The effect of terror attacks on German
stock market returns
Author: Jos Peters - s2701634
Supervisor: Prof. R.E. Wessels
Abstract
This paper studies the effect of terrorist attacks on returns in the German stock market, and its different sectors. Using data on stock returns of different industries in Germany, both parametric and non-parametric models are constructed. Event study methods are extended with GARCH and EGARCH models to obtain more precise results. Furthermore, the reactions of the capital market on the German in-surance industry are examined in more detail. Findings show that major attacks in the beginning of the millennium resulted in significant negative abnormal returns in the German stock market. In contrast, recent attacks did not have significant im-pact on the stock returns. The banking sector seems to have suffered less after the Madrid, London and Berlin attacks, but in contrast to findings in the UK and Spain these effects were not significant. Stock market returns in the insurance industry took a significant large hit after the Brussels attack, possibly due to large claims. However, investors did not alter their beliefs in the ability of insurers to diversify terrorist risk.
University of Groningen MSc Finance Thesis
Contents
1. Introduction 2
2. Review of the literature 4
3. Methods 10
4. Data 16
5. Results 20
6. Conclusion 25
1. Introduction
On a Tuesday in September, the New York stock market did not open (Davis, 2017). Two planes just hit the World Trade Center, leaving the city in chaos. The New York Stock Exchange (NYSE) did not open until the next Monday (Krasny, 2018), resulting in the longest shutdown since the second world war (Davis, 2017). Nevertheless, the NYSE fell over 7% on the first trading day after reopening, starting a week of historical losses (Davis, 2017).
The 9/11 attacks were by far the deadliest attacks worldwide (Johnston, 2018). How-ever, in the decade following these attacks, several other western countries have also been hit by major attacks. Examples include the Madrid train bombings, London sui-cide bombings, and most recently attacks in Brussels, Paris and Berlin.
This paper studies the effect of terrorist attacks on the German stock market, for a variety of reasons. The German stock exchange has not yet been studied on this sub-ject, in contrast to markets in other countries such as Spain (Kollias, Papadamou, and Stagiannis, 2011), Greece (Liargovas and Repousis, 2010), Malaysia (Ramiah, 2012) and the US (Chen and Siems, 2004). The German market is an interesting addition to the long list of economies, as it has some attractive properties.
Katzenstein (1993) argues that the loss of World War II made Germany afraid to be involved in military situations. However, after 9/11, the danger of terrorist attacks in-creased in Germany. Nevertheless, it was difficult to increase security measures, due to the skeptical attitude of Germany towards police interventions. Finally, after a direct threat from al-Qaida, the government managed to introduced two counter-terrorism packages. Katzenstein (1993) argues that Germany since then has become leading in enhancing and widening police cooperation in Europe.
The great impact of terrorist attacks on society and the stock market after 9/11, com-bined with the interesting properties of the German market have let us to study the following hypothesis:
1) Did the individual attacks have any impact on the market value of the German stock market? 2) How long did these effects last?
3) Did the stock prices of the different sectors in Germany react differently to attacks? 4) Does the stock market reaction after attacks change over time?
2. Review of the literature
This section will start with a review of general event study literature. Thereafter, the applications of similar techniques on non-corporate action events will be discussed. In particular, the effects of terrorist attacks on stock market returns in various countries, split into different sectors. Finally, the differences between systematic and unsystem-atic risk will be discussed.
The Efficient Market Hypothesis (EMH) states that all available information is incor-porated in stock prices. Therefore, the share prices represent the fair value of the stock. Fama, Fisher, Jensen, and Roll (1969) decided to examine how stock prices react to new information. They developed a model that could be used to measure the impact of corporate actions on the value of a firm, under the assumption that the EMH is true (Binder, 1998). After the research by Fama et all. (1969), various adjustments have been made to this model. According to Khotari and Warner (2006), it is better to use daily data compared to monthly data, as it leads to more precise measurements. Mackindlay (1997) proposed two ways to calculate the abnormal returns: the constant mean return model, and the market model. The first method uses the statistical properties of re-turns to estimate expected rere-turns, while the latter uses market indices as benchmark for the estimation. De Jong, Kemna and Kloek (1992) argue that the assumptions of the market model are violated for data on stock returns. They made an adjustment using a Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model, that better fits the actual return data, and applied it on the Dutch stock market.
The event study methods have also been used to analyse the economic impact of ter-rorist events. Drakos (2004) discusses that a change in stock prices after an attack can be explained using the fundamental stock pricing formula. This formula explains how, in an efficient market, the price of an asset equals the expected value of the future cash flows. An unexpected event, such as an act of terror, could lead to new information about future cash flows and the likelihood of them being obtained. This will cause the stock price to change, after the new information is incorporated into it. Campbell, Lo and Mackinlay (1997) argue that the change in stock prices is a good information source for the impact of terrorism, as they show the expectations of investors about future profits. The depreciation of the shares give an indication for the devaluation of the firm as a consequence of the attack.
