Heteroscedastic price forecasting for food security management in developing countries

Price forecasting systems are of considerable importance to food security management by governments' and non-governmental organizations. Sparse data availability in low-income economies, however, generally necessitates reliance on reduced form forecasting methods. Relatively recent innovations in heteroscedasticity-consistent time series techniques offer price forecasting tools that are feasible given available data and analysis technologies in low-income economies. Moreover, extended GARCH models exhibit superior out-of-sample forecast accuracy using monthly food price data from Madagascar. These techniques also permit cost reduction in food security operations by more precise estimation of the risk of hitting a critical price level.