Modeling and Forecasting Uganda’s Beef and Cattle Milk Production using the Box-Jenkins Methodology

Abstract

Beef and cattle milk production play a significant role in reducing hunger, malnutrition, and rural poverty, improving rural livelihoods, creating employment opportunities, and supporting the overall development of Uganda's economy. This study was conducted to find a suitable ARIMA model for forecasting Uganda’s beef and cattle milk production using annual time series data from 1961 to 2020, extracted from the Food and Agriculture Organization Corporate Statistical Database (FAOSTAT). Following patterns of the Autocorrelation Function and Partial Autocorrelation Function plots of the differenced series, 4 tentative ARIMA models were identified for milk production, i.e., ARIMA (0,1,0), ARIMA (1,1,0), ARIMA (0,1,1), and ARIMA (1,1,1). While 3 tentative ARIMA models were identified for beef production, i.e., ARIMA (1,1,1), ARIMA (1,1,0), and ARIMA (0,1,1). ARIMA (0,1,0) model was selected to be the most suitable for forecasting cattle milk production because it had the smallest MAPE and Normalized BIC values. On the other hand, ARIMA (1,1,0) was selected to be the best model for forecasting beef production because it had the smallest normalized BIC value and a significant coefficient of the autoregressive component. Forecasts show that milk production will increase at an annual average rate of 1.63%, while beef production will increase at an annual average rate of 0.39% in the five-year forecast period (2021-2025). These findings are important in designing strategies to improve the beef and dairy livestock sub-sectors in Uganda.