Correlation changes between the market and three different
industries in the aftermath of the 9/11 attacks
Bachelor thesis
By:
Pim de Vor
Student number:
10251766
Supervisor:
Marijn Kool
Study:
Economie en Bedrijfskunde
Specialization:
Economie en Financiering
Date:
19-06-2014
Abstract
This paper investigates the effect the terror attacks of 11 September 2001 had on the correlation between the market and the companies of different industries. The industries investigated in this paper are the airline, the defence and the pharmaceutical industry. These industries are chosen, because of their different reaction to the 9/11 attacks in terms of their equity return. Utilizing the beta of the CAPM the correlation between the companies of the different industries and the market portfolio is measured. The betas are calculated for a period of half a year before the attacks and half a year after the attacks as for the whole sample of a year. The Chow test has been done to measure significant breaks in the betas before and after the attacks. The empirical evidence shows that six out of the ten investigated airline companies had a break in their beta, seven out of the ten companies in the defence industry and one out of the ten companies in the pharmaceutical industry. The airline and defence industry were affected the most in terms of their correlation between the market portfolio and the companies in the industries and the pharmaceutical industry almost wasn’t affected, but even the airline and defence industry reacted differently. In the airline industry all the betas of the companies with significant Chow test outcomes were higher after the attacks than before, while in the defence industry three were higher and four were lower. Striking is that in the defence industry the companies that had significant Chow test outcomes were almost all US companies. In the airline industry this was even clearer, because all the US airline companies had a significant Chow test outcome and all the US companies had a shift from defensive stocks before the attacks to aggressive stocks after the attacks, while none of the other airline companies with
significance had this.
Table of contents:
1. Introduction 4
2. Literature Review 5
3. Data and Methodology 6
4. Results 8 4.1 Airline industry 8 4.2 Defence industry 9 4.3 Pharmaceutical industry 11 5. Conclusion 13 References 14 3
1. Introduction
The 11 September attacks caused a lot of damage to the citizens of the city New York and also to the United States as whole. There were more than 3000 deadly victims and there was uncertainty and fear for more attacks. This fear and uncertainty from the 9/11 attacks also affected the economy and the financial markets all over the world, but Wall Street and the New York Stock exchange were affected the most. One hour after the second plane crashed in the second tower of the World Trade Center Wall Street closed to trading. Wall Street was closed until Monday 17 September and when it reopened again the New York Stock Exchange had a negative return of 7.1%. This was the longest shutdown since 1933 and the biggest loss in exchange history for one trading day. Suspicion for the responsibility of the attacks fell on Osama Bin Laden from al-Qaeda, who had hid himself in the mountains of Afghanistan. The United States reacted directly after the attacks by launching the war on terror and later on invaded Afghanistan to depose the Taliban, which harboured Bin Laden’s al-Qaeda. This was the first step in a global war on terror, where the United States were joined by the NATO and many other countries. The 9/11 attacks caused a whole new perspective on terror attacks and terrorism with all her consequences.
The economic consequences of the 9/11 attacks where far greater than any other terrorist aggression in the recent history. Lenain, Bonturi and Koen (2002) said that the destruction of physical assets was estimated to amount to $14 billion for private businesses, $1.5 billion for State and local government enterprises and $0.7 billion for Federal government. Rescue, clean-up and related costs have been estimated to amount to at least $11 billion. Lower Manhattan lost almost 30% of its office space and 200 000 thousand jobs were eliminated or had to be reallocated (p. 6). These were
however only the short term consequences of the attacks. The commercial aviation system suffered from a temporary but complete shut-down in the days after the attacks. The fear and panic of the aviation users caused by the attacks also affected the aviation industry negatively. Ito and Lee (2005) find that the 9/11 attacks caused a negative US airline demand shock of more than 30% that can’t be explained by cyclical, seasonal or other factors (p. 94). Cam (2008) investigated 135 industries and concluded that most industries had abnormal equity returns in the week after the 9/11 attacks. The airline industry had the biggest negative equity shock in the week after the 9/11 attacks, with a negative return of 35.6%. While the defence industry had the biggest positive equity shock in the week after the 9/11 attacks, with a positive return of 13.9%. (p. 115).
