Our Sources and Methodology

Key takeaways

Every dollar figure carries either a source and year, or an explicit statement that it is invented to show arithmetic. What we will not do is print a figure that looks sourced and is not.
We state a source's vintage — several popular insurance datasets are much older than the articles citing them imply.
We do not republish NAIC's tables. They are copyright; we cite figures in prose instead.
We do not build fifty-state tables from inference. Verified states are named; the rest are called unverified.

Why you can trust HomeCoverDesk. Every page is written from primary sources — statutes, regulations, state insurance department material and primary agency documents — quoted and cited inline at the claim, with the part of the document named. A licensed property & casualty producer is being contracted to review every page; that review has not begun, and the byline on each page shows the role as pending. We are paid the same amount whether or not you buy a policy. Read our editorial policy, methodology and how we make money.

The sources, and what is wrong with some of them

Every dataset has limits. Most articles citing them do not mention those limits. We think the limits are part of the fact.

The sources we build on, with their vintageStating a source's vintage is not a formality. Several widely-cited insurance datasets are considerably older than the articles quoting them imply.
Source	What we use it for	Vintage and caveat
Federal and state statutes	What the law actually requires	Cited to the section, as amended, with the date we reviewed it
State insurance department bulletins	Regulator guidance and notice requirements	Cited to the department with the issue date
NAIC market data	Non-renewal rates, premiums, market structure	Cited in prose only — NAIC tables are copyright. The 2026 report excludes surplus-lines carriers and omits New York entirely
US Treasury, Federal Insurance Office	National premium and non-renewal benchmarks	The January 2025 report covers 2018–2022. Public domain and freely reusable
FEMA National Risk Index	County-level hazard exposure	Data version 1.20, December 2025, superseding v1.19.0 of March 2023. We state the version and its date wherever we use it
FEMA / NFIP records	Flood policies, claims and take-up rates	Refreshed roughly monthly. Public domain
NOAA / NCEI Storm Events	Hail, wind and tornado history by county	Monthly, with roughly a 90-day lag. We use 1996 onward only
US Census ACS	Housing stock, values, ownership	Annual. We state the vintage
NOAA Billion-Dollar Disasters	We do not use it	Retired with no updates beyond 2024. Still widely quoted as current. It is not
Named industry research	Market trends where no public equivalent exists	Cited with the firm, the date, and what the sample actually was

The one worth singling out. NOAA's Billion-Dollar Weather and Climate Disasters product was retired, with no updates beyond calendar year 2024. It is still cited across the insurance web as though it were live. The historical archive remains valid as history — but if you see it presented as a current running total, that is wrong, and it is a useful test of whether a source is checking its own citations.

Assumptions behind any figure

A premium figure means nothing without the policy it describes. Where we publish one, these are the assumptions unless the page says otherwise.

The assumption set we would use if we published a modelled premiumOur own assumption set, revised August 2026. **Corrected on that date:** this table previously called itself “the industry-standard assumption set” and claimed our figures were comparable to other publishers'. Neither was true. The personal property figure was about 8% of the dwelling limit, where North Carolina's and Wisconsin's insurance departments both put the standard relationship at around 50%, and no major publisher uses these values. It is now stated as ours and set to the regulators' relationship. Note also that **we do not currently publish any modelled premium**, so this set is not in use on any page.
Assumption	Value
Dwelling coverage	$300,000
Personal property	$150,000 — 50% of the dwelling limit, the relationship North Carolina and Wisconsin both describe as standard
Personal liability	$300,000
Deductible	$1,000

We also distinguish two kinds of average premium that are routinely confused and differ by roughly a factor of two: figures based on premium actually earned across all in-force policies, and figures based on quotes gathered at a fixed coverage level. Both are valid. Mixing them produces nonsense. We label which one we are using.

How we treat uncertainty

Three rules, applied consistently.

If we verified it, we cite it. To the statute, the bulletin, the dataset — with a date.

If we verified some of it, we say which. Several state-level insurance rules are genuinely not published in comparable form anywhere. Where we checked five states and not fifty, we publish five and say so.

If we could not verify it, we do not publish it as fact. Either it is omitted, or it appears with the uncertainty attached. We would rather a page be less complete than quietly wrong.

If a figure on this site is wrong, our corrections policy explains what we do about it.

Methodology and sources

This page describes our own standards. Individual pages carry their own methodology notes for figures specific to them.

Frequently asked questions

Why do you not publish a table for all fifty states?

Because for several topics — roof-age underwriting thresholds, non-renewal notice periods, public adjuster fee caps — no authoritative, comparable fifty-state source exists. A complete-looking table would be inference presented as fact. We publish what we verified and name the rest as unverified.

Can I reuse your figures?

The underlying federal data is public domain and yours to use. Where we have derived something ourselves we say so on the page, and we would appreciate a link. Figures we cite from copyright sources such as the NAIC should be taken from the original.

How current is the data on this site?

It depends on the source, which is why the vintage is stated. FEMA flood records refresh roughly monthly; the National Risk Index is at data version 1.20, published December 2025; census housing data is annual. Where a source is stale, the page says so — and where we have gotten a vintage wrong, we say that too.

Where Our Numbers Come From