Deterministic catastrophe scenario

A deterministic catastrophe scenario is a single event, fully specified (track, intensity, ground footprint), whose loss on a portfolio is computed without attaching a probability to it. It answers "how much if", where a stochastic catalog answers "how much, and how often". The oldest use is the replay. Hurricane Andrew made landfall south of Miami on August 24, 1992, and every reinsurer knows how to rerun what it would cost on today's portfolio, with today's values and today's building stock. Lloyd's turned this into a discipline with its Realistic Disaster Scenarios (RDS): prescribed footprints, sometimes a fixed industry loss, which each syndicate maps onto its exposures and reports gross and net. The math is straightforward. The footprint sets the intensity at each site, and the vulnerability and financial modules do the rest, with no sampling. So there is no mean or standard deviation to read, just one number. The result can be read on its own, or placed on the occurrence exceedance probability (OEP) curve to infer an implied return period, which differs from one portfolio to the next. The uses overlap: accumulation control, regulatory stress testing, a floor for a peril with no model. There is also the conversation with a board, which grasps "an Andrew over Miami" better than a 1-in-200 PML. A practitioner's caveat: a scenario measures the loss from an event its author picked. A portfolio can be organized to keep its RDS losses down while staying heavily loaded twenty kilometers from the footprint, where no scenario goes.