How vulnerable are international financial markets to terrorism? An empirical study based on terrorist incidents worldwide

How vulnerable are international financial markets to terrorism? An empirical study based on terrorist incidents worldwide

Accepted Manuscript Title: How Vulnerable Are International Financial Markets to Terrorism? An Empirical Study Based on Terrorist Incidents Worldwide ...

1MB Sizes 0 Downloads 74 Views

Accepted Manuscript Title: How Vulnerable Are International Financial Markets to Terrorism? An Empirical Study Based on Terrorist Incidents Worldwide Authors: Sanjay Goel, Seth Cagle, Hany Shawky PII: DOI: Reference:

S1572-3089(17)30073-6 https://doi.org/10.1016/j.jfs.2017.11.001 JFS 586

To appear in:

Journal of Financial Stability

Received date: Revised date: Accepted date:

26-1-2017 30-9-2017 1-11-2017

Please cite this article as: Goel, Sanjay, Cagle, Seth, Shawky, Hany, How Vulnerable Are International Financial Markets to Terrorism? An Empirical Study Based on Terrorist Incidents Worldwide.Journal of Financial Stability https://doi.org/10.1016/j.jfs.2017.11.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

How Vulnerable Are International Financial Markets to Terrorism? An Empirical Study Based on Terrorist Incidents Worldwide

SC RI PT

Sanjay Goel (Corresponding Author) School of Business University at Albany, State University of New York 1400 Washington Avenue, Albany, NY 12222 Phone: (518) 956 8337

U

Email: [email protected]

Phone: (845) 514-6508

A

15 Cornell Road, Latham, NY 12110

M

Health Information Exchange of New York

N

Seth Cagle

EP

Hany Shawky

TE

D

Email: [email protected]

School of Business

CC

University at Albany, State University of New York 1400 Washington Avenue, Albany, NY 12222

A

Phone: (518) 956 8337 Email: [email protected]

Impact of Terrorism on World Markets?

Abstract Each year, millions of dollars are reported lost due to terrorist attacks around the world. In this paper, we conduct a systematic examination of the impact of terrorist attacks on

SC RI PT

financial markets. We specifically explore the relationship between world stock market

indices and large-scale terrorist incidents. We consider sixteen incidents outside the U.S.

and thirty-three incidents in the U.S. We use several indices in our analysis, including the

S&P 500, the 10-year US Treasury bond yield, gold prices, and several other domestic and international stock market indices. Given the physical asset losses and the psychological

U

impact on citizens, our expectation was to see a strong correlation between terrorist

N

incidents and financial market valuations. However, to our surprise, with the exception of

A

the September 11, 2001 terrorist attacks, our results show that acts of terrorism do not have

M

a significant or lasting economic effect on stock and bond market returns. Contrary to our

D

starting hypothesis, we find no evidence of ‘flight to safety’ behavior in any of the markets.

Hypothesis

TE

Keywords: Terrorism, Event Studies, Financial Market Analysis, Efficient Market

CC

EP

JEL codes: C, G

1.0 INTRODUCTION Terrorism is a timeworn phenomenon that has plagued humanity for centuries. It not only causes

A

physical and human losses, but also carries with it a great psychological weight and impact. The goals of terrorists have always been the same: to cause damage, drain resources, disrupt services, instill fear, and most importantly, gain attention. With the instantaneous reporting of incidents through electronic media, the psychological impact of terrorist activities spreads outrage and anger more widely, traversing populations like never before. Research has shown that long-term repercussions of terrorism are serious;

2

Impact of Terrorism on World Markets? not only can terrorism affect humans psychologically but it also has important implications for the operation and performance of multinational firms, and the economic outlook of countries (Czinkota et al., 2010; Mobarek, Muradoglu, Mollah and Hou, 2016; Mester, 2017). The economic impact of terrorism is measured in billions of dollars annually, with the most in 2014 at 52.9 billon dollars, surpassing the loss in 2001 after

SC RI PT

the World Trade Center attacks.1

Great societal harm comes from a sense of fear and helplessness within communities, which erodes consumer confidence and consequently stunts economies (Reid, 2001), which can lead to a real slowdown in the economic development of nations (Blomberg et al., 2004; Nitsch & Schumacher, 2004; Abadie & Gardeazabal, 2003). Terrorism may also reduce a country’s ‘investment risk rating’, leading to a reduction

U

in foreign direct investment (Bekaert et al., 2014). Paradoxically, however, entrepreneurial activities

N

flourish under conditions of duress (Branzei & Abdelnour, 2010). Responding to external threats such as

A

terrorism in their operating environment is a complex challenge for multinational firms (Oetzel & Gets,

M

2012; Pragidis, Aielli, Chionis and Schizas, 2015). While there is widespread interest in the relationship between the global economy and terrorism or political conflict (Henisz et al., 2010), detailed studies that

D

quantify that impact are largely missing in the literature, a deficit that this paper addresses.

TE

Given this backdrop, a reasonable hypothesis may be that terrorism is likely to cause financial assets to decline in value due to negative sentiment and fear. Karolyi (2006) asserts that “prices of individual

EP

stocks reflect investors’ hopes and fears about the future […]” and that “investors often flee markets in search of safer financial instruments”. Already in 1984, Robert Shiller (1984) states that “mass psychology

CC

may well be the dominant cause of movements in the price of the aggregate stock market”, which further corroborates our conjecture that terrorism might negatively impact financial markets. To understand this

A

better, this paper first examines the impact of terrorism on stock market prices and on country and global equity and bond market indices. Second, it investigates whether terrorist incidents further exacerbate the negative impact on local economies by inducing a flight of capital to more stable instruments, or flights to

1

https://www.weforum.org/agenda/2015/11/what-is-the-economic-impact-of-terrorism/

3

Impact of Terrorism on World Markets? safer havens outside the country. As we analyze the results we will also reflect on associated conjectures, such as, if the rise in the number of terrorist incidents and excessive media coverage of these events has desensitized the public to these terrorist incidents (Eldor & Melnick, 2004), resulting in a somewhat muted impact on the markets, or if the impact of terrorist incidents is restricted in scope to only localized markets,

SC RI PT

with no bearing on the world economy.

The Efficient Market Hypothesis advanced by Fama (1970), posits that the stock market is informationally efficient, suggesting that all relevant economic information coming to the market is immediately and completely reflected on stock market prices. Consequently, stock market price changes subsequent to an event should reflect the true economic effects of that event, providing the basis for an

U

event study methodology. One area in which negative events have had a measurable impact on company

N

valuations is cyber security breaches. Goel and Shawky (2009) find evidence of significant effects on firms’

A

values from security breaches. We would expect the same kind of result from terrorist attacks. Eldor and

M

Melnick (2004), however, argue that financial markets can absorb the shock of terrorist attacks quite efficiently.2

D

Since the September 11, 2001 (9/11) terrorist attacks, many studies have examined the impact of

TE

terrorism on financial markets. For example, Karolyi and Martell (2006) examined firm-specific terrorism incidents occurring over an eight-year period, while Berrebi and Klor (2010) focus on Israeli companies

EP

and examine the particular impact of terrorism on the defense industry. Drakos (2010) takes a broader view and examines 22 country indices against the Global Terrorism Database (GTD). Kollias, et al. (2011) take

CC

a closer look at the Madrid and London bombing attacks, and Karolyi (2008) performs an assessment of an

A

investment portfolio aimed at countering terrorism-related risks. Given the fact that the United States possesses the most liquid equity markets around the world, it

is reasonable to examine the hypothesis that a ‘flight to safety’ behavior might be the result of terrorism in

2

There has been some criticism of the Efficient Market Hypothesis, primarily based on the cognitive biases of investors in valuing stocks such as confirmation bias, overreaction, negativity bias, anchoring effect, etc. (Nicholson, 1968; Basu, 1977; Rosenberg, 1985). Despite these criticisms, the Efficient Market Hypothesis remains a fundamental normative principle for understanding the influence of events (e.g. announcements, security breaches, terrorism etc.) on financial markets.

