1. Introduction

Terrorism has several adverse economic and financial effects such as a reduction in the physical capital, increased costs of financial and other counter-terrorism regulations, damage of critical infrastructure (power plants, nuclear facilities, chemical factories, bridges, pipelines and water supply), increased economic instability, destruction of market infrastructure and operations and a decrease in investor confidence (Chesney, Reshetar, & Karaman, 2011)

Enders, Sandler, & Gaibulloev (2011) define Terrorism as “a premeditated use or threat to use violence by individuals or subnational groups against noncombatants to obtain a political or social objective through the intimidation of a large audience beyond that of the immediate victims”. This intimidation, together with the destruction of assets, economic growth, and life perception reduction, generate effects in the financial markets. In this regard, Abadie & Gardeazabal (2008) observe a depression in growth of 10% of the Basque country during the 1998-2000 period, similar to what was found by (Karolyi & Martell, 2010), who indicate that every terrorist event generates, on average, a negative reaction of a -0.83% in prices of public companies. In the same line, (Drakos, 2004) observes structural changes in the systematic risk (beta) in the airline business after 9/11.

A terrorist event should affect the price of public companies according to the present value theory Campbell & Shiller (1987), which indicates that the price of an asset depends on the expected cash flow and risk. In this regard, (Drakos & Kutan, 2003) show that a terrorist event affect the cash flow of a company in various forms; diminishes foreign investment, the demand for tourism services and as a consequence, the aggregated demand of the local economy, as well as increasing the risks for business, as the insurance companies exclude from their insurances losses generated by terrorist events. In consequence, beyond the actual terror caused by terrorism, it is reasonable to think that investors would adjust their perceptions of value in front of a specific event, which should motivate them to make investment decisions that affect the prices of public companies.

For this, we should expect to see effects from terrorist attacks of global relevance in Latin-American stocks markets even when these have occurred beyond its borders, as these attacks affect the local economy negatively undermining their growth, investments, and trade, this being relevant for companies in Latin America in regards to economic conglomerates located in the USA, Europe, and Asia. Table 1 shows the total trade of Latin-American countries with the main economics clusters. Chile stands out as an economy with high exposition to international trade (over 35% of the GDP), where 40% of its total trade is concentrated in the East, Asia, and Pacific. Similar situation, although more intense is the one in Mexico, which focuses 65% of its total trade in the USA, that represents 47% of the GPD.

Table 1 shows the percentage of the total trade of the studied Latin American countries compared to three areas of the world for 2017. Column 1 shows the countries studied; columns 2, 3 and 4 indicate the three continents/areas where terrorist attacks took place; column (a) shows the total trade of each country in the East Asia & Pacific area against their gross domestic product (GDP); column (b) represents the total trade of each country in the East Asia & Pacific area against their total trade with all the world, column (c) shows the total trade of each country in the Europe & Central Asia area compared to their gross domestic product (GDP), column (d) represents the total trade of each country in the Europe & Central Asia area in relation to their total trade with all the world, columns (e) y (f) show the total trade of each country in relation to its GDP and the world respectively, in the North America area.

Table 1

Statistics for total trade for countries in the sample

Source: Worldbank official webpage, 2019

A second argument for thinking that terrorist events should significantly affect the Latin-American stock markets is the contagion phenomenon. The technological progress, the rising in foreign investment, and the need to diversify investments has increased the contagion risk of negative shocks. Table 2 shows the 13 terrorist events with the highest impact since the attack on the twin towers of 2001, and its effect over the stock indexes, although in different magnitudes. For instance, the terrorist attacks in the United States led to the most significant stock market crashes, along with an active trading connection between Latin American economies and the United States.

The Table 2 shows the variation observed in the price of the main stock indexes of each country the day of the event in relation to the closing price of the day before. Each row represents the 13 events studied, in the last row the percentage of events with negative returns can be observed. *For the case of Colombia, the effects are studied starting event 6, as there is no data of stock indexed for previous attacks.

