of the airline and leisure & tourism industry show the most negative significant responses towards terror events. In

(7)

2

contrast results of defence industry stock seem to be benefiting from terror events, indicating that investors expect higher future demand of homeland security goods and armaments after a terror event occurs.

Section 2 gives the reader a detailed overview about the existing literature and their empirical results. Based on the provided literature overview I will derive and formulate six hypothesises in section 3. Section 4 contains the methodology used in this thesis and describes the event study approach and the market model. Section 5 deals with the data used in this thesis and presents the statistical summary table, with the aim to give the reader an initial idea about the data used in this thesis. Section 6 finally presents the findings of this study, where I seek to link the findings to existing literature and also interpret the results individually. In section 6.9 the robustness of the analysis is presented by extending the analysis to all 26 industries with additional testing methods. The critical discussion about this thesis is made in the last section including a final conclusion.

(8)

3

2. Literature Review

A growing interest from researchers regarding the effects of terror shocks on macroeconomic characteristics has been emerged due to the increasing number of terror events in recent years. Especially after the 9/11 attacks the number of empirical studies on the effect of terrorism on stock markets have increased substantially. While literature focused mainly on regions, which have a comparatively intense history of terrorist events such as the United States, the United Kingdom, Spain, Turkey, Israel and the Middle East, France seems to be relatively unexploited. Most of these empirical studies suggest that only major terrorist incidents cause a negative downturn on stock returns in the short run, other studies provide evidence that periods with permanent terrorism also cause an overall negative effect on stock returns and GDP in the long run. In the following the most relevant literature about the effect on terrorism on the economy in general and stock markets and single industries in particular will be discussed.

Using data from the Basque Country in Spain during the Basque conflict from 1955 to 1999, Abadie and Gardeazabal (2003) find out that terrorist conflicts have a significant negative impact on the economy. Using the event study approach and comparing a region with terrorism to a synthetic control region without terrorism, they figure out a 10% decrease of the GDP per capita in a region with terrorist activities compared to a similar region without terror in the long run.

Their incentive to study the impact of terrorism in an open economy is founded on the result presented by Becker and Murphy (2001), which says that even attacks with disastrous magnitude like the September 11th do not exert a significant influence on the country’s economic activity in the long term. But in contrast reduced-form estimates of the economic effects of terrorism typically suggest much greater impacts, at least in those areas where the risk of terrorism is constant and severe. The main idea of their paper is that mobility of productive factors is responsible for differences between direct effects and equilibrium effects caused by terrorism. If a country is threatened by local terror, capital will tend to flow out from there into terror-free destinations. This will decrease net foreign investment in economies affected by terrorism. Even if terrorism is a global threat, international investment will respond accordingly to the expected intensity of terrorism. It may happen that variations in the overall global terror risk induce a re-allocation of capital, although the relative intensity of terrorist risk does not change across countries (Abadie and Gardeazabal, 2008).

(9)

4

Abadie and Gardeazabal (2008) have shown how international investors react, when a country faces terrorist attacks. They use a standard endogenous growth model and analyse the impact of increasing catastrophic risk on the net foreign direct investment (FDI) positions of countries. Their model predicts that in an integrated world economy, where international investors are able to diversify other country risk types, terrorism may trigger large capital flows across countries. Terrorist conflicts may have a significant impact on the allocation of productive capital across countries, even if it represents a small fraction of the overall economic risk. They find out that a one standard deviation increase in the intensity of terrorism produces a 5% decrease in FDI.

Eldor and Melnick (2004) investigate the effect of the number of casualties during terrorist events on stock markets in Israel. The results might have a higher degree of relevance for western economies in general, since Israel has more similarity in their economic, social and political characteristics, than results from developing countries. Eldor and Melnick examine 639 terror attacks in Israel using daily time series data from 1990 until 2003 in which 1,121 people were randomly killed and 5,726 people were randomly injured. The data used in this empirical study distinguishes between the sort of attack, the target and the number of induced casualties. They provide evidence that the number of casualties is crucial to affect the domestic stock market and the foreign currency market. They conclude that the higher the number of fatalities and wounded people, the higher the stock return’s variance.

Suicide incidents are believed to cause a softer downturn in stock returns and their variances, since the perpetrators no longer cause further threat as can be seen in the empirical results for the London suicide bombing (Kollias et al., 2011). This insight may also help explain why the non-suicide bombings in Madrid had a larger downward impact on the stock returns and a higher volatility at the beginning. According to Kollias et al. this disparity can be affiliated to the existence of uncertainty about the involved perpetrator group and their hiding location.

Furthermore, Kollias et al. (2011) investigate on the question, whether markets change their reactions on terrorism over time, and if so, whether these reactions are then again influenced by market size and maturity. In order to control for market capitalization, they compared the effects of terrorism on the London stock exchange (high market capitalization) with the Athens stock exchange (low market capitalization). They differentiate between domestic and transnational perpetrators and control for the targeted group (civilian vs. government) in their analysis. Interestingly, their findings show no significant abnormal returns whether neither for large capitalization markets nor for small capitalization markets.

(10)

5

Instead they obtain remarkable results from the volatility analysis. Stock market volatility can be affected significantly, if the extent of exogenous shocks is weighted by the number of fatalities and injuries. The estimated volatility models unveil Greece to react more sensitive to shocks than the UK.

From a global perspective Chen and Siems (2004) compared the reaction of 34 stock markets towards wars and terrorism around the world. They find out that U.S. capital markets are more resistant than in the past and recover sooner from terrorist attacks than other global capital markets. In an event study approach, they examine the US capital market’s response to 14 major terrorist &military attacks dating back to 1915 and global capital markets’ response to two recent events — Iraq’s Invasion of Kuwait in 1990 and the 9/11 attacks in 2001. They postulate an efficiently functioning banking sector of a country to be the key factor of whether capital markets will be able to withstand and suddenly absorb internal and external shocks. The Federal Reserve System reacted to the terrorist attacks on 9/11 by providing liquidity through the banking and financial sector to stabilize the economy.