4

Impact of Terrorism on World Markets? some countries. In seeking ‘safe haven’ investments, international investors might seek US markets, which would in turn, result in significant inflows into S&P 500 stocks, gold, or 10-year U.S. Treasury securities. If, in fact, terrorism leads to ‘flight to safety’ behavior by foreign investors, we would find a rise in asset prices in U.S. markets, measurable either on the day of the event itself, or on the days immediately following

SC RI PT

severe terrorist incidents. Furthermore, if terrorism has a measurable negative effect on markets, we would expect that local country indices would decline in response to terrorism. To investigate whether or not this may be the case, we performed a series of event studies surrounding the onset of terrorist incidents that occurred both inside and outside of the continental United States.

In this paper, we use daily return data to conduct an event-study type methodology to estimate the

U

impact of terrorism on stock and bond markets. We use an exhaustive sample of terrorist incidents and

N

study their impact on a number of domestic and international market indices. We use two different indices

A

in our regressions to estimate expected returns: the MSCI World All Cap Index, and the MSCI World All

M

Cap Index Ex-U.S. We use both indices to control for any distortions that the U.S. equity market size might have on the rest of the World Index given the disproportionate share of the U.S.’s weighting in the world

D

index.

TE

The rest of this paper is organized in five sections: Section 2 presents a review of the literature; Section 3 describes the data sources; Section 4 presents our methodology; Section 5 provides a detailed

EP

analysis and presents our results; Section 6 presents follow-up studies that test our hypothesis; and Section

CC

7 presents our conclusions.

A

2.0 LITERATURE REVIEW Since the September 11, 2001 terrorist attacks on the World Trade Center in New York City,

research interest on the impact of terrorism on financial markets has substantially increased. While several researchers argue that there is a strong correlation between terrorism and financial markets, others suggest that the markets are generally unaffected by terrorist incidents. In this section, we review the extant literature.

5

Impact of Terrorism on World Markets? Chesney, et al. (2011), Hippler and Hassan (2015) and Silva, Kimura, Sobreiro (2017) examine the impact of terrorist attacks, financial crashes, and natural disasters on the behavior of stock, bond and commodity markets. Using an event-study methodology, a non-parametric approach, and a filtered GARCH–EVT model, they find that all industrial, regional, national, and global terrorist attacks have an

SC RI PT

effect on stock markets. Surprisingly, they report that the highest number of attacks affected the Swiss stock market, while the American stock market was affected by the least number of attacks compared to any other market. Airline and insurance sectors showed the highest susceptibility to terrorism, while the banking industry was the least sensitive.

Nikkinen and Vähämaa (2010) analyze the effect of three major terrorist incidents on stock market

U

sentiment (09/11/2001 in NY, DC, 03/11/2004 in Madrid, and 07/07/2005 in London). Focusing on the

N

“behavior of expected probability density function of the FTSE 100 index”, they find that terrorist attacks

A

have a very strong adverse effect on stock market sentiment. Procasky and Ujah (2016) were the first to

M

study the long-term effects of terrorist attacks on debt markets. Their study also tested how terrorist attacks exert different long-term effects on both developing and developed countries. Sedunov (2016) and

TE

of financial institutions.

D

Loveland (2016) and Ribeirto Blanc and Tabak (2016) examine the impact of financial crisis on the stability

Berrebi and Klor (2010) study the impact of terrorism on the Israeli defense industry. They use a

EP

matching procedure to pair each Israeli company with an American counterpart in order to isolate the effect of terrorism from other industry-wide shocks; 65 company pairs were constructed. Their terrorist incidence

CC

dataset spans the period from 1 January 1998 to 10 September 2001. Once they isolate the effect on the defense sector from that of other companies, they find that terrorist attacks have a significantly negative

A

effect on Israeli companies, but a positive effect on Israeli defense companies.3

3

Unfortunately, their study only spans three years, and in fact contains a significant portion of a recessionary period.

6

Impact of Terrorism on World Markets? Arin, et al. (2008) study the effect of terrorism on six countries; Indonesia, Israel, Spain, Thailand, Turkey and the UK. They use a time-series framework to analyze the effects of terrorism on the higher moments of returns. They find statistically significant causality effects, both in mean and in variance, in all of the six countries examined. Finally, Karolyi and Martell’s (2006) study of 75 attacks on 43 individual

SC RI PT

firms is widely cited in the literature. They find a statistically significant negative reaction on stock prices on event-days, amounting to average losses of 0.83%, or $401 million in market capitalization. Unfortunately, their dataset on terrorist attacks is limited to the appendices of the annual “Patterns of Global Terrorism” report of the Counterterrorism Office of the U.S. State Department for 1995-2002.

Given the divergence of the findings in the recent literature regarding the impact of terrorism on

U

financial markets, and considering the data limitations of some of the studies, we comprehensively examine

N

the impact of terrorist incidents on financial markets over a period of twenty years, covering multiple

M

A

markets and using multiple indices.

3.1 Terrorist Incident Data

D

3.0 DATA SOURCES

TE

We use an open source database now managed by the University of Maryland called the Global Terrorism Database (GTD), which is the most complete terrorism incidence database available today. While

EP

the widely cited study of Drakos (2010) investigates the period from January 3, 1994, to December 30, 2004, we examined all terrorist incidents available in the database from January 1, 1991 to December 31,

CC

2010. Our dataset includes all data on the GTD file updated to account for year 1993 data that it once lacked, which contains 54,638 distinct records of terrorism incidents. Of those records, 54,074 incidents occurred

A

outside the United States, leaving the total number of U.S. based incidents at 564. Given the very large number of incidents, careful consideration was given to identifying the most

serious incidents. Inspection of the data shows that the top ten countries with the most terrorist incidents, in order, are: Iraq, India, Pakistan, Colombia, Afghanistan, Algeria, Philippines, Turkey, Thailand, and Russia. Furthermore, not all terrorist incidents included in the database are serious, since many are expected

7

Impact of Terrorism on World Markets? to have minor or no economic consequences. In order to narrow the sample to those terrorist events deemed most severe in terms of damage, the data was sorted to include only ‘Catastrophic (> $1 billion)’ and ‘Major (> $1 million and < $1 billion)’ categories. The sample was sorted again according to the frequency of the terror events. Table 1 lists the number of incidents recorded for each of the ten countries in which the

SC RI PT

highest number of terrorist events occurred, and Table 2 lists the number of incidents after filtering the data by ‘Major’ and ‘Catastrophic’ categories of property damage. Table 1 and Table 2 to go about here

While it is important to note that not every event in the database is classified based on the amount of damage, using this methodology helped ensure that only events known to have been large in scale were

U

included.4 It is also important to note that only the dates on which the greatest number of severe terrorist

N

incidents occurred were tested. Since some dates contained concomitant events, in some instances our event

A

studies on international terrorist incidents are events on dates on which terrorism occurred simultaneously

M

in multiple countries. Thus, we consider these studies to be a true test of the impact of the worst known clusters of terrorist activity. This is not the case with event studies conducted on U.S.-based incidents where

D

each date represents a discrete U.S.-only event.

TE

3.2 Stock Market, Treasury Bond Market, and Gold Data: We obtained the daily closing prices for several country indices from Thomson Reuter’s

EP

DataStream for the period January 1, 1991 to December 31, 2010. We downloaded the daily closing yield data for 10-Year U.S. Treasury bonds from the Federal Reserve’s Data Download Program. Finally, we

CC

obtained the daily gold prices from the London Bullion Market Association (LBMA) website using their closing gold fixing prices in USD. Daily adjusted closing prices on the S&P 500, Dow Jones Industrial

A

Average, the S&P 500 Volatility Index (VIX), were obtained from CRSP. We used these daily closing

We should note one problematic example pertaining to the Oklahoma City bombing in the United States on April 19, 1995. For unknown reasons this event was not labeled ‘Catastrophic’ or ‘Major’ in the database. Nevertheless, the event was included in the sample of studies we considered. 4

8

Impact of Terrorism on World Markets? prices for the entire twenty-year period, even though index information for some countries, such as Russia and Pakistan, were not available for the entire period.

4.0 METHODOLOGY

SC RI PT

To investigate the impact of terrorist incidents on U.S. financial markets, we examine many terrorist incidents and employ the standard event-study methodology (Brown & Warner, 1985). We examined the thirty-two worst terrorist incidents that took place within the continental United States, along with another sixteen dates on which the most damaging incidents of terrorism occurred outside the U.S. We used an estimation period of one calendar year, and an event window of fifteen days, with one week before and one

U

week after the event date. We estimated the abnormal and cumulative abnormal returns and a t-test for

N

significance for each date of the event windows for the S&P 500 stock index, the 10-year Treasury bond

A

yield, the price of gold, against the MSCI World All Cap Index. In total, thirty-two U.S. events, including

M

9/11, were tested. We also tested one additional event, the embassy bombings in Kenya and Tanzania, bringing the total number of U.S.-directed terrorist incidents to thirty-three.

