Applied logistic model to the market auto insurances in the Brazil: calculation of the probability of sinister

Abstract

This paper uses a Logit model to calculate the conditional probability of occurrence of a sinister, for a level of insurance premium, in the market auto insurances, in the Brazil. The results had shown that the conditional probability of sinister, data the premium level, has values many next ones for different states in the five geographic regions with exception for the state of São Paulo in which the probability of sinister occurrence little arrives to be more than the double of the probability of sinister in the state of Rio de Janeiro. Considering that different states possess risks of different sinisters, the fact of the probabilities to be very next indicates the existence of a possible distortion in the determination of the price of the insurance based on the historical data of sinister and this can a determinative factor for volume of insurance premium as fraction of the GDP in the Brazilian case is minor who in countries as United States or Japan. Key words: linear model of logitistic regression, market auto insurances Artigo recebido em 21 de ago. 2009.