The effect of terrorism on the tourism industry is extensively covered in the literature. Empirical studies confirm that terror events lead to an increasing number of travel and flight cancellation particularly after the major 9/11 attacks in New York and Washington D.C. (Floyd, Gray and Thapa, 2003; Kingsbury and Brunn, 2004). However, studies before 9/11 suggest a link between terrorism and tourism. Enders and Sandler (1991) for instance investigated the impact of terrorist events on the Spanish tourism sector, concluding a decrease of 140,000 tourists for an average terror incident in Spain. Another study by D’Amore and Anuza (1986) investigating the international terrorism and the resulting implications and challenges for international tourism. They find that due to increased terrorism in Europe with 16 total terrorist attacks in France, Spain, Italy, Denmark, Belgium and Greece in 1985, 54% of US tourists cancelled their reservation to Europe. Consequently, the World Tourism Organization determined a loss of $105 billion in the tourism sector in the same year (Sonmez and Graefe, 1998).

Drakos (2004) provides an industry-specific view. His research deals with the effects of the 9/11 attacks on a set of airline stocks listed at different international markets. He finds out that the systematic risk of airline stocks has significantly increased since the terrorist attack of 9/11 by testing the stability of their betas (from the CAPM-model) before and after the attack had occurred. Before the 9/11 event, airline stocks used to have betas greater than zero, but significantly lower than unity. This means that a 1% increase in the market portfolio's return would increase the airline stocks on average less than 1% and vice versa. After the 9/11

(11)

6

attacks, airline stocks transformed from less volatile defensive stocks (beta <1) to more volatile aggressive stocks (beta >1). In numbers, the average airline-stock had a beta of 0.76 before the 9/11 attacks, the average airline-stock beta increased to 1.73 after 9/11. Drakos delivers evidence of a sustained shift in the riskiness of airline stocks after the 9/11 attack.

Arin et al. (2008) compare the effects of shocks caused by terrorism of the countries Indonesia, Israel, Spain, Thailand, Turkey and the UK by using a time-series framework for the multi-country sample and analysing their responses towards terror on high return moments. Their research points out that, although the response to terrorist events varies across countries, their overall effect regarding mean and variance is significant. The authors believe that the financial investors in the Spanish and British stock market are more resilient towards terror attacks than in non-European stock markets, since the effects both in mean and variance were less. According to Arin et al. this observation can be explained with the investors´ belief in institutional quality in European democracies. European governments are more capable of absorbing the shocks caused by terrorist attacks emerging economies.

Chesney et al. (2011) investigate the effects of terrorism on stock, bond and commodity markets and compare them to the impacts on these markets caused by natural catastrophes and financial crashes. In order to obtain these results, the authors conduct three different methods (event study, non-parametric-approach and filtered GARCH-EVT) and investigate the impact of 77 terror attacks. About two thirds of the terrorist attacks considered lead to significant negative impact on at least one of the investigated stock market. The Swiss stock market is affected by the highest number of attacks, the American stock market by the lowest. The airline industry and insurance sector exhibit the highest susceptibility to terrorism, while the banking industry is the least sensitive. In opposite, the oil & gas industry showed positive significant response in some cases. They explained this positive effect by taking the market for crude oil into consideration. Since attacks on oil & gas firms or oil production and/or oil transportation facilities could increase the fear of inventors and market participants, the price of crude oil is more likely to increase. Especially in times of a tight oil market, meaning demand is higher than the supply the market could demand a fear premium for oil.

To identify the effects of terrorist attacks on daily stock markets Drakos (2009) uses data from 22 global stock indices during a 10-year period from 1994 to 2004. He investigates on the question whether and how terrorist activities influence investors’ mood and consequently their behaviour on capital markets. The results of his approach show that terrorism leads to significantly lower returns on the day of the attack’s occurrence. Furthermore, terrorist activity amplifies its negative effects, when those incidents cause higher psychosocial

(12)

7

impacts. His approach not only enlightens the mechanism via which terrorism affects stocks, but also provides empirical support for the sentiment effects.

The role of the insurance sector related to terroristic incidents was especially examined by an empirical study conducted by Kunreuther and Michel-Kerjan in 2004. Hereby, the study analyses how government may co-operate with the private sector in order to construct a terrorism risk insurance industry. Otherwise the alternative would be to leave the insurance coverage solely in the hands of the private sector. Another study from Ibragimov et al. (2009) highlighted the necessity of a central agency to coordinate the insurance market to allow for the transference of catastrophic risk. This is being justified by the existence of heavy left tails in risk distributions - a significant possibility of extreme losses. Left trail distributions of returns imply a greater downturn risk which makes losses more likely. These left tails lead to so-called non-diversification traps, which can´t be eliminated or reduced by adding other assets. The loss of a diversification trap is according to Ibragimov et al. (2009) extraordinarily high. The direct welfare loss of a trap in California residential earthquake insurance for example is approximately 3 billion USD per year. In 2004, Bourioux and Scott examined the possibilities to hedge capital markets against terrorist attacks without necessarily the use of insurance companies. Substituting for the insurance market may enable the opportunities to limit the losses caused by the attacks. Especially after the 9/11 attacks the necessity for a substitution arose as insurance companies started to exclude terrorism risk coverage from their policies.

The studies of Mueller and Stewart (2014) concerning counterterrorism policy by the U.S. government and the FBI shows, that expenditure in homeland security and armaments increased by more than $75 billion per year after 9/11 attacks. With respect to the authors’ findings one could assume that higher expenditures in homeland security and armaments after terror events would also boost the stock returns of such defence firms.

(13)

8

3. Hypotheses

In this section six hypotheses are derived based on the insights gained in the literature part. This study examines the effect of terrorism on particular industries. I will focus on six out of 26 industries, which are believed to respond significantly towards terrorism events, namely securities of the industries (1) airlines, (2) leisure& tourism, (3) insurance, (4) oil &gas, (5) media and (6) defence.

As described in the previous section an empirical link could be derived for terrorist events having an overall negative impact on stock index returns in the short run (Abadie and Gardeazabal, 2003). In particular I expect significant negative effects on the stock returns of the industries 1-3; these are stocks of the (1) airline, (2) leisure & tourism and (3) insurances industry. The reason for a significant decline in airline and leisure & tourism stocks caused by terrorist attacks is due to the increased sentiment of fear among the customers. Usually a higher degree of uncertainty about the security situation of a country leads to a downturn in travelling (Enders and Sandler, 1991). Less travelling indicates less tourism, which might lead to a decline in turnover of airline and tourism & leisure companies. The derived implication for a negative response of airline stocks is based on the findings of Darkos (2004). His empirical research shows that shortly after the 9/11-attacks the number of passengers travelling by international airlines declined by 17% compared to the year before. The derived implication for a negative response of leisure & tourism stocks is based on the findings of Enders and Sandler (1991). They find out that there is an average reduction of 140,000 tourists in Spain for each randomly occurring terrorist event.