This paper will test how different industries reacted to the 9/11 attacks in terms of their correlation to the market. The focus will be on three industries, namely the airline industry, the defence industry and the pharmaceutical industry. These industries are selected, because of their positive, negative or neutral reaction in terms of abnormal stock returns to the 9/11 attacks and therefore it should be possible to find differences in the correlation between the market and the firms of these industries. To see if the correlation between the market and the firms of these industries changed before and after the attacks, there will be stock price data collected from a half year before and after the 9/11 attacks. These data will be analysed and tested in Stata and through this it will hopefully be possible to say something about a difference in correlation between the market and the firms in these industries before and after the attacks. The test used to see if the correlation between the companies and the market changed is the Chow test, the betas on the other hand are calculated by ordinary least squares regressions.
To see if the correlation between the market and the firms in these three industries changed before and after the attacks and if there is s difference between the industries, the research question will be: To what extent did the 11 September 2001 attacks affect the correlation between the market portfolio and the stocks of the firms in the airline industry, defence industry and the pharmaceutical industry? To find an answer to this question there will firstly be a literature review of the existing literature about the effect of the 9/11 attacks on the stock market. After this the source of the data is described and there will be discussed which methodology will be used to answer the research
question. After that the results of the investigation will be shown and these results will be discussed.
At last there will be a conclusion which answers the research question and reflects on the results in this paper.
2. Literature Review
The effects that world events have on the financial markets and especially stock markets has been investigated for decades. Interesting for this investigation is to see in which way terrorist attacks affect the stock market. Brounen and Derwall (2009) measured the financial impact of 31 terrorist attacks since 1990 by analysing the price dynamics of international stock markets (p. 586). They found an immediately price reaction on the day of the terrorist attack of -0.34% and they concluded that these results were severely weakened when the 9/11 attacks are left out the sample (Brounen & Derwall, 2009, p. 596). This means that the 9/11 attacks had a big financial impact on the dynamics of the international stock market. Brounen and Derwall (2009) found that only the 9/11 attacks out of the 31 terrorist attacks continued to be a significant source of market disruptions and the most pervasive impact the 9/11 attacks had on the financial market was a change in the systematic market risk. They also concluded that price reactions were strongest in markets and industries that were directly affected by the attacks (p. 597).
Other papers investigated why and how terrorist attacks affected the stock markets. Chen and Siems (2004) investigated the response of global capital markets to terrorist and military attacks. They concluded that terrorist attacks and military invasions have a great potential to effect global capital markets in a short period of time (p. 350). The reason for this is that global capital markets are very inter-linked today and news spreads rapidly with quick spill over or contagion effects (Chen and Siems, 2004, p. 363). The contagion effect occurs when there is a significant economic change in one industry or country and this spreads to other industries or countries, like a medical disease. The spill over effect occurs when there are externalities of economic activity or processes that affect other economies or industries that are not directly involved. Charles and Darné (2006) investigated the effects of the 9/11 attacks on 10 daily stock market indexes. They used an outlier detection methodology to associate the large shocks due the 9/11 attacks to the presence of outliers and to see of these outliers where immediate, temporally or permanent (p. 684). The outcomes of Charles and Darné (2006) were that the international stock markets experienced large permanent and temporary shocks in response to the 9/11 attacks and its aftermath (p. 696). Charles and Darné (2006) concluded like Chen and Siems (2004) that US macroeconomic news announcements can have a great impact on the US and European stock markets, caused by the contagion effect (p. 684).
The papers reviewed till now were about the general effects of terrorist attacks on stock markets. The papers reviewed next are written over the specific industries investigated. Both Carter and Simkins (2004) and Drakos (2004) studied the effect of the 9/11 attacks on the airline industry. It is interesting to see the outcomes of these papers, because it gives more background on how the airline industry reacted to the 9/11 attacks. Carter and Simkins (2004) studied the stock-price reaction of firms in the airline industry after the 9/11 attacks to see if the industry reacted rationally to the event. They tested among other the hypothesis whether the market reaction was the same for each air-transport firm (p. 540). Carter and Simkins (2004) find significant negative abnormal returns for each of the airlines studied and smaller returns for airfreight firms and international airlines. From these findings they conclude that the market expected that the long-term consequences for US airlines are much more significant than for airfreight firms and international airlines (p. 555). No statistical significance was found for firm size, leverage and firm performance or for the airline companies that were involved in the hijacking (Carter and Simkins, 2004, p. 555).