Table 2

Latin-American stock indexes returns on the day of the attack

A third argument is the co-integration between financial markets. When countries are economically and financially integrated, generate a strong correlation, and co-integration in their stock markets (Donadelli & Paradiso, 2014). As a consequence, investors decrease their diversification opportunities and enhance the contagion effect between countries and a higher country risk before a worldwide crisis (Gagnon, Power, & Toupin, 2016). In this regard, Hardy, Magner, Lavin, Cardenas, & Jara-Bertin (2018) find evidence of co-integration between MILA markets when co-integration tests are applied that extend the results delivered by Mellado & Escobari (2015) and Sandoval & Soto (2016).

Despite these conceptual arguments that lead to expect significant effects of terrorist events over the Latin- American stock markets, other might explain the absence of such an impact. The main one is related to their low stock market efficiency or poor reaction, also named as efficiency market hypothesis. A study of the Mexican Stock Market Bhattacharya, Daouk, Jorgenson, & Kehr (2000) find evidence of reduced efficiency, observed through a weak reaction of the stock prices before relevant events. There are four explanations for this phenomenon: (a) the stock market is informationally inefficient; (b) news are not relevant to determine the value of the company; (c) news are entirely anticipated for the market and; (d) the market has an active presence of insider traders. In the same line Hardy et, al., (2018) find evidence that, even after the MILA integration of four important Latin American stock markets (Chile, Colombia, Perú y México), market efficiency is weak because relevant events not generate a reaction of prices.

In this way, our motivation in this paper is to study the effect of a terrorist attack in the main Latin-American stock markets. We think that this is important for two reasons. First, because the majority of the studies focus on the impact of a terrorist attack on the local economy, while a few papers about the indirect effect on other countries, not considering the existence of a global impact and limiting the possibilities of investment risk management through international diversification. Second, literature is focused on the reaction of developed stock markets, while small fraction studies emerging markets and, to the best of our knowledge, absent to Latin-America, even when Arin, Ciferri, & Spagnolo (2008) find that the adverse effect of terrorism is more significant in emerging countries.

To achieve our purpose, we selected six Latin-American Stock markets and collected the returns of stocks with transactions between the 2000- 2018 period. Then, we choose 13 terrorist attacks in areas with a strong relation with Latin-America, and we implement an event study methodology based on the capital asset market model. Finally, we estimate the terrorism impact with the statistical significance of the cumulative abnormal returns for two days before the attack. For management autocorrelation, multicollinearity and heteroscedasticity problems in our expected generation market model (market model OLS, see section 3.2), we include eight statistical tests, five parametric tests, and three nonparametric tests (see section 3.3 for obtaining a detailed explanation). We believe that with this strategy, we can work with the OLS residues obtained by those that present typical problems in their specifications.

The result indicates a weak reaction of the Latin-American stock markets except for US terrorist attacks. For example, during the 9/11 attack, the markets experienced 6,1% of cumulative abnormal returns. Contrary to this, during the London attack, the cumulative abnormal return was 1% positive although with little statistical significance. But the strength of the impact change over the stock markets. In this sense, Brazil is the stock market with the most negative reaction, while Argentina presents a positive one. The results also change between sectors when the most sensitive sectors are energy and industry.

We think that these results have many practical implications oriented to improve the understanding of a new type of catastrophic risk that investors and financial markets must face, quantify and manage. For example, an investor could consider investing their money in Argentina, Colombia, or Mexico in time of terrorism, or increase their exposition in sectors out of energy or industrial sectors.

The rest of the paper is structured as follows: in section 2 we present evidence that shows the effect of terrorist attacks. In section 3 we describe the database and methodology. In section 4 we present results. Finally, in section 5 we conclude.

