Field notes: the renewal where the schedule was short a hundred buildings

Suburban houses along the edge of a wide open field under a cloudy sky

The submission came in on a Tuesday, three weeks before binding, the way they always do. A regional cedant, a mid-sized territory, a schedule of values that had clearly been rolled forward from the prior year with a few new entries tacked on the bottom. Four hundred and twelve structures across the zone. Nothing unusual about the file itself. It had the usual gaps in construction class, a few TIVs that looked copy-pasted from the neighboring line, and occupancy codes that someone had clearly filled in under deadline pressure rather than from an actual survey.

What was unusual was what the imagery showed once we laid a current footprint count over the same boundary. Five hundred and eleven structures, give or take, depending on how you want to treat a few ambiguous outbuildings. A hundred buildings that existed on the ground and nowhere on the cedant's books.

There was no fraud in it, and no broker hiding exposure. A schedule is a snapshot of what someone thought was built, as of whenever the underlying survey or policy data was last touched, carried forward year over year because re-surveying a territory is expensive and nobody budgets for it until something forces the question. The schedule wasn't wrong when it was built. It just stopped being true the day the first new structure went up after the last data pull, and nobody flagged the moment it happened.

Where does a hundred buildings come from in one renewal cycle? In this territory, it was mostly infill. A handful of larger parcels along the zone's eastern edge had been subdivided and built out since the last full survey, the kind of gradual densification that never makes it into a single underwriting file because no one event triggered it. There was also a cluster of rebuilds following a storm two seasons back, structures that came back slightly larger and slightly offset from their original footprints, enough to break any match against an old address list even if the address list had been accurate. And there was the ordinary churn of a growing market: outbuildings, small commercial sheds, additions that never generate a new policy because the risk is already covered under the general TIV.

None of that shows up in a self-reported schedule unless someone goes and looks. It shows up immediately in a current image.

The accumulation implication isn't subtle. A hundred structures missing from a schedule for a single zone in a single renewal is not noise you can wave off with a loading factor. It changes the ground-up exposure base, which changes the PML, which changes what the treaty costs to put up. We ended up running the layer two ways, once on the submitted schedule and once on the corrected footprint count, and the gap between them was the kind of number that gets a second look from the pricing actuary before the slip goes final.

What strikes me, going back over the file, is how ordinary the cause was. No disaster, no scandal, just a built environment that moved faster than the paperwork describing it. That's the normal condition of every schedule more than a year old in a market that's building. Exposure Mapping builds that current footprint count straight from annual satellite imagery before the treaty binds, so the comparison happens at the data layer instead of waiting for a loss to show you where the schedule was wrong.

Next renewal that lands with a schedule that looks a little too clean, it's worth putting it over a current image before you trust the TIV.

Get started

← Back to the blog