Actuarial Summer School 2005
Lectures
1. Statistical Methods of Data Mining (14 lecture hours)
by Jan Mielniczuk, IPIPAN, Poland
2. Hedging of Life Insurance and Pension Guarantees (10 lecture hours)
by Ed Morgan, Milliman, UK and Tamara Burden, Milliman, USA
Organisation
All classes take place from
Monday to Friday (18 -22 July) in the new part of the building of the
Department of Economics of the University of Warsaw, Długa St. 44/50, room B.
Below the daily schedule of classes is inserted.
|
Monday |
Tuesday |
Wednesday |
Thursday |
Friday |
|
830 - 920
|
Mielniczuk |
Mielniczuk |
Mielniczuk |
Mielniczuk |
Mielniczuk |
|
930 - 1020
|
Mielniczuk |
Mielniczuk |
Mielniczuk |
Mielniczuk |
Mielniczuk |
|
1020 - 1050
|
Break |
Break |
Break |
Break |
Break |
|
1050 - 1140
|
Mielniczuk |
Morgan |
Mielniczuk |
Mielniczuk |
Mielniczuk |
|
1150 - 1240
|
Morgan |
Morgan |
Burden |
|
|
|
1240 - 1310
|
Break |
Break |
Break |
|
|
|
1310 - 1400
|
Morgan |
Burden |
Burden |
|
|
|
1410 - 1500
|
Morgan |
Burden |
Burden |
|
|
Syllabuses
Course 1: Statistical Methods of Data Mining by Jan Mielniczuk
- SUPERVISED CLASSIFICATION I
-
Introduction: two- and
multi-class classification problem;
-
Linear discriminant analysis;
-
Bayes approach;
-
Quadratic classification;
-
General issues: choice of a classifier and assessment of performance.
-
SUPERVISED CLASSIFICATION II
-
Logistic discrimination and refinements;
-
Tree classifiers: CART methodology.
-
SUPERVISED CLASSIFICATION III
-
Empirical Bayes methods (kernel and nearest neighbour);
-
Other methods (brief synopsis).
-
Merging of classifiers (boosting method).
-
Dependence analysis and classification for discrete data -loglinear models.
-
REGRESSION METHODS
-
Kernel methods, local linear smoothers, splines;
-
Additive and generalized additive models (logistic additive regression);
- Regression trees, projection pursuit regression and neural nets.
-
MISCELLANEOUS TOPICS
-
Unsupervised classification - cluster analysis;
-
Principal component analysis;
-
Feature selection and extraction.
Course 2: Hedging of Life Insurance and Pension Guarantees
The aim of the course is
to describe how modern financial risk management techniques can be used to
improve the management of a life insurance company or pension funds.
Life companies and pension funds
frequently provide guarantees which can be directly equated to financial
options and evaluated using well established option pricing techniques. They
can also usually be hedged by buying suitable derivative investments.
The talks will describe this theory and go on to explain
some of the practical applications both in the management of existing life
insurance portfolios and in the design of new business.
Part 1: By Ed Morgan
-
THE NATURE OF INSURANCE LIABILITIES WITH INVESTMENT GUARANTEES
-
An overview of the way different insurance
and pension product types provide investment guarantees.
-
What are insurance consumers really looking for
and do these products provide it for them?
-
Can we design the perfect product?
-
RISK MANAGEMENT ISSUES PRESENTED BY INVESTMENT GUARANTEES
-
Risk management issues
-
How life insurance
management techniques and deterministic methods have hidden the
risks
-
What are the options to
manage these risks..
Part 2: By Tamara Burden
-
THEORETICAL ASPECTS OF HEDGING
-
Viewing these guarantees as financial options
-
Risk-neutral valuation of derivatives
-
Extension of risk neutral valuation to insurance products
-
Hedging - what it means and how it works
-
PRACTICAL ASPECTS OF HEDGING
-
Different hedge strategy options
(dynamic, semi-static, structured, optimization techniques)
-
How to select a hedging strategy
(financial projections, tradeoffs of frictional costs vs. reduction in variance)
-
Interaction of hedging with accounting/regulatory issues