Insurance and reinsurance companies have historically traded in annual hurricane risk, transferring risk with annual contracts covering hurricanes that may occur next year. There are many products that use common methods for quantifying risk that allow for the insurance and reinsurance markets to transfer risk between parties. This has led to the development of stochastic catastrophe models that quantify and price annual hurricane risk. Until recently, there have been hardly any products that allow the market to manage risk in the immediate short term. Yet, there has been growing interest in LiveCAT, an alternative risk transfer product that supports the near real-time transfer of risk, for example in the days between a hurricane identification and landfall. LiveCAT products aim to transfer risk associated with known, but still unfolding events and, as such, require analytical tools that combine the strengths of NWP forecasting with stochastic catastrophe models.
Insurance companies have tools in place to estimate the expected post-landfall loss of storms, but there are few tools available to generate pre-loss estimates for a landfalling storm. Current approaches to pricing LiveCAT risk depend largely on identification of stochastic event analogues. If the unfolding event closely matches in track and intensity an event in the historic record, then the associated hazard intensity and loss data may guide pricing. Alternatively, one can search in the stochastic event catalogue associated with a traditional catastrophe model for one or more event analogues. There are two weaknesses to this approach: 1) even with large stochastic catalogues consisting of hundreds of thousands or millions of events, it is often hard to find good matching event analogues, and 2) using historical stochastic analogues may not be suitable to assess the loss due to a specific hurricane given its unique size, intensity, and radius of maximum wind (RMax).
To overcome these challenges, we utilize a risk analysis tool that captures the uncertainty and cost associated with a potential landfalling named storm. Through the use of parametric models, we derive an expected wind field from a set of forecast tracks and storm parameters with a probability distribution. This output generates a small catalogue (500 5,000) of events, each consisting of a track, values for the five key storm parameters, and an associated rate or likelihood of occurrence. This small hazard catalogue is then fed into a catastrophe model to generate a customized expected loss for the landfalling named storm.
The aforementioned tool can be used to increase competitiveness in alternative risk management markets. The LiveCAT market provides a real-time risk management technique allowing parties to use standard forecasts and observation variables to price the risk of insured losses in real-time, which is not possible when using other currently available risk management models. This tool shows how, by incorporating uncertainty around meteorological forecast variables using probability distributions, users can generate a suite of potential wind fields and assess the potential damage of the storm to exposed risk.
Supplementary URL: http://www.flagstonere.com/index.aspx