D

We use the MSCI index as a global benchmark to observe relative movements in other countries

TE

and markets before and after an occurrence of an event. We use both the Global MSCI index, and MSCI index excluding the U.S. as our global benchmark to measure changes in all other international markets. It

EP

should be noted that these two global indices that were used in our analysis are the theoretically most comprehensive indices and, thus, most accepted in the literature to measure global economic activity. Our

CC

interest in this research was to examine the impact of terrorist incidents on particular countries, thus, we were focused on trying to identify and measure the reactions of specific country market indices to these

A

events. The best benchmark we could identify to measure such effects was the MSCI Global Market index that we used to measure relative changes, i.e. we measured the reaction of these markets relative to the Global Market index. The use of the MSCI can be considered an adjustment for the ‘all other things remain the same’ assumption.

9

Impact of Terrorism on World Markets? 4.1 Procedure McWilliams and Siegel (1997) point out that a well-crafted event-study must clearly define the event that provides the new information with sufficient theoretic justification. They further note that one can only achieve confidence in the results when the abnormal returns associated with the event are correctly

SC RI PT

specified. Such identification relies on the following assumptions: (1) markets are efficient; (2) the event was unexpected by the market; and (3) there are no confounding effects during the event window.

In the context of this study, the new information that is unanticipated by the market is the occurrence of a significant terrorist act. When this new information is released to the market, the true economic impact of such an incident should be quickly and fully reflected on the country’s index through

U

its constituent firm’s stock prices.5 Brown and Warner (1985) show that the value adjustment process by

N

the stock market would include any expected cost and benefit related to the new information. The third

A

assumption pertaining to potential confounding effects is quite important, although this can be largely

M

addressed by choosing a very short event window, which is used in this study.6 In addition, we verify that there are no confounding events by examining the countries’ records six months surrounding the event. For

D

an incident to enter in our analysis, we confirm that there were no other major confounding events such as

TE

the occurrence of any other major economic, political or terrorist event in that country around the terrorist incident. 7

EP

We use the standard single index market model in estimating expected returns: (1)

CC

Rt  RtT Bill    1  RMRFt   t

where Rt is the return on the specific company or country market index i in period t, RtT-Bill is the return on

A

US treasury-bills in month t, and RMRFt, is either the CRSP value-weighted market portfolio minus the US T-bill rate in the case of the US market, or the MSCI World All Cap Index and the MSCI World All Cap

5

Ball and Brown (1968) investigated the impact of earnings announcements on security prices, and Fama, Fisher, Jensen, and Roll (1969) examined the market reaction to stock splits. 6 The longer the event window, the more difficult it would be to control for events that might cause a confounding effect. 7 Brown and Warner (1980, 1985), and more recently MacKinlay (1997), provide a comprehensive review of event-study methodology, and articulate several essential steps for the successful implementation of this approach.

10

Impact of Terrorism on World Markets? Ex-U.S Index when examining equity market index for other countries. Having estimated the expected return, abnormal returns can then be calculated by subtracting the normal return from the actual return so that, ARit = Kit - Rit , where ARit is the abnormal return on the security or index i in period t, and Kit is the actual return on security or index i in period t. We have chosen the length of the estimation window (255

SC RI PT

days) to be sufficiently long so that under the null hypothesis, the distribution of the sample abnormal returns of a given observation in the event window is approximately normally distributed:

ARit  (0, 2 ( ARit ))

Although the event study structure is relatively simple to implement, some statistical issues need

U

to be carefully addressed, especially when the event window is long (typically greater than 12 months). Kothari and Warner (1997, 2006) show that long-horizon studies typically lack the ability to detect

N

abnormal performance, and are particularly sensitive to the return generating process. However, short-

A

horizon event windows studies such as those employed in this study, generally do not suffer from these

M

limitations. Other possible sources of bias that can influence event-study results include non-synchronous

D

trading, thinly traded securities, and some further liquidity issues related to small capitalization stocks. Fortunately, we use equity and bond market indices that are actively traded on major exchanges. Finally,

TE

the choice of the appropriate market index is also critical. In this study, we use both the MSCI World All

EP

Cap and the MSCI World All Cap ex USA Index.8 For each of the identified terrorist incidents, we estimated the expected returns and the abnormal

CC

returns for each asset across the event window and generated a t-test for significance. We compare the number of observations and a count of significant abnormal returns at both the 5% and 10% levels of

A

significance against those of other assets, taking note of any differences resulting from having run the event study against the two different benchmarks. In addition to the first set of tests on the originally identified

8

The MSCI is widely accepted as the most comprehensive world index that is updated regularly to represent the respective market value weights of all countries. Due to the disproportional size of the US market in the index, it is common to use the MSCI World index EX USA when examining events or markets outside of the United States.

11

Impact of Terrorism on World Markets? events, we conduct further robustness tests on other events known for their high level of severity. Specifically, we examine in this second round of tests the following set of events; the November 2008 terrorist attacks in Mumbai; the July 11, 2006 train bombings in Mumbai; the September 10 and 13, 1999 apartment bombings in Moscow; the Moscow Theater hostage crisis that began on the night of October 23, 2002; the London subway bombings of July 7, 2005; the March 11, 2004 train bombings in Madrid; the February 26, 1993 World Trade Center bombing in New York; the Oklahoma City bombings of April 19, 1995; the U.S. embassy bombings on August 7, 1998 in Kenya and Tanzania.

SC RI PT

        

5.0 ANALYSIS AND RESULTS

U

We first present the argument that terrorism has little, if any, statistically significant impact on asset

N

valuations. In the second part of our analysis and results, we illustrate the cases that suggest the opposite,

A

in particular both the September 11, 2001 attacks, and the London tube bombing of July 7, 2005, which

M

point to a significant negative impact on financial markets as a consequence of terrorism.9 However, even in these two incidents, the significant negative results appear only in our second event study on the UK

D

incident, in which we used the regular MSCI World Index to estimate residuals instead of the All Cap index.

TE

5.1 Market Impact Associated with U.S.-Based Terrorist Events We now present results for the major terrorism incidents that have occurred in the United States

EP

and internationally. Because we consider the effects from the 9/11 attacks to be a special case, the U.S.based terrorist event results discussed here refer to 30 of the 32 main tests that were conducted, with event

CC

studies eighteen and nineteen removed due to overlapping event windows in the proximity of 9/11. Further, we made sure to remove any duplicate results when tallying the number of significant abnormal returns, as

A

a few other event windows shared the same dates. In total, we present results for 30 U.S. events, and 13 Ex-U.S. events.10

10

The last two Ex-U.S. studies contained more than one date on which a cluster of severe terrorist attacks occurred. Many attacks were carried out in Iraq and Thailand within these event windows.

12

Impact of Terrorism on World Markets? Table 3a. provides summary statistics for the 30 U.S.-based terrorist events, after controlling for the 9/11 event. From the right to the left, the table shows the total number of abnormal returns that are significant at the 5% level for each asset class, the total number of abnormal returns that were positive at the 5% significance level, and the number of negative abnormal returns that are significant at the 5% level.

SC RI PT

Finally, the table lists the total number of positive and negative returns that were significant at the 5% level that occurred on the event study date, t=0. Table 4b shows the same tally for abnormal returns significance, but at the 10% level for all assets examined within their respective event windows. Table 3 to go about here

The results in Tables 3a and 3b show that the number of positive abnormal returns exceeds the

U

number of negative returns at both the 5% and 10% significance levels. To arrive at the correct sum, note

N

that a negative return in the 10-year US Treasury Bond yield means an increase in the 10-year Treasury’s

A

price. We note that gold must not be considered as a U.S. asset, and consequently, its abnormal returns

M

should be removed from the totals meant to demonstrate an effect on U.S. markets. This is important, since gold has the highest number of significant positive abnormal returns that occur on the event dates.

