The Power of first differencing in addressing non stationarity time series: Empirical evidence from price of rice using SARIMA model, Tanzania

Abstract

The main purpose of this study was to forecast the price of rice in Tanzania using a secondary univariate time series from January 2005 to September 2024. We used the seasonal ARIMA (3, 1, 2) (2, 0, 0) [12] model with seasonal data. The analysis revealed that the price of rice in Tanzania will exhibit a consistent upward trend from October 2024 to a peak in March 2027 reflecting a total increase of approximately 121.4% over this period. From April 2027, prices will begin to decline steadily reaching low in August 2029, a reduction of 44.3% from the March 2027 peak. Prices typically dip in late summer, with the sharpest decline observed in August 2025, when prices fall by 32.8% from June 2025. The decline beginning in April 2027 reflects a significant market shift, with an average monthly decrease of approximately 2.1% from April 2027 to August 2029. This may be attributed to increased production or effective market interventions. The policy issues should focus on the areas of price stabilization programmes, strengthening trade and export policies putting in place mechanisms to support farmers by giving them access to agro inputs and ensuring the technical services are availed to farmers through agricultural extension officers. This will increase productivity and address the problem of demand which will in turn solve the entire problem of price volatility. This shall be achieved by influencing consumer behaviour and moderating demand, the interventions that help reduce the volatility that arises from sudden spikes in demand, contributing to more stable pricing over time.