Modeling and Forecasting of All India Monthly Average Wholesale Price Volatility of Onion: An Application of GARCH and EGARCH Techniques

Abstract

The study utilized log returns of all India monthly average wholesale prices(Rs/Q) of onion over period Jan-2010 to Dec-2021 and employed the autoregressive integrated moving-average (ARIMA), generalized autoregressive conditional heteroscedastic (GARCH), exponential GARCH (EGARCH) and threshold GARCH (TGARCH) modeling techniques with different error distribution such as normal and student-t. Lagrange multiplier test has been applied to detect the presence of autoregressive conditional heteroscedastic (ARCH) effect. The Ljung-Box test has been used for testing the autocorrelation exists in a time series. A comparative study of the above models has been done in terms of root mean square error (RMSE), mean absolute percentage error (MAPE) and R-square. The residuals of the fitted models have been used for diagnostic checking. The study has revealed that the ARMA (2,1) model is the best fitted modeling the mean equation for the log returns whereas in the variance equation, basic GARCH (1,1) and EGARCH (1,1) models with student-t innovations are appropriate in describing the symmetric and asymmetric behaviors of the log returns on the basis of smaller value of AIC (Akaike information criterion) and BIC (Bayesian information criterion).