Understanding and modelling Tropical Cyclone risk in Oman

In recent years Oman and the MENA region have been significantly impacted by Tropical Cyclones (TC) which, on top of affecting the society in various aspects, have also led to unprecedented (re)insurance losses. Notable cyclones include Gonu (2007), Mekunu (2018), and Shaheen (2021). For instance, this last developed from the remnants of TC Gulab and made landfall on the coast of Al-Musannah, Oman, on 3^rd October 2021 as a Category 1 Cyclone, while causing strong winds and heavy rainfall also around the capital city of Muscat and, in turn, deaths and widespread damage to both public and private properties. It is thus of increasing importance to accurately understand and reproduce TC risk in Oman. In general, this can serve to predicting and preparing for any event and, in the context of the (re)insurance industry, to avoid poor risk assessment and weak financial protection.

Reproducing and quantifying TC risk in Oman results still very challenging, mainly because it can be considered as an unmodelled country, i.e., it is not part of the domain of main catastrophe model vendors. In this study it is shown how Antares Global, under Qatar Insurance Company, the main insurance in the region, has faced this challenge in developing its own TC view of risk and catastrophe model for Oman. The study has mostly focused on the Wind component of the model, which consists of 10,000 years of stochastic catalogue relying on IBTracs data; a claim-based vulnerability module for main line of business, i.e., commercial, residential, and industrial, adjusted to four recent historical events; and a financial module that considers the conditional probability of having a loss, also in this case calibrated to the same recent historical experience.

It is shown how it has been possible to converge to a robust View of Risk with best combining the three main components of the model and adjusting them to the input exposure. Claim data has also required a detailed analysis to isolate the windstorm component from the flood and efficiently use it for the validation. Ongoing work is looking at expanding the framework to include the TC and extra tropical cyclones flood components.