Krugman (2004) compares the societal cost of terrorism to that of crimes. These costs are divided into three categories. First, the direct losses as a result of the crime. In the case of terrorist attacks these can for example be the losses of properties as a result of bomb attacks. Second, the increased costs of law enforcement. After a terrorist attack, the costs associated with more anti-terror security measures could increase. Last, the costs of changing consumer behavior due to the concern for new crimes. As an exam-ple associated with the last categories, Krugman (2004) argues that demand for airline tickets decreases due to the longer waiting time as a result of stricter security. Mark-oulis and Katsikides (2018) further analyze the behavioral dimension to stock returns. Stock traders are human after all, which implies that they do not always follow ratio-nal behavior, but can be influenced by emotions. The panic after acts of terror could have impact on the stock market decisions, and thereby have negative impact on stock prices. All in all, there are four major direct and indirect ways in which terror attacks could have impact on stock prices.
specifically studied the impact of these attacks on the airline sector. As the returns in this sector are not independently and identically distributed, Carter (2004) decided to use a multivariate regression model, instead of standard event-study methods. The air-line industry was affected differently after 9/11 compared to other sectors, for example due to flight bans and bailout legislation. They found great negative abnormal returns for airlines, and smaller negative returns for air cargo firms. Karolyi and Martell (2010) studied the effect of terrorist attacks that were aimed at firms, on their stock price. They examined more than 70 attacks that took place between 1995 and 2002. The average market value of firms that were targeted equaled 58 billion dollars. By calculating the reduction in market capitalization after an attack, they find that on average firms lost over 400 million dollars in value per attack.
Br ¨uck and Wickstr ¨om (2004) stress the relevance of evaluating the economic conse-quences of terrorism. They summarized various studies about the financial reactions to terror, and concluded that these vary between industries and countries. Therefore, the next section will outline the results of studies aimed at particular countries.
increase in value during turbulent times, such as terror attacks (Chen, 2019). Ramiah (2012) studied the impact of terrorist attacks on the Malaysian stock market. Again, event study methods are used to calculate the abnormal returns after each of five ma-jor attacks, including 9/11, the Madrid bombings, and the London attack. In contrast to earlier literature, he did not find much impact after most of the attacks. In line with other research, the 9/11 attack did have impact, in all of the Malaysian industries. He suggests further research in other countries, to test the hypothesis that stock markets are responsive to terror attacks.
The most recent paper studies the impact of 11 attacks, starting with the 9/11 attacks and ending with the Westminster attack in the beginning of 2017 (Markoulis and Kat-sikides, 2018). They use an event study approach to obtain their results. Similar to Ramiah (2012), they find earlier attack to have less impact compared to more recent ones. They provide three explanations for this effect. First, the financial system became more stable in recent years. Furthermore, investors learned from previous attacks that the market overreacts after acts of terror. Therefore, they have learned how to assess the impact of attacks, and started to follow a calmer behavior. Finally, they suggest that investors have started pricing the risk of terror into stock prices. While the treats of terror kept being at a high level, the attacks were less unexpected. In line with Chen and Siems (2002), they find that the abnormal returns did not last long.
So far, studies have used event study methods to estimate the impact of attacks on the stock market returns. However, for investors, a perspective aimed at the riskiness of the stocks is more relevant. The risk of a stock can be divided into two categories. Systematic risk is not diversifiable, and therefore affects the whole market and not an individual firm. In contrast, unsystematic risk is risk that is specific to certain firms or sectors and can be eliminated through diversification (Chen, 2017). For the first risk, investors will ask a compensation, as they cannot diversify it. The height of this risk depends on the correlation between the stock and the market, which can be measured with beta. On the contrary, an investor can not expect any reward for bearing unsys-tematic risk as it is diversifiable.
with unsystematic risk. There will not be compensation for this risk, on the contrary, if the risk is high enough the investor could be tempted to insure himself again it. The subsequent premiums could result in negative returns. In contrast to the capital market, insurers are able to diversify the risk originated by terror attacks. Due to risk adversity, people are willing to pay a premium above the actuarial fair price in order to transfer the risk to the insurers, which will result in a positive return for the insur-ers. Hence, the capital market does not provide the mechanisms to insure one against terror risk.