For insurance company’s terror attacks would lead to higher costs for the damage caused by the attack at the first sight. Not only the resulting damage may cause a decline for insurance stocks, but also the existence of sharp left tails in risk distributions – a significant possibility of extreme losses. Left tail distributions of returns imply a greater downturn risk which makes losses more likely. These left tails lead to so-called non-diversification traps, in which case the value of diversification of this type of risk is decreased, as discussed before. The loss of a diversification trap according to Ibragimov et al. (2009) is high. The direct welfare loss of a trap in California residential earthquake insurance for example is approximately 3 billion USD per year. Therefore, I assume that insurance stocks decrease after the occurrence of a terrorist event. The implications for stocks of industry 1-3 suggest a

(14)

9

negative relationship between terror events and the abnormal and cumulative abnormal returns. Hence, the first three hypotheses are:

H1: Domestic terror events cause significant negative abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French airline industry during the event window.

H2: Domestic terror events cause significant negative abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French leisure & tourism industry during the event window.

H3: Domestic terror events cause significant negative abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French insurance industry during the event window.

For the industries 4-6, which are (4) oil & gas, (5) media and (6) defence there are reasons that the stock returns would respond positively towards terrorist events. As mentioned in the literature section Chesney et al. (2011) find evidence about a positive relationship between terrorist events and stock returns of oil & gas stocks. They explain this positive effect by the crude oil prices, since successful terror attacks could increase the hypothetical chance of an attack on oil production facilities, which again would boost up the traded crude oil price. Thus, I believe that terror events lead to significant positive abnormal and cumulative abnormal returns for stocks of the oil & gas industry.

I also believe that stocks from media companies will react positively, since it seems intuitive to me that the demand of information and broadcast increase during times of terrorist events. However, there is no literature dealing with the effect of terrorist events on media firms or media stock returns.

In contrast to media stocks there is evidence that terrorism often increases expenditures of homeland security and military goods according to the research of Mueller and Stewart (2014). I assume that higher spending in the military sector and home security would increase the turnover of defence firms and would lead to an increase of stock returns consequently. In other words, terror events lead to significant positive abnormal and cumulative abnormal returns for stocks of the defence industry. The implications for stocks of industry 4-6 suggest a positive relationship between terror events and the resulting abnormal and cumulative abnormal returns. Hence, the next three hypotheses are:

(15)

10

H4: Domestic terror events cause significant positive abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French oil & gas industry during the event window.

H5: Domestic terror events cause significant positive abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French media industry during the event window.

H6: Domestic terror events cause significant positive abnormal returns or/and negative cumulative abnormal returns for stocks associated to the French defence industry during the event window.

(16)

11

4. Methodology

The methodology part aims to explain the underlying econometric approach and established model. In the first part of this section the event study approach is explained in a general sense as well as on this particular research. In the second part of this section the market model is illustrated and all relevant equations of the economic model are presented.

4.1 The Event Study Approach

The event study approach is an empirical method to examine, whether and how certain events, such as share repurchase programs, earning announcements or mergers and acquisitions influence the value of companies or their securities (Mitchell and Netter, 1994). The impact of events can only be exposed by comparing observable or actual returns with expected returns on a specific date. Latter can be estimated on the basis of historical data. The Theory of Efficient Markets was developed by Eugene Fama (1970) and builds up the theoretical background of this method. The Efficient Market Hypothesis (EHM) assumes that prices of securities reflect all available information immediately. A change in security prices can be clearly attributed to the considered event, if the EMH is accepted and the absence of confounding events is affirmed. Thus, the new security price can be interpreted as a result of its re-evaluation taking all new information into account (Fama, 1970). The application of the event study method can be split up in eight steps (Campbell et al, 1997).

• Define the Event (1)

• Identify the exact event date (2)

• Remove events that became apparent with other relevant incidents at the same time (3)

• Collect and obtain historical data (4)

• Determine the model to estimate the expected returns (5)

• Set up the time frame for the estimation and event window (6)

• Compute abnormal returns (7)

• Test for statistical significance (8)

First it is important to define the type of event (1) that is of interest. It should be clear which event is considered to cause abnormal returns. For the aim of this research only terrorist attacks in France from 1986 until 2016 will be used and tested for inducing excess returns. Terrorism is defined as the premeditated use or threat of violence by individuals or

(17)

12

groups to achieve political and social aims by the intimidation of a large audience (Enders and Sandler, 2014). Based on this definition for terror eight relevant events with exact dates can be extracted (2), which will be demonstrated in the following section.

Removing events that became apparent with other relevant incidents at the same time is essential for the sake of the fit’s goodness (3). It is insufficient to assume that terrorist attacks were the only drivers for excess returns on a certain day, because other news may have had an impact as well. Omitting these variables would lead to biased estimators due to confounding effects and all the variation would end up in the error term (Greene, 1993). Controlling for macroeconomic variables would be a solution but there are there is a lack of data on a daily frequency. Furthermore the literature is inconclusive on the sign of these relationships. For this research, it is assumed that when terrorist planning an attack, they do not take macroeconomic conditions into account. As a result omitted variable bias in not expected in this empirical work.

Furthermore, there could not be found any other mentionable occurrences than the examined terror events for causing the effects on the relevant dates. Other financial event studies have not controlled for this either.

To estimate the daily expected returns for each industry daily stock prices of all the French stocks are used. They are drawn from the Database WRDS, that provides information about historical stock prices (4). The data sample about France’s terror events from 1986 to 2016 contains eight events and is obtained from the global terror database (GTD). Moreover, the overall selection of these eight terror events required the number of casualties to be ten in total at least.

