Modeling Domestic Price Volatility for Cereal Crops in Ethiopia

Abstract

The volatility in the domestic prices of maize and teff crops has been found to vary over time from month to month. Thus, Families of time series models namely, ARCH with their extensions to generalized ARCH, GARCH and EGARCH models with ARIMA mean equations were considered to the data. The best fitting model among each family of models was selected based on how well the model captures the variations in the data. The optimal lag specification for the models are accessed via AIC and SBIC. Comparisons of the symmetric and asymmetric selected models were carried out based on the significance of asymmetric term in the EGARCH model. Thus, statistically significance of asymmetric term and least forecast error from the model established that the EGARCH model with GED for residuals was superior to the GARCH model. Therefore, the ARIMA(2,0,3)-EGARCH(1,1) and ARIMA(0,0,3)-EGARCH(2,3) were chosen to be the best fitting models among the ARIMA(p, d, q)-GARCH(P, Q) family for monthly domestic price volatility of maize and teff crops, respectively. However, the volatility in the domestic price of wheat and barley was found to be not changing over time. Hence, the variance of the ARIMA process was used as the measure of volatility in the prices of these two crops which were 0.00112 and 0.0004, respectively. Moreover, it was found that from candidate exogenous variables, import prices for maize crop, fuel oil price, exchange rate (dollar-birr), inflation from non-food items, past shock and volatility on the domestic price had statistically significant effect on the current month domestic price volatility for maize and teff crops.