About this study

What it is for, what it takes no position on, exactly what was done, and what the figures can and cannot show.

Counties 66 with their own state filings (Duval files with the City of Jacksonville) Fiscal years FY2015–FY2025 (baseline FY2019; comparisons on FY2024) Published October 15, 2026

Why this exists

This is an exercise in analysis and transparency. It takes no position on how much counties should spend, on property tax policy, or on Amendment 3 on the November 2026 ballot.

Its premise is a plain one: the people who make important decisions about any organization — a business, or a county — make better ones with all of the information in front of them. Here those people are local officials and the voters who choose them.

So every figure comes from a public record, every calculation is described, and every page says what the numbers can and cannot show. Where the numbers raise a question, the page asks it and leaves the answer to the people who can give it.

Findings in both directions get the same treatment. A county whose property tax grew slower than its population and inflation is reported as plainly as one that grew faster, and a county that chose to go past its own limit is described with the reasons on record.

What was pulled

Four public sources, used as published and not edited by hand.

County revenues and expenditures, FY2006–FY2025 (expenditures from FY2005). The Florida Legislature's Office of Economic and Demographic Research publishes, for each county, one spreadsheet of revenues and one of expenditures, by account code and fund type, compiled from the Annual Financial Report every county must file with the Department of Financial Services under s. 218.32, Florida Statutes. These are the county's own figures for money actually received and spent — actuals, not adopted budgets. 66 counties file their own reports; Duval's are consolidated with the City of Jacksonville. 42 counties had filed FY2025 when the files were pulled; the rest run to FY2024. Each file also carries the countywide population estimate the state used that year. Source.
Prices. The Consumer Price Index for All Urban Consumers, All Items, South region (BLS series CUUR0300SA0), monthly, via FRED; averaged over each October-to-September fiscal year. Source.
Property tax rolls and rates, 2017–2025 tax rolls. The Florida Department of Revenue's annual County Profiles: each county's taxable value, the County Commission's countywide operating millage (prior year, rolled-back, majority-vote maximum and adopted), the vote required, and the taxes levied. 585 county-rolls were used; 9 whose printed figures were internally inconsistent were left out and are named on the county's page. Source.
A bounded search of public records for the largest changes. For the 25 largest flagged changes statewide, at most three searches each of the county's own documents and named news outlets. Each produced one sentence with a named source and a link, or "No public explanation located." Nothing in between. The Brevard feature page draws on the county's charter, budget documents, Charter Review Commission records, Florida House and Senate staff analyses and named news coverage, each linked where it is used.

What was calculated

Every figure on every page is produced by the same script from the sources above. The definitions:

Scope. Governmental funds only: general, special revenue, debt service, capital projects and permanent funds. Enterprise, internal service, fiduciary and component-unit columns are excluded. "Revenue" excludes transfers in, debt proceeds and special items (accounts 37x–39x). "Operating spending" is spending by function in the general, special revenue and permanent funds before transfers out. "County outlays" is all governmental-fund spending including transfers out, minus transfers received from other governmental funds and minus payments to refunded-bond escrow, so a county that funds its sheriff by transfer is counted the same as one that reports it by function.
Population. The countywide estimate published with each year's filing. Estimates were rebased to the 2020 Census beginning with FY2021 filings, so the FY2020-to-FY2021 step mixes real change with the rebase; for each county the real change that year is taken as the median of its other yearly changes in the period, and FY2020-and-earlier estimates are rescaled so that is the step. Each county's page states its two numbers.
The baseline. For any figure and any later year: the FY2019 figure × (population in that year ÷ population in FY2019) × (CPI in that year ÷ CPI in FY2019). FY2019, the year ended September 30, 2019, is the last full fiscal year before the pandemic. "Growth above baseline" is the county's actual growth minus that, in percentage points. "At the baseline" for a year is what the amount would be at exactly that growth; "above/below baseline" is the actual minus it; "cumulative above the baseline path" sums the yearly difference from FY2020 to the latest year. The statewide figure is the sum of each county's own cumulative figure, over the counties that report the series in every year. These describe arithmetic. They are not an estimate of what a county could or should have spent or collected.
Per resident. Each year's figure divided by that year's population. "In FY2019 dollars" divides by the CPI ratio as well.
Property tax rates. Operating levy = taxable value × adopted countywide operating rate. Baseline levy = the FY2019 operating levy grown by population and inflation. Baseline rate = baseline levy ÷ that year's taxable value. The rolled-back rate is the Department of Revenue's figure (s. 200.065, F.S.). The homeowner illustration applies the adopted, rolled-back and baseline rates to fixed taxable values of $150,000, $300,000 and $500,000; it is the county operating levy only, not a tax bill. The 2025-roll profiles print a rolled-back rate computed from the prior year's maximum majority-vote rate rather than from the adopted rate, so FY2026 is left out of the rolled-back count. Where the operating rate moved 15% or more in a year while the county's reported property tax revenue moved 20 or more points differently from the operating levy, rates are not compared across that year.
Debt. Debt-service payments are account 517, net of any refunding payoff run through it. New borrowing (account 384) and refunding bonds (385) are reported separately; an issue whose proceeds went to escrow or to a payoff in the same year is classed as refunding.
Net change in fund balance. Everything into the governmental funds (revenues, transfers in, bond proceeds) minus everything out (expenditures, transfers out, escrow), summed from FY2019 to the latest year and labelled with the number of years.
Account lines. Every line of $1 million or more in FY2019 or the latest year, or 1% or more of operating spending in any year shown, grouped under its function; revenue lines renumbered by the state in 2021 are shown as one family so the series is continuous. Where a line and another line in the same function moved opposite ways, the page reports the pair, because lines are often recategorized between years.
Flags and data notes. Rules, not judgment, decide what is marked: a category up 25%+ or down 30%+ in a year (and at least $1M or 2% of operating spending); a category 25+ points above baseline over FY2019–latest; $5M+ in debt proceeds; a jump that coincided with federal grants or disaster relief; a drop and a matching rise in the same year (possible recategorization); and data notes for a missing filing year, a series doubling or halving in a year, property tax absent while taxes are reported, a filing that collapsed to a few lines, a transfer-heavy year, or a discontinuous FY2019 population estimate. 125 data notes appear across the 66 pages; a county with a note touching FY2019, FY2023 or FY2024 carries a dagger wherever it is compared.

