Methodology: Per-1,000-homes rates and Texas county rank

Every figure carrying this label on a county page is computed exactly as described below, from public government data. Nothing is estimated or editorialized.

What it measures	A raw count of anything mostly measures how many people live somewhere, so every headline count on a county page is also shown per 1,000 homes (ACS housing units) and ranked against the other Texas counties. The ranking population is all 254 Texas counties. Rows in the data that are not counties are excluded: the state-level catch-all bucket that holds policies whose county could not be resolved, and counties outside the states this build covers. Counties with no NFIP policies are INCLUDED at a rate of zero — there are 26 such counties in Texas, and dropping them would make the site rank against 227 counties while calling them Texas. Rank is competition-style: counties with the same rate share a rank. A county's own rate is published only when it has at least 500 housing units, because a rate over a few dozen homes moves too far on one policy to print as a finding; such counties are still ranked. Claim rates compare like with like — the county's claims and every other county's claims are summed over the SAME loss-year window, the one shown on the chart.
Unit	—
Source dataset	FEMA — OpenFEMA NFIP v2 + US Census — American Community Survey (5-year)
Data version / accessed	through 2026-07 (cube build 2026-07-21b, generated 2026-07-21)
Exact source fields	`policyCount`, `countyCode`, `amountPaidOnBuildingClaim`, `dateOfLoss`, `B25001_001E (housing units)`
Formula	—
Predicates / filters	A raw count of anything mostly measures how many people live somewhere, so every headline count on a county page is also shown per 1,000 homes (ACS housing units) and ranked against the other Texas counties. The ranking population is all 254 Texas counties. Rows in the data that are not counties are excluded: the state-level catch-all bucket that holds policies whose county could not be resolved, and counties outside the states this build covers. Counties with no NFIP policies are INCLUDED at a rate of zero — there are 26 such counties in Texas, and dropping them would make the site rank against 227 counties while calling them Texas. Rank is competition-style: counties with the same rate share a rank. A county's own rate is published only when it has at least 500 housing units, because a rate over a few dozen homes moves too far on one policy to print as a finding; such counties are still ranked. Claim rates compare like with like — the county's claims and every other county's claims are summed over the SAME loss-year window, the one shown on the chart.
Date semantics	—
Window boundaries	not windowed
Rounding	one decimal below 10 per 1,000; whole numbers at or above 10
Suppression rule	no rate and no rank when the county has fewer than 500 housing units, or when fewer than 30 counties can be ranked
Minimum sample (min_n)	—
Geography	US county (FIPS)

Internal build query

This query runs against our internal data cube (private cube_*/fct_* tables built from the public sources below). It is exact but not runnable by a reader — use the public reproduction recipe that follows.

SELECT p.county_fips,
       1000.0 * COALESCE(p.policies_in_force, 0) / p.acs_housing_units
FROM cube_county_profile p
JOIN dim_county d ON d.county_fips = p.county_fips
WHERE p.county_fips LIKE '48%';   -- the join is what makes it a county

Reproduce from the public source

This recomputes the figure from the raw public dataset. Public data is live and may run ahead of our snapshot (see the data version above); numbers can differ by rows added since our ingest vintage.

OpenFEMA FimaNfipPolicies v2 (policyCount by countyCode) divided by
ACS 5-year table B25001 (total housing units by county), x 1,000.
Rank the result across all 254 Texas county FIPS codes.

Worked example: Harris County: about 99 NFIP policies per 1,000 homes, against a Texas county median near 6 — the gap a raw count of 190,470 hides.

Change log

2026-07-21 (r3) — window boundaries corrected (no windowed metric reads "not windowed"); LIMIT/ordering/geography/status predicates documented; the query relabeled the internal build query and paired with a public raw-source reproduction recipe.
2026-07-21 — record published with source URL, exact fields, predicates, window boundaries, rounding, suppression, and a worked example. Prior records disclosed a formula only.

Corrections

Spotted an error? Corrections: support@countyguard.com. We publish the fix and note it in this change log.

Every figure in this record is computed from the public dataset named above and independently re-checked against it before publication. This record documents the data method only; it is not insurance, legal, or engineering advice.

Published by Neha Malik, Publisher — Eldhurst LLC, who is accountable for what this site publishes. Countyguard is not a licensed adjuster, contractor, engineer, or attorney. Corrections and questions: support@countyguard.com.