Fagseminar Skade

08:55

Velkommen

Walter Neuhaus, leder i Fagkomité Skade

09:00 – 10:00

Chain Ladder Parameterisation with Supervised Learning

Abstract:

 This session explains how we can the bridge between traditional actuarial methods and machine learning modelling mindsets and workflows. Using this approach, a new technique was developed for the automated tuning of traditional reserving method parameters, such as chain link pattern selections in the Chain Ladder method. The technique involves defining an objective function which measures reserve efficiency with the addition of reserving class context specific constraints. We show how this objective function can be used to automatically tune the pattern, speeding up the pattern selection workflow, reducing reliance on subjective judgment and mitigating the likelihood of manual calculation errors.

Stephan Marais

10:00 – 10:20

Pause

10:20 – 11:25

Niels Henrik Abel 

Abstract:

I foredraget vil vi se litt på Niels Henrik Abels vei inn i matematikken belyst ved nye funn som viser hva den unge Abel leste. Videre markerer vi at det i år er 200 år siden Abel leverte sitt hovedverk til vurdering i Vitenskapsakademiet i Paris. Arbeid ble oversett og lagt til side – og manuskriptet har gjennom årene hatt en kronglete og spennende historie, inkludert da det skulle til Norge i forbindelse med Abeljubileet i 2002. Til slutt tar vi for oss den mytiske Abel som alt mens han var i live var tiltenkt en nasjonsbyggende rolle, og vi avslutter det hele med hvordan en pris til Niels Henrik Abels minne ble opprettet i 2002 – Abelprisen.

Nils Voje Johansen

11:30 – 12:30

Lunsj

12:30 – 13:35

Aktuarstudiet i Danmark m.m. 

Abstract:

The actuarial programme at the Department of Mathematical Sciences, University of Copenhagen (KU), is currently experiencing considerable success. Admission requires excellent high school grades, reflecting the strong demand for places. While high salary prospects are undoubtedly an important factor, the programme's strong academic reputation and visibility also play a significant role. As a result, we likely graduate more actuaries each year than the rest of the Nordic countries combined. In this talk, I will present the structure and key features of our programme and discuss the factors that have contributed to its growing popularity. I will also address some of the challenges the programme is currently facing. If time permits, I will further discuss the advantages and disadvantages of applying machine learning methods in non-life insurance pricing.

Jostein Paulsen 

13:35 – 14:25

Aktuarstudiet ved Universitetet i Bergen 

Abstract:

This section will talk about the Actuarial education offered by the Statistics and Data science group at the Department of Mathematics, University of Bergen.  Our Actuarial education was established in 1990s and has existed as a separate specialization within the 2-year master program in Statistics. In 2014, to highlight the actuarial profession, a 5-year integrated master's program "Actuarial Data Analytics" was created and it is the only 5-year integrated program in Actuarial science and data analysis in Norway. From spring 2026, this program is renamed as "Integrert master i finans, forsikring og data science".  This program provides candidates with broad knowledge in mathematics, statistics, finance/insurance and data science. The program also offers candidates Actuarial competence and follows the requirements for actuarial education of European Association of Actuaries.

Yushu Li 

14:25 – 14:45

Pause

14:45 – 15:50

The Event Interlocking Mechanism in a Multi-Layer 
Excess of Loss Reinsurance Program i skadereservering

Abstract: 
To enhance risk retention management, (international) insurance companies have 
largely shifted to non-proportional reinsurance structures. These structures allow for 
the design of reinsurance programs where an insured loss event can activate multiple 
underlying policies or portfolio segments. These segments are (a) reinsured separately 
for smaller events and (b) combined for larger events. An interlocking mechanism 
aggregates all claims from a single large event to facilitate recoveries under the 
combined protection. This presentation outlines the implementation of such an event 
interlocking clause and provides some modelling indications. 

Norbert Haible 

15:50

Avslutning