Excess returns can be estimated by subtracting expected returns from actually realized returns within the event window. To estimate the expected returns several methods are available. The market model approach, which is basically a regression analysis constructed of historical data of stock prices and a corresponding stock index, seems to be the most common technique (MacKinlay, 1997). Following the literature (Drakos, 2004) for this particular

one-factor-model the S&P 500 Index has been selected, which serves as a proxy variable for the

market portfolio (5).

The estimation procedure can be described as setting up the framework for the event study. Here the parameters for the linear regression model (market model approach) have to be defined. Moreover, it is essential to determine an estimation window, which means to set up a time frame before the actual event occurs. For the estimation window data of daily returns from 252 days’ prior the event window will be taken into account. The event window

(18)

13

happens to be 21 days in total, namely 10 days before the actual event and 11 days after it (6). The reason why the event window does not exceed the estimation window is based on a simple explanation. Returns are expected to be abnormal during the event window and including them into the estimation window would lead to distorted estimators for the expected normal returns. As can be seen in the figure below the event window is followed by the post-event window, which is limited to be 10 days in total.

Figure 1: Event study timeline

The objective of interest contains the estimation and calculation of expected and excess returns. Based on data from the estimation window (from and ) and applying the market model approach the expected daily returns for each industry will be estimated. The time period within the potential impact of a certain terrorist event is expected, is called event window ( − ). To compute daily abnormal returns (AR) expected daily returns have to be subtracted from actual daily returns. Latter can be observed on the French stock market. It should be noted that the difference between actual and expected returns will be zero, in case the terrorist attack did not occur. If the impact is supposed to extend over several trading days, the total impact can be determined by computing cumulative abnormal returns ( )over many days within the event window. can be computed by summing up over a certain yield of time (7).

Applying a standard t-test (8) will give information about the existence of statistical significance for the obtained results (Brown and Warner, 1985). There are also other parametric and nonparametric methods, which can be used for this purpose such as the Standardized Residual Test (Patell, 1976) and the Generalized Sign Test (Cowan, 1992). The common aim of all these tests is basically to check out the validity of the underlying null hypothesis = 0, which sets the effects of and equal to zero.

(19)

14 4.2 The Market Model

The market model approach is based on the assumption of a constant and linear relationship between individual asset returns and the return of a market index1.

=∝ + , + , ℎ _, = 0 _, = _!#_"

The notation represents the expected returns of the observed stock on date . As previously mentioned eight relevant event dates could be identified and 240 stocks out of 26 different industries were observed (section 5). For the sake of this particular research only six industries will be analysed, which are believed to be affected significantly. The returns of the

Standard & Poor’s 500 Index are utilised to approximate the theoretical assumption of a

perfect market portfolio. The market model parameters are ∝ (intercept), (slope) and , = !#"(standard error), where ∝ stands for the daily realised return of stock . The

actual daily returns for all the stocks and the S&P 500 are computed on the basis of the logarithmic return formula,

= $ (_%%, , & )

where %_, is the price of stock and %_{, &} the price of stock one trading-day before (Brown and Warner,1980). The model parameters are estimated by ordinary least squares regressions and based on estimation window observations. To gain unbiased estimators the

Gauss-Markov assumptions have to be met and will be explained in the following. The first

assumption requires linearity in the parameters alpha and beta. Since the dependent variable , is a linear function of one independent variable , and a random error component , the first assumption is satisfied. Further it is implied that the expected value of the error term equals zero for all observations. Thirdly, the conditional variance of the error term must be constant for each stock over time. Error terms are regarded to be homoscedastic as confounding events were removed from the data sample. Moreover, the error term is assumed to be identically and independently distributed, because stock prices are supposed to follow a random walk as well (Fama, 1965). The expected returns for each stock are unconditional on the event but conditional on a separate information set. This fact will also preserve the parameters∝ and from distortion.

1

(20)

15

Abnormal Returns are the crucial measure to assess the impact of an event. The general idea of this measure is to isolate the effect of the event from other general market movements. The abnormal return of firm and event date is defined as the difference of the realized return and the expected return given the absence of the event.

, = − ( , )

By CAR across time yields aggregated abnormal returns can be measured,

= ' ,

()

*(+

where indicates the starting point of the event window and _#indicates two, five and ten days after the event day( =0) in this particular analysis. To obtain ,,,,,,, we can aggregate the

across industries for each event date. For n industries we have:

,,,,,, ( , #) =1' ( , #) .

*

whereby we use the assumption that the event windows for each event do not overlap to set the covariance terms to zero. For testing whether an event has a significant impact on stock returns or not, we use the simple -test. The t-statistics basically test the significance of the economic impact through normal variation:

− /0 = ,,,,,, ( , #)

1 2,,,,,, ( , #)3~5(0,1)

Under the null hypothesis, the cumulative average abnormal return is equal to zero. The variance estimator of this statistic is based on the cross-section of abnormal returns.

2,,,,,, ( , #)3 = #( , #) =₅1_#' #( , #) 6

*

Brown and Warner (1980) show that the cross-sectional t-test is robust to an event-induced variance increase. However, Boehmer, Musumeci and Poulsen (1991) provide evidence that their standardized cross-sectional test (requiring an estimation window) exhibits a comparable size, but is more powerful.2

2

(21)

16

5. Data

The aim of this section is to give the reader a feeling for the data by describing the source and illustrating the descriptive statistics in tabular form. In order to select the correct event dates a definition of the term terrorism is required. As mentioned in the previous section for this thesis the widely used definition of Enders and Sandler (2014) is chosen and supposed to be as follows: “terrorism is the premeditated use or threat to use violence by individuals or sub-national groups to obtain a political or social objective through the intimidation of a large audience beyond that of the immediate non-combatant victims”. To collect the relevant event dates the Global Terrorism Database (GTD) is used. This database is provided by the National Consortium for the Study of Terrorism and Responses to Terrorism at the University of Maryland and is the largest publicly accessible database. The GTD includes more than 150,000 entries of global terrorist attacks from 1970 onward. The database gives detailed information about date, location, number of casualties, perpetrator group and type of attack. Additionally, the database allows filtering the data for specific regions and also for time frames. For this particular topic all relevant information could be extracted from this database. For France 75 terrorist events could be observed in the generated sample, which had occurred in the time period from 1986 until 2016. In order to capture only the relevant terrorist events, the filter for casualties was set up to the number of at least ten in total. By applying this restriction to the data eight events could be obtained as presented in table 1.