D

Perhaps the most significant observation in several of our tables is the general lack of significant

TE

abnormal returns observed on event days. The S&P 500, for example, registers no significant abnormal returns at the 5% level on the event date when tested using MSCI All Cap World Index as the benchmark;

EP

this number changes to two negative and one positive significant abnormal returns on the event date when applying the MSCI All Cap World Index Ex USA as the benchmark. It is worth noting that even without

CC

controlling for the effects of the 9/11 attacks, the percentage of positive and negative abnormal returns at the 5% significance level for the U.S. events are still: 52.56% positive and 47.44% negative for the first

A

benchmark; and 58.11% positive and 41.89% negative against the second benchmark. It is interesting to note the rather even number of both positive and negative abnormal returns from

which few meaningful conclusions may be drawn. The distribution generally hints at a lack of an existing pattern. It is not surprising to find that only few incidents, other than September 11, 2001, have significantly affected U.S. markets, since there are only a few truly severe terrorist incidents that are available for study.

13

Impact of Terrorism on World Markets? Indeed, on closer inspection of the U.S.-based events in our sample, we find that many of the events that are classified as ‘Major’ and ‘Catastrophic’ in the Global Terrorism Database (GTD) are events that could be classified as eco-terrorist events by groups such as the Earth Liberation Front. Such events included the burning of a luxury home, a destroyed private mansion, and an attacked Chevrolet dealer, which,

SC RI PT

understandably, may not be of interest to a great number of investors.

However, one would certainly expect to see an effect resulting from two major and memorable events of U.S. domestic terrorism; the World Trade Center bombing on February 26, 1993, and the Oklahoma City bombing on April 19, 1995. Surprisingly, these incidents, which are orders-of-magnitude greater in severity to other incidents, reveal no significant reactions in U.S. financial markets. Both events

U

occurred at times when the market would have had time to react, with the World Trade Center bombing

N

occurring in the morning before the U.S. market opened, and the bombing of the Alfred P. Murray Federal

A

Building in Oklahoma City occurring a short time after 12:00 p.m. In fact, we observe no significant

M

negative abnormal returns for these events either on, or after, the event date.11 In sum, for the 1993 World Trade Center bombing, the S&P 500 Index and Dow Jones Industrial

D

Average actually finished positive for the day on February 26, 1993, while the S&P 500 Volatility Index

TE

(VIX) closed lower that day. For the Oklahoma City bombing, the S&P 500 Index finished the day fractionally lower on April 19, 1995, while the Dow Jones Industrial Average finished higher with the VIX

EP

finishing lower. Again, we would not have expected such a result if terrorism had a negative impact on stock markets.

CC

5.2 Market Impact Based on International Terrorist Incidents Table 4a. and 4b. present the results for international terrorist incidents. Examining the S&P 500

A

abnormal returns that are significant at the 5% level, we notice that there are almost twice as many positive abnormal returns as there are negative abnormal returns against both benchmarks. This result may lead us

11

Chen and Siems (2004) examine the Dow Jones Industrial Average and find a positive abnormal return on the event day for the Oklahoma City bombing, as well as positive 6- and 11-day CARs for this attack. They also study the February 26, 1993 World Trade Center bombing event, and find negative abnormal returns on the event-day and negative 6- and 11-day CARs, but the results are not statistically significant.

14

Impact of Terrorism on World Markets? to wonder whether the U.S. market may have benefited from international terrorism. However, the results are not really as convincing when we check the number of times the S&P 500’s returns were positive and significant on event-days. For the 95% significance level reported in Table 4a, we find that only two positive results against the MSCI, and two positive and one negative results against the MSCI (ex. U.S.) are

SC RI PT

significant. At the 90% significance level reported in Table 4b, the proposition deteriorates further as we notice the significant abnormal returns for the S&P 500 on event days now total two for negative returns, and zero for positive returns using the MSCI as the benchmark. Table 4a and 4b to go about here

Further, the 10-year US Treasury bond returns show more positive abnormal returns than negative

U

at both the 5% and 10% levels of significance, with three positive abnormal returns significant at the 5%

N

level on the event day. This suggests that US Treasuries were mostly sold on these event days, or at least

A

were sold more than they were bought. These results provide evidence against ‘flight to safety’ behavior

M

by investors wishing to purchase U.S. Treasury securities. In the case of gold, abnormal returns are more positive than negative during the event windows. Further, there are no positive abnormal returns on the

D

event days at the 10% significance level, and only one significant abnormal return occurring on an event

TE

day, leading us to conclude that gold may not be a true safe haven asset after all. 5.3 Market Behavior Following the World Trade Center Attacks (9/11)

EP

Table 5 shows selected event-study results conducted on U.S. terrorist incidents. For the September 11, 2001 attacks, no stock market return data was available for the event day itself, because U.S. exchanges

CC

were closed. The event window contains the date of the market’s reopening, which reveals statistically significant results for the September 17, 2001 date for all assets when using the MSCI All Cap World Index.

A

Every category was significant at 95% against the benchmark. On the date of the market’s reopening, the S&P 500 experienced a significant negative impact. The

10-year yield rose, signifying the selling of U.S. Treasury bonds on that day. Gold showed unique behavior on both the event day itself, where it gained 5.7%, and on the day of the market’s reopening, gaining 2.62% (see Table 5 Panel C). Both days exhibited abnormal returns that are significant.

15

Impact of Terrorism on World Markets? Table 5 to go about here The event window for 9/11 does not provide enough of a time span to evaluate the cumulative abnormal returns of a full week of trade on the U.S. market’s reopening. Fortunately, we are able to compensate for this, as the nineteenth U.S. event was September 20, 2001, and the two event windows

SC RI PT

naturally overlap. The window of this study spanned September 13, 2001 to September 27, 2001. The cumulative abnormal returns during the 9/20/2001 event window using the MSCI All Cap World Index are estimated as follows: -1.91% for the S&P 500; +0.08% for the 10-year Treasury Bond; and +3.78% for gold. 5.4 Market Behavior Following the London Tube Bombings (7/7/05)

U

We next consider the results of the London tube bombings of July 7, 2005, which offer one of the

N

only cases where a terrorist incident appears to significantly affect both a foreign country’s index and a

A

U.S. index (in opposite directions). First, Table 6 reports the estimated results conducted for this incident,

M

using the MSCI UK Index and the Dow Jones Industrial Average. We observe significant results on the actual event day, with the UK Index declining and the Dow Jones rising. This result comes from our second

D

series of studies, and has a longer event window (+14 and -14 days). However, interpreting the result of

TE

this incident as evidence that U.S. markets benefited from an international terrorist event is problematic. First, the cumulative abnormal returns for the UK index are actually positive, while the Dow Jones index

EP

is negative (see Table 6). Second, a closer look at the actual percentage price change in the Dow reveals that the gain in the Dow that day was only 0.003%, hardly an economically meaningful gain.

CC

Table 6 to go about here

Table 7 illustrates that while the S&P 500 experienced a statistically significant positive return on

A

the event day when measured against the MSCI All Cap World Index, the significance completely disappears when the MSCI All Cap Ex USA World Index is used. Further, we note that while the MSCI UK Index declined by 1.37% on the event day, it rebounded by 1.44% the very next day. This market move represents a market recovery that is consistent with the observation of Kollias, et al. (2011), who found that all three FTSE indices rebounded one trading day after the event.

