Evaluating the Volatility Behaviour in Irish ISEQ Overall Index Using GARCH Models

Abstract

This paper aims to model the volatility of the Irish stock market (ISEQ) index price; data used in this article include the daily closing prices of ISEQ overall index, from January 1st, 2008 to March 28th, 2014. The data are observed to be naturally divided into three time frames; the first period from January 1st, 2008 to December 31st, 2009, the second from January 1st, 2010 to December 31st, 2011, and the third from January 1st, 2012 to March 28th, 2014. The volatility is modelled using symmetric and asymmetric Generalised Autoregressive Conditional Heteroscedasticity (GARCH) models including GARCH, EGARCH, TGARCH, PGARCH and FIEGARCH under the assumption that data follow a Normal distribution. Comparisons, both of estimations and forecasts of the volatility between GARCH family models have been performed. In general, for the whole time frame (1/1/2008–28/3/2014) and for the first period, (as specified above), the FIGARCH (1,1) model performs better than the others but, for the second period, the PGARCH (1,1,1) model is preferred. For the third period, the best model is found to be EGARCH (1,1), so that the ISEQ overall index of volatility does appear to exhibit different behaviour in different periods. The empirical findings summary, for the GARCH forms that apply for the overall and three sub-periods, indicate that so-called explosive volatility is present in the ISEQ Overall index returns over the extended period.