Some authors have addressed this issue, by adding an analysis about the systematic and unsystematic risk. After analyzing the impact of 9/11 on the market value of air-line stocks, Drakos (2004) used a market model to differentiate between the systematic and unsystematic risk. He tested the hypothesis that the beta did not change, in which case the effect would be unsystematic. In addition, he used the volatility of the returns to measure the total risk of the stocks. By comparing the ratio of systematic risk to total risk, the difference between systematic and unsystematic risk became clear. He found a change in the betas, which indicate a change in systematic risk. Moreover, he also found an increase in unsystematic risk, which has large implications for portfolio managers as they now need to adjust their portfolio to diversify their risk.
Studies about the different effects on systematic and unsystematic risk do not stay lim-ited to the airline industry. After the event study, Ramiah (2012) also included a regres-sion to estimate the change in systematic risk of the Malaysian stock index. He used a CAPM model with a dummy variable for the terror attacks, to see how much the beta changed after the attacks. He finds that most industries have a significant change in systematic risk after 9/11, but this effect is not present after subsequent terror attacks. He suggests that investors have increased their risk factor after the first major attack, and kept their new risk level at that level as long as there are terror threats. The fol-lowing attacks therefore did not result in a new change, as this was already accounted for after 9/11.
3. Methods
First, the effect of attacks on German stock returns will be examined. As standard prac-tice in the literature, we will use event study methods to compare the actual returns after the attacks with the expected returns. To estimate the expected returns, we can choose between an economic and statistical approach. An economic approach uses a market portfolio as benchmark, to predict the returns in the case that the attack did not happen. This study could use a world index, as benchmark for returns in the German index. However, these indices are likely to be correlated, and could both be affected by terror attacks. Therefore, this method could lead to misleading results, and a statistical approach is preferred in this case.
This approach uses a strong statistical assumption, that is likely to be violated in our case. It states that the returns are jointly multivariate normal, and independently and identically distributed. First, we will test if this assumption is indeed violated for our data. A Jarque-Bera test is used to test if the returns of the DAX are normally dis-tributed. As the null hypothesis of a normal distribution is rejected at a significant level of <1%, we conclude that there is indeed strong evidence that the series is not normally distributed. Despite the violation of the assumption, the parametric test is still useful, as Mackinlay (1997) states that inferences using this model tend to be ro-bust to violations of this assumption. Therefore, we include it in this study to get an impression of the effect of the attacks on the German stock index. However, as a ro-bustness check we will also include a nonparametric test, which does not rely on the normality assumption imposed to the parametric model.
Constant mean model
The statistical model uses a constant mean return as a predication for future returns. This leads to the following equation for the return at time t, around attack i:
Rit =µi+eit (1)
Hence, µ is a constant, and e is the error term which is assumed to be normally dis-tributed. Although this model is extremely simplistic, Brown and Warner (1985) argue that the results following this model are often very close to those of more complex models. As suggested by Chen and Siems (2004) and Kollias, et all. (2011), the mean for attack i is calculated using an estimation window of twenty days. Hence,
µi = 1 20 −6
∑
t=−25 Rit (2)Now, the abnormal returns can be calculated as the difference between the actual re-turns and the estimated rere-turns following the constant mean model:
ARit =Rit−µi (3)
We can test the null hypothesis that the terrorist attack did not have impact on the returns. If that would be the case, the abnormal returns would have an expected value of zero, with variance σi2. To test if the effect lasted multiple days after the event, we make use of the cumulative abnormal returns (CAR), which equals the sum of the abnormal returns. The variance of the CAR depends on the length of the event window. To estimate the effect of approximately a week after the attack, we use a 5-day event window to calculate the CAR. This will result in a cumulative standard deviation of 5 times the individual standard error, according to Mackinlay (1997).
Nonparametric model
Ti =
(rank(Ri0) −125, 5) S(rank(Ri))
) (4)
The standard deviation, denoted by S(rank(AR))above, is given by:
S(rank(Ri)) = v u u t 1 250 5
∑
t=−244 (rankit−125, 5)2 (5)GARCH market model
Once we have measured the impact of attacks on the German stock market as a whole, we can compare the reaction of different sectors using a market model. The returns of the various sectors can be compared with the market portfolio, represented by the DAX index. Again, we face the problem that the returns are not normally distributed. Therefore, the assumptions of the basic OLS regression proposed by Mackinlay (1997) are rejected. To get a more reliable estimation, we will adjust the market model to fit our data better, following de Jong, Kemna and Kloek (1992).
The return data of the DAX in the period since 2000 is shown in figure 1 below.
We observe a common characteristic in stock returns, that is, there exist periods of high and periods of low volatility. Therefore, we expect the data to not have a constant vari-ance. For statistical support of this presumption, we will carry out a modified Levene test. We can use this test, as it does not require the error term to be normally dis-tributed. In this test, we construct a group for each year of stock returns. We find that during some periods, such as the financial crisis in 2008, the variance is more than 3 times as high as in other years. The null hypothesis of the modified Levene test, which states that the variance is constant, is rejected at a significance level of<1%.
