Forecasting Volatilty of Returns of Soy Oil Futures Using Garch Models
Published: 2016
Author(s) Name: Alok Kumar Sahai and Brijendra Pratap Singh |
Author(s) Affiliation: Asst. Prof., Sri Sri University, Odisha, India and Asst. Prof., GLA Univ., Mathura, U.P., India
Locked
Subscribed
Available for All
Abstract
The purpose of this paper is to model and forecast the volatility of returns for soy oil futures prices
using the GARCH family of models. Non linear models from GARCH family, specifically the TGARCH
and EGARCH models are used to assess the role of time varying volatility of soy oil futures prices.
The results reveal that soy oil futures do not exhibit different volatility responses to negative and
positive shocks. The shocks to the conditional variance showed a high persistence. GARCH (1, 1)
model was the most efficient for modeling the volatility based on Log likelihood, Akaike Information
Criteria (AIC) and Schwarz Criteria (SC).
We tested forecasting efficiency of the three variants of GARCH using four different criteria of
root mean square error (RMSE), mean absolute error (MAE), mean absolute percent error (MAPE)
and Theils inequality coefficient (TIC). While the models do not differ much on RMSE and MAE,
GARCH (1, 1) performs best on TIC while TGARCH (1, 1) model performs best on the MAPE
criteria.
Keywords: GARCH, TGARCH, EGARCH, Soy Oil, Conditional Volatility.
View PDF