ARIMAX Model to Forecast Grain Production under Rainfall Instabilities in Brazilian Semi-Arid Region

Abstract

The state of Ceará has most of its area in Brazil's semi-arid region. Initially, the research segmented Ceará's annual rainfall into 6 periods: very rainy, rainy, normal-humid, normal-dry, drought and very drought. This segmentation was based on the annual rainfall in the state between 1901 and 2020. The research estimated the average rainfall and instability of both the annual rainfall in the state during the period and those estimated for the periods in which the rainfall was segmented. The research then developed forecast models for harvested areas, yields, production values and average annual grain prices between 1947 and 2020, the years in which this information is available. To make these forecasts, the research used the ARIMAX model, which is an extension of the Box-Jenkins model, with the addition of an exogenous variable. The exogenous variable included in the model was the annual rainfall observed between 1947 and 2020, assuming that this variable influences these forecasts. The results showed that the state's rainfall has a high level of instability and that the adjusted models proved to be parsimonious and robust from a Statistical point of view.