Project CEMAPRE internal
|Title||Optimal reinsurance with dependencies|
|Participants||João Andrade e Silva, Alfredo Egídio dos Reis, Manuel Guerra, Alexandra Moura (Principal Investigator), Carlos Oliveira|
|Summary||The project focuses on the optimal reinsurance of dependence risks, continuing the 2021 project with|
the title "Optimal reinsurance of dependent risks with applications". There are still many open
questions regarding the optimal reinsurance problem with dependencies. Indeed, in spite being a
classical field of research in Actuarial Science, only recently dependence has been considered and
it is an active field of research. The interest in studying reinsurance under dependencies is still
increasing, driven by the need for real, robust and reliable quantitative risk models, even though
data is frequently very hard to obtain.
We aim continuing studying the reinsurance problem in the presence of dependencies of the underlying
risks and under general optimality criteria and general premium calculation principles. Dependencies
are considered both between risks and in time.
One interesting aspect of optimal reinsurance design is that in practice the insurer transfers part
of each risk, independently of other risks, even if there is awareness that there may be
dependences. So, the first insurer could somehow play with this fact. On the other hand, there are
usually several constraints in practice that are not commonly accounted in analytical works on
optimal reinsurance. One of such constraints is for instance the fact that no reinsurer will
indemnify an unlimited amount of the losses. Also, real reinsurance contracts usually include type
of contracts that may not be considered in fundamental works. We intend to study the optimal
reinsurance problem in the presence of dependence of the underlying risks, from both an analytical
fundamental point of view and an application perspective. Here, the main challenge is to obtain
insurance data, namely loss data. For that reason, simulated data will be also investigated.