Testing for causality in variance for world stock exchange indexes

The main aim of this study was to determine the nature of the relations between selected stock exchange indexes in the world (ATX, DAX, NASDAQ, NIKKEI, FTSE and WIG20), with special emphasis on the causality in variance. Due to the characteristics of financial variables (the daily closing rates of analyzed indexes) such as: focusing on volatility, volatility of the conditional variance, skew and leptokurtic, GARCH models and Cheung-Ng test were used to study the relations between selected capital markets. First part of the paper contains a brief theoretical introduction about GARCH model and the description of the Cheung-Ng test. In the next part were the analyzed time series described, including the testing for ARCH effect. The last and the most comprehensive empirical part contains the results of tests for different combinations of three levels of significance: α = 0.1, α = 0.05 and α = 0.01 and for time intervals: m = 1, m = 3 and m = 5. The results of analysis demonstrate that selected world's capital markets are strongly linked with each other, and the volatility of one financial series has an impact on others. It may be explained by the increasing integration and liberalization of financial markets, globalization and technological advances in information flow. The results of Cheung-Ng test indicated that among the analyzed indexes of stock markets the greatest impact on the world markets has the NASDAQ index and the lowest – DAX, FTSE and WIG20. In addition, tests showed that the most vulnerable to foreign influence is the NIKKEI index, and the most independent – NASDAQ

Autoregressive conditional heteroskedasticity · Capital market · Conditional variance · Econometrics · Economics · Financial economics · Granger causality · Stock exchange · Stock market · Stock market index · Variance decomposition of forecast errors · Complex Systems and Time Series Analysis · Financial Risk and Volatility Modeling · Stock Market Forecasting Methods · Finance