Fortunately, Batchelor and Orakcioglu (2003) argue that the Generalized Autoregres-sive Heteroskedasticity (GARCH) model is suitable for these kind of data. Abnormally high or low returns will in this model result in high expected variances. The most sim-ple model they propose, is the GARCH(1,1) model, as described in formula (6) and (7) below.
Rit =αi+Rmt+uit, uit∼ N(0, σt2) (6)
σt2 =α0+α1u2t−1+βσt2−1 (7)
Both equations will be estimated using the maximum likelihood technique. In equa-tion (6), the returns of sector i at time t (Rit) are predicted using an intercept a,
mar-ket return Rmt, and an error term. We assume the error term in (6) to be normally
However, many studies, such as Christie (1982), have shown that in financial time se-ries a negative shock will result in higher volatility changes than positive shocks of the same magnitude. Therefore, we will include a model which allows for asymmetric variance shocks, to see if it fits our data better. The EGARCH model, as designed by Nelson (1991), allows for such asymmetric shocks. The variance in this adjusted model includes similar terms as before. However, this time it also includes a term that allows for asymmetry. The new variance equation is given by:
ln(σt2) = ω+γzt−1+α(|zt−1| − r 2 π) +βln(σ 2 t−1) (8) where zt = eσtt.
α and β measure the same effects as in the previous model, however this time γ
mea-sures the asymmetry effect. As we find that γ < 0, in our case negative shocks will cause a higher increase in volatility than positive shocks. The latter model fits bet-ter for most of our data. However, this does not necessarily have to be the case for all sectors. In some cases the original GARCH model might give a better estimation. Therefore, we will compare the first with the second model for all the sectors by com-paring the log likelihoods. In the case that the log likelihood gives a preference to the original GARCH model, it could still be useful to use the EGARCH model in case the non-negativity constraint of the variance is violated. In the EGARCH model, the neg-ative variance is eliminated, since we are working with the logarithm of the expected variance. In the case of the metal and mining, real estate and telecommunication sec-tor, the GARCH model fits better and the non-negativity constraints are not violated. Therefore, we will stick to the original model for these sectors.
Regression analysis
Where Ritequals the return of sector i in time t. Rf t is the German short term interest
rate at time t, which is used as a proxy for the risk free rate. Rmt represent the market
return at time t, and uses the DAX index as benchmark. α and β can be estimated using the OLS technique, and e represents the error term.
To estimate the reaction of the market after the attacks, we include a dummy vari-able D, that takes on value 1 on the day of the attack, and 0 otherwise. The regression now looks like this:
Rit−Rf t =α+β1(Rmt−Rf t) +β2(Rmt−Rf t) ∗D+eit (10)
If the market correctly estimated the ability of the insurers to diversify the risk of terror,
β2should be insignificant. This would imply that the attack did not alter the perception
4. Data
The event-study requires a benchmark, which can be used as comparison to the actual returns after the event has occurred. For the constant mean model, we make use of the Deutscher Aktienindex (DAX). This is a capitalization-weighted stock index, consist-ing of the 30 major tradconsist-ing companies at the Frankfurt Stock Exchange. Data of this index is extracted from Thomson Reuters Eikon.
In this study, the following sectors are examined: banking, real estate, insurance, metal & mining and telecommunication. The real estate, and metal & mining sectors were selected, because they were also used in studies of similar countries, such as Spain, the UK and Malaysia, which makes them suitable for comparison. The banking, insurance and telecommunication sector were used, as they yielded contrasting results in simi-lar studies. Some industries, such as commodities, were not included due to a lack of data-availability in the German stock market.
11 September 2001, New York
3000 people were killed during the 9/11 attacks in New York, 6000 victims were injured (Plumer, 2013). Two hijacked planes flew into the Twin Towers, one airplane crashed into the Pentagon, and another plane crashed in a field (Huffpost, 2016). It was the deadliest attack in human history (Johnston, 2018), and had enormous impact on so-ciety. The attack happened around 10 AM local time. Meanwhile the German stock exchange, the Frankfurter Wertpapierb ¨orse, was still open. Therefore, we compare the closing price at 11 September with the opening price at that day to compute the abnor-mal returns. Responsibility was claimed by Islamic terrorist network Al-Qaeda (CBC News, 2004).
11 March 2004, Madrid
On 11 March 2004, multiple bombings took place on trains in Madrid. 191 people were killed, and around 2000 were injured (Peral Gutierrez de Ceballos, Turgano-Fuentes, Perez-Diaz, Sanz-Sanchez, Martin-Llorente and Guerrero-Sanz, 2005). The social im-pact was great, as the attack took place three days before Spanish political elections (Montalvo, 2011). Moreover, it was among the terrorist attacks in Europe (Johnston, 2018). Al-Qaeda again claimed responsibility, however in contrast to the 9/11 attacks no evidence has been found of their involvement (Nash, 2006). As the attack occurred in the morning at local German time, we again use the abnormal returns of that partic-ular trading day.