16

Impact of Terrorism on World Markets? Table 7 to go about here Figures 1a-c show comparison charts of the FTSE 100 Index and the Dow Jones Industrial Average for the period spanning the event study window, as well as a longer one-year period. A casual reading of the charts reveals that over the longer period, the FTSE appears resilient and outperforms the Dow for that

SC RI PT

year. To help visualize both the FTSE 100’s and the Dow Jones’s price movement on the study date, July 7, 2005, we add a candlestick chart. We note that the FTSE 100 did experience large price swings on the event date, but ended well off the lows of the day, as Kollias, et al. (2011) also found.12 Figure 1 to go about here 5.5 Madrid Train Bombings

U

This is a final example from our second series of event studies that demonstrates the unlikely

N

influence that a terrorist event might have on financial markets. We conducted an additional study of the

A

Madrid train bombings of March 11, 2004, looking at the following indices: the MSCI All Cap Europe

M

Index in USD, MSCI Germany, MSCI France, MSCI Italy, Spain’s index from DataStream in Euros, and the Dow Jones Industrial Average. We used the MSCI World Index in this case. Within these longer

D

windows (-14, +14 days), we note the following: the Spanish market’s return on the event day was not

TE

significant at either the 90% or the 95% levels of confidence. There were three abnormal returns at the 5% significance level during the event window for the Spanish market, two negative and one positive, with

EP

neither on the event day itself. The worst day for the Spanish market was March 15, when it fell a

CC

statistically significant 4%, a full four calendar days and two trading days after the attacks. The German index had only one significant abnormal return, which was positive, and occurred far

A

in advance of the incident on March 2, 2004. France’s index contained only one abnormal return at the 5% level of significance; it was positive, and occurred the day before the attack on March 10. The Italian market had exactly the same result. The MSCI All Cap Europe Index had only one statistically significant result, which occurred on March 3, far in advance of the incident, similar to the German market. Throughout the

They state that the FTSE 100 Index “fell by about 200 points in the 2 h after the first bomb attack …” but that “By the time the market closed the FTSE 100 had recovered to only 71.3 points down (-1.36%)” (p. 538). 12

17

Impact of Terrorism on World Markets? entire event study window, however, the Dow Jones Industrial Average exhibited no statistically significant movements whatsoever. On the day of the attack, we find it interesting that in terms of the percentage price declines, the Spanish market fared much better than did some of its European neighbors. For Spain and its surrounding

SC RI PT

neighbors, the following returns were observed: Spain’s market declined by -1.83%; Germany’s market was off -3.42%; France was down -2.93%; Italy lost -2.12%; the Dow Jones Industrial Average lost -1.64%; and the MSCI All Cap Europe Index was down by -2.45%. Surprisingly, on a day when quite a severe terrorist attack occurred in Madrid, Spain, the stock markets of France, Italy, Germany and the European region as a whole all lost more value than Spain’s.

U

Figures 2a-c chart the Dow Jones Industrial Average and the Spanish Ibex 35 Index in a larger

N

context with a line drawn indicating when the 3/11 attacks occurred. We also include a chart that roughly

A

spans the dates of the event window, which depicts the movements in the German DAX, France’s CAC,

M

Spain’s FTSE MIB Index, along with the Dow Jones Industrial Average. We believe that an argument that advances these markets’ inter-linkage and co-movement during this period stemming from other market

TE

Figure 2 to go about here

D

forces represents a better overall explanation for the similar patterns observed.

EP

6.0 FOLLOW-UP STUDIES

In addition to the extra cases already mentioned, we conducted a number of follow-up tests on

CC

selected incident dates when severe terrorist events occurred. These additional tests provide robustness and also examine dates meriting further investigation.

A

6.1 The 2008 Mumbai Terrorist Attacks A series of terrorist attacks began in Mumbai, India on the night of Wednesday, November 26,

2008, and continued into Saturday. By November 29, this incident resulted in the killing of at least 100

18

Impact of Terrorism on World Markets? people and injuring 250 in India’s financial capital.13 We looked for effects on the Dow Jones Industrial Average and the MSCI India Index denominated in local currency. Because the events began on the night of Wednesday 26, we expected to see significant results reflected in the markets on November 28 (no data is available for Nov. 27). We used a slightly longer event window spanning -14 days before November 26

SC RI PT

and +14 days after November 28 to allow for the multiple days of the attacks. Yet, we found no significant abnormal returns for any of the event days, at either the 5% or the 10% levels of significance for either the MSCI India Index or the Dow Jones Industrial Average. Surprisingly, both indices finished with positive gains on Friday, November 28. Further, the only statistically significant returns that occur within the event window were both positive. The DJIA’s results are similar. Figures 3a-c show India’s Nifty Index during

U

the event window, and contrast its performance against the DJIA by overlaying the Nifty Index on the DJIA

N

during the same period.

A

Figure 3 to go about here

M

One hundred and fifty people lost their lives in the Mumbai train bombings which occurred during the evening rush hour on Tuesday, July 11, 2006. Both the MSCI India Index and the DJIA Index were

D

used to measure market impact. However, the MSCI India index registered a positive abnormal return –

TE

significant at 5% – on the next trading day, while the DJIA finished lower on the same day. The percentage price gain in the MSCI India index on July 12 was +2.97%, and cumulative abnormal returns for this index

EP

were positive for both +6 days after the event, and at the end of the event window at +14 days. 6.2 Russia Apartment Bombing

CC

We examined the Russian Apartment bombing event that occurred in Moscow just after midnight

on Friday, September 10, and again on Monday, September 13, 1999, at about 5:00 a.m. The MSCI Russia

A

Index and DJIA were used to measure market impact, with significant results expected for Friday, September 10 and September 13. Once again, no statistically significant abnormal returns were observed

See for example: Sengupta, Somnini. “At Least 100 Dead in India Terror Attacks.” www.nytimes.com. 26 November 2008. Available from: http://www.nytimes.com/2008/11/27/world/asia/27mumbai.html 13

19

Impact of Terrorism on World Markets? for either of the two indices during the entire event window, which spanned the period 8/27/1999 to 9/27/1999. However, cumulative abnormal returns for the MSCI Russia Index were negative 17.01% by the end of the event window, with the DJIA’s registering a negative 4.41% during the same period. The Russia index was indeed down by -6.83% on Monday, September 13, 1999, but the drop was not statistically

SC RI PT

significant.

We take a final look at the Russian market during the Russian Theater Hostage Crisis that began in Moscow on Wednesday night, October 23, 2002, and lasted until Saturday, October 26. Once again, the MSCI Russia index and the DJIA index were used to measure impact, on October 24 and October 25, 2002 as the event dates. Again, no significant abnormal returns were observed for the Russia Index during the

U

event window. We surmise that this may be due to the extra volatility in the Russian market, which when

N

smoothed, would require even larger jumps in returns to stand out.14 As for the DJIA, we observe a negative

A

abnormal return significant at the 5% level for October 24, the day after the event. While the cumulative

M

abnormal returns for the MSCI Russia index do look very high, at -14.5% for the event window as a whole, it is worth noting that they were already high and negative (-8.2%) before these events. With the DJIA

D

average down significantly on the day after the Russian Hostage Crisis event began, and again down after

TE

the event period, and with only one positive abnormal return after the event, we can reasonably conclude

EP

that the U.S. markets did not benefit from terrorism in this case. 7.0 SUMMARY AND CONCLUSIONS

CC

The Efficient Market Hypothesis suggests that in an efficient capital market, all available public information is fully and instantaneously reflected on stock prices. Our initial hypothesis suggests that

A

terrorist incidents elicit a strong emotional reaction ranging from anger, fear, outrage, and helplessness, which would eventually result in a strong negative effect on financial markets. However, the results generally did not lend themselves to our stated hypothesis. The exception to this is the September 11, 2001

14

Charles and Darné (2014), for example, used a modified iterative cumulative sums of squares (ICSS) algorithm to identify subperiods of changed volatility levels in the DJIA after acknowledging that some extreme returns in the DJIA are not seen as outliers due to the heightened volatility of some periods (pp. 194-5).

20

Impact of Terrorism on World Markets? attack on the World Trade Center, which did result in a significant negative impact on the U.S. and World markets. We also argued that terrorist incidents in foreign countries might lead to flight of capital into safer havens such as gold and US Treasuries, and to safer countries such as the United States. However, our results have not borne out either of these hypotheses.

SC RI PT

Overall, our results lead us to argue against a causal relationship between terrorist events and movements in financial and asset markets. We have found many instances, in which both positive and negative returns occur, with no significant predictable pattern present. We found that in most of the existing studies, factors other than terrorist attacks can explain the negative abnormal returns. Drakos’s (2010) study showed that national returns are significantly lower on days when terrorist attacks occur; however, using

U

the same database of terrorist events, we do not find any evidence of such a relationship.

N

Based on our results we conjecture that even though terrorism has a high emotional impact on

A

investors, there is limited material impact on the overall markets in general. While a specific company or

M

industry may get economically impacted by a terrorist incident, by and large, the market stays un-impacted. We believe that the high frequency of terrorist incidents and extensive media coverage has desensitized the

D

public to terrorist incidents. In rare cases the psychological impact may be so severe as to impact the long-

TE

term mood of investors. For instance, in the case of 9/11 and the bombings of the World Trade Center, there was a deep psychological impact that crashed the market and also caused a long term danger to the market.

EP

Indeed, we encourage researchers to remain agnostic on the impact of terrorism on financial markets until

CC

the empirical evidence confirms a causal relationship.

Acknowledgements:

A

The authors would like to thank Annika Heffter for a thorough editing of the paper.

21

Impact of Terrorism on World Markets? REFERENCES [1] Arin , K.P., Ciferri, D., and Spagnolo, N. (2008). The price of terror: the effects of terrorism on stock market returns and volatility. Economics Letters, 101(3),164–167.

International Business Studies, 45(4), 471-493.

SC RI PT

[2] Bekaert, G., Harvey, C. R., Lundblad, C. T., Siegel, S. (2014). Political risk spreads. Journal of

[3] Berrebi, C., and Klor, E.F. (2010). The impact of terrorism on the defence industry. Economica, 77(307), 518–543. doi: 10.1111/j.1468-0335.2008.00766.x

[4] Blomberg, S.B., Hess, G.D., and Orphanides, A. (2004). The macroeconomic consequences of terrorism. Journal of Monetary Economics, 51(5), 1007–1032.

U

[5] Branzei, O., & Abdelnour, S. (2010). Another day, another dollar: Enterprise resilience under terrorism

N

in developing countries. Journal of International Business Studies, 41(5), 804-825.

A

[6] Brown, S.J., and Warner, J.B. (1985). Using daily stock returns the case of event studies. Journal of

M

Financial Economics, 14(1), 3–31.

[7] Chesney, M., Reshetar, G., Karaman, M. (2011). The impact of terrorism on financial markets: An

D

empirical study. Journal of Banking & Finance, 35(2), 253–267.

TE

[8] Czinkota, M. R., Knight, G., Liesch, P. W., & Steen, J. (2010). Terrorism and international business: A research agenda. Journal of International Business Studies, 41(5), 826-843.

EP

[9] Drakos, K. (2010). Terrorism activity, investor sentiment, and stock returns. Review of Financial

CC

Economics, 19(3), 128–135. [10] Eldor, R., Melnick, R. (2004) Financial markets and terrorism. European Journal of Political Economy, 20(2), 367-386.. http://ssrn.com/abstract=2357312.

A

[11] Fama, Eugene, (1970), Efficient Capital Markets: A Review of Theory and Empirical Work, Journal of Finance, 25(2), 383-417.

[12] Goel, S., and Shawky, H. (2009). Estimating the Impact of Security Breaches on Stock Valuations of Firms, Information & Management, 46(7), 404–410.

22

Impact of Terrorism on World Markets? [13] Henisz, W. J., Mansfield, E., & Von Glinow, M. A. (2010). Conflict, security, and political risk: International business in challenging times. Journal of International Business Studies, 41(5), 759-764. [14] Hippler, W.J., Hassan, M.K., (2015). The impact of macroeconomic and financial stress on the U.S. financial sector, Journal of Financial Stability, 21, 61-80.

SC RI PT

[15] Karolyi, G. A., and Martell, R. (2006). Terrorism and the stock market. Available at SSRN: http://dx.doi.org/10.2139/ssrn.823465

[16] Karolyi, G.A. (2006). The consequences of terrorism for financial markets: What do we know? Available at SSRN: http://ssrn.com/abstract=904398 or http://dx.doi.org/10.2139/ssrn.904398

[17] Karolyi, G.A. (2008). [Abstract]. An assessment of terrorism-related investing strategies. The

U

Journal of Portfolio Management, 34(4), 108-123.

N

[18] Kollias, C., Papadamou, S., Stagiannis, A. (2011). Terrorism and capital markets: The effects of the

A

Madrid and London bomb attacks. International Review of Economics & Finance, 20(4), 532–541.

M

[19] Kothari, S. P., Warner, J. B., (1997). Measuring Long-Horizon Security Price Performance, Journal of Financial Economics 43(3), 301-339.

D

[20] Kothari, S., Lewellen, J., Warner, J., (2006). Stock Returns, Aggregate Earnings Surprises, and

TE

Behavioral Finance, Journal of Financial Economics 79(3), 537-568. [21] Loveland, R. (2016) How prompt was regulatory corrective action during the financial crisis? Journal

EP

of Financial Stability, 25, 16-36.

CC

[22] McWilliams, A., & Siegel, D. (1997). Event Studies in Management Research: Theoretical and Empirical Issues. The Academy of Management Journal, 40(3), 626-657. Retrieved from

A

http://www.jstor.org/stable/257056

[23] Mester, L. J., (2017). The nexus of macroprudential supervision, monetary policy, and financial stability, Journal of Financial Stability, 30, 177-180.

23

Impact of Terrorism on World Markets? [24] Mobarek, A., Muradoglu, G., Mollah, S., Hou, A. J., (2016). Determinants of time varying comovements among international stock markets during crisis and non-crisis periods, Journal of Financial Stability, 24, 1-11.

45(2), 263-275.

SC RI PT

[25] Nikkinen, J. and Vähämaa, S., (2010) Terrorism and Stock Market Sentiment. Financial Review,

[26] Nitsch, V., Schumacher D. (2004). The economic consequences of terror. European Journal of Political Economy, 20(2), 423–433.

[27] Oetzel, J., & Getz, K. (2012). Why and how might firms respond strategically to violent conflict?. Journal of International Business Studies, 43(2), 166-186.

U

[28] Pragidis, I.C., Aielli, G.P. Chionis, D., Schizas, P. (2015) Contagion effects during financial crisis:

N

Evidence from the Greek sovereign bonds market, Journal of Financial Stability, 18, 127-138.

A

[29] Procasky, W. J. and Ujah, N. U., (2016), Terrorism and its impact on the cost of debt, Journal of

M

International Money and Finance, 60, 253-266.

[30] Reid, W. H, (2001), Psychological Aspects of Terrorism, Journal of Psychiatric Practice, 7(6) , 422-

D

425.

TE

[31] Ribeiro, J. B., Vilmunen, J., Tabak, M.B., and Barroso, B., (2016). Risk, Financial Stability and Banking, Journal of Financial Stability, 25, 113-114.

EP

[32] Sedunov, J., (2016). What is the systemic risk exposure of financial institutions?, Journal of Financial Stability, 24, 71-87.

CC

[33] Shiller, R. J., Fischer, S., & Friedman, B. M. (1984). Stock prices and social dynamics. Brookings papers on economic activity, 1984(2), 457-510.

A

[34] Silva, W., Kimura, H., and Sobreiro, V. A., (2017). An analysis of the literature on systemic financial risk: A survey 28, 91-114.

24

SC RI PT

Impact of Terrorism on World Markets?

D

M

A

N

U

Figure 1a. Comparison of FTSE 100 and Dow Jones Industrial Average (6/23/2005 – 7/21/2005)

TE

Figure 1b. Comparison of FTSE 100 and Dow Jones Industrial Average Candlestick Charts (6/23/2005

A

CC

EP

– 7/21/2005)

Figure 1c. Comparison of FTSE 100 and Dow Jones Industrial Average (12/31/2004 – 12/30/2005

25

SC RI PT

Impact of Terrorism on World Markets?

TE

D

M

A

N

U

Figure 2a: Spain’s IBEX 35 Index (10/31/2003 – 9/1/2004)

A

CC

EP

Figure 2b: Dow Jones Industrial Average (10/31/2003 – 9/1/2004)

Figure 2c: Key World Indices (Spain’s FTSE MIB with DAX, CAC, and DJIA) 2/26/2004 – 3/25/2004

26

SC RI PT

Impact of Terrorism on World Markets?

M

A

N

U

Figure 3a: India’s Nifty Index end closing values (10/31/2008 – 12/15/2008)

A

CC

EP

TE

D

Figure 3b: India’s Nifty Index Candlestick Chart (11/10/2008 – 12/12/2008)

Figure 3c:

Dow Jones Industrial Average with Nifty Index Overlaid (10/27/2008 –

12/15/2008)

27

Impact of Terrorism on World Markets?

Table 1: Top ten countries sorted by terrorist incident, January 1, 1991 to December 31, 2010.

N

U

of

A

CC

EP

TE

D

M

Iraq India Pakistan Colombia Afghanistan Algeria Philippines Turkey Thailand Russia

Number Terrorist Incidents* 6445 4592 4153 3502 2562 2399 1845 1785 1592 1523

A

Country

SC RI PT

Our dataset includes all data on the GTD file (updated to account for year 1993, data that it once lacked), which contains 54,638 distinct records of terrorism incidents. Of those records, 54,074 incidents occurred outside the United States, leaving the total number of U.S.-based incidents count at 564. This table lists the number of incidents recorded for each of the ten countries in which the highest number of terrorist events occurred.

