There are two distinct analyses needed to underwrite property: risk assessment and accumulation. The first is the evaluation of a risk based on information specific to the risk, so that it can be accepted and rated, or rejected. The second limits the writing of new risks within a defined geographical area to prevent excessive losses from a single event. Both are important for underwriting cat risks: assessment because of the potential of losses up to the full value of a property (and beyond), and accumulation because cat events cause damage to swaths of properties at once.
One example of generalized data being used for underwriting is the NFIP A Zones. In this article from Brink, the findings of a comprehensive study of NFIP claims (cheers to the authors, Carolyn Kousky and Erwann Michel-Kerjan) are described, and conclusion #3 states:
In the U.S., much communication about flood risk centers on the 100-year floodplain line. This creates a false sense of safety for residents outside the 100-year line and does not indicate that risk can vary dramatically within the 100-year floodplain as well.
Data that misses the risk outside such a zone, and does not indicate the variety of risk within a zone, is generalized and inadequate for risk assessment. An underwriter should be able to segment and price 10-year vs 20-year vs 50-year vs 100-year vs 250-year vs >250-year risks, and not just write 10-year risks the same as 100-year risks because they are in the same generalized zone.
There has to be a better reason to write a risk than “there is room in that aggregation zone.” For goldfish, though, that’s good enough… but who wants to be a goldfish? Especially when writing cat risks.