Drakos (2004) studied the effect of terrorism on airline stocks in terms of their fundamental risk. This is relevant for this present paper, because Drakos used the Chow test to see if there was a break in systematic risk which comes close to the research in this paper. Drakos said that the 9/11 attacks are of particular interest for the effects on airline stocks, because the attacks revealed major deficiencies in airport securities and therefore extra spending for these companies was needed.
Secondly, the attacks led to the war on terror that affected traveling plans and thirdly, it was the first time airplanes were hijacked by terrorist to commit suicide attacks, rather than using the hijacking itself as purpose (p. 437). All these three reasons had a big impact on the airline industry and
therefore the 9/11 attacks caused an interesting field to study. The conclusion of Drakos (2004) study was that systematic risk of airline stocks had significantly increased since the 9/11 attacks and more than doubled compared to the level before the 9/11 attacks. Drakos (2004) also found a structural break in systematic risk after the 9/11 attacks for the companies investigated in the airline industry (p. 445). This is in line with Brounen and Derwall (2009) which also concluded that there was change in systematic risk after the 9/11 attacks (p. 597). Also the volatility has increased in the post 9/11 period, reflecting the increased uncertainty in the airline industry (Drakos, 2004, p. 445).
There are also a few papers about the effect of terrorism on the defence industry. Berrebi & Klor (2005) studied the effect of terrorism in Israel on Israeli companies that are traded on American markets (p. 1). The paper of Berrebi and Klor will possibly come closest to this investigation, because they looked among others if terrorism affects the defence industry differently than other industries. The main contribution of the paper is that it shows that the impact of terrorism varies across companies in different industries. In their results they find evidence that strongly suggest that terrorism has a positive effect on the stock market valuation of among others companies in the defence industry (Berrebi & Klor, 2005, p. 20). Berrebi and Klor (2005) also find that terror attacks in Israel are positively correlated with Israeli defence exports even after controlling for the level of terrorism abroad and the observed defence expenditures of the importing countries (p. 2).
This paper contributes to the existing literature by looking at the correlation between the market and firms of different industries. Drakos (2004) looked at correlation between the market and the airline industry, but didn’t look if other industries had the same reactions or changes. This is the first way this paper will contribute, because in this study the airline, defence and pharmaceutical industry will be tested if their correlation with the market changed after the 9/11 attacks. There are numerous studies done about the effect of terrorism and also specifically the 9/11 attacks for the economy and stock returns, but none of them looked if these different industries reacted differently in terms of their correlation with the market. This is the second way this paper will contribute to the existing literature, because in this study the effects of the 9/11 attacks will be measured in terms of the correlation between the market and companies from different industries.
3. Data and Methodology
The aim of this study is to test if and how different industries reacted different towards the 9/11 attacks in terms of their correlation with the market. To see if this has happened three industries are investigated that all reacted different to the 9/11 attacks. For one of the industries there has to be a negative effect, for one a positive effect and for one a neutral effect on their stock prices towards the 9/11 attacks. Cam (2008) showed in her study that from the 135 industries she investigated the airline industry had the biggest negative abnormal equity return in the week after the 9/11 attacks (p. 123). This could best be explained by the fact that this was the first time terrorist used airplanes to commit their suicide attacks, which caused aversion to flying for companies and individuals. The defence industry showed the biggest positive abnormal equity return in the week after the 9/11 attacks (Cam, 2008, p. 128). The reason for this could be that after the 9/11 attacks the United States launched the war on terror which was supported by many countries and the NATO. Together with that the US made plans to invade Afghanistan, this could be a reason for the positive returns equity returns for the defence industry. For the third industry that will be used as a control variable it is necessary that it reacted neutral to the 9/11 attacks, because then it is possible to see of the different industries reacted different in terms of their correlation to the market. Cam (2008) shows also thirty industries which did not have abnormal equity returns in the week after the 9/11 attacks (p. 129). One of these was the pharmaceutical industry. From these thirty industries the
pharmaceutical industry is one of the biggest and oldest and therefore the firms in the
pharmaceutical industry probably will almost not be affected by other events and market disruptions that happened in the investigated time period. This is important to study the effects of the 9/11 attacks on this third neutral reacting industry.