Table 1: Major terror events in France from 1986-2016

Daily stock data and index data from different French industries are generated from the

Event Date Description Location Fatalities Injuries

1 17.09.1986 Bomb attack at the Tati store on rue de Rennes Paris 7 55

2 24.12.1994 Hijacking of Air France Flight 8969 Marseille 7 25

3 25.07.1995 Bomb attack at the Paris regional train network Paris 8 140

4 03.12.1996 Bomb attack at Port Royal metro station Paris 4 170

5 19.03.2012 Shootout at the Jewish school Toulouse 7 5

6 07.01.2015 Armed assault attack against Charlie Hebdo Paris 12 11

7 13.11.2015 Mass shooting and bombing in multiply locations Paris 137 352

8 14.07.2016 Truck attack including shootout Nice 86 303

(22)

17

Wharton Research Data Services (WRDS) database. Within the WRDS the database

Compustat contains data over 99,000 global securities and indices. By filtering out all French

securities from 1986 until 2016 this paper examines 240 securities, which make 1,285,244 daily return observations in total. Since this thesis aims to find the response of daily returns on terrorist events among specific French industries, it is necessary to set these 240 French securities up into various industrial groups. Therefore, these securities were subsequently categorized into 26 industries using the Global Industry Classification Standard (GICS), which is a widely used international norm to categorize different firm sectors. In this research the focus is on six out of the 26 industries, which are believed to response towards terrorist events, as mentioned in section 3.

Table 2: Descriptive statistics

Name Mean Min. Max Stand.Dev. Kurtosis Skewness Observ.

*Airlines 0.041% -26.44% 20.52% 3.94% 37.70 1.57 7,480 Automobiles 0.055% -19.17% 22.69% 2.42% 8.52 0.42 42,970 Banks 0.025% -39.77% 41.57% 2.37% 22.92 0.50 91,354 Beverage/Tobacco 0.042% -18.89% 17.75% 2.05% 6.62 0.04 63,765 Capital Goods 0.049% -21.11% 25.63% 2.62% 24.25 1.11 219,398 Commercial Services 0.066% -20.85% 18.50% 2.12% 6.15 0.21 20,252 Consumer Durables 0.015% -23.20% 32.23% 2.71% 27.75 -0.33 51,986 *Defence 0.050% -11.86% 32.64% 2.54% 24.64 1.13 115,667 Diversified Financials 0.024% -33.77% 22.45% 2.22% 13.44 -0.36 55,800 Energy (others) 0.046% -15.94% 34.14% 2.52% 11.95 0.13 6,509 Food 0.031% -12.76% 23.46% 2.02% 13.50 -0.07 30,811 Health Care 0.068% -25.93% 18.83% 2.42% 22.65 1.22 13,664 Household Products 0.093% -18.24% 20.40% 2.15% 20.70 1.06 9,177 *Insurance 0.027% -47.25% 20.98% 2.36% 13.98 -0.04 35,804 *Leisure/Tourism 0.031% -19.83% 32.43% 2.53% 12.61 0.61 39,604 Materials 0.045% -19.75% 46.25% 2.55% 16.87 0.79 146,564 *Media 0.041% -11.52% 27.14% 2.08% 9.95 0.24 48,909 *Oil/Gas 0.055% -29.95% 29.79% 2.39% 11.38 0.03 56,437 Pharmaceuticals 0.063% -24.79% 33.67% 2.51% 15.83 0.74 22,718 Real Estate 0.033% -37.28% 47.14% 2.03% 42.41 0.58 76,443 Retailing 0.094% -16.34% 21.48% 2.36% 8.83 0.30 18,051 Software/Services 0.029% -52.75% 42.86% 2.99% 28.59 0.34 42,832 Technology 0.044% -36.90% 32.69% 3.05% 10.90 0.18 20,575 Telecommunication 0.027% -19.24% 35.23% 2.85% 12.01 0.31 10,112 Transportation 0.057% -24.65% 21.01% 2.95% 56.49 1.87 19,017 Utilities 0.050% -26.72% 25.70% 2.44% 17.31 0.10 19,345

(23)

18

After the procedure of collecting and preparing the data is explained accurately, the descriptive statistics listed in table 2 will be described. The daily average stock return estimates based on data from 1986 until 2016 reported in the first column vary between 0.015% and 0.094% for the various industries. This indicates that all 26 French stocks have been experiencing an average positive return over the last 20 years. Their expected daily returns are-between 0,025% and 0,055%. Their standard deviation varies mainly from 2% to3% except of the airline industry, which has the highest standard deviation of 4%. In the fifth column the values for the kurtosis are listed and they are all far above three. A positive excess kurtosis is important, when examining historical returns of a stock or portfolio, for example. The higher the kurtosis coefficient is above the normal level (Kurtosis > 3), the more likely it is that future returns will be either extremely large or extremely small. This result is of advantage for the underlying research question. As expected for the positive kurtosis to exist there must be more extreme values of returns, before and after the terrorist events occurs (Campbell et al, 1993).

(24)

19

6. Results

In this section the main results of this empirical work will be presented and interpreted. The aim is to link the obtained findings with the previously discussed literature from section 2. Furthermore, this section checks the validity of the six derived hypothesis from section 3. The results are presented in tables showing each of the eight events. Every table contains six industries ordered and constructed identically with eight rows and five columns illustrating the estimated abnormal and cumulative abnormal returns.