7 July 2005, London
13 November 2015, Paris
The next major attack happened a decade later in Paris. Suicide bombings outside a football stadium were followed by shootings at bars and restaurants in the city (CNN, 2018). The attack ended at a theater, where the attackers were eliminated. 130 people were killed, and 494 were injured (CNN, 2018). The attack started on a Friday evening after the German stock index had closed. Therefore, we use data from the first follow-ing tradfollow-ing day on Monday 16 November 2015. Responsibility was claimed by ISIS (CNN, 2018), an Islamic terrorist organization with as main goal the foundation of an Islamic caliphate (Elbaum, 2018).
22 March 2016, Brussels
Two suicide bombings happened at the Airport of Brussels, and one attack was con-ducted at a metro station in the center of the capital of Belgium. 32 people were killed, and 340 were injured (BBC, 2016). Responsibility was claimed by the same terrorist group that was responsible for the Paris attack, named ISIS (BBC, 2016). The attack happened just after the opening of the German stock index, which implies we can use that particular trading day for our results.
24 July 2016, Ansbach
Fortunately, this attack was not as deadly as the ones described above. The only person killed was the suicide bomber itself. 15 people were injured (The Guardian, 2016). Despite the low number of fatalities, this attack is interesting to study, as it was the first suicide bombing in the country due to Islamic terrorism (Troianovski and Buell (2016), S ¨uddeutsche Zeitung (2016)) The attacker was a refugee who was acting in the name of the Islamic State (The Guardian, 2016). The bombing happened on a Sunday evening. Therefore, we look at the excess return of the following Monday.
19 December 2016, Berlin
5. Results
This section shows the outcomes of the study. It starts with the results of the constant mean model, followed by the nonparametric model. Thereafter, the results of the mar-ket models are shown.
Constant mean model
Following the constant mean model, table 1 shows the abnormal returns and cumu-lative abnormal returns with an event window of 5 days, and their respective signif-icance levels. The attacks are listed in chronological order, starting with 9/11 in 2001 and ending with the Berlin attack in 2016.
Table 1: Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) of the DAX after terrorist attacks, following the constant mean model.
Attack AR CAR New York -8,27%*** -6,78% Madrid -3,53%*** -3,81% London -1,93%*** 1,08% Paris -0,87% -0,81% Brussels 0,01% -1,06% Ansbach 0,28% 0,75% Berlin 0,11% -0,74%
Note: Significance levels are indicated by: ***p<0.01, **p<0.05, *p<0.1
results. Moreover, the last two attacks were conducted in Germany, which could po-tentially have more social impact because they were domestic. The market reaction seems to be temporary after each attack, as none of the 5-day cumulative abnormal returns were statistically significant.
These findings appear to be in line with the literature, as there is much evidence that 9/11 had worldwide impact on stock markets. Moreover, other markets also show temporary effects, and reactions that are less abnormal after more recent attacks. Mark-oulis and Katsikides (2018) provide three possible explanations for the decreasing im-pact of terrorism on the stock market over time. Banking systems became more stable, investors learned how to assess the impact better, and they became used to the risk of terror.
Nonparametric model
The data did not satisfy the normality condition of the constant mean model. There-fore, the nonparametric test is included as a robustness check. The daily returns during the 250-day event windows were ranked from high to low. Table 2 shows the ranks of the returns on the day after the attack, in chronological order.
The New York attack resulted in the lowest stock return of the whole event window,
Table 2:Ranks of the abnormal returns of the DAX after terrorist attacks using a 250-day event window.
Attack Returns rank
New York 250* Madrid 246* London 246* Paris 136 Brussels 91 Ansbach 92 Berlin 101
Note: Significance levels are indicated by: ***p<0.01, **p<0.05, *p<0.1
attacks. These outcomes confirm the previous results, although they are less significant due to the method of testing.
GARCH market model
For each sector a GARCH or EGARCH model is used to estimate the returns after the attack. The difference between the estimated returns, and the actual returns is dis-played in table 3 below.
Table 3: The Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) of different sectors in Germany after terrorist attacks, with the DAX index as benchmark.
