This paper examines the influence of political instability and terror on Pakistan stock market returns between 1997 and 2010. The study constructs three variables that quantify political instability and terror and examine the effect on country stock return. This study seeks to apply the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model to assess the impact of these variables on stock market returns and volatility using daily time series data for KSE. Results for KSE showed strong support for the hypothesis that bad news exerts more adverse effect on stock market volatility than good news of the same magnitude. Furthermore, terror and regime have significant negative impact while war has positive but insignificant effect on stock market volatility.
JEL Classification: O40, C32.
Keywords: Terror, Regime, political instability, growth, ARCH/GARCH.
Many people agree that stock prices sometimes behave in bizarre ways. Markets are pretty tough and quite difficult. In the world of today’s no one can negate the importance of stock markets. Stock market acts as a barometer for any country’s economy. In today’s information-oriented world, news travels very fast and contagion can spread quickly and capital markets become more flexible and are absorb shocks brought on different news such as terrorism, political instability etc. Stock market of Pakistan is going through quite rough patch from many years. The change of political government and later on the terrorists’ attacks have badly affected the stock market and make the Pakistan Stock Market unreliable place for investment. As by seeing the overall scenario of Pakistan’s stock market during that time period it was not difficult for prices to follow certain patterns that support the rejection of Random Walk Hypothesis.
This paper examines the impact of change in government, war and terror on economic growth in the Pakistan. Pakistan is one of those episodic-democratic countries who are facing continuous upheavals and socio-political disruptions since their inception. Military interventions could be witnessed in the political history of Pakistan. More over intervallic wars with India, strikes, antigovernment demonstrations and most importantly the ongoing war on terror have popped Pakistan to prominence on the socio-political platform. Such sociopolitical flux, terrorist attacks and other disruptions can have serious implications for stock price movement because stock prices reflect investors’ expectations about the future and these stock price movements on aggregate can generate a surged wave of activity.
There has been an extensive work on study of stock market returns and volatility with respect to the fundamental variables and the macroeconomic variables but a diminutive work has been done so far to study the impact of socio-political factors on the stock market volatility in Pakistan. The existing literature on impact of socio-political factors on stock returns volatility is quite inadequate especially if we talk in context of Pakistani market. Masood & Sergi (2008) analyzed Pakistan’s political risks and events that have affected the Pakistani stock market since its independence but their study chiefly covers the political events. Terrorism and strikes which have recently become the matters of intense interest and the source of unrest in the economy are the missing part there.
The Karachi stock market is rapidly converting into a volatile market. If we see figure below it showed that there are high volatility during 1997 to 2010. This cannot be viewed as a positive sign for this emerging markets like Stock market of Pakistan. Though heavy fluctuations in stock prices are not an unusual phenomena and it has been observed at almost all big and small exchanges of the world. But focusing on the reasons for such fluctuations is instructive and likely to have important policy implications. The efficient market hypothesis argued that changes in stock prices are mainly dependent on the arrival of information regarding the expected returns from the stock and risk associated with that stock.
(See Figure 1.1)
So the purpose of our study is to examine empirically the impact of socio-political instability on Pakistani stock market.
This study examines the three factors and their impact on the Pakistani stock market; the political instability due to military interventions, 1999 Kargil war, and terrorism.
A number of theoretical and empirical articles argue that these factors hinder economic growth of a country. Cutler, Poterba and Summers (1989) claimed that the sock prices move in response to the information other than about the fundamental values. They estimate the fraction of stock returns that can be accredited to various kinds of economic and non-economic events including assassinations of important political or national figures, war, invasions, raids and major policy change but their findings suggests a very small effect of non-economic news on the share price.
Most of the studies have found a significant impact of political news or events on the stock market behavior. Chan & Wei (1996) studied the impact of political news on the stock market volatility in Hong Kong and using GARCH-M model they found the strong evidence of the impact of political news on stock market volatility inferring that unfavorable political news is correlated to negative returns for the Hang Seng Index and vice versa. Mei & Guo (1999) examined the impact of political insecurity on the financial crises in emerging markets and they observe that market volatility increased during political election and transition periods and political uncertainty could be a major contributory factor to financial crisis.
Similarly Kim & Mei (2001) infered through empirical analysis using GARCH(1,1) filter that the political risk affect the stock market volatility but this impact of political events or news is asymmetric, with bad news having a greater influence on volatility relative to good news. However Voth (2001) have argued that the impact of political factors in studies on German market has been over stated. He argued that the majority of events escalating political uncertainty had a minute or no effect on the value of German assets and the volatility of their returns. Instead, it was inflation that is mainly responsible for most of the variability in stock returns. He suggests that there is no direct linkage between the political factors and the stock market, however through channel it impacts. But Voth (2002) in a panel study of a set of 10 countries using panel regression confess that during great depression political risks changed dramatically over the period, and are adequate to account for a large part of the boost in stock price volatility.