6.1 Tati store bombing

Table 3: Event study results of event 1

Table 3 presents the findings of the event “bomb attack at the Tati store on rue de Rennes” in

Paris at the 17th of September 1986. During this event, perpetrators placed a bomb in the city centre of Paris which killed seven and injured 55 people. At the first sight one can observe that the stocks of airline, leisure &tourism, insurances and defence have a negative abnormal return on the event day, while stocks of oil &gas and media show a positive abnormal return at the event day. Nevertheless, the abnormal and cumulative abnormal returns among all six industries during this event are not significant at all. The number of trading days for the six

(1) (2) (3) (4) (5)

Industry Event day AR 3-day CAR 6-day CAR 11-day CAR Days of rebound

1 Airlines -4.77% -0.82% -5.64% 1.18% 5 (-1.0658) (-0.1843) (-1.2604) (0.7917) 2 Leisure/Tourism -2.09% -1.27% -1.00% 1.13% 4 (-0.09919) (-0.3474) (-0.2733) (0.3098) 3 Insurance -2.25% 3.76% 2.90% 1.18% 1 (-0.7902) (1.3203) (1.0189) (0.4139) 4 Oil/Gas 1.55% 1.68% -2.06% -0.08% 5 (0.873) (0.9454) (-1.1617) (-0.0474) 5 Media 1.69% 2.61% 0.15% 0.02% 3 (1.0292) (0.5725) (0.0326) (0.9973) 6 Defence -4.65% -1.18% -1.08% 0.33% 5 (-0.9456) (-0.2405) (-0.2203) (0.0681)

(5) Number of trading days for the market index to return to pre-attack level. * Statistically significant at the 0.10 level.

** Statistically significant at the 0.05 level. *** Statistically significant at the 0.01 level.

(25)

20

industries varies between one and five days to return to the pre-attack level, which is relatively fast compared to other events as we are going to see further. Since the number of fatalities for this event is lower compared to the sample mean of 35 fatalities, the non-significant results seems to be plausible. Furthermore, the attack did not target any locations associated with the investigated industries such as airports, tourist hotels or oil refinery. For this event all six hypotheses can be rejected.

6.2 Air France Flight 8969 Hijack

Table 4 presents the findings of the event “Hijacking of Air France plane” at the 24th of

December 1994. The Flight 8969 was hijacked by four perpetrators from the Houari Boumedienne Airport in Algiers and landed at the Marseille Provence Airport. Three days later the rescue mission caused seven fatalities in total and 25 injured victims. Stocks of airline, leisure & tourism, insurances and oil & gas show a negative abnormal return on the event day, while stocks of defence and media show a positive abnormal return at the event day. As expected these results are not significant due to the relatively small number of casualties, except for the airline and leisure &tourism industry. With an abnormal return of -5.78% (row 1 column 1) airline stocks experienced the largest decline in excess returns at the

(1) (2) (3) (4) (5)

1 Airlines -5.87%* -3.57% -2.60% 0.53% 7 (-1.7799) (-0.9602) (-0.6983) (0.8863) 2 Leisure/Tourism -1.28% 1.34% 0.96% -0.32%* 1 (-0.9806) (0.5105) (0.366) (-1.6509) 3 Insurance -1.14% -4.15% -0.99% -2.64% 12 (-0.4048) (-1.4792) (-0.3524) (-0.939) 4 Oil/Gas -1.31% -0.12% 1.55% 1.42% 1 (-0.4785) (-1.2121) (0.7902) (0.7613) 5 Media 0.23% 0.40% 1.30% 0.47% 6 (0.0952) (0.3895) (0.5323) (0.1921) 6 Defense 0.11% -0.48% 0.19% 1.30% 4 (0.0689) (-1.07) (0.7819) (0.1081)

(26)

21

event day. This result is statistically significant at the 0.10 level. The leisure & tourism industry shows a significant cumulative abnormal return of -0.32% (row 2 column 4) at the 11th post-event day. The fact that only the airline and leisure & tourism industry show a significant responds seems to be quite intuitive taking the type of attack and location into consideration. The type of the attack contained the hijacking of a passenger plane of the largest French airline. Thus, it is most likely that this event would affect this particular airline stock immediately. The location in this event is an airport, which is linked to travellers in general.

A negative response of stocks associated with the leisure &tourism industry indicates that owner of these stocks tend to sell their shares at a higher amount after terror events. It could be plausible that stockholders expect less travellers visiting France after the hijacking of an Air France machine due to irrational sentiments. It can be concluded, that for this event the hypotheses 1 & 2 can be confirmed, while the remaining four hypotheses are rejected.

6.3 Paris Métro and RER bombings

Table 5 presents the findings of the event “Bomb attack at the regional train network” in Paris

at the 25th of July 1995. The self-constructed explosive device of the terror group

(1) (2) (3) (4) (5)

1 Airlines -2.64% -0.28% -2.46% 1.98% 4 (-0.4957) (-0.0528) (-0.4621) (0.3702) 2 Leisure/Tourism 0.13% 2.21% 1.33% 0.04% -(0.057) (0.9347) (0.5613) (0.018) 3 Insurance -0.63% -0.39% 0.94% 1.84% 2 (-0.2454) (-0.1511) (0.3683) (0.7202) 4 Oil/Gas -0.94% 0.66% 1.94% 0.78% 2 (-0.3785) (0.0786) (0.5669) (0.1756) 5 Media -0.49% -0.60% -1.76% -1.26% 1 (-0.1836) (-0.2244) (-0.6571) (-0.4708) 6 Defense -2.89% -0.74% 1.52% 1.46% 6 (-0.9891) (-1.2794) (0.519) (0.4979)

(27)

22

GroupeIslamiqueArmé (GIA) killed eight and injured 157 people in total. At the event day,

stocks of all industries listed negative abnormal returns except of the leisure &tourism industry, which had an abnormal return close to zero. Surprisingly all the results were not significant, although the number of casualties and especially the number of injured people were relatively high. One reason why this event does not cause any significant responses may be the historical background of this attack. The terror attack happened in the context of France's involvement in the Algerian civil war from 1991 until 2002. Therefore, this terror event was not completely unexpected since two smaller attacks had happened before - the assassination of two persons opposed by the GIA a month and a half before the actual event and the murder of a police officer 15 days later. Both attacks occurred before the investigated event and for which the GIA was suspected as being the perpetrating group. It may be possible that the market already anticipated such terror events due to the past tensions with the GIA. Consequently, all hypotheses are rejected for this event.