Out of the airline, defence and pharmaceutical industry this study made use of the ten biggest companies in terms of yearly revenue. This done to obtain the least biased results, since larger companies are most likely less vulnerable for outside influences compared to smaller companies. Daily stock returns have been calculated for all the ten companies of these three industries. The daily unadjusted prices of the stocks are collected from DataStream. The prices are collected from 11-03-2001, half a year before the 9/11 attacks until 10-03-2002, half a year after the 9/11 attacks. Most asset studies use monthly data for asset price studies. Lewellen and Nagel (2006) used in their research weekly or daily stock returns since the intervals of their regression were short (p. 297). This raises some concern compared to the monthly data, but according to Lewellen and Nagel (2006) the effects this causes are tiny and can be ignored (p. 298). Lewellen and Nagel say that the effects of using daily stock returns compared to monthly stock returns can be ignored. The effects of the 9/11 attacks on stocks are most interesting just before and after the attacks. For this reason and the outcomes of Lewellen and Nagel this study will make use of daily stock returns. This also makes it possible to get enough observations for the short regression interval used, namely half a trading year before and after the 9/11 attacks.
There are some adjustments made in the dataset used. The ten biggest airlines at this
moment are different than they were in 2001. The International Airline Group the sixth largest airline at the moment was formed in 2011 by a merger between Iberia and British Airways. In the dataset there is made use of British airways instead of the International Airline Group, because British Airways was the largest of the two mergers at the time of the merger in 2001. The same problem came up for American Airlines and United Continental Holdings. In these cases there is made use of AMR Group instead of American Airlines and there is made use of Continental Airlines instead of United Continental Holdings in the dataset. For the pharmaceutical firms there are also made some adjustments. This because Johnson and Johnson, Roche and Novartis did stock splits in the period 11-03-2001 till 10-03-2002. To avoid biased results the data points of the dates when the stock splits occurred are left out the dataset for these three companies.
Berk and Demarzo (2011) say that it is common practice to use the S&P500 index as an approximation for the market portfolio (p. 317). The S&P500 is seen as the most reliable index to see developments in the stock market and therefore this investigation will make use of the S&P500 to calculate the market portfolio return. This study will make use of the CAPM Beta to measure the correlation between the market portfolio and the stock of the firms in the three industries. According to Berk and Demarzo (2011) the CAPM Beta of a security can be used to measure the systematics risk of a security by calculating the sensitivity of the security’s return to the return of the market
portfolio. The Beta of a security is the expected percentage change in its return given a 1 percent change in the return of the market portfolio (p. 317). In formula the CAPM Beta look like this: 𝛽𝛽 = 𝐶𝐶𝐶𝐶𝐶𝐶(𝑅𝑅𝑆𝑆&𝑃𝑃500,𝑅𝑅𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓)
𝑉𝑉𝑉𝑉𝑉𝑉(𝑅𝑅𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓) or the covariance of the market return and the security return divided by the
variation of the security return. In this study the market portfolio will be replaced by the return on the S&P500 index and the securities will be the individual stock returns of the ten firms of each industry. This CAPM beta will be useful for this investigation, because it look at the correlation between individual firms and the market. Berk and Demarzo (2011) say that beta measures how sensitive the stock’s underlying revenues and cash flows are to general economic conditions (p. 319). For this study this means that the beta measures how sensitive the underlying revenues and cash flows of the stock from the firms for the different industries react on the 9/11 attacks.
To calculate the CAPM Beta all the unadjusted stock prices from the in total thirty companies and the S&P500 are collected from DataStream. In excel the return of the stocks of all these firms are calculated and the return of the S&P500 is been calculated. Also a dummy variable ‘period’ is created in excel to separate the returns before and after the attacks, where period 1 relates to the returns before the attacks and period 2 relates to the returns after the attacks. The point where period 1
goes over in period 2 is 9-11-2001, the day of the 9/11 attacks. To test if the beta of period 1 differs from period 2 the Chow test is used in Stata. The Chow test tests for the presence of a structural break in a time series analysis. This very useful for this study, because from the outcome of the Chow test it is possible to say if there was a significant different correlation between the market and the companies tested. To find this structural break the beta for period 1, period 2 and the whole sample are calculated for all companies in the three industries and from there the Chow test is used to see if there is a structural break in the beta at the reference point. For each company there has been done a regression with the S&P500 returns to calculate the beta before and after the attacks and for the whole sample. After the regression there has been done a Chow test to see if there was a structural break in the betas. The results of these regressions and tests will be presented in the next section.
