OPTIMUM CONDITIONAL DISTRIBUTION FOR ESTIMATING THE VOLATILITY OF STOCK INDICES

Abstract

This study aims at minimizing the excess kurtosis observed in the financial data especially in stock price by obtaining an optimum conditional distribution for estimating the linear GARCH (p, q) model. The study used the daily stock prices of First Bank Nigeria Pic. traded in the Nigeria Stock Exchange from January 2001 to July 2008 and the results from the statistical properties of the daily returns and that of the Jarque-Bera test showed that the returns series is leptokurtic. The selected GARCH model was compared for estimation based on the Normal,

Student-t distribution and Generalized Error distribution (GED). The optimum distribution was selected using the Bayesian information criterion (BIC) and the Likelihood values and the results indicated that the kurtosis displayed was reduced when GED is used in the estimation. For stationarity. the parameters sum is less than one and this implies that the parameters satisfy the second order stationary (weakly stationary) conditions.