Data Compression Algorithms, Marine Liability Modeling, and Hierarchical Risk Aggregation in Reinsurance


Ressource 1 Under indefinite embargo.
UNIL restricted access
State: Public
Version: After imprimatur
License: Not specified
Serval ID
PhD thesis: a PhD thesis.
Data Compression Algorithms, Marine Liability Modeling, and Hierarchical Risk Aggregation in Reinsurance
Guevara Alarcón William Miguel
Albrecher Hansjörg
Institution details
Université de Lausanne, Faculté des hautes études commerciales
Publication state
Issued date
Global reinsurers are companies that operate worldwide providing protection for the most extreme and complicated risks. Therefore, these companies continuously face new challenges. Particularly, the constant increase in computer processing power and available digital information requires mechanisms to efficiently store and transfer the influx of data. Technological advances produce an increasing accumulation of risks, for example, bigger construction projects, taller skyscrapers or longer vessels. The aggregate risk of global reinsurers must be estimated in a precise and reliable manner, considering appropriately the dependencies between different portfolios, to allow for adequate risk management. The present thesis discusses models and methodologies related to these particular challenges. Firstly, algorithms to reduce the amount of information from large data sets, while keeping the shape of their distributions and controlling the approximation error, are developed. Secondly, we focus on two issues related to the marine liability insurance market. The expected number and cost of a set of large claims is analyzed and modeled. Moreover, the optimal participation percentages in reinsurance contracts under restrictions in the total accumulation of risk are studied. Finally, the precision and variability of the simulations of a hierarchical risk aggregation dependency model under different methodologies are examined.
Create date
22/05/2019 19:04
Last modification date
17/12/2019 6:23
Usage data