Beaulieu, Cosset & Essaddam (2002) examined the impact of political risk in Canada on the volatility of stock returns, covering important political events in the country. Their study suggested that political news performs a significant role in the volatility of stock returns. Moreover the volatility of stock returns also depends on the degree of how much a firm is exposed to political risk i.e. the structure of its assets and the level to which there is foreign involvement. Kutan & Perez (2002) also found a significant impact of social and political factors on stock return volatility in their study conducted on Colombian stock market.
Bautista (2003) applied Regime-switching-ARCH regression on Philippine stock returns to estimate its conditional variance and the estimated volatility was then related to major political and economic events. Their study revealed that the Philippine stock market is sensitive to radical changes in the political situation. Moreover the series of military takeover attempts during late 1980s in Philippines lead to hefty fluctuations in stock market index.
Masood & Sergi (2008) analyzed political risks and events that have affected the Pakistan’s stock markets since its foundation. They have found that Pakistan’s political risk carries a significant risk premium of between 7.5% and 12%. They made forecasts using Bayesian hierarchical modeling and Markov Chain Monte Carlo (MCMC) techniques and found that there is relatively high probability of occurrence of events with an average arrival rate of approximately 1.5 events per year.
Many others also wrote that political instability warped the future path of investment decisions (Calvo and Drazen (1997), lessened public investment leading to a shift of government budgets from capital spending to government consumption (Darby, Li and Muscatelli (1998), and makes governments less inclined to make improvements to the legal system (Svensson (1993)
Wars and unrest at the borders creates instability and panic among the investors that could affect the stock market movement at large. The affect of war has been analyzed in many studies including Cutler, Poterba and Summers (1989), Aggarwal, Incaln & Leal (1999) and in Pakistan Masood & Sergi (2008).
Aggarwal, Incaln & Leal (1999) examined the sort of events that cause large swings in volatility of emerging stock markets. For this purpose they examine various social, political and economic events both at global and domestic level to find out their explanatory power in context of the returns volatility in the emerging markets including the impact of gulf war. Though at small scale but the impact of gulf war was felt in those emerging markets. Similarly Masood & Sergi (2008) found that among other factors that they studied, wars with India, 1948, 1965, 1971 and 1999 kargil war negatively influenced the Pakistani stock market.
Evia et al. (2008) examined the affect of socio-political conflict in Bolivia on economic performance. Factors studied widespread during the conflicts as strikes, demonstrations, road blockades, and conventional rent-seeking. Their results showed that economic growth due to external factors is positively related to conflict while growth due to productive investment is negatively related to conflict.
Terrorism is another as put that has been studied in relation to economic activity. Many studied in this distance; produced conflicting results as Becker and Murphy (2001) argue that economic performance are not much affected, because terrorist attacks usually devastate only a small portion of the overall stock of capital in a country. By contrast, Abadie and Gardeazabal (2005) repeated that terrorism shape overall economic risk in a country and lead to the economic shakiness in the country. They also conclude their study that higher level of terrorism risks results into the lower levels of foreign direct investment (FDI). Almost all studies on terrorism and its influence on stock prices limited to only on a single or few events, such as the 11 September 2001 attacks, as considered by Hon et al. (2004) & Chen and Siems (2003) study.
Chen & Siems (2003), used event study methodology to capture the aftermath of terrorism on global capital markets. They studied on the reaction of U.S. capital markets in response to terrorist attacks. Their results showed that capital markets of US are more resilient & flexible than in the past and recover quicker from terrorist attacks than other global capital markets. Their study suggests this increased market resilience to be partially explained by a stable financial sector in US that provides adequate liquidity to support market stability and reduce the spread panic.
Methodology and Data Description
Stock index data is taken from Karachi Stock Exchange, Yahoo Finance. This is a well known and reliable source of business information in Pakistan. The daily closing value of KSE-100 index is used for calculating the daily returns. The continuously compounded annual rate of return is used to measure the returns for the specific period as;
Rt = ln (Pt / Pt-1)
The closing prices of KSE-100 index for Karachi Stock Exchange are taken for the period July 2, 1997 to Oct 13, 2010. Our proxies are TERROR, a dummy variable of terrorist incidents during this period; REGIME, a dummy variable for government changes from fully democratic government to Marshal Law or democratic under such condition; a dummy variable for the period of the Kargal War in 1999. We applied regression model and Arch/Garch technique to capture the results.
ARCH/GARCH Study Models
This section presents the methodology of the paper. Daily data for Karachi stock markets were obtained from Yahoo finance and data for terror, kargal war and regime were obtained from South East Asia Terrorism Portal, and Different News Paper of Pakistan. Study apply ARCH/GARCH tools to see the long term relationship of these variable taking stock return as dependent variable and terror, regime and kargal war as independent variables.
