EXACT CREDIBILITY FOR WEIGHTED OBSERVATIONS

作者: Rob Kaas , Dennis Dannenburg , Marc Goovaerts , None

DOI: 10.2143/AST.27.2.542053

关键词:

摘要: This note generalizes Jewell's theorem on exact credibility from the classical B/Jhlmann model to (weighted) Bfihlmann-Straub model. A well-known of Jewell (1974) states that credibility, which is concurring Bayesian estimator (posterior mean) with a contract mean, found for class examples includes many common situations. In nutshell, obtains when observations are drawn frorn distributions in exponential family, natural conjugate prior risk parameter. Surprisingly, pertains only Bfihlmann model, and does not hold case different variances allowed, as BfihlmannStraub this contribution we prove well, thus allowing be averages varying numbers observations, also Poisson Binomial distributions. The parametrization used coincides one theory Generalized Linear Models. original form theorem, ours rather cumbersome reparametrizations required ordinary like Gamma special cases theorem. remedied Gerber (1995) by choosing more convenient parametrization.

参考文章(8)
Peter McCullagh, John Ashworth Nelder, Generalized Linear Models ,(1983)
D.R. Dannenburg, R. Kaas, M.J. Goovaerts, Practical actuarial credibility models IAE. ,(1996)
J.A. Nelder, R.J. Verrall, CREDIBILITY THEORY AND GENERALIZED LINEAR MODELS Astin Bulletin. ,vol. 27, pp. 71- 82 ,(1997) , 10.2143/AST.27.1.563206
Hans U. Gerber, A Teacher's Remark on Exact Credibility Astin Bulletin. ,vol. 25, pp. 189- 192 ,(1995) , 10.2143/AST.25.2.563247
A.E. van Heerwaarden, T. Bauwelinckx, R. Kaas, M.J. Goovaerts, Effective actuarial methods North-Holland. ,(1990)
William S. Jewell, Credible Means are exact Bayesian for Exponential Families Astin Bulletin. ,vol. 8, pp. 77- 90 ,(1974) , 10.1017/S0515036100009193
William S. Jewell, Regularity Conditions For Exact Credibility Astin Bulletin. ,vol. 8, pp. 336- 341 ,(1975) , 10.1017/S0515036100011260
P. McCullagh, J. A. Nelder, Generalized Linear Models Springer US. ,(1989) , 10.1007/978-1-4899-3242-6