Geostatistical models for parametric insurance triggers

This project is one of a number that are in competition for funding from the NERC Great Western Four+ Doctoral Training Partnership (GW4+ DTP). The GW4+ DTP consists of the Great Western Four alliance of the University of Bath, University of Bristol, Cardiff University and the University of Exeter plus five Research Organisation partners: British Antarctic Survey, British Geological Survey, Centre for Ecology and Hydrology, the Natural History Museum and Plymouth Marine Laboratory. The partnership aims to provide a broad training in earth and environmental sciences, designed to train tomorrow’s leaders in earth and environmental science. For further details about the programme please see http://nercgw4plus.ac.uk/

Location: University of Exeter, Streatham Campus, Exeter EX4 4QJ.

Supervisory team:
Dr Ben Youngman Department of Mathematics, College of Engineering, Mathematics and Physical Sciences University of Exeter.
Dr Theo Economou Department of Mathematics, College of Engineering, Mathematics and Physical Sciences University of Exeter.
Geoffrey Saville, Willis Towers Watson

Project details:
Reliably quantifying and predicting natural disaster risk is vital for the exposed/vulnerable population and insurance companies. Reinsurers that underwrite such risk typically rely on catastrophe models for quantification. Parametric insurance is a cost-effective alternative to reinsurance, by avoiding development costs of catastrophe models, which gives developing countries opportunity to promptly receive financial support following natural disasters. One insurer, CCRIF SPC, paid Caribbean countries approximately $31.2 million soon after Hurricane Irma.

The rise of parametric insurance highlights its importance and the benefit of parsimonious, cost-effective modelling. Most parametric insurance products trigger a payout when simple criteria are met, e.g. a flood payout if rainfall exceeds some threshold, irrespective of whether a flood occurs. Basis risk describes the mismatch between payouts and disaster occurrence. This project’s research will aim to help minimise basis risk by building statistical models able to represent natural hazard events.

Project Aims and Methods
This project will focus on the meteorological component of natural disasters, i.e. rainfall for flooding, or wind speeds for hurricanes. It will involve the development of geostatistical models to support the parametric insurance industry. Such models allow fast, efficient and robust simulation of meteorological hazards associated with natural disasters, such as floods or hurricanes. Compared to typical vendor-operated catastrophe models, geostatistical models require significantly less computing resource and hence expense. In fact such models can be translated into open-source user-friendly computer code, which offers the industry transparency. Both sides of the industry can benefit: parametric insurers can validate and/or modify criteria to reduce basis risk, while customers can better understand and compare products. Ultimately, reinsurance could even become more cost-effective.

The majority of the project will involve the development of cutting-edge statistical methodology and accompanying software to efficiently combine geostatistical and extreme-value models for use with Big Data. This will enable fast simulation of natural disasters at high resolution without the need for supercomputation. The project will therefore produce statistical catastrophe models that are accessible to developing countries. Not only will work be on the frontier of academic research, but significant collaboration with partner Willis Towers Watson will helps the work’s impact outside of academia.

Training
The successful applicant will be encouraged to attend four Academy for Postgraduate Training in Statistics (APTS) courses. Attending relevant ad-hoc courses relevant to statistics, programming, meteorology or natural hazards will also be encouraged (examples include Introduction to Catastrophe Modelling, by Oasis Loss Modelling Framework, or Big Data: tools and statistical methods, by RSS).