28

Impact of Terrorism on World Markets?

Table 2: Top ten countries sorted by the number of severe terrorist incident count over the period January 1, 1991 to December 31, 2010.

N

U

Severe

A

CC

EP

TE

D

M

Colombia Iraq India Pakistan United States Northern Ireland Russia Spain Thailand Philippines

Number of Incidents 121 103 46 44 35 34 30 28 27 21

A

Country

SC RI PT

The GTD database files classify ‘severe’ terrorist incidents in terms of damage. Specifically, severe incidents are classified as those included only in the ‘Catastrophic (> $1 billion)’ and ‘Major (> $1 million and < $1 billion)’ categories. This table lists the number of incidents after filtering the data by ‘Major’ and ‘Catastrophic’ categories of property damage.

29

Impact of Terrorism on World Markets?

Table 3: Aggregate CAR Results at the 95% and the 90% significance levels for U.S-based Terrorist Incidents We use the standard single index market model in estimating expected returns:

Rt  RtT Bill    1  RMRFt   t

SC RI PT

where Rt is the return on the specific company or country market index i in period t, RtT-Bill is the return on US treasury-bills in month t, and RMRFt, is either the CRSP value-weighted market portfolio minus the US T-bill rate in the case of the US market, or the MSCI World All Cap Index and the MSCI World All Cap Index Ex-U.S when examining equity market indices for other countries. Abnormal returns can then be calculated by subtracting the normal return from the actual return so that, ARit = Kit - Rit , where, ARit is the abnormal return on the security or index i in period t, and Kit is the actual return on security or index i in period t. Panel A: Aggregate CAR Results at the 95% significance level for U.S-based Terrorist Incidents

Gold

9 17 25 18 30 27

7 10 13 6 12 11

2 7 12 12 18 16

0 1 1 0 3 3

U

10-yr. Yield

MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.)

N

S&P 500

0 2 1 1 0 0

M

A

“MSCI” indicates the MSCI AC World Index is used, and the “MSCI (ex. U.S.)” indicates that the MSCI World AC Index ExUSA is used.

Panel B: Aggregate CAR Results at the 90% significance level for U.S-based Terrorist Incidents

10-yr. Yield

10% Pos. 6 5 4 8 4 4

CL

10% Neg. 7 2 3 4 2 7

CL

# Positive 10% CL Returns on actual date t=0 1 1 0 1 0 0

# Negative 10% CL Returns on actual date t=0 0 0 0 0 0 1

EP

Gold

MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.)

Tot. 10% CL 13 7 7 12 6 11

D

S&P 500

Benchmarks

TE

Index

A

CC

“MSCI” indicates the MSCI All Cap World Index, while the “MSCI (ex. U.S.)” indicates that the MSCI All Cap World Index ExUSA is used.

30

Impact of Terrorism on World Markets?

Table 4: Aggregate CAR Results at 95% and 90% significance level for International Terrorist Incidents We use the standard single index market model in estimating expected returns:

Rt  RtT Bill    1  RMRFt   t

SC RI PT

where Rt is the return on the specific company or country market index i in period t, RtT-Bill is the return on US treasury-bills in month t, and RMRFt, is either the CRSP value-weighted market portfolio minus the US T-bill rate in the case of the US market, or the MSCI World All Cap Index and the MSCI World All Cap Index Ex-U.S when examining equity market indices for other countries. Abnormal returns can then be calculated by subtracting the normal return from the actual return so that, ARit = Kit - Rit , where, ARit is the abnormal return on the security or index i in period t, and Kit is the actual return on security or index i in period t. Panel A: Aggregate CAR Results at 95% significance level for International Terrorist Incidents

10-yr. Yield Gold

5%

5% Pos. 8 9 6 6 9 8

CL

5% Neg. 5 4 4 4 5 5

CL

# Positive 5% CL Returns on actual date t=0 2 2 3 3 1 1

U

MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.)

Tot. CL 13 13 10 10 14 13

N

S&P 500

Benchmarks

A

Index

# Negative 5% CL Returns on actual date t=0 0 1 0 0 0 0

D

M

“MSCI” indicates that the MSCI AC World Index is used, and the “MSCI (EX- U.S.)” indicates that the MSCI AC World Index Ex-USA is used.

S&P 500 10-yr. Yield

MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.) MSCI MSCI (ex. U.S.)

CC

Gold

Benchmarks

EP

Index

TE

Panel B: Aggregate Results for the 90% significance level for International Terrorist Incidents Tot. 10% CL 5 4 6 7 5 5

10% Pos. 1 1 4 4 3 4

CL

10% Neg. 4 3 2 3 2 1

CL

# Positive 10% CL Returns on actual date t=0 0 0 0 0 0 0

# Negative 10% CL Returns on actual date t=0 2 0 0 0 0 0

A

“MSCI” indicates that the MSCI AC World Index is used, and the “MSCI (ex. U.S.)” indicates that the MSCI AC World Index ExUSA is used.

31

Impact of Terrorism on World Markets?

Table 5: Selected Results from event study on September 11, 2001, using MSCI All Cap World Index. Missing event dates are weekend days with no trading activity. We use the standard single index market model in estimating expected returns:

Rt  RtT Bill    1  RMRFt   t

SC RI PT

where Rt is the return on the specific company or country market index i in period t, RtT-Bill is the return on US treasury-bills in month t, and RMRFt, is either the CRSP value-weighted market portfolio minus the US T-bill rate in the case of the US market, or the MSCI World All Cap Index and the MSCI World All Cap Index Ex-U.S when examining equity market indices for other countries. This table shows selected event-study results conducted on U.S. terrorist incidents. For the September 11, 2001 attacks, no stock market return data is available for the event day itself, since U.S. markets were closed and only reopened for trading on September 17. The event window contains the date of the market’s reopening, which reveals statistically significant results for the September 17, 2001 date for all assets when using the MSCI All Cap World Index. An asterisk indicates the coefficient is significant at the 1% level.

Date

Event Date

9/4/01

t = -7

E(r) 0.0037

9/5/01

t = -6

-0.0089

0.0079

9/6/01

t = -5

-0.0228

0.0004

9/7/01

t = -4

-0.0167

-0.002

9/10/01

t = -1

-0.0042

9/11/01

t=0

9/12/01

t=1

A

9/17/01 9/18/01

-0.0066

-0.0052

CAR 0.0282* (2.6176) 0.0268 (-0.1319) 0.0108 (-1.4857) 0.003 (-0.7227) 0.0127 (0.8996) 0.0127

0.0124

-0.002

0.0127 0.0127

A

M

D

TE

0.0104

E(r) 0.0006

AR 0.0282

-0.0026

-0.0014

-0.0061

-0.016

-0.0046

-0.0078

-0.0014

0.0097

N

-0.0193

CAR -0.0043 (-0.7853) 0.0036 (1.4358) 0.004 (0.0745) 0.002 (-0.3582) 0.0124* (1.8887) 0.0124

t=2

0.0062

0.0124

0.0013

t=3

-0.0111

0.0124

-0.0031

-0.012

t=6

-0.032

-0.0172

-0.0048*

-0.0085

0.0216

0.0223*

t=7

-0.0043

-0.0015

(-3.1264) -0.0063 (-0.269)

-0.0014

0.0208

(2.004) 0.0432* (1.934)

CC

9/14/01

10 Year US Govt. Bond Yield

AR -0.0043

EP

9/13/01

S&P 500

U

Panel A: Event study results for S&P 500 and 10-Year Yield results:

0.0007 (-1.1096)

32

Impact of Terrorism on World Markets? Panel B: Event study results for Gold prices around the September 11, 2001 event. Date

Event Date

Gold

9/4/01

t = -7

E(r) 0.0001

AR -0.004

CAR -0.004 (-0.4906)

9/5/01

t = -6

-0.0002

-0.0034

-0.0074

9/6/01

t = -5

-0.0006

0.0034

9/7/01

t = -4

9/10/01

t = -1

9/11/01

t=0

9/12/01

t=1

-0.0002

-0.0277

9/13/01

t=2

0.0002

0.0043

9/14/01

t=3

-0.0003

0.0199

9/17/01

t=6

-0.0009

0.0271

9/18/01

t=7

-0.0001

-0.013

(-0.4268)

SC RI PT

-0.004 (0.4186)

-0.0004

0.0061

0.0021 (0.7618)

-0.0001

-0.0076

-0.0055 (-0.9397)

-0.0005

0.0576

0.0521* (7.1470)

M

A

N

(-3.4376) 0.0287 (0.5335) 0.0487* (2.4709) 0.0758* (3.3633) 0.0627 (-1.6157)

U

0.0244*

Date

Event Date

9/4/01

TE

D

Panel C: Daily returns for MSCI All Cap World Index, MSCI All Cap World Index EX US, S&P500, gold prices surrounding the September 11, 2001 event.