4. Results
In this section the results of the regressions and Chow tests for all companies in each industry will be presented and be discussed. First the results of the regression and Chow test for all companies will be presented in two tables for each industry. The first table shows the betas before and after the attacks, as well of the whole sample with corresponding t-values for each company. The second table shows the outcomes of the Chow tests with corresponding F-values for each company. The results will be analysed and discussed after they are presented. Both the results and discussion will be done for each industry apart to see if there are similarities or differences in the outcomes in the industry.
4.1 Airline industry Betas:
Whole sample Before the attacks After the attacks
Beta Beta Beta
Lufthansa 0.6083*** 0.3097* 0.9407*** AMR 1.4744*** 0.8564*** 2.2272*** Delta Airlines 1.5374*** 0.7954*** 2.4394*** Air France-KLM 0.6700*** 0.4395** 0.9212*** British Airways 0.8065*** 0.5121*** 1.1419*** ANA Holdings 0.2875*** 0.0432 0.5842*** Southwest Airlines 1.0216*** 0.7344*** 1.3719*** Qantas Airlines 0.5567*** 0.4367* 0.7028***
China Southern Airlines 0.6182*** 0.5588** 0.6873**
Cont. Airlines 1.8409*** 0.7734*** 3.1441***
Where *** means significant at a 1% significance level, ** significant at a 5% significance level and * significant at a 10% significance level
In the table with betas it can be seen that the betas for whole sample, but also the betas before and after the attacks of the airline companies are significant. Only the beta of ANA Holdings in the period before the attacks is insignificant at a significance level of 10%. Possible reason for this can be that the time period of half a year with 132 trading days, which might be relatively small to obtain only significant betas. Still the insignificant beta is the best possible outcome according to the OLS-regression there is been done in Stata, so these results are shown above and will be discussed later on. From the table with all the betas it is notable that all the airline betas after the attacks are higher than the betas before the attacks. Possible reason for the increase in betas is that the 9/11 attacks caused much uncertainty towards and in the airline industry and therefore the stock returns became more sensitive towards the market portfolio after the 9/11 attacks.
The table with the Chow test outcomes show that six of the ten companies in the airline industry had a significant break in their beta with an alpha of 5%. Therefore it can be said that six of the ten airline companies had a significant different correlation with the market before then after the 9/11 attacks with an alpha of 5%. An interesting observation is that all four US airlines in the sample are part of the six significant Chow test outcomes. Only Lufthansa out of Germany and ANA Holdings out of Japan have significance chow test outcomes too. The four non-significant outcomes are all of airlines from non-US countries. Possible explanation for this is that the hijacked planes were from AMR and Continental Airlines and these companies are both US companies, also the 9/11 attacks were in and against the US. This could have caused extra risk and fear towards US airlines and customers in contrast to non-US airlines and customers and therefore contribute to a significant break in the betas of the US airline companies.
Five companies changed from defensive stocks to aggressive stock, namely AMR, Delta Airlines, British Airways, Southwest Airlines and Continental Airlines. From these five companies four have significant Chow test outcomes, only British Airways not. Defensive stocks are stocks with a beta below one, while aggressive stocks have a beta larger than one. For these stock returns this means that if the market portfolio goes up by 1% these companies are now going up by more than 1%, where they first were going up less than 1%. This is about systematic risk, which beta measure by calculating the sensitivity of the stock return to the return of the market portfolio. This increased sensitivity towards the market portfolio after the 9/11 attacks could also be a possible reason for the significant Chow test outcomes. The four companies that changed from defensive to aggressive stocks and have a significant Chow test outcome are al US companies. From these observation and that four out of the six significance Chow test outcomes are of US companies it may be possible to say that the US airline companies reacted more heavily towards the 9/11 attacks than non-US airline companies.