What was produced

66 county pages, one per county that files its own report, each built to the same plan: snapshot and baseline, revenue, property tax rates and the homeowner, spending by function and by account line, per-resident figures, net change in fund balance, debt, peers, observations, questions, data notes, what public records say, and sources. A section is left out only where its own rule says so (for example, no homeowner comparison where a county's operating rate changed structure), and the page says why.
One comparison page with every county on FY2024, the most recent year every county has filed: the statewide figure, rankings by growth above baseline, by dollars above the baseline path per resident and by the homeowner difference, sortable tables, key spending lines per resident, and the counties whose own charter sets a limit on the levy.
One feature page on Brevard County, the one county located whose charter caps growth in property tax revenue, built from the same figures plus the county's own records.
This page.

All text on the county and comparison pages is generated from a fixed set of sentence templates and a word list that bars characterizations ("waste", "excessive", "surge", "should" and the like). Observations state figures; questions are phrased so a county official could answer them. The pages name no individual official.

How it was checked

Every parsed file was cross-footed against its own totals (zero mismatches across 66 × 2 files).
Every figure on the county and comparison pages was recomputed from the source files by an independent, automated review that had not seen how the pages were built; its findings were fixed and the recomputation repeated, and the last pass found no differences in any table cell, tile, chart point or ranking. The feature page's figures were checked the same way and its sourced statements were checked against the research notes; the findings were fixed.
Every page passes the banned-word check and a structural check (no missing sections, no malformed figures).
Where a county's filing changed what it included from year to year, the page says so rather than treating the change as a trend.

What it cannot show

Why a figure changed. The records show the change; they do not show the reason. Hurricanes, state mandates, insurance costs, new facilities and new demands are all legitimate answers to the questions the pages ask, and the pages leave those answers to the counties.
What a county could or should have spent. The baseline is a yardstick, not a budget.
Adopted budgets (the figures are actuals; the Brevard feature quotes county budget documents only where it says so), reserves as a county reports them in its audited statements, or anything in enterprise funds (utilities, airports, ports).
Cities, school boards, independent special districts or the state. The property tax figures in the financial sections cover every county levy (operating, MSTUs, dependent districts and voted debt); the rate sections cover the countywide operating rate only, and each page quotes both figures so the difference is visible.
Duval County, whose government is consolidated with the City of Jacksonville and files as a city; and property tax for Dixie County, which is not reported in its FY2023 and FY2024 filings.

Who made it

Beachside Automation, with software-assisted analysis and human review. The pipeline — fetch the public files, parse them, compute the measures, write the pages, check the pages — runs end to end by script, so the study can be rebuilt as counties file new years. Feedback is welcome by email at [email protected]. If a county believes a figure misstates its filing, the filing is the reference and the page will be corrected to match it.