2. Terrorism and financial markets

The effects of terrorism in the financial markets have been widely documented (see a survey in Kollias, Kyrtsou, & Papadamou (2013)). However, only a few catch the attention in the literature, especially the 9/11 attacks of 2001 in the United States. Terrorist attacks have been a dark side of Latin American history since the ’70s. However, following the 9/11 attacks, these events have concentrated in the Middle East, since the invasion of Afghanistan in 2001 and Iraq in 2003, and they account for more than 3.000 deaths, besides from the ones in the United States and Europe (Global Terrorism Database, 2019). In fact, since the 9/11 attacks, terrorist events have not only been a concern in terms of safety or fear, but also for their impacts on the economy. In this regard, (Goel, Cagle, & Shawky, 2017) calculate that the economic impact of terrorism is measured in billions of dollars annually, with the most in 2014 at 52.9 billion.

The psychosocial nature of the markets explain the adverse financial effects of the terrorist attacks, where investors see a negative impact in the valuation for the panic generated by terrorism (Hirshleifer, 2001; Boubaker, Farag & Nguyen, 2015). In this sense, and despite the critical argument of the physical losses of assets, Johnston and Nedelescu (2005) evidence that the financial markets are not only confronted with significant disruptions caused by the massive damage to property and communication systems, but also by the high levels of uncertainty and market volatility. Complementary, Nikkinen, Omran, Sahlstrom & Aijo (2006) say that technological progress causes severe reactions in front of news that may threaten the future of their investments.

From this perspective, Drakos (2010) suggests that terrorist attacks lead to significantly lower returns on the day a terrorist attack occurs. However, there is no complete consensus on the literature. In this regard, Kollias et al., (2013) found opposite results. For instance, following the Madrid terrorist attack, abnormal returns can be verified in most sectors of Spain, while in the London attacks these effects are not statistically significant.

Concerning the duration of the effects, it is possible to classify between temporary effects, those that occurred within the following days to the attack, and long term effects, those that are related to changes in the systematic risk. Drakos (2004) observes that following the 9/11 attacks in the USA companies from the airline’s sector suffered high losses in the stock prices and changed their relationship with the stock index significantly increasing their beta. In contrast, Kollias et al., (2013) studied the reaction of public companies of the Spain and UK and observed negative effects on the first days following the Madrid and London attacks, respectively, although said effects disappeared after a few days.

The effect on the stock markets is not constant through all the sectors. More specifically, Chesney et al., (2011) find that airline, travel, tourism, accommodation, restaurant, postal and insurance industries are particularly susceptible to increased terrorism risks and, in consequence, economies with more concentration for these industries hurt most from falls in output and employment.

3. Latin-American security market daily returns

3.1 Data

The dataset employed in our analysis consists of the daily closing prices for 115 stocks³ covering the 2001 - 2018 period (See table 3). Then, we consider 13 terrorist attacks with global communicational effects outside Latin-America (see table 4).

The Table 3 shows the descriptive statistics by country including: Numbers of companies considered in the study for each country; Average abnormal return (AAR) for the event window; Standard deviation of the abnormal returns; Average of abnormal cumulative return(CAR); Standard deviation of average abnormal cumulative return; Standard deviation of average abnormal cumulative return on absolute value; and percentage of zero returns for each country that has the database, on average, the days that price of companies registered non trading.

Table 3

Descriptive statistic

The following table shows the 13 events considered in this investigation. Columns 1 and 2 show the number and name of the event; columns 3 and 4 show the location of the event, country, and continent, respectively; column 5 shows the date of the terrorist attack.

Table 4

Detail of studied events

3.2 Event study methodology

The event study methodology (See MacKinlay (1997) for a survey and Corrado (2011) for a good update and recommendations) has two stages: abnormal returns estimation process and event study test statistics for abnormal returns. Since Fama, Fisher Jensen, & Roll (1969), abnormal (excess of) return is typically calculated from the market model (especially in the daily analysis). In this sense, $A{R}_{i,h}$ is the abnormal return of a security $h$ on day $i$, $R_{i,h}$ is the realized return and $R{M}_{i,h}$ is the market index⁴. As the market model is a linear regression (estimated by OLS), the parameters ${\alpha }_i$ and ${\beta }_i$ are the intercept and slope estimates, respectively.