Banking Real estate Insurance Metal&Mining Telecommunication
Attack AR CAR AR CAR AR CAR AR CAR AR CAR
New York -0,46 -1,01 - - -1,47 0,85 - - - -Madrid 1,39 0,83 - - 0,43 -1,16 - - - -London 0,80 0,77 0,04 -1,80 -0,17 -0,17 - - - -Paris -0,47 -1,74 1,43 2,89 0,05 -2,11 0,55 1,21 0,38 0,11 Brussels -0,75 -5,76 -0,47 1,55 -2,11*** -3,45 -0,35 3,47 -0,78 0,04 Ansbach 0,06 -3,47 0,53 3,00 -0,18 -0,03 0,46 2,63 -0,43 -1,91 Berlin 1,52 -0,37 -0,31 0,90 -0,46 -0,93 0,21 0,64 0,23 0,25
Note: Significance levels are indicated by: ***p<0.01, **p<0.05, *p<0.1
banking sector. This can be attributed to the fact that they tested the returns on the do-mestic markets, which suffered from a higher social impact. The reason that the bank-ing sector suffered less, could be that this sector is relatively safe in turbulent times. This safe haven function is discussed by Liargovas and Repousis (2010) and included in the literature review. A good functioning financial sector could prevent chaos and harm to the economy after attacks (Johnston and Nedelescu, 2005). Therefore it is good to closely monitor the behavior in this sector after attacks, and take measurements to prevent chaos if necessary.
In the real estate sector, the Paris attack shows high positive abnormal returns, al-though these are also not significant at a 10% level. There are no similar results avail-able from other studies related to this attack, but the result is in contrast with most results of other attacks in different countries. In these countries, the real estate sector shows negative abnormal returns for other attacks. Only a terrorist attack in Mumbai had a positive impact on the real estate sector in Malaysia (Ramiah, 2012). A further study could be dedicated to this particular event and sector to find possible explana-tions.
The insurance sector is hit hard after the attacks in Brussels in 2016. The abnormal returns are lowest among all studied sectors and attacks, and the test statistic is highly significant. A possible explanation lies in the fact that the losses from this attack were covered by a special terrorism insurance fund (Veysey, 2016). Approximately 95 per-cent of insurers operating in Belgium were part of this insurance pool, which plausibly hits German insurers operating in Belgium directly. Besides this direct loss, the pool also puts a claim of 300 million on reinsurers across Europe. This way, German insur-ers could have suffered both directly and indirectly.
Regression analysis
The excess returns above the risk free rate in the insurance industry are used in the re-gression. Table 4 shows the results of the model, that estimates the impact of the event on risk in the insurance sector, as it is perceived by investors. The dummy variable indicates if there is a change in this risk after a terror attack.
Table 4: Results of the regression that estimates the impact of attacks on the riskiness of the German insurance sector. The β is represented by Coefficient. After the attack, the β equals the sum of Coefficient and Attack Dummy.
Attack Intercept Coefficient Attack Dummy
New York -0,23%*** 0,87%*** 0,04% Madrid -0,23%*** 0,87%*** 0,87% London -0,23%*** 0,87%*** 0,03% Paris -0,23%*** 0,87%*** 0,28% Brussels -0,23%*** 0,87%*** -1,91% Ansbach -0,23%*** 0,87%*** 0,02% Berlin -0,23%*** 0,87%*** -0,23%
Note: Significance levels are indicated by: ***p<0.01, **p<0.05, *p<0.1
6. Conclusion
This paper studies the effect of terrorist attacks on German stock market returns. The findings show that attacks are followed by significant negative abnormal returns in the beginning of the century. More recent attacks did not have a significant impact on stock market returns. None of the 5-day cumulative abnormal returns were signif-icant, therefore we believe that the effects were transitory. The insensitivity to recent attacks could be attributed to the fact that most of them were less deadly, and had less social impact than earlier attacks. However, recent attacks that were relatively deadly, also had insignificant results. Other explanations include a more stable financial sys-tem, calmer reactions of investors, and the incorporation of terror risk in stock prices. These findings are in line with previous literature in other countries, and further re-search could explore the effects in more sectors and countries. Some studies could be extended to larger regions such as continents, to prevent a lack of data availability. Although the current statistical methods yield useful results, their reliability could be improved. Multi-factor models could for example be used for more precise estimates of returns.
7. References
Batchelor, R., Orakcioglu, I., 2003. Event-related GARCH: the impact of stock divi-dends in Turkey. Applied Financial Economics 13.
BBC, 2005. London bomber: text in full. Retrieved from:
http://news.bbc.co.uk/2/hi/uk news/4206800.stm on 13 May 2019.
BBC, 2015. 7/7 London bombings: what happened on 7 July 2005? Retrieved from: https://www.bbc.co.uk/newsround/33401669 on 13 May 2019.
BBC, 2016. Brussels explosions: what we know about airport and metro attacks. Re-trieved from: https://www.bbc.com/news/world-europe-35869985 on 13 May 2019. Binder, J., 1998. The event study methodology since 1969. Review of Quantitative Finance and Accounting 11, 111-137.