6.4 Paris Métro bombing 1996

Table 6 presents the findings of the event “Bomb attack at Port Royal metro station” in Paris

at the 3th of December 1996. In total, four people died and another 91 people got injured

(1) (2) (3) (4) (5)

1 Airlines -3.82% 0.31% -2.93% 0.15% 2 (-0.4023) (0.0228) (-0.1649) (0.0069) 2 Leisure/Tourism -1.85% -2.41% -1.32% -1.13% 9 (-1.0429) (-1.3594) (-0.7414) (-0.6362) 3 Insurance -0.52% -4.69%*** -4.76%* 2.38% 8 (-0.3397) (-3.0724) (-1.6674) (1.5579) 4 Oil/Gas 0.35% -1.98% -4.61% -5.93% 19 (0.1426) (-0.5675) (-0.9984) (-0.9815) 5 Media 0.52% 1.25% 1.94% 0.48% 2 (0.2442) (0.5885) (0.916) (0.2249) 6 Defence 0.98%** 1.89% 1.66% 2.67% 5 (1.9959) (0.7544) (0.6638) (1.0668)

(28)

23

during the explosion. As expected stocks of the airline, leisure & tourism and insurance industry jointly showed negative abnormal returns at the event day. While the results for the airline and leisure &tourism industry are not significant the insurance stocks show significant negative results in row 3, column 2 and column 3. The 3-day cumulative abnormal return for insurance stocks is -4.69% (row 3 column 2) with a high significance at the 0.01 level. Further, the 6-day cumulative abnormal return is -4.76% (row 3 column 3) and statistically significant at the 0.10 level. The significant negative cumulative abnormal returns for insurance stocks indicate that investors of insurance stocks tend to sell their shares after the event and also confirms hypothesis 3.

Like the stocks for airlines and leisure& tourism, the oil &gas stocks and media stocks do not show significant results. Therefore, hypothesis 1, hypothesis 2, hypothesis 4 and hypothesis 5 can be rejected for this event. In contrast hypothesis 6 can be confirmed for this event, since defence stocks show a positive abnormal return of +0.98% (row 6 column 1) at the event day with statistical significance at the 0.05 level. This indicates that the market may have expected higher demand in homeland security, which appears plausible after the third major terror act within three years.

6.5 Toulouse and Montauban shootings

(1) (2) (3) (4) (5)

1 Airlines 1.97% -0.74% -1.26% -3.30% 3 (0.487) (-0.1293) (-0.1663) (-0.3325) 2 Leisure/Tourism -2.76% -1.58% 3.33% 0.80% 5 (-0.8243) (-0.4721) (0.9951) (0.2395) 3 Insurance -1.38% -1.40% -0.93% -2.70% 1 (-0.3077) (-0.313) (-0.2073) (-0.6028) 4 Oil/Gas 1.56% 1.31% -1.04% -3.47% 7 (0.8791) (0.5212) (-0.3138) (-0.7978) 5 Media 0.46% 0.18% 1.01% -0.08% 7 (0.2054) (0.0805) (0.4493) (-0.0373) 6 Defence -0.51% 0.11% 0.95% 0.85% 4 (-0.9083) (0.5028) (0.7876) (0.8978)

(29)

24

Table 7 presents the findings of the event “Shootout at a Jewish school” in Toulouse at the

19th of March 2012. The assassin shot seven people and wounded five persons in total over a period of eleven days. The act of terror includes two sub-events with one fatality before and two fatalities after the main event (19th of March in 2012). The reason for choosing the 19th of March as the event date is justified by high death toll, since four people got shot at this event. Surprisingly each observed industry shows negative but also positive abnormal and cumulative abnormal returns during the event window. This indicates that the attacks did not push the stocks in one certain direction. This may be a reason why the event does not affect the stocks of the industries in any significant manner.

As discussed in the literature section the number of casualties caused by a terror attack is linked with the abnormal and cumulative abnormal returns (Eldor and Melnick, 2004). Since the terror event consisted of three sub-events, which again were distributed over eleven days the intensity of the event may have weakened. All six hypotheses can be rejected for this event.

6.6 Charlie Hebdo attack

(1) (2) (3) (4) (5)

1 Airlines -9.4%** -5.33% -1.76% -2.85% 4 (-2.6365) (-1.0547) (-0.264) (-0.3263) 2 Leisure/Tourism -0.24% -2.35% -1.94% -4.26%** 14 (-0.1439) (-1.4013) (-1.1576) (-2.53) 3 Insurance 2.60% -0.64% 1.12% 0.61% 1 (1.4624) (-0.3581) (0.6292) (0.3443) 4 Oil/Gas 3.16%* 2.21% 5.73% 1.40% 5 (1.7141) (0.8479) (1.66089) (0.7613) 5 Media 0.73% 2.17% 4.65%* 3.57% 14 (0.5077) (1.0661) (1.7287) (1.0145) 6 Defence 1.49% 3.06%* 8.52%*** 10.03%** >20 (0.8494) (1.7749) (2.5949) (2.3342)

(30)

25

Table 8 presents the findings of the event “Charlie Hebdo shooting” in Paris at the 7th of

January 2015. The armed attack by a group linked to al-Qaida caused 23 casualties, including 12 fatalities. When analysing the results of the Charlie Hebdo attacks it is crucial to know regarding a best possible interpretation of the results, that two days after this event a hostage-taking occurred in Paris. Both attacks were conducted by the same terror group. The following incident caused four fatalities and injured nine people. Since the perpetrators of both events were linked to each other and a gap of two days’ delay is relatively small, it seems reasonable for me to analyse these events together. Industries with significant negative abnormal and cumulative abnormal returns during these events are stocks from airline and leisure & tourism industry. Stocks of the French airline industry suffered the largest negative abnormal return at the event day of -9.40%, which is statistically significant at the 0.05 level. Consequently, hypothesis 1 can be confirmed for the Charlie Hebdo event. On average the number of trading days for the airline stocks to return to the pre-attack level were four (row 1 column 5), which shows that the airline stock prices recovered relatively quickly. This could be interpreted as an overreaction of the market (Farma, 1998).

While airline stocks could have experienced a negative overreaction and a quick recovery afterwards, leisure &tourism stocks did not show significant results shortly after the two events until ten days. The 11-day cumulative abnormal return is -4.26% (row 2 column 4) and highly significant. The lagged reaction of the leisure &tourism stocks also corresponded with the days of rebound, since the number of trading days for the leisure &tourism stocks to return to the pre-attack level lasted 14 days (row 2 column 5). Hypothesis 2 can also be confirmed for the Charlie Hebdo event.

