Home > Department of Statistics > Events > abstracts > Credibility for Phase-type Distributions with a Single Risk Parameter

 

Credibility for Phase-type Distributions with a Single Risk Parameter

 

When Monday 7th December at 5.15pm
Where B617, Leverhulme Library, Columbia House
Presentations  
Speaker David Stanford
From University of Western Ontario
Abstract Recently, interest has developed as to whether exact credibility results (wherein the optimal Bayesian credibility premium takes on a linear form) are applicable to phase-type distributions. This talk establishes that this is in general not the case, although it is true for special cases such as Erlang distributions. At the same time, by exploiting the fact that every phase-type distribution can be expressed as an infinite mixture of Erlang distributions of increasing order, we are able to establish that the optimal linear credibility premium can be calculated in a straightforward fashion. It is unclear at present as to whether the optimal Bayesian credibility premium can be calculated for any given phase-type distribution in a tractable manner, as we illustrate through a selection of examples.
For further information Sabina Allam (Postgraduate Administrator) Ext. 6879
Department of Statistics, Columbia House
Share:Facebook|Twitter|LinkedIn|