A sinistralidade no seguro de responsabilidade civil geral

Abstract

This study investigates which observable factors in the insurance market explain the dynamics of loss ratios in line 0351 – General Liability Insurance – in Brazil. The empirical analysis is based on monthly data reported by insurance companies to the Brazilian insurance regulator (SUSEP), covering the period from November 2016 to May 2025. Initially, a SARIMA time series model was estimated in order to identify temporal and seasonal patterns in market loss ratios. Subsequently, a vector autoregressive (VAR) model was applied to examine the interaction between loss ratios and variables representing the operational structure of insurance companies. In addition, generalized linear models (GLM) were estimated under different distributions from the exponential family. The Tweedie distribution provided the best fit according to the Akaike Information Criterion (AIC). A clustering algorithm was also employed to identify distinct segments among insurers operating in the market. The results indicate that loss ratios exhibit temporal persistence and are associated with variables such as risk retention, participation in technical reserves, and market premium share. The findings suggest that the dynamics of loss ratios in the analyzed line can be explained by observable factors related to insurers’ operational structure and market conditions