Brown, S.J., Warner, J.B., 1985. Using daily stock returns: the case of event studies. Journal of Financial Economics 14, 3-31.
Br ¨uck, T., Wickstr ¨om, B., 2004. The economic consequences of terror: guest editors’ introduction. European Journal of Political Economy 20, 293-300.
Bryson, P.J., 2001. Economy and new economy in the United States and Germany. Intereconomics 36.
Campbell, J.Y., Lo, A.W., Mackinlay, A.C., 1997. The econometrics of financial markets. Princeton University Press.
Carter, D., Simkins, B. J., 2004. The markets reaction to unexpected, catastrophic events: the case of airline stock returns and the September 11th attacks. The Quarterly Review of Economics and Finance 44, 539-558.
CBC News, 2004. Bin Laden claims responsibility for 9/11. Retrieved from:
https://www.cbc.ca/news/world/bin-laden-claims-responsibility-for-9-11-1.513654 on 13 May 2019.
Christie, A. A., 1982. The stochastic behavior of common stock variances: value, lever-age and interest rate effects. Journal of Financial Economics 10, 407-432.
CNN, 2018. 2015 Paris Terror Attacks Fast Facts. Retrieved from:
Chen, J., 2017. Unsystematic risk. Investopedia. Retrieved from:
https://www.investopedia.com/terms/u/unsystematicrisk.asp on 13 May 2019. Chen, J., 2019. Safe haven. Investopedia. Retrieved from:
https://www.investopedia.com/terms/s/safe-haven.asp on 13 May 2019.
Connolly, K., Oltermann, P., 2016. Police investigate deadly Berlin truck crash as ’pre-sumed terrorist attack’. The Guardian. Retrieved from:
https://www.theguardian.com/world/2016/dec/19/berlin-truck-crashes-into-christ mas-market on 13 May 2019.
Corrado, C.J., 1989. A nonparametric test for abnormal security-price performance in event studies. Journal of Financial Economics 23, 385-395.
Davidson, W.N., Worrell, D.L., Dutia, D., 1993. The stock market effects of CEO suc-cession in bankrupt firms. Journal of Management 19.
Davis, M., 2017. How September affected the US stock market. Investopedia. Re-trieved from: https://www.investopedia.com/financial-edge/0911/how-september-11-affected-the-u.s.-stock-market.aspx on 13 May 2019.
Deutsche Welle, 2016. ’Islamic State’ claims responsibility for Berlin attack. Retrieved from: https://www.dw.com/en/islamic-state-claims-responsibility-for-berlin-attack/ a-36844638 on 13 May 2019.
De Jong, F., Kemna, A., Kloek, T., 1992. A contribution to event study methodology with an application to the Dutch stock market. Journal of Banking & Finance 16, 11-36.
De Santis, R.A., 2012. The euro area sovereign debt crisis: safe haven, credit rating agencies and the spread of the fever from Greece, Ireland and Portugal. Working Paper 1419. ECB, Frankfurt am Main, Germany.
Drakos, K., 2004. Terrorism-induced structural shifts in financial risk: airline stocks in the aftermath of the september 11th terror attacks 20.
Fama, E.F., Fisher, L., Jensen, Roll, R., 1969. The adjustment of stock prices to new information. International Economic Review 10.
The Guardian, 2016. Germany bomb blast: what we know about Ansbach attack. Re-trieved from: https://www.theguardian.com/world/2016/jul/25/germany-bomb-bl ast-what-we-know-so-far-ansbach-attack on 13 May 2019.
Hoggarth, G., Milne, A., Wood, G.E., 2001. Alternative routes to banking stability: a comparison of UK and German banking systems. Financial Competition, Risk and Accountability, 11-32. Anglo-German Foundation for the Study of Industrial Society. Palgrave Macmillan, London.
Huffpost, 2016. A minute-by-minute breakdown of what happened on 9/11. Retrieved from:
https://www.huffpost.com/entry/9-11-timeline n 57d300d8e4b06a74c9f48c09?gucc ounter=1&guce referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLm5sLw&guce referrer sig=A QAAAGnmfaCQ2Equge0D00hvwMU78qgDm8R7pQJG20MbLDjoEI2yJUYXHVxd17u Ss7AmrY11eLeCcE3dgLWahXy0-8riPC4CW3gUca5wJAOyuYJa6pInxRYS2R0mVaWU wrmqDUZuMowAlXYQYPqRFq2MnvUT1IUvoZQJBXT WvxCv4vz on 13 May 2019. Hofmann, F., 2017. Berlin Christmas market attack: We werent prepared. Deutsche Welle. Retrieved from: https://www.dw.com/en/berlin-christmas-market-attack-we-werent-prepared/a-41847714 on 13 May 2019.
