What is an aggregation zone, and why structure counts set its edges

Aerial view of a suburban street grid intersection lined with houses, trees, and parked cars.

An aggregation zone is the boundary you draw around exposure before you ask how bad a single event could get inside it. Draw it too wide and you dilute a real concentration into a number that looks manageable. Draw it too narrow and you split one cluster of buildings across two zones, and the model never sees the accumulation at all. Either way, the treaty gets priced on a shape that doesn't match the ground.

Most zones start life as something administrative: a CRESTA zone, a postcode group, a cedent's own territory split. Those lines are convenient because everyone already has them. They have nothing to do with where buildings sit. A CRESTA boundary can run straight through the middle of an industrial park that went up in the last five years, putting half the warehouses in one zone and half in the next, with no relationship to how a hailstorm or a flood would move across that site.

Where the zone boundary comes from

An aggregation zone should track where structures actually cluster on the ground: three hundred rooftops packed into a four-block radius, then open fields for a mile past that. The zone edge belongs at the point where the density drops off, regardless of where the zip code line runs underneath it. That's the definition that drives PML and CAT limit decisions. A zonal accumulation only means something if the boundary reflects a real break in built density, a point where the exposure genuinely thins out.

The problem is that most schedules don't tell you where that break is. A statement of values gives you a count and a TIV per location, rolled up by whatever zone the cedent already uses. It doesn't tell you that two subdivisions sit a half-mile apart with nothing but a drainage channel between them and should be split, or that a new logistics corridor has filled in what used to be the gap between two zones and should now be treated as one continuous exposure concentration area. Schedules describe what was reported. They don't describe what's standing.

Why structure counts define the edges

Structure count is the variable that locates the edge. Where counts per unit area climb, you're inside the zone. Where they fall off, you've found the boundary, whether or not it lines up with the administrative line underneath it. This is a different exercise from scoring individual risk. It's a density read across a whole territory, and it's the kind of thing you can only do reliably with a current, wall-to-wall view of what's built, not a sample of reported addresses.

That's also why these edges move. A zone drawn five years ago off a cedent's schedule doesn't account for the subdivision that filled in since, or the rebuild after the last event that replaced forty single-family footprints with a denser infill pattern. The administrative boundary never changes. The actual concentration does, every renewal cycle, in whatever direction the local build cycle is going. If your aggregation zones are still keyed to a boundary set years ago, you're carrying a pricing error that has nothing to do with the model and everything to do with the input.

A current built-inventory layer pulled from annual satellite imagery gives you something closer to the real edge: structure footprints and counts across the full treaty territory, not just the addresses a cedent happened to report. You still make the zoning call yourself, but you're making it against a map of what's there this year, which is a different exercise than making it against a schedule that was accurate when it was submitted and has been going stale ever since.

None of this replaces a cat model's own aggregation logic. It replaces the guesswork in the step before the model runs, where someone has to decide where one zone ends and the next begins. Getting that boundary closer to the real density pattern doesn't change what the model does with the exposure. It changes whether the exposure you hand it reflects the territory you're underwriting.

If you're renewing a treaty and the zone edges in your current schedule look more administrative than real, Exposure Mapping builds that current built-inventory view straight from annual high-res imagery, ready to load into your model.

Get started

← Back to the blog