How we counted: Florida homestead exemption screen

Last updated 28 July 2026.

This page explains where our headline figure comes from, what it means, and — just as importantly — what it does not.

---

The claim

617,235 parcels across Florida appear owner-occupied and carry no homestead exemption on the current tax roll, carrying in aggregate $284.6M of first-year exemption value at county-wide millage rates.

That is a records screen, not a determination that anyone is eligible.

Absence of a homestead exemption on the roll is not evidence of entitlement. A rental, a second home, a property held in trust, an estate and a company-owned house can all look identical to this screen. We are counting parcels that look owner-occupied in the public record and show no exemption — nothing more.

We do not say that 617,235 homeowners are missing $279 million, because we don't know that any particular owner is eligible. And the dollar figure is what the exemption would be worth on those parcels if granted, which is a different sentence from what any individual would receive.

---

The screen

A parcel is counted when all three are true:

1. No homestead exemption appears on the roll for that parcel. 2. The property use code is single-family or condominium. Vacant land, commercial and agricultural parcels are out of scope. 3. The owner's mailing address matches the property address, and the owner's name doesn't look like a company. This is the owner-occupancy proxy — someone who lives in the house usually receives mail there.

That third test is the one doing the work, and it is also the one with the most give in it. See why county figures aren't comparable, below.

What was read, and what was in scope

The screen reads every parcel carrying a situs address and can only count the ones whose use code puts them in scope. Both numbers matter, and only the second is the denominator for 617,235.

parcels read, all 67 counties

10,658,310

in scope — single-family and condominium use codes

7,726,306

counted by the screen

617,235

share of in-scope parcels counted

7.50%

---

Which records, and when

All 67 counties, using each county's 2026 preliminary tax roll as published by the Florida Department of Revenue, loaded 28 July 2026. No county fell back to an older roll.

That matters more than it sounds. An earlier version of this screen ran against a year-old roll, and roughly one in six of the parcels it flagged had already been resolved by the time we looked. Every figure here is measured on the newest roll that exists.

---

How often the screen goes stale

We measured this rather than estimating it, across all 67 counties.

Taking the parcels flagged on the 2025 final roll and checking them against the 2026 roll:

flagged on the earlier roll

561,698

showing an exemption a year later

97,401

rate

17.34%

of those, same owner on both rolls

87.6%

So about one in six flagged parcels shows an exemption within a year, and in the large majority of those the owner didn't change.

Note what is measured and what is not. We observe an exemption appearing on the later roll for a parcel that had none. We do not observe anyone filing. The same-owner share is reported because it is the fact that makes filing the natural reading — but it remains a reading, and the roll does not record it.

What this does and doesn't establish. It measures the cost of letting a screen go stale. It is not an error bar on today's figure, which is computed on the most current roll available. That number's own staleness can't be measured until the next roll publishes.

---

The dollars

Not a single per-parcel estimate multiplied across the state. For every counted parcel, the first-year exemption value is computed from that county's own millage, with the school and non-school split preserved — because the first $25,000 of the homestead exemption applies to all levies, while the second $25,000 applies to non-school levies only.

aggregate first-year exemption value

$284.6M

median per counted parcel

$451

aggregate divided by counted parcels

$425

TAX_AUTH_CD resolvable on counted parcels 617,235 / 617,235 (100%)

The median sits above the simple average because most counted parcels carry the full exemption while a tail of low-value parcels carries only part of it.

This understates, by design. The calculation uses county-wide levies only. Municipal and special-district levies that apply to some parcels and not others are excluded, so a typical property would save more than these figures show. For a number that appears on a public page, we'd rather be low than high.

---

Portability

The same approach doesn't work for portability, and we don't pretend otherwise.

Homestead value depends only on the parcel, so it can be computed for every one. Portability value depends on the previous property, which the tax roll doesn't identify. We publish counts and no statewide dollar figure, because any such number would be an assumption about hundreds of thousands of prior properties nobody has identified.

Two different counts, and they are not interchangeable:

parcels whose record is consistent with a transferred benefit not yet claimed

780,376

of those, the ones our checker shows a visitor

303,184

The narrower number is what a visitor actually sees, because the checker only raises portability where the parcel already carries a homestead exemption — the situation where a transfer is the remaining step. The wider number includes parcels where that is not yet true. Neither is a statement about any owner's entitlement; the Property Appraiser determines every transfer.