Johnston, R., 2018. Worst terrorist strikes worldwide. Retrieved from:
http://www.johnstonsarchive.net/terrorism/wrjp255i.html on 13 May 2019
Johnston, R.B., Nedelescu, O.M., 2005. The impact of terrorism on financial markets. IMF Working Paper.
Karolyi, G.A., Martell, R., 2010. Terrorism and the stock market. International Review of Applied Financial Issues and Economics 2.
Katzenstein, P.J., 1993. Coping with terrorism: norms and internal security in Germany and Japan. Routledge, London.
Khotari, S.P., Warner, J.B., 2006. Econometrics of event studies. Handbook of Corporate Finance 1, 3-36.
Krasny, R., 2018. Trump says grasso opened NYSE the day after 9/11. He didnt. Bloomberg. Retrieved from: https://www.bloomberg.com/news/articles/2018-10-27/trump-says-grasso-opened-nyse-the-day-after-9-11-he-didn-t on 13 May 2019. Krugman, P., 2004. The costs of terrorism: what do we know? Princeton University, New Jersey.
Liargovas, P., Repousis, S., 2010. The impact of terrorism on Greek banks stocks: an event study. International Research Journal of Finance and Economics 51.
Liargovas, P., Repousis, S., 2011. The impact of mergers and acquisitions on the perfor-mance of the Greek bankings sector: an event study approach. International Journal of Economics and Finance 3.
Mackindlay, A.C., 1997. Event studies in economics and finance. Journal of Economic Literature 35, 13-39.
Markoulis, S., Katsikides, S., 2018. The effect of terrorism on stock markets: evidence from the 21st century. Terrorism and Political Violence. Routledge, Cyprus.
Montalvo, J.G., 2011. Voting after the bombings: a natural experiment on the effect of terrorist attacks on democratic elections. The Review of Economics and Statistics 93, 1146-1154.
Nash, E., 2006. Madrid bombers were inspired by Bin Laden address. The Indepen-dent. Retrieved from:
https://web.archive.org/web/20080706184144/
http://news.independent.co.uk/europe/article1961431.ece on 13 May 2019.
Nelson, B. D., 1991. Conditional hetroskedasticity in asset returns: a new approach. Econometrica 59, 347-370. The Econometric Society, New York.
Peral Gutierrez de Ceballos, J., Turgano-Furgano-Fuentes, F., Perez-Diaz, D., Sanz-Sanchez, M., Martin-Llorente, C., Guerrero-Sanz, J.E., 2005. 11 March 2004: The ter-rorist bomb explosions in Madrid, Spain an analysis of the logistics, injuries sustained and clinical management of casualties treated at the closest hospital. Critical Care 9, 104-111.
11/nine-facts-about-terrorism-in-the-united-states-since-911/?utm term=.5b9c5d91a8 f2 on 13 May 2019.
Ramiah, V., 2012. The impact of international terrorist attacks on the risk and return of malaysian equity portfolios. Review of Pacific Basin Financial Markets and Policies 15.
Ryan, M., Faiola, A., Mekhennet, S., 2015. Fear of new terrorist plots spreads; German stadium evacuated after threat. Washington Post. Retrieved from:
https://www.washingtonpost.com/world/france-launches-new-airstrikes-on-syria-and-more-anti-terrorism-raids-in-france/2015/11/17/4b84e27e-8cc9-11e5-934c-a369c 80822c2 story.html?utm term=.ed2830ab3bbc on 13 May 2019.
Schwert, G., 1981. Using financial data to measure effects of regulation. Journal of Law and Economics 24.
Siems., T.F., Chen., A.H., 2002. An empirical analysis of the capital markets response to cataclysmic events. Paper presented at the German Institute for Economic Research, Berlin.
Siems., T.F., Chen., A.H., 2004. The effects of terrorism on global capital markets. Eu-ropean journal of political economy 20. 349-366.
S ¨uddeutsche Zeitung, 2016. Erster Selbstmordanschlag in Deutschland - Bundesan-waltschaft bernimmt ermittlungen. Retrieved from:
https://www.sueddeutsche.de/bayern/anschlag-in-ansbach-erster-selbstmordan schlag-in-deutschland-bundesanwaltschaft-uebernimmt-ermittlungen-1.3094000 on 13 May 2019.
Troianovski, A. And Buell, T. (2016). Ansbach Bombing in Germany Believed to Be Islamist Terror Attack. The Wall Street Journal. Retrieved from:
https://www.wsj.com/articles/syrian-asylum-seeker-blows-himself-up-outside-concert-in-germany-1469447416 on 13 May 2019
Veysey, S., 2016. Brussels terrorism-related property losses covered by state-run pool. Business Insurance. Retrieved from:
