CAUSALITY ANALYSIS OF STOCK MARKETS: AN APPLICATION FOR ISTANBUL STOCK EXCHANGE

CAUSALITY ANALYSIS OF STOCK MARKETS: AN APPLICATION FOR ISTANBUL STOCK EXCHANGE Aysegul Cimen Research Assistant, Department of Business Administration Dokuz Eylul University, Turkey Address: Dokuz Eylul University Department of Business Administration, Buca/IZMIR/TURKEY Abstract: This study examines the short run relationship between Istanbul Stock Exchange and twelve stock markets classified into groups as American, Asian-Pacific and European. Daily closing prices in terms of natural logarithm are used for the countries of Argentina, Brazil, China, Germany, France, Hong Kong, Japan, Korea, Netherlands, Turkey and the USA after the global financial crisis. As a result of Granger causality tests, Istanbul Stock Exchange causes all stock markets at Asia-Pacific, Europe and America with one exception of China stock market. Besides, Istanbul Stock Exchange is bidirectionally caused by Nikkei, Netherlands, France and Brazil equity markets. Keywords: Causality, Istanbul Stock Exchange, Financial Crisis 1. Introduction Risk minimization and return maximization are the main aims of all investors. To success these aims investors diversify their portfolios, namely manage the risks of portfolios. Since 1980s, borders between countries in terms of capital markets have removed by globalization. In this borderless world, information about stock markets flow easily among individual and professional investors. After the globalization, the free movement of information about capital markets gives insight to investors to invest in different stock markets in order to diversify portfolios. Investing in different stock markets is not only enough for risk minimization. The correlation coefficients of stock markets and causality among them are important for right decision making. Linkages of stock markets both regionally and globally are to be analyzed to diversify portfolio risk. Transactions have been done since 1986 at Istanbul Stock Exchange and studies are conducted about ISE since the beginning of 1990s. The aim of (ISE) is to provide transparent and secure market for local and foreign investors (www.ise.org). Approximately 70 % of transactions at ISE are held by foreign investors. The aim of this study is to find out whether the investors of Istanbul Stock Exchange have opportunity to optimize their portfolios by investing at American, Asian-Pacific or European stock markets in the short run. Numerous studies held about developing countries and relationships between developing countries. Limited number of studies is applied to Turkish stock market because of being a developing country. The contribution of this paper is analyzing the linkage of Istanbul Stock Exchange with other markets classified as American, Asian-Pacific and European post-crisis period. 2. Literature Review Prior studies focused on both short run and long run relationships between equity markets. Short term linkages measured by Granger (1969) causality tests and long run interdependence are measured by Dickey and Fuller (1981), Engle and Granger (1987) and Johansen (1988) econometric models. Some studies just focus on short run relationships between equity markets whereas others analyze long term dynamics between stock markets. Besides, numerious studies investigate not only short run but also long run relationships of stock markets for portfolio diversification. Yüce and Şımga Mugan (1996) studied the short run and long run relationship between Istanbul Stock Exchange and Turkish lira price of US dollar. Result showed that there was not any relationship in the long-run, however, the foreign exchange market Granger caused the stock market. Gerrits and Yuce (1999) tested the interdependence between stock prices in Germany, the UK, the Netherlands and the US. Daily closing prices were used for the period from March 1990 to October 1994. COPY RIGHT 2013 Institute of Interdisciplinary Business Research 124

