Gestão de riscos e dependência espacial em seguros agrícolas: uma análise da produtividade da soja

Abstract

Studies on spatial correlation agricultural productivity can generate relevant information for the risk management of insurance companies regarding geographical distribution strategies of customers. The objective of this study was to evaluate the spatial dependence structure of a municipal average soybean yield for the State of Paraná, Brazil, from 1990 to 2013. The analysis was conducted by means of the Moran spatial correlogram. A strong spatial autocorrelation was found for recent years. The distance at which the spatial dependence was no longer significant increased from 89.37km in 1990 to 121.87km at the end of the reporting period. Therefore, it was concluded that the Moran’s correlogram was sufficient to describe the spatial relationships of dependence of the average soybean productivity in Paraná. Moreover, this correlation confirmed the breakdown of the actuarial assumption that the studied units are spatially independent. Also, it is recommended that insurance firms use more than one spatial correlation analysis method to validate its estimates for the definition of risk management strategies of the geographical distribution of the customer portfolio.