Geographic resolution of exposure data
The level of precision at which a portfolio locates its insured properties, from a building's coordinates to a CRESTA zone or a country, which caps what a cat model can do with it.
Geographic resolution of exposure data is the level of precision at which a portfolio locates its insured properties: building coordinates, street address, postal code, municipality, CRESTA zone, region or country. It decides what a catastrophe model can do with the data. At point level, the model reads the hazard at the site: distance to the coast, elevation, soil type, water depth. At postal code level, it only knows a box and spreads the value inside it on an assumption, often in proportion to built area or population. At CRESTA zone level, it smears the portfolio over hundreds of square kilometers. Perils do not all react the same way. European windstorm, which hits wide, tolerates postal codes fairly well for a homeowners book. Flood, storm surge and surface water vary over a few tens of meters: two houses on the same street can be one dry, the other under a meter of water. Reinsurance treaties are still often quoted on submissions aggregated by CRESTA zone, because the cedent has nothing better or does not want to hand over more. Models compensate with wider uncertainty, which usually fattens the tail of the exceedance curve. The direction of the error is not guaranteed, though: disaggregation can also move toward a river a value that actually sits on high ground. Geocoding produces the resolution, and exposure formats carry it in a precision field. Underwriters look at that field before the average annual loss, because a portfolio 40% geocoded at municipality level does not read like one coded to the building, even with the same headline number.
Textbook case, made-up figures: a portfolio of 3,000 homes along a river, 600 million euros insured. Submitted by postal code, the model spreads it across the whole built area of those codes, and the 1-in-100 flood loss comes out at 45 million. Geocoded to the address, it turns out 400 houses sit less than 200 meters from the bank: the 1-in-100 loss rises to 78 million. The reverse case happens too, when the homes are clustered on high ground and the uniform spread had pulled them toward the water.
exposure resolution, geocoding resolution, geographic granularity of exposure, location resolution
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.