Let $h=0$ denote the day of a hypothetical event for a specific security. Besides, a 200-day estimation period is required; therefore, the event window is defined as $h\in \left(0;-2\right)$ for each event date and security combination.

Also, n denotes the length of the event window and m the sample size. We define $AA{R}_{n_i}$ as the average abnormal return and $PE\left(AA{R}_{n_i}\right)$ its prediction standard error as in Salinger (1992):

Here, $AR{M}_{m_i}$ represents the average return of the market index for the estimation window and ${\sigma }_{m_i}$ the standard error from the estimation window regression. Under a null hypothesis of no abnormal return, $\frac{AA{R}_{n_i}}{PE\left(AA{R}_{n_i}\right)}$ is distributed t-Student with $m_i-2$ degrees of freedom (assuming identical, independent and normally distributed data).

3.3. Event study tests

For this research, we applied parametric and nonparametric tests. We included more tests because the standard event study tests cannot be used correctly in Latin-American stocks markets with a higher degree of variability in liquidity and return distributions, until an empirical specification study. Kearney (2012) present that emerging stock markets are significantly less informationally efficient than their developed counterparts. He explain that their limited transparency, scarce corporate information, weak regulation, low auditing standards, lax disclosure requirements, dominant players who can move prices away from intrinsic values, incompletely developed “culture(s) of equity” that reduce the reaction times of market participants to new information, and the resulting illiquidity that impedes the market’s ability to accommodate orders.

3.3.1. Parametric tests

1. a. T-test

   1. The t-test is the simplest statistic test uses for single firms in each time point $t$ .

1. 1. Where $S_{A{R}_i}$ is the standard deviation of the abnormal returns in the window. Then, we calculate the statistics of the accumulated abnormal returns (CAR) for each company.

1. b. Patell t-test (Patell, 1976)

1. 1. Under the null hypothesis of no abnormal returns, $T_{Patell}$ must theoretically distribute t-student with N degrees of freedom, thus, asymptotically as standard normal.

1. c. Boehmer test (Boehmer et al., 1991)

   1. The Boehmer test of standardized residuals corrected for event-induced changes in volatility. This test is similar to the Patell t-test, but it considers an adjustment due to induced volatility in the cross-sectional event window.

1. 1. Where

1. 1. The $T_{Adjusted}$ distributes asymptotically under the null hypothesis of no abnormal return as standard normal

1. d. Adjusted standardized Cross-section test (Kolari and Pynnon, 2010)

   1. The Kolari test of standardized residuals corrected for event-induced changes in volatility and cross-correlation. This test proposes a modification to the Boehmer-test to account for cross-correlation of the abnormal returns. The context of this investigation is important because the events are on the same dates for all the stocks in the market. The correlation between financial stocks is a stylized fact in the literature, and it is important to leave out that the results are dragged by correlations between assets. We use the standardized abnormal returns and defining r as the average of the sample cross-correlation of the estimation period abnormal returns.

1. e. Crude dependence (Brown and Warner et al., 1980,1985)

   1. CDA is a parametric test, which measures if the average abnormal return time series during the event was statistically different from that of the estimation window. To execute it, a single variance estimate for the entire sample is assumed, so the test does not account for the unequal return variances across securities. In 1985, Brown and Warner suggested that the standard deviation of average residuals should be estimated from the time series of the average abnormal returns over the estimation period. The advantage from CDA is that it avoids the potential problem of cross-sectional correlation of security return.

3.3.2 Non-parametric test

1. a. Rank test (Corrado, 1989)

   1. The Corrado test is a non-parametric test which considers the ranks of abnormal returns (an adaptation of the popular rank test). Let’s consider Rank( $A{R}_{i,h}$ ) as the rank of the abnormal return in the complete sample (estimation window and event window). Then, under the null hypothesis, each rank is a uniform random drawing of integers from 1 through the complete sample. The Corrado statistic is calculated as follows:

1. b. Adjust Rank Test (Corrado and Zivney, 1992)

   1. The Corrado and Zivney rank test corrected for event-induced volatility of rankings. For cumulative (average) abnormal returns, the aggregation formula in Cowan (1992, p. 346) is applied. For this, it is necessary to define the Rank ( $SA{R}_{i,h}$ ) as the new ranking of the abnormal returns considering the complete sample, that is, for the estimation window and the event window.

1. c. Sign test

   1. The main issue in the sign test is the non-symmetric distribution of abnormal returns (Brown and Warner, 1985). In emerging markets, large skewness values are expected. We consider Sign( $A{R}_{i,h}-x_i$ ) the sign of the difference between the abnormal return and a specific value of x. In addition, we consider p as the proportion of positive signs in the estimation window for each security.

1. 1. where

   2. $Sign(A{R}_{i,h}-x_i)=\mathrm{ }\mathrm{ }1\mathrm{ }\mathrm{ }if\mathrm{ }\mathrm{ }A{R}_{i,h}>x_i$ and $Sign(A{R}_{i,h}-x_i)=0$ otherwise

   3. As evaluated in Cowan (1992), the first generalization of the sign test establishes $x_i=0$ (by using the mean of abnormal returns for the estimation period, which by construction, is zero for the market model). In this sense, $p_i$ is the proportion of abnormal returns greater than 0.

4. LATAM stock market reaction and terrorist attacks

To research the effect of terrorism in Latin-American stocks markets, we calculate CAAR for 115 companies in the day of the terrorist attack and the next two days. To expand our results, we repeated this process for 13 international terrorist attacks and applied eight parametric and non-parametric test to determine the stock market reaction. Additionally, we estimate the overall all our results using a GARCH(1,1) model for generating non-normal returns to evaluate the dependence of the non-normal generating model (Bacmann and Dubois (2003) for a survey). Table 5 indicates that the most critical event for LATAM stock markets was the 9/11 terrorist attack (CAAR: - 6,1%). Besides the magnitude, this event caused a significant reaction in all markets, independent of the test applied. We observe a similar effect in the Boston attack, evidence of the strong dependence between US and LATAM stock markets. Appendix 2 shows the same results but generate with a GARCH(1,1) model. In this sense, the events with more significance are 9/11 attack and attack in Boston, but the attack in Bombay are significant too.

Table 5

T-statistics by event

The first row shows the average cumulative abnormal return of the 13 terrorist events included in the sample; the following rows show the result of each statistical test specified in the methodology. The grey boxes indicate the statistically significant values to 10%, where the null hypothesis is rejected.

Table 6 shows the cumulative abnormal returns in the event window between the day of the terrorist attack and the next two days. On average, we show that the Latin-American companies react with positives (returns 0,2% on average). However, the direction of the reaction is different between countries. Mexico, Argentina, and Colombia show a positive response, while Chile, Peru, and Brazil present an adverse reaction. These results are similar when we change the non-normal return model. Appendix 3 shows the reaction for all test by countries using a GARCH (1,1) model. Argentine has the major percentage of reaction, and Mexico has less percentage. To evaluate the statistical significance, we applied eight tests. The results are mixed and statistically insignificant. On average, the percentage of reaction is 15, 6% with a minimum rate of the 4,8% for the sign test, and 29,8% for the Patell test.

Table 6

Percentage of reaction of companies by country and type of test

Table 6 shows the percentage of companies that react to the events, classified by country. Row (2) indicates the number of people considered by country, row (3) shows the average cumulative abnormal return by country; the following rows show the percentage of companies that react significantly to 10% according to the different statistical tests described in the methodology. The last row shows the average reaction percentage by country; likewise, the last column shows the average reaction by statistical test.

The effects of the terrorist attacks are different between industries. Table 7 shows CAAR for seven industries according to GICS classification. Sector energy and industry have the most reaction (17 and 16% of reaction, respectively), while the Materials and Communication services have the lowest reaction rates (7% of response on average). However, the impact can be changed with the test used. For example, rank test adjusted shows that sectors consumer discretionary and communication service have the most significant reaction. The differences within sectors are in line with the literature. Raby (2003) finds that airline, travel, tourism, accommodation, restaurant, postal, and insurance industries are particularly susceptible.

