Research · Free to cite

Storm risk, housing age, and who actually pays.

Original analysis built on FEMA's National Risk Index, Census American Community Survey housing data, and the NOAA Storm Events database. Every study publishes its method and its limitations. Journalists and researchers are welcome to cite or reproduce any of it.

83,200 census tracts
161,403 storm events
Free to cite
01Studies

Three questions about storm risk, answered with public data.

STUDY 01

Where America's oldest housing meets the worst hail

28.8 million Americans live in high-hail neighbourhoods where the typical home was built before 1980. The overlap is concentrated in a belt running from Iowa through Ohio.

READ THE STUDY →
STUDY 02

The households most exposed to hail are the least able to pay the deductible

Americans in the lowest hail-risk neighbourhoods have a median household income roughly $14,000 higher than those in the highest-risk ones. Storm exposure and financial cushion run in opposite directions.

READ THE STUDY →
STUDY 03

FEMA's hail risk score and actual hail events agree less than you would expect

Across 1,354 counties, the correlation between FEMA's modelled hail risk and observed NOAA hail events over five and a half years is 0.24. The two measure different things — and a homeowner reading 'low risk' may not realise it.

READ THE STUDY →
02Data

What these are built on.

83,200

census tracts with both a FEMA risk score and ACS housing data

161,403

NOAA storm events, 2021-01-01 to 2026-08-01

64,876,851

US building footprints held for roof-area work

Sources are FEMA's National Risk Index, the US Census Bureau American Community Survey, and NOAA's Storm Events Database — all public. Our own roof measurements are deliberately not used in these studies: the sample is far too small and too self-selected to support national claims. Underlying aggregate data is published at /datasets.

03For press

Using this work.

All of it is free to cite with attribution to RoofTap. County-level and tract-level breakdowns beyond what is published here are available on request, including custom cuts for a specific market.

Request data or an interview →