As aggregate uncertainty may be a function of political instability, we proceed to model uncertainty directly. It is natural to look at the conditional variance of output. Thus, we examine GARCH processes, in a more general framework than in the previous section. The model estimated here is a GARCH (1,1) process. Engle (1982) argue that in high frequency data large and small disturbance errors appear in group therefore error term variances can be shown as a function of their lagged values. He calls it Autoregressive conditional Heteroskedasticity (ARCH). As an investor or policy maker, we might be interested in investigating the returns and variance financial assets over observable period of time (conditional) rather than long run estimate of variance (unconditional). Engle (1982) shows that it is possible to describe the conditional mean and conditional variance of a financial asset using information set of previous period;
Where is the return of financial asset in time t conditional on the information set at time t-1. E represents the expected value in statistics.
Consider the simple model
Where the rate of is return and are the regression parameters. A typical ARCH model can be written as follows:
Conditional Mean Equation;
where ‘v’ is the part of variance which is homoskedastic and is the conditional variance which is Heteroskedasticity. This conditional variance can be shown as ARCH Conditional variance Equation, i.e.
where and are non negative.
Engle (1982) has also derived a Lagrange Multiplier (LM) based principle to test the hypothesis of.
Another useful variant of ARCH methodology, proposed by Bollerslev (1986) is the generalized ARCH or GARCH model. Bollerslev (1986) argues that conditional variance in financial series is not only the function of its lagged error term but also the function of its lagged conditional variances.
Therefore, GARCH (1, 1) process would be
So GARCH model helps to explain the conditional variance with the help of past squared error term and conditional variance lag value. Which also means that conditional variance at time‘t’ would be function of long run variances and also variances conditional on past information set (short run) or observed shocks i.e. .
Testing for ARCH/GARCH effects:
Before estimating Arch/Garch techniques, it is first important to check for possible presence of Arch effect in order to know which model is requires the ARCH estimation instead of OLS (Ordinary Least Squire). The presence of ARCH effects in a regression model does not invalidate OLS estimation. However it implies that there is more efficient nonlinear estimator than OLS.
(See Table 1.1)
Obs*R-Squared is 147.26 and has a probability limit of 0.000. This clearly suggested that ARCH effect is present and presence of Heteroskedasticity suggested that ARCH/GARCH is appropriate model for this type of time series data. So we can apply ARCH/GARCH model on this data instead of ordinary least squire regression.
Result of GARCH effects:
The results of GARCH are presented in Table 1.3. The first column presents the regression results when we include as independent variables dummy values of the regime, terror, and war. In most of the cases, the variables enter with the anticipated signs, but not all of them are consistently significant at the 0.05 level. We can see an evidence of significant negative impact of terror, regime that show due to bomb blast in Pakistan and change in government negatively impact the country stock return in long run while insignificant positive impact of war on the country stock return.
The results can further explained that stock return volatility every day is explained by approximately 71% of the previous month’s return volatility for Karachi stock exchange. This is significant for KSE returns. The coefficient of return innovation are statistically significant for market implying that new information arrival into the markets has significant impact on predicting next day’s stock market volatility. Because, the constant term in the variance equation for KSE is significant.
The results of GARCH (1,1) are presented in Table 1.3
The model can be written as;
= 0.001188+ 0.064048* R_KSE(-1)
GARCH = 4.01E-05 + 0.20721*ARCH+ 0.713458 GARCH(-1) – 1.21E-05*Terror + 1.93E-05*War -1.48E-05*Regime
The persistence parameter for KSE Durbin-Watson stat = 1.943, which is > 1. This show a very explosive volatility in KSE returns. It also demonstrates the capability of past volatility to explain current volatility (Engle and Bollerslev, 1986) and because it is very high, the rate at which it diminishes is rather very slowly.
For ACRH/GARCH, conditional standard deviation and conditional varience graph were as shown in figure 1.2 and 1.3;
The GARCH coefficient is both statistically significant and conforms to expectation. This implies that past variances exert significantly positive effect on stock return volatility in KSE. On the basis of these results, it is evident that there is significant time varying volatility in Pakistan stock market returns during the sample periods.
Conclusions and Recommendations
In this paper, we have estimated a nonlinear GARCH model for daily stock returns volatility and terror, Kargal war and regime in Pakistan. Data for the estimation of GARCH (1,1) models was obtained from Yahoo finance and South Asia Terror Portal and news paper of Pakistan. The asymmetric effect of terror, war and regime on stock returns and volatility was investigated. Preliminary investigation into the nature of the data reveals that study had to employ ARCH/GARCH techniques for data analysis.
Firstly, results show evidence of time varying volatility in stock market returns across the market and from the asymmetric model, results indicate that bad news has larger impact on stock volatility than good news in the KSE. The result for KSE showed that terror and regime has negativity impact on returns of KSE while war has positively effect, it may be due to short term period of the war. All three variable are significantly have their impact on the returns.