Chow test outcomes:
Lufthansa ANA Holdings
F( 2, 256) 4.54 F( 2, 256) 3.41
Prob > F 0.0115 Prob > F 0.0346
AMR Southwest Airlines
F( 2, 256) 10.16 F( 2, 256) 4.28
Prob > F 0.0001 Prob > F 0.0148
Delta Airlines Qantas Airlines
F( 2, 256) 12.91 F( 2, 256) 0.39
Prob > F 0.0000 Prob > F 0.6800
Air France-KLM China Southern Airlines
F( 2, 256) 2.53 F( 2, 256) 0.08
Prob > F 0.0819 Prob > F 0.9231
British Airways Cont. Airlines
F( 2, 256) 1.94 F( 2, 256) 18.36
Prob > F 0.1463 Prob > F 0.0000
4.2 Defence industry Betas:
Whole sample Before the attacks After the attacks
Beta Beta Beta
Lockheed Martin 0.1668* 0.3405*** -0.0593 Boeing 1.1967*** 0.8572*** 1.5985*** BAE Systems 0.3381*** 0.2592** 0.4347** Raytheon -0.0925 0.3427** -0.6474*** General Dynamics 0.4857*** 0.5888*** 0.3594** Northrop Grumman 0.0246 0.4454*** -0.5049*** Airbus 0.8209*** 0.4428*** 1.2676*** United Technologies 1.3920*** 0.9388*** 1.9398*** Finmeccanica 0.9088*** 0.6990*** 1.1514*** L3 Communications 0.2184 0.8230*** -0.5537*
Where *** means significant at a 1% significance level, ** significant at a 5% significance level and * significant at a 10% significance level
In the table above with the defence industry betas there are some insignificant betas in the whole sample as one in the after the attacks sample with a significance level of 10%. For the insignificant after the attacks beta it is possible that as with the insignificant airline industry betas that the sample is relative small and therefore the number of observations is too low to get only significant betas. For the insignificant betas of the whole sample there could be a different explanation, namely the whole sample betas that are insignificant are al of companies that had before the attacks small but positive betas and after the attacks small but negative betas. This means at a 10% significance level that before the attacks Raytheon, L3 Communications and Northrop Grumman had a positive correlation with the market, but after the attacks they had a negative correlation with the market. For the whole sample this could be problematic in order to get a significant beta. Lockheed Martin also had a shift from positive to negative, but the after the attacks beta is insignificant and so it’s not possible to conclude the same for Lockheed Martin with significance.
The Chow test outcomes show that seven of the ten companies in the defence industry had a significant break in the betas with an alpha of 5%. This means that seven of the ten defence
companies have a different correlation with the market before the 9/11 attacks compared to the after the 9/11 attacks period with an alpha of 5%. The three companies that didn’t have a significant break in their beta are BAE Systems, General Dynamics and Finmeccanica. BAE Systems and
Finmeccanica are together with Airbus the only non-US companies in the sample. Two of the three non-US companies have an insignificant Chow test outcome, while only one of the 7 US companies have an insignificant Chow test outcome. It may be possible to say that again US companies reacted more heavily in terms of their stock returns on the 9/11 attacks than non-US defence companies did, but in the defence industry it is not so clear as it is in the airline industry.
In the airline industry all the betas after the attacks where higher than the betas before the attacks. At the defence industry the betas became higher and lower after the attacks than before the attacks. From the five companies that have a lower beta after the attacks than before even four companies shift from a positive beta to a negative beta. All these four companies that shift to a negative beta after the attacks have significant Chow test outcomes. A possible reason for this could be that before the attacks these companies moved with the market, but after the 9/11 attacks when the market had negative returns these companies had positive returns. The attacks caused fear for more attacks and war, which could be the reason for the opposite reaction after the attacks of the market and the defence industry. This fear had a negative effect for the stock market returns, but a positive effect for the stock returns of the defence companies through the launched war on terror and the US invaded Afghanistan at 7 October 2001.
A possible reason for the betas that were higher after the attacks in the defence industry is that they are all large companies that not only produce defence products. The companies that have higher betas after the attacks than before the attacks are: Boeing, BAE Systems, Airbus, United Technologies and Finmeccanica. Three of this five companies also produce airplanes or parts of
airplanes, namely Boeing, Airbus and United technologies and these three companies have all significant Chow test outcomes. It could be possible that companies that are related to the airline industry also share in the risk of the airline industry and that this is the reason for the upshifting betas. The relation to the airline industry for these companies could have bias the results, but these companies were the biggest defence producers and therefore included in this paper. It can’t be said for sure that this is the reason for the upshifting betas, because BAE Systems and Finmeccanica had also upshifting betas and are not related to the airline industry, but the Chow test outcomes of these two companies are not significant.
