Influence of Terrorist Activities on Financial Markets: Evidence from KSE

This paper investigates the influence of terrorist activities taking place in Pakistan on the KSE (Karachi Stock Exchange) for the period of 01/2005 to 12/2010 using the GARCH & GARCH-EVT to identify the relationship between these two variables; the study establishes that terrorist activities adversely affect the financial markets and in the case of KSE, it is a highly significant relation. The reason why the negative relationship exists is because of the foremost increase in the number of terrorism attacks in Pakistan.


Introduction
Terrorism is the major cause affecting the economy of Pakistan and a curse affecting international trade, investments and financial institutions. Terrorism is a political issue nowadays, the stock exchange can be directly or indirectly affected by terrorism activity (IMF 2005). This study is about terrorism attacks and its negative effects on the Karachi stock exchange (KSE); unfortunately since 2005 to 2010 there has been an increase in terrorist activities in different areas of Pakistan. Although there has not been much relevant studies on this topic but some studies have alreadly been done on the impact of terrorism on the financial market after 9/11. The New York 9/11 terrorist attack has introduced the definition of terrorist. Before then it was not a major topic and prominent issue (Wilkinson and Jenkins, 2003). Terrorist attacks happening anywhere in the world especially Asia pacific or any other countries in the outer rims of the world affects the stock exchange of every country directly or indirectly. Chen and Siems (2004) highlight that global financial market are strongly inter-connected so news from any event spread like fire across countries (particularly shocking news).
Unfortunately, terrorist activity in Pakistan since 2001 have been increasing, so this study's main concern is how terrorist attacks in Pakistan affects the stock Financial Assets and Investing 6 exchange; the volatility of stock exchange is a result of many other factors but terrorism activities affects stocks severely. Karolyi (2006) further illustrates that after the 9/11 attack, terrorism became a major geopolitical threat for the global financial markets as well as for the stability of stock markets. Our study differs from the extant literature on two main points, I) Prior studies are confined to the extent of the 9/11 attack, Madrid bombing, London bombing or other attacks, but our study is purely about the major and critical terrorist attacks in Pakistan affecting our stock market. II) No prior study done on the attacks of Pakistan affecting its stock market. We are doing it on the bomb blasts in Pakistan and its impact on the stock exchange KSE, though stocks show a negative trend after the major bomb blast as studied before by many scholars.

Literature Review
Terrorist attacks in the past few years has shown an increasing trend and so the importance of its existence and need for the study to be conducted in this field has to be increased too. It is a main issue affecting every economy nowadays. Sandler and Enders (2002) define terrorism as a premeditated use, "threat of violence to obtain political objectives through fear directed at public or civilians". Literature in this field of study is still emerging as Karolyi and Martell (2006) stated that "Not every theory has an intuition behind it for conducting an exercise but a synthesis of research in different areas of this field". Chen and Siems (2004), investigates the effects of the 9/11 catastrophe and 14 other major terrorist assaults impact on the global and US stock prices and compare its impact on the political and economic activities with the help event study method. The conclusion after the tests is that, after 9/11, financial markets were severely or ruthlessly crashed and stock prices showed a negative down fall but prior to that event it was in a better condition. Berrebi and Klor (2005) conducted a similar study to evaluate the impact of terrorism on the stock market prices of Israeli companies using the same event method study. Conclusion and results of the study shows that, companies which have involvement or deals in stock of defense, security or antiterrorism security company measurers show a positive outcome but other companies show a negative trend. So it was concluded that terrorism attacks do have a negative impact on stocks and equity markets.
Further Carter and Simkins (2004), also examined the catastrophic 9/11 event and their impact on airline stocks with the help of a multivariate regression model. As the market was closed for about six days, so they only evaluated the stock prices on the first trading day after the major event and their results were stated that it has a different effect on different airline firms because the congress of US passed the Air transport safety and System stabilization Act (Sep 18,2001). The effect of the terrible catastrophe of 9/11 was also studied by Darkos (2004); in his study, he studied the impact of various airline stock listed in different stock markets with the help of a Market Model and his study concluded that measuring with market Beta (ß) shows that systematic risks have more value than it is on average. The market risks of the airline stocks in different stock markets showed a rising trend after the event.
Eldor and Melnick (2004)  Nevertheless result of the study indicated that the 9/11 attack, London bombing and Madrid bombing shows an abnormal, negative and no effect on Greek banks' stock respectively. The 9/11 attack shows a huge and abnormal effect because of the dominancy of the US economy over the world's economy. Our study's main concern is to identify the impact of terrorism activities (Bomb B lasts) on the KSE 100 index.

The GARCH Model
Bollerslev (1986) Whereas (p,q) are order of the GARCH and ARCH term respectively. The variance term σ is the conditional variance at time "t" and ψ indicates constant, whereas ψ and ϕ are the parameters, ε is the indicator of previous squared shocks and σ reflects prior variances. Various studies employed GARCH (1, 1). Brooks (2008) indicates that a GARCH (1, 1), in most cases is enough to grasp the volatility clustering and that higher order is very rarely used in the field of finance. Negative variance possibility is very rare; limitations have to be generally specified for these parameters particularly. Therefore the GARCH model successfully captures various number of features of the financial time series, such as volatility clustering and thick tailed returns. The GARCH model becomes stationary when the total of alpha and beta are less than one (α + β < 1). On the other hand, if α + β = 1,, , the process is still stationary because the variance is infinite. The GARCH models applicable in this study will estimate according to maximum likelihood criteria. The εt is assumed to be normally distributed approximately with an average value of zero and time-varying variance is expressed in this manner (εt ~ N (o,σ )). Nelson (1991) introduced the Exponential GARCH model. This model is quite purposeful and useful in comparison to the GARCH because it permits good news and bad news to have a different impact on the volatility. Moreover it also permits big news to have a higher impact on volatility. This particular model works in two stages; firstly it takes into consideration the mean and secondly the variance component. The EGARCH (p, q) model can be defined in this manner:

The EGARCH Model
Whereas ϕ, λ, and ω indicates parameters for conditional variance estimation and λ shows the effect of the previous period measures on the conditional variance.
In the case if λ is positive, it means a positive change in the stock price and is related with more positive change and vice versa. ϕ co-efficient measures the impact of last period information set and narrates the prior standardized residuals impact on the present volatility. Moreover, ω $ indicates an asymmetric effect in the variance and negative ω $ interprets that bad news has a greater impact on stock volatility than good one having the same magnitude. EGARCH models indicate the logarithmic time-varying conditional variance, where concerned parameters are permitted to be negative. So this element shows that the model does not require any non-negativity limits in the parameters. Therefore, the lack of non-negative limits makes the model more attractive than GARCH. The stationary constraint for an EGARCH (1, 1) model is that the beta should be less than one (λ < 1). Hence in the case of symmetry, where the amount of positive and negative shocks is equally impacting on the variance, ω will be equal to zero.
On the other hand, if ω < 0 the strength of a negative (positive) shock will reason the variance to increase (fall) and if ω > 0 positive and negative shocks will reason the variance to rise or fall respectively. The natural logarithm of the conditional variance is modeled in EGARCH(1,1), and it is calculated as,

Results
The data is analyzed in steps; first of all, the descriptive statistics are checked for the KSE indices the results are mentioned in the Table 1   In the below tables the heteroskedastic quality of the data is confirmed via the BDS test and ARCH-LM test in Table 2 and 3 respectively. The BDS test signifies that the data is highly significant in all dimensions, same as that for the ARCH-LM test which confirms that the data has the above mentioned characteristics.