Stocks with significant positive abnormal and cumulative abnormal returns during these events are stocks of the oil & gas, media and defence industry, which confirm the assumptions towards the reaction on the six industries from the hypotheses section. The stocks of oil &gas show a positive significant abnormal return of +3.16% (row 4 column 1) at the first event day, which indicates that this event might boost the purchase of oil &gas stocks shortly after the first event. The stocks of the media industry had a positive significant 6-day cumulative abnormal return of +4.65% (row 5 column 3) at a 0.10 level. It seems logical that the huge public interest for the Charlie Hebdo attack caused a higher demand for media content, such as print media's, online web pages, televisions and radio broadcasts after the event had occurred. This may explain why the media stocks experienced positive abnormal returns for two weeks, which also corresponded with the day of rebound, namely 14 days (row 5 column 5).

(31)

26

In terms of cumulative abnormal returns, the stocks from the defence industry are the biggest winner out of the six examined industries during this event. They show significant results for the cumulative abnormal returns of +3.06% (row 6 column 2), +8.52% (row 6 column 3) and +10.03% (row 6 column 4). The number of trading days for the defence stocks to return to the pre-attack level was more than 20 days (row 6 column 5). The reason for these enormous positive responses of defence stocks could be explained by the political consequences caused by this attack. After this attack the President of France announced the war against terror and a higher international engagement against terrorist groups. It can be possible that the market associated higher future expenditures in home security and armaments goods and thus higher revenues for defence firms. The only industries without any significant respond are stocks associated with the insurance industry. Therefore, hypothesis 3 is the only rejected hypothesis for the Charlie Hebdo events.

6.7 Paris November attacks

Table 9 presents the findings of the event “Paris mass shooting and bombing” at the 13th of

November 2015, which is the greatest terror attack in the history of France in terms of casualties. The attacks contained multiple suicide bombings and shootings in six different

(1) (2) (3) (4) (5)

1 Airlines -5.22%* -0.11% -6.53%** -0.13% 4 (-1.9212) (-0.0408) (-2.4016) (-0.0474) 2 Leisure/Tourism -4.3%* -3.80% -6.05%** -3.26% 14 (-1.7888) (-1.4874) (-2.5291) (1.3619) 3 Insurance 0.78% -2.15% -3.18% -0.31% 1 (0.3499) (-0.9592) (-0.7677) (-0.1386) 4 Oil/Gas 3.23% 2.12% 0.62% 1.45% 13 (1.4369) (0.6687) (0.147) (1.0932) 5 Media 1.06% 1.61% 0.47% -0.68% 6 (0.5426) (0.8244) (0.2429) (-0.3498) 6 Defence 5.92%*** 1.04% 1.18% -0.57% 14 (2.907) (0.5083) (0.5778) (-0.281)

(32)

27

locations in Paris with 137 fatalities and 352 injuries in total conducted by the terror group

ISIL. Before getting into the analysis it is worth mentioning that this event occurred on a

Friday evening, so stock markets were closed during this event. Because stock markets stay closed during weekends and public holidays in general, for this specific case Monday the 16th is set to be the event date.

Industries with significant negative abnormal and cumulative abnormal returns during these events are stocks from airline and leisure &tourism industry. Similar to the previous events, airline stocks again show significant negative results, namely an abnormal return of -5.22% (row 1 column 1) after the event day and a cumulative abnormal return of -6.53% (row 1 column 3) five days after. As observed for the other events, stocks of the airline industry show the largest negative response towards terrorist events out of all six industries. These negative results regarding airline stocks correspond with hypothesis 1.

For the leisure &tourism industry similar results as for the airline industry can be obtained. Stocks associated with the leisure & tourism industry show an abnormal return of -4.30% (row 2 column 1) at the event day and a cumulative abnormal return of 6.05% (row 2 column 3), confirming hypothesis 2. The results for the airline and leisure & tourism industry are negative as anticipated for this event, which seems to be plausible when looking at the huge number of casualties. In contrast to the airline and leisure & tourism industry, stocks associated with insurance companies do not show any significant results, although most of the excess returns are negative. Therefore, hypothesis 3 can be rejected for this event.

Hypothesises 4 and 5 can be rejected for this event as well, since for stocks associated with the oil & gas and media industry no significant positive excess returns are obtained. However, hypothesis 6 can be confirmed in this event due to an abnormal return of +5.92% at the event day, which is also statistically significant at the 0.01 level.

(33)

28 6.8 Truck attack in Nice

Table 10 presents the findings of the event “Truck attack with shootout” in Nice at the 14th of

July 2016. The attack caused 389 casualties in total, including 86 fatalities and was conducted by a single perpetrator. As observed in all the previous events before, stocks associated with the airline industry show negative abnormal returns. But none of the events caused such an intensive impact towards airline stocks, then the truck attack in Nice. Stocks associated with the airline industry show overall negative excess returns during the event window. The cumulative abnormal returns from the second until the tenth day after the event reach from -7.81% to almost -15% and all the results are statistically significant. With more than 20 trading days for airline stocks to return to the pre-attack level the truck attack caused the most extreme results in this entire research. These extreme negative CARs for the airline industry might be explained due to the high amount of foreign tourists among the victims, because 31 out of 84 killed3victims were travellers. That could lead to less travelling to France in the future, what is likely to cause a decline of the turnover in the airline industry. These implications also correspond with the reaction of stocks associated with the leisure & tourism

3

29.07.2016 from the news provider “theatlantic”: http://www.theatlantic.com/news/archive/2016/07/victims-nice-attack/491480/

(1) (2) (3) (4) (5)

1 Airlines -2.16% -7.81%* -13.71%** -14.99%* >20 (-0.6489) (-1.6515) (-2.1915) (-1.8304) 2 Leisure/Tourism -2.79%* -2.30% -2.39% -3.75% 5 (-1.8326) (-0.4978) (-0.3903) (-0.4681) 3 Insurance -0.80% -0.02% -1.61% -1.24% 3 (-0.2802) (-0.0075) (-0.5639) (-0.1731) 4 Oil/Gas -1.61% -1.38% -4.45% -4.77% 4 (-0.5361) (-0.3258) (-0.7913) (-0.6487) 5 Media -0.24% -0.16% 1.86% -0.05% 4 (-0.106) (-0.0723) (0.8345) (-0.0206) 6 Defence 1.57% 0.32% 1.68%** 1.29% 8 (0.1345) (0.143) (2.1019) (0.5812)