GARCH modelling

This paper examines whether the use of the 24 hours separately generates more accurate volatility estimates and forecasts. We have used the GARCH(1,1), GARCH(2,2), GJR- GARCH(1,1), GJR-GARCH(2,2), and EGARCH(1,1). We have tried to solve the problem of explosive volatility process by performing several methods. The results show that the use of the 24 hours separately does not generate more accurate estimates of the conditional volatility. One-day ahead forecasts obtained with the different GARCH models for the daily spot prices are the worst forecasts when comparing with the one-day ahead volatility forecasts of the 24 hours separately. The results also show that the average one-day ahead volatility forecasts of the 24 hours separately generates significant better forecasts for the daily volatility of the returns than using the one-day ahead volatility forecasts of the daily spot prices.