Other States
California
City and county financial data comes from the California State Controller's Office, which
collects a Financial Transactions Report from every city and county and publishes it as open
data. We hold cities and counties for fiscal years 2003 to 2025. California's fiscal year runs
July 1 to June 30; "2024" means the year ending June 30, 2024. San Francisco, a consolidated
city and county, files as a city. Transfers and other financing sources are left out of
revenue, as they are for New York. Each time we import the Controller's published data into
our database, we then compare our totals with the Controller's own published totals, to the
dollar, for every year the Controller has published them: through 2024 as of November 7, 2026,
so 2025 is not yet checked. School districts and special districts are not included: they
report to other agencies or in separate datasets.
Florida
City and county financial data comes from the Florida Department of Financial Services, which
receives an Annual Financial Report from every local government. We hold fiscal years 2013 to
2025. Most Florida local governments end their fiscal year on September 30; "2025" means the
year ending September 30, 2025. The Department lists every municipality as a city; we show each
as a city, town or village following the Census Bureau. Where the Department publishes an
account both as one figure and broken down by fund, we store the breakdown, and every import
reports any account whose figure differs from its breakdown's sum by a dollar or more. Every
import also compares each government's revenue and spending for the year with the
Department's own totals for it (its report of total revenues, expenditures and debt) and
names any government whose figures differ from them by a dollar or more. The
Department's file prints some figures twice (from fiscal year 2021 on, every impact fee); we
count a figure once however many times it is printed, and an import stops if two printings of
one figure disagree (until November 10, 2026, each printing was added; see Corrections).
A Florida government's figures count the funds it runs for itself. The Department's file
reports every account by fund group, twelve in all. We count eight (general, special revenue,
debt service, capital projects, permanent, enterprise, internal service and component units)
and leave out the four that hold money for others: custodial funds, such as the taxes a county
collects for its school district and cities, and pension, investment and private-purpose trust
funds. The Department's own totals for fiscal year 2024 leave out the same four. The rule
covers revenue, spending and balance-sheet figures in every year we hold, and an import stops
if a figure in the file is filed under a fund group outside the twelve (until November 10,
2026, every fund was added; see Corrections).
Under the heading "Other Sources" the Department files several kinds of money. We count the
interest and grants that a government's enterprise and internal service funds receive, and
chart them under the Department's heading for them, "Proprietary Non-Operating Sources". We
leave out the rest: transfers between a government's own funds and from a county's elected
officers, contributions from its enterprise operations, debt, lease and refunding proceeds,
sales of capital assets and payments for their loss, capital contributions, other
non-operating sources, and extraordinary and special items (until November 14, 2026, the
interest and grants were left out too; see Corrections).
An import also stops if a figure of a government it reads is filed under an account that is
not on our list of the Department's accounts (until November 11, 2026, such a figure was left
out without notice; see Corrections). It reads each government we hold that is in the
Department's main file for the year; from fiscal year 2021 on, six accounts come from a
second request, and a government found only there is reported and none of its figures is
read.
School districts and special districts are not included.
North Carolina
City, town, village and county financial data comes from the Annual Financial Information
Report collected by the North Carolina Department of State Treasurer. We hold fiscal years 2012
to 2025; the fiscal year ends June 30. Census-designated places are unincorporated and file no
report, so they carry no figures of their own; a community whose town has dissolved keeps the
years that town filed. The Treasurer publishes spending two ways, by function
and by type of expense. We add up the by-function figures and keep the other breakdown for
reference, so no dollar is counted twice. Every import ends with an error if our lines and the
Treasurer's own totals differ by more than $1. School districts and special districts are not
included.
Property Tax Burden
Property-tax comparisons answer two related but different questions: what a household
should expect to pay at a specific address, and how much a government collects from
the property-tax base serving that address. We keep those measures separate.
