Bivariate Dependence Modelling of a Non-Life Insurance Company: Pair-Copula Construction

This study demonstrates the aggregate losses and risk estimation. The purpose of this study is to test the bivariate dependence modelling between losses among different business lines. Subsequently, this research estimate the risk of a non-life insurance company by using Value-at-Risk (VaR) and Tail-Value-at-Risk (TVaR). The risks were calculated by using incurred claims and premiums data from nine business lines based on Malaysia non-life insurance company for loss data. R software, an open source software is used to conduct the analysis in this study. Two models from pair-copula construction were used; C-Vine and D-Vine copula model. D-Vine copula model was found has stronger dependence structure than the C-Vine model based on the value of Loglik, Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Lastly, this paper visualize the trend of incurred claims data of the non-life insurance company. This study believe the trend is based on Malaysian behaviour. This research observe that Motor insurance has the layout claims, which may translate to higher road accidents in Malaysia.