Results indicated that US market had significant effect on European markets. Diversification among European stock markets did not diversify portfolio risk. Yüce and Sımga Mugan (2000) investigated short run and long run relationships between Eastern European stock markets. Data set consisted of Prague, Budapest, Moscow, Warsaw, Istanbul, London, Frankfurt, New York, and Tokyo stock markets. Weekly closing prices were analyzed from September 20, 1994 to December 31, 1999. The existence of one bivariate cointegration was found between stock markets. Erdal and Gunduz(2001), investigated linkages between Istanbul Stock Exchange and G-7 countries and Israel, Jordan, Egypt and Morocco before and after the Asian crisis. Causality between Istanbul Stock Exchange and G-7 countries were found. The impact of crisis on stock markets are also analyzed in order to diversify portfolio during crisis period. Tokic (2003) investigated emerging markets before Asia-Pacific crisis between 1995-1997 period. The results indicated that there was long-run relationship between Hong Kong, Indonesia and Malaysia stock markets with the US stock market. Kim (2005) examined the stock market linkages in the Australia, Japan, Hong Kong, Singapore with the US stock markets. Daily index prices and trading volumes of US and Japan as the leaders were used for the analysis. Granger causality tests showed that the US market Granger caused Asia Pacific countries, U.S. volume also Granger caused both the returns and volatilities of the Asia-Pacific markets. The Japanese returns had a less significant effect than US market. Küçükçolak (2008) measured the long run relationship between Turkish equity market and UK, Germany and France by using cointegration and Eagle-Granger tests. Data consisted of daily returns from January 2001 to December 2005. Analysis showed that Turkish stock market was not cointegrated with EU countries whereas Greek was cointegrated in the long run. Ergun and Nor (2010) studied the the dynamic relationship and between the stock markets in Turkey and the United States under the conditions for Turkey's accession to the European Union. Results showed that there were strong dynamic linkages between the Istanbul Stock Exchange and NASDAQ after the Custom Union Agreement between Turkey and the European Union. Shamiri and Isa (2010) investigated the volatility transmission of US and Japan stock markets to Asia- Pacific markets using daily stock market return from 1991 to 2004. It was found that only US market influenced Asia Pacific stock markets.kospi, Singapore and Hong Kong stock markets were so cointegrated with US market that they move in the same direction. Khan, Hassan and Ahmad (2011) analyzed the cointegration and causality among Asian markets. Augmented Dickey Fuller (ADF), Multivariate Cointegration and Eagle-Granger Causality tests were applied to data that consists of monthly data for the period from August1998 to August 2008. Results indicated that developed stock markets such as Hong Kong, Singapore, Tokyo, Shanghai, and Bombay were cointegrated with each other and large stock markets granger caused small stock markets. David (2011) studied the causality between New york, London, Tokyo and Hong Kong stock markets during the global financial crisis period. Daily prices of S&P 500, Nikkey 225, Hang Seng and FTSE100 were used for the study. According to Granger causality results, Nikkei225 and Hang Seng Indexes were affected by S&P 500 and FTSE100, whereas S&P 500 and FTSE100 affect each other and they were not affected by Nikkei225 and Hang Seng Indexes. 3. Data and Methodology 3.1. Data This study consists of daily closing prices in terms of natural logarithm used for the countries of Argentina (MERV), Brazil (BVSP), China (SSEC), Germany (GDAXI), France (FCHI), Hong Kong (HIS), Japan (N225), Korea (KS11), Netherlands (AEX), Turkey (ISE) and the USA (SNP). Data set focuses post-crisis period from September 1, 2009 to December 31, 2012. All stock prices are in terms of local currencies to avoid risks due to exchange rate transformation. The data of stock market indices were downloaded from the Yahoo Finance website (http://finance.yahoo.com). [TABLE 1] COPY RIGHT 2013 Institute of Interdisciplinary Business Research 125

3.2. Methodology 3.2.1. The Correlation Test Firstly, correlation test is used for determining the direction between Istanbul Stock Exchange and other stock markets divided into three groups. The correlation matrices show the relationship between ISE and America group, ISE and Asia-Pacific group and ISE and Europe group. Correlation coefficients always range between -1 and 1. For risk minimization and portfolio optimization, investors invest in stock markets either negatively correlated or low correlated. 3.2.2. Granger Causality Test Granger causality test is applied to analyze short run causality between Istanbul Stock Exchange and other twelve stock markets. According to Granger (1969), causality test gives answer whethet X causes Y or vice versa. The pairwise Granger causality test equation is as follows: Y Y t 0 1 t 1... iyt i 1X t 1... X i H0: β 1 =β 2 = 0 (X does not Granger cause Y) H1: At least one of the β 1 0 4. Empirical Results 4.1. Descriptive Statistics The descriptive statistics of the thirteen stock markets are in Table 2. FCHI, FTSE and N225 have the lowest standard deviation which means that risk in these markets are low. BVSP, KS11, ISE and MERV have the highest risk as well as the return. If we look at the results in terms of standard deviation, we can classify the low risk countries are developed and high risk countries are developing. Skewness of all countries are negative which means these are left tailed. 5 stock indices out of 13 have Kurtosis less than 3 which is an indicator of low peak distribution and thin tails. [TABLE 2] 4.2. Correlations Correlation of Istanbul Stock Exchange with other stock markets is shown in Table 3.a, 3.b and 3.c. [TABLE 3] Table 3.a shows that all of the correlation coefficients are positive. Namely, stock indices of group America move in the same direction. MERV and SNP have the highest correlations with ISE. Table 3.b shows that all stock indices at group of Asia-Pacific are positively correlated. KS11 and HSI have very high correlations whereas correlation coefficients of N225 and SSEC with ISE are low. Table 3.c shows that all stock indices are highly correlated with ISE with one exception. The correlation between ISE and FCHI is less than 0.50. 4.3. Granger Causality Granger causality test is applied to the data set to analyze the integration between stock markets in the short run. Pairwise causality test is conducted to thirteen stock markets from three territories. groups. The Granger causaliy tables show the relationship between ISE and America group, ISE and Asia-Pacific group and ISE and Europe group. In each group there are five indices which means 10 pairwise hypothesis exist. Table 4.a. shows pairwise Granger causality between Istanbul Stock Exchange and American countries. Bidirectional causality exists between ISE and BVSP. Also, ISE causes DJI, MERV and SNP unidirectional. Causalities among American countries show that all countries Granger cause each other with one exception. SNP and DJI do not cause each other. Table 4.b shows pairwise Granger causality between Istanbul Stock Exchange and Asian Pacific countries. There is unidirectional causality between ISE- KS11 and ISE- HSI. Neither SSEC nor ISE cause each other whereas there is bidirectional causality between N225 and ISE. Unidirectional causality exists between KS11 and HSI, N225 and HSI, HSI and SSEC, N225 and KS11, KS11 and SSEC. Bidirectional causality exists between N225 and SSEC. Table 4.c shows pairwise Granger causality between Istanbul Stock Exchange and European countries. Bidirectional causality exists between ISE-AEX and ISE- FCHI. ISE Granger causes FTSE and GDAXI. [TABLE 4] t i t COPY RIGHT 2013 Institute of Interdisciplinary Business Research 126