MSCI WORLD

MSCI World USA

t = -7

0.0027

9/5/01

t = -6

9/6/01

t = -5

9/7/01

t = -4

9/10/01

t = -1

9/11/01

10-Yr. Yield's

Gold

0.0041

-0.0006

0.0289

-0.0038

-0.0076

-0.0152

-0.0011

-0.004

-0.0037

-0.0188

-0.0145

-0.0224

-0.0221

0.0028

-0.0138

-0.0086

-0.0186

-0.0123

0.0057

-0.0037

-0.0153

0.0062

0.0083

-0.0077

t=0

-0.0159

-0.0322

0.0571

9/12/01

t=1

-0.0057

-0.0116

-0.0279

9/13/01

t=2

0.0047

0.0097

0.0045

9/14/01

t=3

-0.0093

-0.0191

9/17/01

t=6

-0.0263

0.0002

9/18/01

t=7

-0.0038

-0.0026

A

CC

EP

S&P 500 Daily

EX

-0.0151

0.0196

-0.0492

0.0131

0.0262

-0.0058

0.0194

-0.0131

33

Impact of Terrorism on World Markets?

DOW JONES -0.0157

E(r) -0.0035

AR 0.0062

6/24/05

t = -13

-0.0064

-0.0073

-0.0119

-0.0032

-0.0041

6/27/05

t = -10

-0.0024

-0.0068

-0.0007

-0.001

-0.0058

6/28/05

t = -9

0.0059

0.009

0.0112

0.0036

0.0054

6/29/05

t = -8

-0.0001

0.0034

-0.003

0.0002

0.0032

6/30/05

t = -7

-0.0043

0.0008

-0.0096

-0.0021

0.0029

7/1/05

t = -6

-0.0009

0.0097

0.0028

-0.0002

U

7/4/05

t = -3

-0.0003

0.0042

7/5/05

t = -2

0.0039

0.0008

0.0066

0.0025

-0.0017

7/6/05

t = -1

-0.0017

0.0079

-0.0097

-0.0007

0.0086

7/7/05

t=0

-0.0042

-0.0137

0.0031

-0.002

-0.0117

7/8/05

t=1

0.0101

0.0145

0.0143

0.0059

0.0086

7/11/05

t=4

0.01

0.002

0.0068

0.0058

-0.0038

7/12/05

t=5

7/13/05

t=6

7/14/05

t=7

7/15/05

t=8

CC

DJIA Index

0.0098

N

A

0.0001

D

6/23/05

TE

Event Date

0.0041

0.0046

-0.0054

-0.0006

0.0028

-0.0083

-0.0019

0.0052

0.0041

-0.0008

0.006

0.0035

0.0021

0.0068

0.0022

-0.0001

-0.0015

-0.0056

0.0011

-0.0005

-0.0051

EP

Date

7/18/05

t = 11

-0.0026

-0.0034

-0.0062

-0.0011

-0.0023

7/19/05

t = 12

0.0023

-0.003

0.0068

0.0015

-0.0045

7/20/05

t = 13

0.0025

0.0024

0.004

0.0017

0.0007

7/21/05

t = 14

0.0023

0.0012

-0.0057

0.0016

-0.0004

A

CAR 0.0062 (1.3548) 0.0022 (-0.8819) -0.0036 (-1.25) 0.0018 (1.1711) 0.005 (0.6904) 0.0079 (0.6251) 0.0177* (2.1377) 0.0218 (0.8934) 0.0201 (-0.3637) 0.0288* (1.8734) 0.0171* (-2.5358) 0.0257* (1.87) 0.0219 (-0.8206) 0.0136 (-1.7999) 0.0196 (1.3039) 0.0196 (-0.0175) 0.0145 (-1.1075) 0.0122 (-0.4976) 0.0077 (-0.9806) 0.0084 (0.1508) 0.0079 (-0.0923)

E(r) -0.0076

AR -0.0081

-0.0071

-0.0047

-0.0029

0.0022

0.0061

0.0051

-0.0004

-0.0026

-0.0049

-0.0047

-0.0012

0.004

SC RI PT

MSCI UK 0.0028

UK Index

t = -14

MSCI WORLD -0.0068

M

Table 6: Regression estimates of the MSCI UK Index and the Dow Jones Industrial Average for the London Tube Bombing event on July 7, 2005, for event window (-14 days to +14 days). An asterisk indicates that the coefficient is statistically significant at the 1% level. Missing event dates are weekend days with no trading activity.

-0.0006 0.0039

0.0027

-0.0022

-0.0076

-0.0048

0.0079

0.0106

0.0037

0.0104

-0.0037

0.0047

-0.0052

-0.0023

0.0065

0.0035

0.0033

-0.0019

0.003

-0.0031

-0.0031

0.0021

0.0046

0.0024

0.0016

0.0022

-0.008

CAR -0.0081* (-2.1392) -0.0129 (-1.2403) -0.0107 (0.569) -0.0056 (1.334) -0.0082 (-0.6832) -0.0129 (-1.2259) -0.0089 (1.0461) -0.0089 -0.0062 (0.7119) -0.0138* (-1.9904) -0.0059* (2.0723) -0.0022 (0.9717) -0.0059 (-0.9687) -0.0111 (-1.3673) -0.0046 (1.696) -0.0013 (0.8703) 0.0017 (0.7903) -0.0014 (-0.823) 0.0032 (1.214) 0.0048 (0.4166) -0.0032* (-2.0915)

34

Impact of Terrorism on World Markets?

Table 7: Actual, expected and cumulative returns for the S&P 500 on July 7, 2005. Two alternative benchmarks, the MSCI World AC Index, and the MSCI World AC Index ExUSA were used in the regressions. An asterisk indicates that the coefficient is statistically significant at the 1% level. Missing event dates are weekend days with no trading activity. Event Date

6/30/05

t = -7

7/1/05

t = -6

7/4/05

t = -3

-0.0007

7/5/05

t = -2

0.0037

7/6/05

t = -1

MSCI World AC Index E(r) S&P 500

AR S&P 500

-0.0047

-0.0024

MSCI World AC Index Ex-USA CAR S&P 500 -0.0024

E(r) S&P 500

AR S&P 500

-0.0003

-0.0068

(-0.65) -0.0007

0.0033

(-1.0482)

0.0009

-0.0011

(0.8884) 0.0009

-0.0002

0.006

-0.0006

(1.3921) -0.0066

-0.0006

0.0018

N

-0.0017

0.0037

-0.0031

0.0094

-0.0049

0.0073

0.0067*

A

t=0

-0.0035

-0.0102

0.0107

0.001

0.0078

M

t=1

0.006

t=6

7/14/05

t=7

-0.002

EP

7/13/05

0.0056

-0.0053

D

t=5

0.0115

-0.0034

TE

7/12/05

t=4

0.0026

0.0091

-0.0014

(1.4032) 0.005

0.0013

(-1.4275) -0.0009

0.0028

-0.0005

(-0.3724)

0.012 (-0.0794)

-0.0012

0.0021

0.0018

0.0009

(0.771) 0.0006

0.0125 (0.2013)

(-0.9053) 0.002

0.0112

0.0141 (0.3201) 0.015 (0.1328)

A

CC

0.004

0.0029

0.0025

0.0022 (0.9252)

(0.2757)

7/11/05

-0.0038 (-1.5695)

(1.9809) 7/8/05

0.0063

(1.4488)

(-1.793) 7/7/05

-0.0031

(0.5766)

U

0.0052

CAR S&P 500 -0.0068

SC RI PT

Date

35