For deepen our evidence, and approach the capital market model limitations, we use a Garch (1,1) method and replicate all our analysis. Appendix 4 shows that energy and industrial sectors have a high rate of reaction, similar to material and consumer staples sectors. This test is evidence that the global results do not depend on the non-normal return generating method.

Table 7 shows the percentage of companies that react to the events, classified by their respective sectors, according to the Global Industry Classification Standard. Row (2) indicates the number of companies considered by sector, row (3) shows the average cumulative abnormal return by sector; the following rows show the percentage of companies that react significantly to 10% according to the different statistical tests described in the methodology. The last row shows the average reaction percentage by sector; likewise, the last column shows the average reaction by statistical test.

Table 7

Percentage of reaction by sector according to the eight types of tests

5. Conclusions

We observe opposite effects of global terrorist attacks in Latin-American stock markets. In specific, we show a negative and significant impact of the US terrorist attacks and weak and non-significant effect when the attacks occur outside the US.

This effect varies through different countries and sectors. The most sensible markets were Brazil, Peru, and Chile, while the most sensitive sectors were energy and communications.

We have some explanation for this contradictory behavior. The adverse effect of US terrorist attacks can be understood by the strong correlation between the Latin American Stock Markets and those located in North America. Another explanation is that investors adjust their expectations when an attack occurs anticipating a drop in the cash flow of Latin American companies traded in local markets with economic connection with the United States. Unfortunately, a limitation of our work is that our methodology does not allow to distinguish between these explanations.

For the other hand, we explain the poor reaction for the other terrorist attacks with the low efficiency of Latin American Stocks Markets. Complementary to this, we considerate that the errors in the statistical tests derived from liquidity problems and abnormal returns present in emerging markets can justify an insignificant response. As before, our methodology does not allow us to separate these explanations.

An application of these results is that the low reaction in Latin American markets makes them a good diversification option for international investment funds in front of terrorism, due to their weak response and low probability of a terrorist attack.

We think that study the effect of the terrorist attacks contributes to understanding better the phenomenon in the financial markets. Their violent, unexpected, sudden, and viral nature make them an attractive laboratory to evaluate how information influences financial markets; after all, is difficult to think that markets would not react to terrorist attack based on explanations related to the use of privileged information (it is not impossible, but it is difficult to think that an investor would know of an attack beforehand) or the lack of attention (the news explode in all media)

An extension of this work is to continue studying in depth the relation between social phenomena, financial ones, and understand psychosocial variables, such as fear and terror, and how these influence investment decisions, especially in contexts of emerging economies where the evidence is scarce.

References

Abadie, A., & Gardeazabal, J. (1999). The economic costs of conflict: A case study of the Basque country. American Economic Review, 93(1), 113-132. https://doi.org/10.1257/000282803321455188

Arin, K. P., Ciferri, D., & Spagnolo, N. (2008). The price of terror: The effects of terrorism on stock market returns and volatility. Economics Letters, 101(3), 164-167. https://doi.org/10.1016/j.econlet.2008.07.007

Bacmann, J. F., & Dubois, M. (2003). Event studies with conditionally heteroscedastic stock returns. In Proceedings of the European Financial Management Association Conference (pp. 1-29).

Bhattacharya, U., Daouk, H., Jorgenson, B., & Kehr, C. (2000). When an event is not an event: the curious case of an emerging market. Journal of Financial Economics, 55, 69-101. https://doi.org/10.1016/S0304-405X(99)00045-8

Boehmer, Musumeci, & Poulsen. (1991). Event-study methodology under conditions of event-induced variance. Journal of Financial Economics, 34, 253-272.