5. Conclusion This study investigated the linkage between Istanbul Stock Exchange and 12 stock markets from America, Asia-Pacific and Europe in the short run in order to inform investors about portfolio optimization and risk minimization at post-crisis period. Correlation tests and Granger causality tests are applied to the data starting from September 1, 2009 to December 31 2012. The study is limited to four stock indices from each of three territories and relationships of them with Istanbul Stock Exchange are shown separately. According to correlation test results, MERV and SNP have the highest correlations with ISE whereas BVSP and DJI have the lowest correlation with ISE by 0.86. In group of Asia-Pacific, all stock indices are positively correlated. KS11 and HSI have very high correlations with ISE in contrary to N225 and SSEC. In group of Europe, stock indices are highly correlated with ISE with one exception. The correlation between ISE and FCHI is less than 0.50. It is found that Istanbul Stock Exchange has not causal relationship with SSEC. Namely, investors of ISE can invest in China stock market in order to diversify their portfolios and minimize risks. ISE has bidirectional causality with BVSP, N225, AEX and FCHI which means that the returns of these stock markets affect each other in the short run. Unidirectional causality exists between ISE and rest of the seven stock markets. It is found that ISE causes all these stock markets, namely returns of ISE impact returns of DJI, MERV and SNP at America, HSI, KS11 at Asia-Pacific and FTSE and GDAXI at Europe. 6. Future Research The contribution of this paper is analyzing the linkage of Istanbul Stock Exchange with other markets classified as American, Asian-Pacific and European after the global financial crisis. Short run relationships between territories and stock indices analyzed. Long run linkages or pre-crisis and post-crisis analysis may be analyzed in the future. COPY RIGHT 2013 Institute of Interdisciplinary Business Research 127

