Impact of spatial correlations in earthquake catastrophe modelling
John Douglas ( Academic Supervisor )
Goran Trendafiloski ( Industrial Supervisor )
Catastrophe models are important tools providing robust assessment and financial management of earthquake-related emergencies, which still face the largest protection gap amongst the main perils. These models comprise three main components: (1) earthquake hazard, (2) exposure and, (3) vulnerability. Assessment of the hazard over a geographical region requires quantification not only of the expected ground shaking at a single location but also on how this shaking could vary over distances of a few kilometres. This variation is captured within spatial correlation models. The aim of this project is to improve the modelling of spatial correlation within catastrophe models. We are currently investigating the impact of different correlation modeling assumptions on the predicted losses to building portfolios and infrastructures in future earthquakes. The results of this project will have implications for (re)insurance companies evaluating such losses for their clients.
The main reason for choosing Strathclyde was my interest in the project and the experience of the main supervisor. An interesting project within the field of engineering seismology, the ability to collaborate with an industrial partner and the opportunity to live abroad in such an international department, university and city drew my attention. Aside from these main motivations, choosing the University of Strathclyde has other benefits. Specifically, it provides postgraduate students with a framework that allows them to develop invaluable transferable skills that go beyond the research itself. Through the activities offered by the University, I have had the opportunity to strengthen my problem solving and critical thinking as well as my creativity and project management skills. Finally, working with Impact Forecasting has allowed me to see the results of my research implemented within their catastrophe models to provide more robust risk assessment analyses.
This joint research project between the University of Strathclyde and Impact Forecasting, Aon’s catastrophe model development centre of excellence, has provided me with industrial training on the development and use of catastrophe models. It has also given me the possibility to understand the outcomes of the project not only from an academic point of view but also from the viewpoint of industry. I have had the chance to work with different software, including commercial packages, for seismic hazard and risk assessment ,which will be beneficial for my future career. The interaction with the industrial partner has been a great opportunity to apply my research in practice and to interact with a wide variety of people. Attending Impact Forecasting workshops has given me the opportunity to keep up to date with what is going on not only in the academic context but also in the private sector. Finally, my communication skills both in written and oral forms have improved through the participation in project meetings and by writing reports and papers.
In an increasingly risky world, insurers and reinsurers need more sophisticated tools to quantify and manage the risks their businesses have to face. Aon's catastrophe model developer, Impact Forecasting, enables firms to analyse the financial implications of catastrophic events and achieve a greater understanding of their risks. There is a market need for earthquake catastrophe models that take into account spatial correlation in order to provide more robust earthquake shaking estimated losses for spatially distributed building portfolios. As a result of this collaboration, we aim at implementing spatial correlation models into Impact Forecasting catastrophe models and understanding their effects on potential losses. We have currently developed custom spatial correlation models for Italy for the purposes of improving the seismic hazard and risk assessment for our upcoming Italy earthquake catastrophe model. Finally, we have investigated the effects of spatial correlation on-per-event loss estimates and in-location loss estimates for underwriting purposes.