Application of SARIMA Model on Forecasting Wholesale Prices of Food Commodities in Tanzania: A Case of Maize, Rice and Beans

Abstract

This research used a time series model called the Seasonal Autoregressive Integrated Moving Average (SARIMA) technique to model and forecast  wholesale prices of Tanzania`s key food crops, notably maize, rice, and beans. The SARIMA model was selected due to its ability of fitting data with  seasonality. Monthly wholesale prices data of the three crops between February 2004 to October 2021 in Tanzania were retrieved from the website  of the Bank of Tanzania (BoT), resulting in 213 observations on each crop. The data from February 2004 to October, 2020 were used to fit a SARIMA  model and data of November 2020 to October 2021 were used to validate the model. The results show that SARIMA (0,1,2) (1,0,1)12, SARIMA (0,1,0)  (1,1,1)12 and SARIMA (0,1,0) (0,1,1)12 are the most suitable models for forecasting wholesale prices of maize, rice and beans respectively. The  model’s accuracy was tested using Mean Absolute Percent Error (MAPE), and the results were found satisfactory. The results reveal that maize, rice,  and beans will all have higher peak prices in February 2022, with TZS 54,083/=, TZS 167,258/=, and TZS 180,117.68/= per 100kg, respectively.  Therefore, SARIMA (0,1,2)(1,0,1)12, SARIMA (0,1,0)(1,1,1)12 and SARIMA (0,1,0) (0,1,1)12 models could serve as a useful tool for modelling and  forecasting monthly wholesale prices of maize, rice and beans respectively in Tanzania.