REFERENCES David, N. B. (2011). Causality Among New York. London. Tokyo and Hong Kong Stock Markets. Australian Journal of Business and Management Research 1 (4), 122-131. Dickey,D. A. & Fuller, W. A. (1981). Distribution of the Estimators for Autoregressive Time Series with a Unit Root. Econometrica 49, 1057-72. Engle, R. & Granger, C. W. J. (1987). Cointegration and Error Correction: Representation, Estimation and Testing. Econometrica 55 (2), 251-276. Erdal, F. & Gunduz, L. (2001). An Empirical Investigation of the Interdepence of Istanbul Stock Exchange with Selected Stock Markets, Global Business and Technology Association International Conference Proceedings. Gerrits, R. & Yuce, A. (1999). Short and Long Term Links Among European and U.S. Stock Markets, Applied Financial Economics, 9, 1-11. Granger, C. W. J., (1969). Investigating Causal Relations by Econometric Models and Cross-spectral Methods. Econometrica 37, 424-438. Johansen, S. (1988). Statistical Analysis of Co-integrating Vectors. Journal of Economic Dynamics and Control 12, 231-54. Khan, M., Hassan, I. & Ahmad, M. S., (2011). Co-integration and Causality Analysis among Asian Stock Markets. Interdisciplinary Journal of Contemporary Research in Business 3 (3), 632-649 Kim, S. (2005). Information Leadership in the Advanced Asia-Pacific Stock Markets: Return, Volatility, and Volume Information Spillovers from the US and Japan, Journal of the Japanese and International Economies, 19 (3), 338-365. Küçükçolak, N. (2008). Co-integration of the Turkish Equity Market with Greek and Other European Union Equity Markets. International Research Journal of Finance and Economics 13, 58-73. Shamiri, A. & Isa, Z. (2010). Volatility Transmission: What do Asia-Pacific Markets Expect?.,Studies in Economics and Finance, 27 (4), 299-313. Tokic, D. (2003). Emerging Markets Before the 1997 Asia Pacific Financial Crisis. Asia Pacific Business Review, 9 (3), 105 15. Ugur, E. & Nor, A. H. (2010). The Stock Market Relationship Between Turkey and The United States Under Unionization. Asian Academy of Management Journal of Accounting and Finance. 6 (2), 19-33. Yüce, A. & Şımga Mugan, C. (2000). Linkages Among Eastern European Stock Markets and the Major Stock Exchanges. Russian and East European Finance and Trade, 36 (6), 54-69. Yuce, A. & Simga-Mugan, C. (1996): An Investigation of the Short- and Long-term Relationships Between Turkish Financial Markets, The European Journal of Finance, 2 (4), 305-317. www.ise.org COPY RIGHT 2013 Institute of Interdisciplinary Business Research 128

Annexure Table 1. List of Stock Markets Asia- Pacific European American Abbreviation Name of stock Market Country HSI Hang Seng Index Hong Kong KS11 Kospi Composite Index Korea N225 Nikkei 225 Japan SSEC Shanghai Composite Stock Exchange China AEX Amsterdam Stock Exchange Netherlands FCHI CAC 40 Index France FTSE Financial Times Stock Exchange United Kingdom GDAXI Deutscher Aktienindex Germany ISE Istanbul Stock Exchange Turkey BVSP Bovespa Stock Index Brazil DJI Dow Jones Industrial Average USA MERV Buones Aires Stock Exchange Argentina SNP S&P 500 Index USA TABLE 2 Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis Jarque-Bera AEX 2.49 2.51 2.62 2.30 0.06-0.96 3.41 178 BVSP 4.76 4.78 4.86 4.47 0.08-1.28 3.99 336 DJI 4.04 4.05 4.13 3.82 0.07-0.71 2.68 96 FCHI 3.54 3.54 3.66 3.40 0.05-0.31 2.36 36 FTSE 3.72 3.74 3.78 3.55 0.05-1.13 3.25 235 GDAXI 3.78 3.79 3.88 3.56 0.07-0.72 2.90 97 HSI 4.30 4.31 4.40 4.04 0.07-1.38 4.37 431 ISE 4.71 4.76 4.90 4.33 0.14-1.22 3.35 277 KS11 3.23 3.25 3.35 2.97 0.08-1.11 3.50 234 MERV 3.35 3.38 3.56 2.92 0.15-0.95 3.18 160 N225 3.97 3.98 4.11 3.85 0.04-0.03 3.37 6 SNP 3.06 3.07 3.17 2.83 0.07-0.77 2.85 109 SSEC 3.41 3.41 3.54 3.23 0.07-0.24 2.14 44 COPY RIGHT 2013 Institute of Interdisciplinary Business Research 129

Table 3. Correlation Matrices Table 3.a Correlation Matrix of Group America ISE BVSP DJI MERV SNP ISE 1 BVSP 0.86 1 DJI 0.86 0.6 1 MERV 0.94 0.86 0.8 1 SNP 0.87 0.64 0.99 0.82 1 Table 3.b Correlation Matrix of Group Asia-Pacific ISE HSI KS11 N225 SSEC ISE 1 HSI 0.88 1 KS11 0.95 0.86 1 N225 0.35 0.61 0.3 1 SSEC 0.39 0.63 0.36 0.57 1 Table 3.c Correlation Matrix of Group Europe ISE AEX FCHI FTSE GDAXI ISE 1 AEX 0.8 1 0.81 0.88 0.87 FCHI 0.44 0.81 1 0.55 0.6 FTSE 0.92 0.88 0.55 1 0.93 GDAXI 0.89 0.87 0.6 0.93 1 TABLE 4 Table 4.a.Granger Causality of Istanbul Stock Exchange and American Countries Null Hypothesis: Prob. Remarks BVSP does not Granger Cause ISE 0.0327 ISE does not Granger Cause BVSP 0.0427 ISE <=> BVSP DJI does not Granger Cause ISE 0.1778 ISE does not Granger Cause DJI 2.E-07 ISE => DJI MERV does not Granger Cause ISE 0.4794 ISE does not Granger Cause MERV 5.E-06 ISE => MERV SNP does not Granger Cause ISE 0.1384 ISE does not Granger Cause SNP 5.E-09 ISE => SNP DJI does not Granger Cause BVSP 3.E-33 BVSP does not Granger Cause DJI 0.0124 DJI <=> BVSP MERV does not Granger Cause BVSP 1.E-06 BVSP does not Granger Cause MERV 0.2238 MERV => BVSP SNP does not Granger Cause BVSP 4.E-34 BVSP does not Granger Cause SNP 0.0062 SNP <=> BVSP COPY RIGHT 2013 Institute of Interdisciplinary Business Research 130

