High-Definition Modeling: A Game Changer for Catastrophe Risk Management
Chesley WilliamsMay 01, 2020
The RMS high-definition (HD) models are the latest generation of the RMS probabilistic modeling suite, incorporating features that allow flexible catastrophe risk management to meet the requirements of the evolving catastrophe risk market. But what does HD mean for the users of the models, and their metrics and analytics?
Innovation driving quality of modeling insights
RMS has been building catastrophe risk models for more than 30 years. The RiskLink® DLM models you use today, the first models built to accept detailed portfolios, are very efficient to run and serve the market well. They are an excellent example of how RMS takes on the challenge to build solutions to advance customer outcomes: more detailed and richer data leads to better analytics and insights. This dedication to using innovation to address industry needs and challenges continues today, with the development of the high-definition modeling and the HD models.
Over the last several years, the RMS model development team has been focused on developing a suite of RMS next generation HD probabilistic models, covering a range of perils from flood and wildfire to earthquakes and tsunami, and geographies. These HD solutions, featuring a shift to a high-fidelity, simulation-based framework for modeling event frequency and severity, and represent a major step forward in the quality of catastrophe risk quantification.
A new approach to simulation modeling
Simulation approaches for catastrophe models have been around for a long time. So, what’s different about our approach to simulation? For HD modeling, we use simulation not just temporally but also at the finest level of exposure in a portfolio. Let’s take a deeper look.
Firstly, temporal simulation: HD models have a more realistic representation of hazard event frequency. The temporal simulation allows for explicit modeling of the time-dependencies of events, including seasonality, clustering, and antecedent conditions, which are particularly important for flood modeling and the time-dependent recurrence of earthquakes.
Typhoon seasonality in the western North Pacific provides a good example of why capturing seasonality enhances typhoon and inland flood risk modeling. While typhoons in this region can occur in any month of the year, there is a clear seasonality with more typhoons in the summer. Capturing this seasonality allows for a better representation of antecedence conditions for the flood modeling; the state of antecedent conditions has proven to be very important in the recent typhoons impacting Japan. Additionally, temporal simulation allows modeling of special policy and treaty conditions such as Nth event covers, multi-year contracts, and contracts that include hours clauses.
Secondly, exposure loss simulation: the RMS HD models use a novel and unique approach to simulating ground-up losses leading to a more realistic representation of loss severity. Losses are simulated at the location coverage level, by peril and sub-peril, from the ground up for each event. This ensures that the models represent how policies will take loss in real life; these losses are then propagated through the financial model.
From a vulnerability point of view, HD modeling also brings improvements. The RiskLink DLM engine uses a beta distribution to quantify the secondary uncertainty on the mean damage ratio (MDR), which works well in many situations but does not capture that many locations will experience 0% or 100% loss from an event. In the HD models, a damage distribution that reflects the reality of loss distributions is achieved through the use of 4-parameter uncertainty modeling; this specifically captures the probability of complete loss or no loss, i.e. 0% or 100% MDR.
During reconnaissance trips, RMS researchers observed first-hand the importance of capturing locations with complete loss (see Figure 1) and with no loss. This observation has been backed up by claims data analysis. At low hazard levels when MDRs are low, there will be a large number of locations with no damage in close proximity to locations with damage. At high hazard levels, there are often catastrophic structural failures resulting in complete losses. This more realistic distribution of MDRs is captured in the vulnerability module of HD models.
More realistic losses lead to superior business outcomes
These simulation approaches represent a fundamental enhancement in the quality and resolution of modeling output, with resultant improved transparency in loss analytics. In conjunction with the RMS Contract Definition Language (CDL), users can express all possible terms and conditions in an insurance or reinsurance contract and use the HD financial model to analyze losses from any contract specification. The HD models provide more transparent and accurate calculations of risk metrics. That said, this does not mean we are converting every risk model to HD. This is because the existing RiskLink models continue to work very well for many peril regions, however, we believe HD models are game changers in the markets for which they are available and will drive differential competitive advantage. Can you afford not to have them?
To learn more about the RMS HD models suite, we invite you to join us for RMS Exceedance 2020 Virtual Event– This Changes Everything where we will have several sessions on HD models, HD modeling, the RMS platform, Risk Intelligence, and modeling application, Risk Modeler.
If you haven’t registered yet, click here to find out more about Exceedance 2020.
Chesley manages the commercial development of the RMS earthquake and tsunami models for the Asia-Pacific region. Chesley joined the Model Development Group at RMS in 1995, with expertise in developing seismic source models.
Through her tenure at RMS, Chesley was a model developer for key products including the 2018 RMS Japan Earthquake and Tsunami HD model and earthquake models for Europe, Asia, and Latin America. Chesley has recently shifted roles from model development to product management to use her years of experience at RMS to facilitate strategic product development, product marketing and product management.
Chesley holds a master's degree in geophysics from Stanford University, where she researched ground deformation associated with the 1989 Loma Prieta earthquake.