This paper compared new error innovation distribution in estimating volatility models. A new error innovation distribution called Exponentiated Generalized skewed student t distribution (EGSSTD) was developed and compared with the existing error distributions using an empirical dataset of daily returns of Nigeria Stock Exchange (NSE) index return from 2007 to 2017. The result of the stationarity Statistic shows that the data is stationary without transformation while the ARCH effect statistic using ADF statistic shows the presence of heteroscedasticity. The estimate of the volatility models were significant with probability values at 0.01 for the new error distribution and the existing distributions. The results obtained show that GARCH (1, 1) with EGSSTD error distribution performed better than the other models having the least AIC value. In terms of forecasting performance, GARCH (1, 1) with ESSTD error distribution outperformed other volatility models and error distributions with the least RSME.
Digital Object Identifier (DOI)
Agboola, Samson; Garba Dikko, Hussaini; and Ebenezer Asiribo, Osebekwin
"On A New Exponentiated Error Innovation Distributions: Evidence of Nigeria Stock Exchange,"
Journal of Statistics Applications & Probability: Vol. 7
, Article 9.
Available at: https://dc.naturalspublishing.com/jsap/vol7/iss2/9