Premium forecasting methods for Crop Insurance and allocation of federal resources to the Crop Insurance Premium Subsidy Program

Abstract:

The forecast of crop insurance premiums has a relevant impact on the budget dimensioning of the Crop Insurance Premium Subsidy Program (PSR). From the perspective of responsibility and public expenditure planning, it is necessary to estimate with relative precision how much is intended to be spent on the program in the short and medium-term (Constitutional Amendment no. 95, 2016). The present study makes a comparison of methods for the projection of the agricultural insurance premium, by region. The SARIMA models and the NNAR, TBATS, MAPA, and ELM algorithms were used, with and without a subsidy covariate. The methodologies were applied to the monthly data of the premium volume of crop insurance in the South, Southeast, Midwest, and Northeast regions of Brazil between 2006 and 2018. It was observed that the univariate SARIMA model showed better results in the Midwest regions and Northeast, while SARIMA and ELM with the covariate were higher in the South and Southeast, respectively. From these results, it was possible to discuss the relevance of the subsidy for the expansion of insurance in the regions analyzed.

Keywords:
crop insurance; premium; forecast; subvention