Resident-Facing Estimates
The preferred resident-facing presentation is an estimated annual residential bill
for the user's scenario, broken out by levy authority. We also show an
effective tax rate where the required inputs exist: estimated net bill
divided by the property's market value. This is more comparable across jurisdictions
than a raw statutory rate because assessment ratios, phase-ins, exemptions, tax classes,
credits, caps, and special districts vary by state and locality.
Every estimate should disclose its assumptions: market value, assessed value, taxable
value, class, credits or caps applied, and whether parcel-specific districts or fees are
included. We also show source freshness for each component, such as current, stale
latest-available, non-ACFR current actual, missing, or parcel-required.
When we label a row modeled pre-credit, the number is not a
parcel-exact bill and should not be read as a net household liability. For the
Metro-DC Maryland pilot, we combine the official State of Maryland, county, and
municipal rates published by SDAT for July 1, 2025 tax bills, then multiply that
combined rate by SDAT's county residential weighted assessment ratio. Homestead
caps, exemptions, credits, taxable-value phase-ins, and parcel-specific districts
are disclosed but not applied until we have parcel or scenario inputs that support
them. DC and Northern Virginia modeled rows are shown on an assessed-value basis:
DC uses the OTR residential class rate, and NOVA rows combine the applicable
county, independent-city, and incorporated-town rates where the official source
exposes them. Special districts, service charges, relief programs, credits,
exemptions, and parcel-specific assessment facts remain excluded unless separately
modeled.
Aggregate Civic Burden
Aggregate civic burden uses audited or official property-tax revenue and levy rows,
divided by population or household denominators, to describe what a government collects
from the tax base. This is useful fiscal context, but it does not predict what a
particular homeowner will owe. Aggregate revenue can include commercial, industrial,
rental, personal-property, utility, penalty, and interest amounts depending on the
source's taxonomy.
When aggregate rows are shown, we identify the levy authority, reporting or billing
authority, denominator geography, fiscal year, and selected source row. We do not
collapse state, county, municipal, school, or special-district rows into one authority
unless the source explicitly defines that total.
Source Priority
-
Parcel exact — actual parcel assessment, tax bill, or official tax
record, including exemptions and special districts.
-
Modeled effective rate — official current rates plus official
assessment, taxable-value, credit, and cap rules for a representative scenario.
-
Assessment-rate estimate — official rates and assessment semantics
when the full credit or cap model is not available.
-
Aggregate civic burden — audited or official property-tax revenue or
levy rows by authority and denominator geography.
-
Typical reported burden — ACS or similar household survey metrics,
clearly labeled as reported historical context rather than the user's bill.
Source Notes
Raw rates are source facts, not comparable burden measures by themselves. For example,
Maryland property-tax bills are based on assessment multiplied by state, county, and
municipal rates, with reassessment phase-in and credits such as the Homestead credit
affecting the amount billed. Other states use different assessment ratios, caps, and
local taxing structures.
Our cross-jurisdiction property-tax presentation follows the same practical lesson used
in property-tax comparison research: compare net bills and effective rates where possible,
and keep valuation assumptions visible because assessment accuracy affects the distribution
of tax burdens.
Corrections
When we find an error in a published figure, we fix it and list it here.
November 21, 2026 — Tax calculator: earned income credits counted a child of any age
From August 27, 2026, when the tax calculator began asking each child's age, until November 21,
2026, it counted every child toward the earned income credit whatever age was entered. The
federal credit counts a child under 19 at the end of the year, with exceptions for some
full-time students under 24 and for a permanently and totally disabled child (the IRS's
qualifying-child rules), so a household that entered a child aged
19 or older could be shown a larger earned income credit than the age test gives, and larger
state, county or city credits built on it. The amount did not always change: it stayed the same
where the household's other children already reached the credit's largest band, or at incomes
where the credit is not paid. This touched the calculator and its income chart for places whose
state, county or city has a credit built on the federal one. In the calculator's estimate for a
single filer earning $30,000 with one child aged 20, Washington, D.C. showed a refund of
$2,565.33 where the tax is $620.18, and Rockville, Maryland a refund of $734.88 where the tax is
$1,557.38. Credits that do not depend on the earned income credit, and exemptions for
dependents, were not affected. We found the error on November 21, 2026, in a check of the
calculator.