Boubaker, S., Farag, H., & Nguyen, D. K. (2015). Short-term overreaction to specific events: Evidence from an emerging market. Research in International Business and Finance, 35, 153-165. https://doi.org/10.1016/j.ribaf.2014.10.002

Brown, S., & Warner, J. (1980). Measuring security price performance. Journal of Financial Economics, 8, 205-258. https://doi.org/10.1016/0304-405X(80)90002-1

Brown, S., & Warner, J. (1985). Using daily stock returns: The case of event studies. Journal of Financial Economics, 14, 3-31. https://doi.org/10.1016/0304-405X(85)90042-X

Campbell, J., & Shiller, R. (1987). Cointegration and Tests of Present Value Models. Journal of Political Economy, 95(5), 1062-1088. https://doi.org/10.1086/261502

Chesney, M., Reshetar, G., & Karaman, M. (2011). The impact of terrorism on financial markets: An empirical study. Journal of Banking and Finance, 35(2), 253-267. https://doi.org/10.1016/j.jbankfin.2010.07.026

Corrado, C. J. (1989). A nonparametric test for abnormal security-price performance in event studies. Journal of Financial Economics, 23, 385-395. https://doi.org/10.1016/0304-405X(89)90064-0

Corrado, C. J. (2010). Event studies: A methodology review. Accounting and Finance, 51, 207-234. https://doi.org/10.1111/j.1467-629X.2010.00375.x

Corrado, C. J. (2011). Event studies: A methodology review. Accounting & Finance, 51(1), 207-234.

Corrado, C. J., & Zivney, T. L. (1992). The specification and power of the sign test in event study hypothesis tests using daily stock returns. Journal of Financial and Quantitative Analysis, 27(3), 465-478.

Cowan, A. R. (1992). Nonparametric event study tests. Review of Quantitative Finance and Accounting, 2(4), 343-358. https://doi.org/10.1007/BF00939016

Donadelli, M., & Paradiso, A. (2014). Does financial integration affect real exchange rate volatility and cross-country equity market returns correlation? North American Journal of Economics and Finance, 28, 206-220. https://doi.org/10.1016/j.najef.2014.03.001

Drakos, K. (2004). Terrorism-induced structural shifts in financial risk: Airline stocks in the aftermath of the September 11th terror attacks. European Journal of Political Economy, 20(2), 435-446. https://doi.org/10.1016/j.ejpoleco.2003.12.010

Drakos, K. (2010). Terrorism activity, investor sentiment, and stock returns. Review of Financial Economics, 19(3), 128-135. https://doi.org/10.1016/j.rfe.2010.01.001

Drakos, K., & Kutan, A. M. (2003). Regional effects of terrorism on tourism in three mediterranean countries. Journal of Conflict Resolution, 47(5), 621-641. https://doi.org/10.1177/0022002703258198

Enders, W., Sandler, T., & Gaibulloev, K. (2011). Domestic versus transnational terrorism: Data, decomposition, and dynamics. Journal of Peace Research, 48(3), 319-337. https://doi.org/10.1177/0022343311398926

Espinosa-Méndez, C., Gorigoitía, J., & Vieito, J. (2017). Is the Virtual Integration of Financial Markets Beneficial in Emerging Markets? Evidence from MILA. Emerging Markets Finance and Trade, 53(10), 2279-2302.

Fabozzi, F. J., & Focardi, S. M. (n.d.). Trends in quantitative finance.

Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. (1969). The adjustment of stock prices to new information. International economic review, 10(1), 1-21.

Gagnon, M. H., Power, G. J., & Toupin, D. (2016). International stock market cointegration under the risk-neutral measure. International Review of Financial Analysis, 47, 243-255. https://doi.org/10.1016/j.irfa.2016.08.002

Goel, S., Cagle, S., & Shawky, H. (2017). How vulnerable are international financial markets to terrorism? An empirical study based on terrorist incidents worldwide. Journal of Financial Stability, 33, 120-132. https://doi.org/10.1016/j.jfs.2017.11.001

Hardy, N., Magner, N., Lavin, J., Cardenas, R., & Jara-bertin, M. (2018). Small consequences of a major agreement: the MILA case. Academia Revista Latinoamericana de Administración. https://doi.org/10.1108/ARLA-12-2017-0357

Hirshleifer, D. (2001). Investor psychology and asset pricing. The Journal of Finance, 56(4), 493-494. https://doi.org/10.1111/0022-1082.00379

Karolyi, G. A., & Martell, R. (2010). Terrorism and the Stock Market. International Review of Applied Financial Issues and Economics, (2), 285-314.

