Time Series Analysis of Stock Market Volatility in Pakistan

Abstract

The stock market in an emerging country like Pakistan has been volatile from the earliest times. This paper studies the volatility of Pakistan Stock Exchange (PSX) (using Karachi Stock Exchange 100 Index (KSE-100) as a proxy) through the application of Generalized Autoregressive Conditional Heteroscedasticity (GARCH) family models. The sample period consists of 4831 daily observations for the 19 year trading period (from 2000 to 2019). Symmetric GARCH (2, 1), asymmetric EGARCH (1, 1), GJR-GARCH (1, 1) and APARCH (1, 1) models were used under Gaussian distributional assumptions. The results validate the empirical findings of previous studies conducted in Pakistan that log returns of KSE-100 Index are characterized by volatility clustering, time-variability, leptokurtic distribution with dominant ARCH and GARCH effects. An interesting feature of Pakistan Stock Exchange revealed by asymmetric models (used in the study) is that PSX is more volatile to good news than bad news. Moreover EGARCH (1, 1) outperforms all other models of the study on the basis of AIC/BIC criterion. However the comparison of correlations of variances predicted by three asymmetric models reveal that correlations among them are very high, with minimum correlation being 98%. This essentially means that all three asymmetric models provide a good fit to PSX.