Robust Loss Development Using MCMC

49 Pages Posted: 8 Nov 2009 Last revised: 17 Feb 2010

Date Written: February 15, 2010

Abstract

A Bayesian model of developing aggregate loss triangles in property casualty insurance is introduced. This model makes use of a heteroskedastic and skewed t-likelihood with endogenous degrees of freedom, employs model averaging by means of Reversible Jump MCMC, and accommodates a structural break in the consumption path. Further, the model is capable of incorporating expert information in the calendar year effect. The model, which has been compiled into the R package lossDev, is applied to two widely studied General Liability and Auto Bodily Injury Liability loss triangles.

Keywords: Loss development, skewed Student distribution, Reversible Jump MCMC

JEL Classification: G22

Suggested Citation

Schmid, Frank A., Robust Loss Development Using MCMC (February 15, 2010). Available at SSRN: https://ssrn.com/abstract=1501706 or http://dx.doi.org/10.2139/ssrn.1501706

Frank A. Schmid (Contact Author)

AIG

80 Pine Street
6th Floor
New York, NY 10005
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
295
rank
97,461
Abstract Views
905
PlumX Metrics