Surface water flood forecast-based loss estimation for resilient finance

点击翻译按钮获取中文摘要

Following initial supervision of climate related financial risks, The Bank of England requires UK banks and insurers to stress test their assets and liquidities against climate risks biennially; the next stress test is planned for summer 2023\. Participating insurers will need to respond to questions around their strategic responses to the three CBES scenarios. Central to these are the abilities of banks to have sufficient reserves for climate shocks. Estimating damage from flooding has always been reliant on understanding the depth and duration of events. Use of depth-damage curves has become industry standard for finance and insurance modelling. However, before an event, no system has been able to supply such data as existing technologies only show areas at risk based on generic event simulations. By simulating predicted depth times of floods ahead of events, Previsico has created the ability to produce pre-event automated loss estimations, for the first time, globally. Specifically, loss estimates for surface water and small watercourse flooding provide critical insights for the financial sector to support decision making, driving climate resilience. Further, loss estimates delivered alongside real-time risk assessments for assets can 1) help insurers reserve enough capital in advance of events so that they are able to pay-out their customer's claims in a timely and cost-effective manner and 2) support improved flood response capabilities for both the financial sector and associated stakeholders, helping deliver direct and indirect savings to assets and critical infrastructure, and mitigate impacts on the wider economy by supporting business continuity.During Phase 1, a MVP - representing the combination of flood forecasts and nowcasts and forecasted losses on a single dashboard and email - was developed, tested and then demonstrated in a non-production environment. Phase 2 will correspond to the operationalisation, live trialling and then refinement of the developed MVP in partnership with financial sector partners/clients with different specialisms and therefore, applications/uses for the technology. Specifically, the developed MVP will be extensively trialled across the UK, with stakeholder feedback on the form and functionality of the developed MVP collected. Furthermore, and through focused feedback sessions with selected partners, feedback on algorithm performance will be sought and used to significantly improve the accuracy of outputs, the granularity of forecasting losses, and the way forecasts are communicated. An integral part of this work will be accessing the partners portfolios of loss data to back-test past events, comparing their claims experience against Previsico's loss predictions.