---

Why county figures aren't comparable

County rates range from about 13% down to effectively zero, and most of that spread is how a county formats its records, not what its housing market looks like.

The owner-occupancy test compares two address fields. That works cleanly where a county records both fields the same way. In several counties it doesn't — one uses the mailing-address field for a care-of line, another for a second owner's name, a third formats the two fields differently often enough that an exact comparison almost never matches. Those counties contribute close to zero, for reasons that have nothing to do with their homeowners.

county

flagged rate

reading

Clay

13.37%

Hernando

12.53%

Marion

11.62%

Charlotte

0.00% → 9.62%

address formatting, normalised

Santa Rosa

0.00% → 9.59%

address formatting, normalised

Levy

0.57% → 8.96%

ordinal formatting, normalised

Pasco

0.02% → 7.71%

address formatting, normalised

Flagler

2.00% → 6.95%

street-type spelling, normalised

Gilchrist

0.83% → 5.41%

ordinal formatting, normalised

Franklin

1.21%

unreadable — RED

Resolved 2026-07-31. Nine counties whose rate was suppressed by address convention are now read through jv_address_county. The screen had carried its own copy of the comparison, labelled "VERBATIM from build_checker_db" — it was not, and had not been since the checker learned per-county conventions. There is one implementation now and the screen calls it. The count moved 579,853 → 617,235 (+37,805): Pasco +16,629, Charlotte +10,462, Santa Rosa +6,854, Flagler +2,928, Levy +637, Gilchrist +144, Dixie +70, Hamilton +41, Liberty +40.

Gilchrist was previously listed here as "genuine composition." That reading was wrong. The county writes 4599 SW 90 CT where the mailing address says 4599 SW 90TH CT; normalising numeric ordinals took its homesteaded mail-equals-situs rate from 6.0% to 70.1%. The genuine composition in Gilchrist is the 12% of homesteaded owners who mail to a PO box — real, and an order of magnitude smaller than the formatting effect it was mistaken for.

617,235 is still a floor, and the gap has two different kinds of cause

These are not the same kind of uncertainty and collapsing them into one sentence hides which way the next revision moves.

Unmeasured — the screen cannot run. DeSoto and Franklin are classified RED: their homesteaded mail-equals-situs rates (34.8% and 20.6%) sit below the floor at which the comparison can be trusted, and the normalisation that explains the other nine does not explain them. Their true contribution is unknown, not low. Saying they are low would assert something about their homeowners; saying they are unreadable asserts something about their records, and only the second is supported. This gap cannot be sized from held data — it can only be closed by diagnosing those two counties.

Measured — real composition. The residual inside the counties that DO read correctly is genuine and quantified: owners who mail to a PO box (12% of homesteaded parcels in Gilchrist, 17% in Levy and Dixie, 21% in Liberty), genuine second-home and seasonal ownership, and records whose two address fields disagree on the directional — Hamilton writes mailing 7468 SE 191ST AVE against situs 7468 NE 191ST AVE. That last one is a roll data-quality problem, and it is deliberately not normalised away: a rule that joined those addresses would manufacture the false negatives this whole method exists to avoid.

So the predicted direction of the next revision is up, by an unknown amount from two counties, and not by much from anything else. The composition residual is not recoverable — it is what the method correctly declines to count.

Two further consequences:

County rates should not be ranked against each other as a market signal without normalising

the address comparison first.

The low end of the table is not what it looks like. Some genuinely low counties reflect real

second-home and seasonal ownership. Others are unreadable by this method, and the table now distinguishes the two.

Roughly 8% of counted parcels — 50,447 — carry a trust or estate in the owner name. We don't exclude them, because a revocable trust or a life estate can hold a homestead — but we can't tell from the roll whether a particular one does, which is why our checker shows those owners a caution rather than a conclusion.

---

What this number does not mean

It does not mean 617,235 people are entitled to an exemption.

It does not mean any of them will receive $451, or any amount.

It is not a count of eligible homeowners, of unclaimed benefits, or of money anyone is owed.

It is a count of parcels that look owner-occupied on the public record and carry no homestead exemption — and a reason for any homeowner to check their own record, which is free and takes a few seconds.

---

Our checker's coverage

The screen above covers all 67 Florida counties, and so does our address checker (67 counties, read from the roll it was built from). Comparable-sales analysis is narrower and is named on every page that offers it.