Chow test outcomes:
Lockheed Martin Northrop Grumman
F( 2, 256) 3.21 F( 2, 256) 13.79
Prob > F 0.0420 Prob > F 0.0000
Boeing Airbus
F( 2, 256) 7.71 F( 2, 256) 4.76
Prob > F 0.0006 Prob > F 0.0094
BAE Systems United Technologies
F( 2, 256) 0.35 F( 2, 256) 12.12
Prob > F 0.7027 Prob > F 0.0000
Raytheon Finmeccanica
F( 2, 256) 7.92 F( 2, 256) 2.32
Prob > F 0.0005 Prob > F 0.1003
General Dynamics L3 Communications
F( 2, 256) 0.73 F( 2, 256) 12.19
Prob > F 0.4844 Prob > F 0.0000
4.3 Pharmaceutical industry Betas:
Whole sample Before the attacks After the attacks
Beta Beta Beta
Johnson & Johnson 0.2427*** 0.2728*** 0.2080***
Pfizer 0.4955*** 0.5815*** 0.3888*** Roche 0.3998*** 0.2206** 0.6049*** GlaxoSmithKline 0.2775*** 0.3617*** 0.1696 Novartis 0.3279*** 0.3406*** 0.3019*** Sanofi 0.2222** 0.3125** 0.1208 Astrazeneca 0.3263*** 0.3552*** 0.2844** Abbot 0.3777*** 0.5062*** 0.2217** Merck & Co 0.3540*** 0.5169*** 0.1550 Bayer 0.5731*** 0.4628*** 0.6895***
Where *** means significant at a 1% significance level, ** significant at a 5% significance level and * significant at a 10% significance level
In the table above with the betas of the companies in the pharmaceutical industry there are some insignificant betas at the after the attacks outcomes with an alpha of 5%. This insignificance can again be explained through the relative low number of observations in the sample. All the insignificant betas are calculated in the period after 9/11 attacks a possible explanation for this is that the market itself fluctuated more than it did before the attacks. This could have had his impact on the insignificance of some of the betas after the attacks.
The Chow test outcomes show that only Roche had a significant break in the beta before the attacks and after the attacks with an alpha of 5%. This means that according to the Chow test only Roche had a significant different correlation with the market before the attacks then it has after the attacks with an alpha of 5%. Striking is to see that in the pharmaceutical industry only one company has a significant different correlation to the market. This in contrast to the airline industry with six companies and the defence industry with seven companies. To find explanation for the outcome of the Chow test for Roche is difficult, but a possible reason for the significant Chow test outcome of Roche could be that competitor Novartis started buying Roche shares beginning in 2001 with the idea of a merger or collaboration. This action of Novartis could have caused extra fluctuation of the shares of Roche and therefore contribute to the significant Chow test outcome of Roche.
All the betas before the attacks, after the attacks and in the whole sample are between zero and one, so on the first sight there are no big changes in the betas of the firms in the pharmaceutical industry. Still within in these range there are some notable changes in the betas. The beta of Roche is almost three times higher after the attacks than it was before the attacks and this reflected by the significant Chow test outcome. That the betas of the whole sample are al significant could be explained as that the pharmaceutical industry was pretty stable in the time period over which the betas are calculated. This in contrast with the defence industry where there were some insignificant whole sample betas and the airline industry where all betas had large upwards changes. Eight of the ten companies had a lower beta after the attacks then they had before the attacks, but most changes in betas are very small and the Chow test shows that most of these changes in betas are insignificant. Chow test outcomes:
Johnson & Johnson Sanofi
F( 2, 255) 0.17 F( 2, 256) 0.79 Prob > F 0.8424 Prob > F 0.4530 Pfizer Astrazeneca F( 2, 256) 0.79 F( 2, 256) 0.33 Prob > F 0.4530 Prob > F 0.7211 Roche Abbot F( 2, 255) 3.97 F( 2, 256) 2.12 Prob > F 0.0201 Prob > F 0.1225
GlaxoSmithKline Merck & Co
F( 2, 256) 0.76 F( 2, 256) 2.49 Prob > F 0.4670 Prob > F 0.0846 Novartis Bayer F( 2, 255) 0.69 F( 2, 256) 1.42 Prob > F 0.5036 Prob > F 0.2428 12
5. Conclusion
According to the Chow test outcomes in section 4 in the airline industry six out of ten companies, in the defence industry seven out of ten companies and in the pharmaceutical industry one out of ten had a significant other correlation with the market before the 9/11 attacks then after the 9/11 attacks. In particular it is possible to say that the companies that had a significant Chow test outcome had a break in their beta before and after the reference point of 9-11-2001 looking at the whole sample. In the airline industry all the US companies are part of the significant Chow test outcomes and this could imply that the US airlines companies reacted more heavily on the 9/11 attacks then non-US companies did. All the US airlines had a beta that was before the attacks between zero and one and after the attacks higher than one. The only non-US company that has this too is British Airways, but British Airways does not have a significant Chow test outcome. In the defence industry seven companies are founded in the US and six of them have significant Chow test outcomes, while only one of the three non-US companies in the defence industry had a significant Chow test
outcome. From these two industries it may be possible to say that US companies reacted more heavily on the 9/11 attacks than non-US companies did. In the pharmaceutical industry only Roche had a significant Chow test outcome and this could possibly be explained by the effort of Novartis to merge or collaborate with Roche started in 2001, which possibly caused more than normal stock return fluctuations.
