MATH
COLLOQUIUM, THURSDAY, MARCH 16,
INVITED Dr. Vytaras
Brazauskas, Associate Professor (Actuarial Science),
SPEAKER: Department of Mathematical Sciences, University of Wisconsin-Milwaukee
TITLE: Robust Statistical Methods for
Actuarial Modeling
ABSTRACT: Statistical methods play an important role in the pricing and
risk measurement of insurance portfolios. While standard approaches
(e.g., parametric and empirical nonparametric procedures) are
quite
successful in predicting behavior of typical risks, their performance
is poor when portfolio contains non-typical excessive claims, also
known as `outliers'. To alleviate this problem, we propose to use
robust
statistical techniques because they are designed to "deal" with
outliers.
In this talk, we will present a sample of actuarial modeling
problems
from reinsurance (e.g., proportional, excess-of-loss, largest claims),
risk measurement (e.g., proportional hazard transform, conditional
tail
expectation, value-at-risk), and credibility (e.g., Buhlmann-Straub
model). Also, some "success stories" of robust statistical
methods in
these areas will be discussed.
REFRESHMENTS: 11:00 am, STARR 138
http://www.ferris.edu/htmls/colleges/artsands/Math/MATH_COLLOQUIUM/ColloquiumWeb/index.html