MERV does not Granger Cause DJI 0.0047 DJI does not Granger Cause MERV 9.E-10 DJI <=> MERV SNP does not Granger Cause DJI 0.7385 DJI does not Granger Cause SNP 0.5672 SNP = DJI SNP does not Granger Cause MERV 6.E-10 MERV does not Granger Cause SNP 0.0015 SNP <=> MERV Table 4.b.Granger Causality of Istanbul Stock Exchange and Asian-Pacific Countries Null Hypothesis: Prob, Remarks HSI does not Granger Cause ISE 0.0896 ISE does not Granger Cause HSI 5.E-08 ISE => HSI KS11 does not Granger Cause ISE 0.5915 ISE does not Granger Cause KS11 1.E-09 ISE => KS11 N225 does not Granger Cause ISE 0.041 ISE does not Granger Cause N225 0.0006 N225 <=> ISE SSEC does not Granger Cause ISE 0.5109 ISE does not Granger Cause SSEC 0.7974 SSEC = ISE KS11 does not Granger Cause HSI 6.00E-06 HSI does not Granger Cause KS11 0.4398 KS11 => HSI N225 does not Granger Cause HSI 5.E-06 HSI does not Granger Cause N225 0.0548 N225 => HSI SSEC does not Granger Cause HSI 0.6162 HSI does not Granger Cause SSEC 3.E-10 HSI => SSEC N225 does not Granger Cause KS11 4.E-05 KS11 does not Granger Cause N225 0.1384 N225 => KS11 SSEC does not Granger Cause KS11 0.3409 KS11 does not Granger Cause SSEC 1.E-05 KS11 => SSEC SSEC does not Granger Cause N225 0.0045 N225 does not Granger Cause SSEC 0.0355 N225 <=> SSEC Table 4.c.Granger Causality of Istanbul Stock Exchange and European Countries Null Hypothesis: Prob. Remarks AEX does not Granger Cause ISE 0.0086 ISE does not Granger Cause AEX 3.00E-11 ISE <=> AEX FCHI does not Granger Cause ISE 0.0252 ISE does not Granger Cause FCHI 0.0074 ISE <=> FCHI FTSE does not Granger Cause ISE 0.0574 ISE does not Granger Cause FTSE 4.00E-16 ISE => FTSE GDAXI does not Granger Cause ISE 0.0684 ISE does not Granger Cause GDAXI 1.00E-09 ISE => GDAXI FCHI does not Granger Cause AEX 0.244 COPY RIGHT 2013 Institute of Interdisciplinary Business Research 131

AEX does not Granger Cause FCHI 3.00E-07 AEX => FCHI FTSE does not Granger Cause AEX 3.00E-07 AEX does not Granger Cause FTSE 0.7552 FTSE => AEX GDAXI does not Granger Cause AEX 0.071 AEX does not Granger Cause GDAXI 0.0014 AEX <=> GDAXI FTSE does not Granger Cause FCHI 0.0469 FCHI does not Granger Cause FTSE 0.0851 FTSE => FCHI GDAXI does not Granger Cause FCHI 1.00E-32 FCHI does not Granger Cause GDAXI 0.0135 GDAXI <=> FCHI GDAXI does not Granger Cause FTSE 0.0005 FTSE does not Granger Cause GDAXI 7.00E-06 GDAXI <=> FTSE Note: = indicates no Granger causality, => indicates uni-direction Granger causality from returns of one stock market to another and <=> means bi-directional Granger causality between two stock markets. COPY RIGHT 2013 Institute of Interdisciplinary Business Research 132