Multivariate loss reserving using factor copulas


  • Marcelo Orgler


07/05/2020 - 14:00


Via Zoom


This work proposes using a copula implied by a latent factor structure in order to model the dependence between multiple run-off triangles. We consider a cell-wise dependence between triangles and use Simulated Method of Moments for the estimation of the factor copula parameters and GLM regressions to estimate the marginal distributions. We use data from two major auto insurance group in the US in order to perform an empirical study. We compare the results from the factor copula with the well known Gaussian Copula model and observe that the presence of tail dependence impact the risk of the portfolio.

*Texto enviado pelo aluno. 

Membros da banca: 

  • Rodrigo dos Santos Targino (orientador) - FGV EMAp
  • Eduardo Fonseca Mendes (co-orientador) - FGV EMAp
  • Yuri Fahham Saporito - FGV EMAp
  • Reinaldo Antonio Gomes Marques - UNIFAL