So in what extent did the 9/11 attacks affect the correlation between the market portfolio and the stocks of the firms in the airline industry, defence industry and the pharmaceutical industry? In the airline industry six, in the defence industry seven and in the pharmaceutical industry one company had a significant other correlation with the market before the attacks than after the attacks. The airline and defence industry were affected the most in terms of correlation between the market portfolio and the firms in the industries and the pharmaceutical industry almost wasn’t affected, but even the airline industry and defence industry reacted differently. In the airline industry all the betas of the companies with significant Chow test outcomes were higher after the attacks than before, while in the defence industry three were higher and four were lower. In conclusion the airline and defence industry were most affected by the 9/11 attacks, but all three industries reacted different on the 9/11 attacks in terms of their correlation between the market portfolio and the firms in these industries.
For further research it would be interesting to test with a sample with more observations, so that less betas will be insignificant and the outcomes of the Chow test will have more power. It will be also interesting to investigate which underlying causes there are for the insignificant chow test outcomes in the airline and defence industry and the significant chow test outcome in the
pharmaceutical industry. To see if these results are affected by the firm specifics or other events that occurred in the time period used.
References:
Berk, J., & DeMarzo, P. (2011). Corporate Finance. Harlow: Pearson Education Limited.
Brounen, D., & Derwall, J. (2010). The impact of terrorist attacks on international stock markets.
European Financial Management, Vol. 16 (No. 4), 585-598.
Cam, M. (2008). The impact of terrorism on United States industries. Economic Papers, Vol. 27, 115-134.
Carter, D. A., & Simkins, B. J. (2004). The market’s reaction to unexpected, catastrophic events: the case of airline stock returns and the September 11th attacks. The Quarterly Review of
Economics and Finance, Vol. 44, 539-558.
Charles, A., & Darné, O. (2006). Large shocks and the September 11th terrorist attacks on international stock markets. Economic Modelling, Vol. 23, 683-698
Chen, A. H., & Siems, T. F. (2004). The effect of terrorism on global capital markets. European Journal
of Political Economy, Vol. 20, 349-366.
Chesney, M., Reshetar, G., & Karaman, M. (2011). The impact of terrorism on financial markets: An empirical study. Journal of banking and finance, Vol. 35, 253-267.
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, Vol. 20, 435-446.
Ito, H., & Lee, D. (2005). Assessing the impact of the September 11 terrorist attacks on U.S. airline demand. Journal of economics and business, Vol. 57, 75-95.
Lenain, P., Bonturi, M., & Koen, V. (2002). The Economic Consequences of Terrorism (Working Paper No. 334). Retrieved from OECD Economics Department Working Papers website:
http://dx.doi.org/10.1787/511778841283
Lewellen, J., & Nagel, S. (2006). The conditional CAPM does not explain asset-pricing anomalies.
Journal of Financial Economics, Vol. 82, 289-314.
The Maurice Falk Institute for Economic Research in Israel. (2005). The impact of terrorism across
industries: an empirical study. Jerusalem: Berrebi, C., & Klor, E. F.