Kearney, C. (2012). Emerging markets research: Trends, issues and future directions. Emerging Markets Review, 13(2), 159-183. https://doi.org/10.1016/j.ememar.2012.01.003

Kolari, J. W., & Pynnönen, S. (2010). Event study testing with cross-sectional correlation of abnormal returns. Review of Financial Studies, 23(11), 3996-4025. https://doi.org/10.1093/rfs/hhq072

Kollias, C., Kyrtsou, C., & Papadamou, S. (2013). The effects of terrorism and war on the oil price-stock index relationship. Energy Economics, 40, 743-752. https://doi.org/10.1016/j.eneco.2013.09.006

Mackinlay, A. C. (1997). Event Studies in economics and finance. Journal of Economic Literature Journal of Economic Literature, 35140104(1), 13-39.

Mellado, C., & Escobari, D. (2015). Virtual integration of financial markets: a dynamic correlation analysis of the creation of the Latin American Integrated Market. Applied Economics, 47(19), 1956-1971. https://doi.org/10.1080/00036846.2014.1002892

Nedelescu, O. M., & Johnston, M. R. (2005). The Impact of Terrorismon Financial Markets (No. 5-60). International Monetary Fund.

Nikkinen, J., Omran, M., Sahlström, P., & Äijö, J. (2006). Global stock marketreactions to scheduled U.S. macroeconomic news announcements. Global Finance Journal, 17(1), 92-104. https://doi.org/10.1016/j.gfj.2006.06.003

Patell, J. M. (1976). Corporate forecasts of earnings per share and stock price behavior: Empirical test. Journal of Accounting Research, 14(2), 246. https://doi.org/10.2307/2490543

Raby, G. (2003). The Costs of Terrorism and the Benefits of Cooperating to Combat Terrorism, Department of Foreign Affairs and Trade (Australia), Economic Analytical Unit.

Salinger, M. (1992). Standard errors in event studies. The Journal of Financial and Quantitative Analysis, 27(1), 39. https://doi.org/10.2307/2331297

Sandoval, E., & Soto, M. (2016). Mercado Integrado Latinoamericano: Un Análisis De Cointegración (Integrated Markets of Latin American: A Cointegration Analysis). Revista Internacional Administración & Finanzas, 9(1), 1-17.

Appendix 1. List of stocks companies

Appendix 2. T-statistics by event with GARCH(1,1) model.

The first row shows the average cumulative abnormal return of the 13 terrorist events included in the sample; the following rows show the result of each statistical test specified in the methodology. The grey boxes indicate the statistically significant values to 10%, where the null hypothesis is rejected.

Appendix 3. Percentage of reaction of companies by country and type of test with GARCH (1,1) model.

Table shows the percentage of companies that react to the events, classified by country. Row (2) indicates the number of people considered by country, row (3) shows the average cumulative abnormal return by country; the following rows show the percentage of companies that react significantly to 10% according to the different statistical tests described in the methodology. The last row shows the average reaction percentage by country; likewise, the last column shows the average reaction by statistical test.

Appendix 4. Percentage of reaction by sector according to the eight types of tests

Table 7 shows the percentage of companies that react to the events, classified by their respective sectors, according to the Global Industry Classification Standard. Row (2) indicates the number of companies considered by sector, row (3) shows the average cumulative abnormal return by sector; the following rows show the percentage of companies that react significantly to 10% according to the different statistical tests described in the methodology. The last row shows the average reaction percentage by sector; likewise, the last column shows the average reaction by statistical test.

Notes

Author notes

*Email: nmagner@uft.cl. Escuela de Negocios, Universidad Finis Terrae Pedro de Valdivia 1509, Providencia, Santiago, Chile. Postal code 7501015**Email: croam@uft.edu