November 14, 2026 — Florida cities and counties: enterprise funds' interest and grants were left out of revenue
From April 3, 2026, when Florida's local finances first reached our site, until November 14,
2026, our Florida revenue figures left out the interest and grants that a government's
enterprise and internal service funds receive (an enterprise fund runs a service such as a
utility).
The Department files these four accounts under its heading "Other Sources", and our readers
left that heading out whole, because in the New York reports we hold it is the heading for
transfers between a government's own funds. On 18 of our 476 Florida pages (14 cities, 3
towns and 1 county) the "Top Revenue Sources" list left out a category that belongs among
its five, and on 2 of those (Umatilla and Wewahitchka) it is the largest. Arcadia's page left
out $8,530,692 of these interest and grants for fiscal year 2025, its second-largest source;
Pensacola's and Broward County's left out $57,143,034 and $84,251,167 for that year. The
pages print no revenue total, and no other figure on them changes.
Our MCP server, which reads the same rows, gave Florida totals that were too low for as long
as it has served Florida: its total revenue, and revenue per resident, for 1,333 of the 6,015
Florida governments and years we hold (821 for cities, 186 for towns, 34 for villages and 292
for counties), by $6.46 billion over fiscal years 2013 to 2025. 180 of those left out a
twentieth or more of the government's revenue, and 31 a quarter or more; the most was the
Town of White Springs, fiscal year 2021, with 65.5% left out. Arcadia's revenue per resident
for fiscal year 2023 was given as $2,468, where it is $3,225.
We had known since October 30, 2026 that these accounts were left out, and from November 1,
2026 our MCP server's Florida revenue answers carried a note saying so. We fixed it by filing
the four accounts under the Department's own heading for them, which our readers count. The
two accounts left under "Other Sources", transfers to a county's commission from its tax
collector and its supervisor of elections, stay left out, and a check now fails if any other
Florida account we store is filed there.
November 14, 2026 — New York compliance page and place pages: annual-report filing and Fiscal Stress Monitoring scores were overstated
From February 22, 2026 until November 14, 2026, our New York compliance page said that
counties, towns, villages and school districts had all filed every report, and from
March 2, 2026 its overview table printed 100% in the annual-report column for each of them.
The figure was typed into the page, not counted. For fiscal year 2025, the Comptroller holds
reports from 56 of 57 counties, 833 of 933 towns, 468 of 531 villages and 689 of 689 school
districts, so the page overstated filing for counties, towns and villages. From February 20, 2026, its city
counts included New York City, which files no report with the Comptroller, as a filer: for fiscal
year 2025 the page said 52 of 62 cities had filed, where 51 of 61 had; the percentage it printed,
84%, rounds the same. We found the errors on November 14, 2026, while checking every page's wording
against our data; the column now counts each kind's reports, cities without New York City, and the
sentences are built from those counts.
The same page's Fiscal Stress Monitoring System column, from March 2, 2026, counted a government
as scored when the Comptroller's workbook listed it at all, including the governments it lists
as "Not Filed" because it could not score them, and it counted New York City as scored. For
fiscal year 2025 it showed 925 of 933 towns, 527 of 532 villages and 62 of 62 cities scored, where
the Comptroller scored 819 towns, 429 villages and 50 of 61 cities (New York City is not in the
system). Its counties' figure, 57 of 62, counted New York City's five borough counties, which the
system does not cover, as never scored: 56 of 57 counties were scored. Its headline, "703 have
never been scored", counted Census-designated places, which the system does not score; the
governments it lists add up to 29. The same error reached place pages from March 1, 2026: at the
fiscal year 2025 scores, it left the note that a place is missing from the latest scores off the
pages of 218 governments (106 towns, 98 villages, 11 cities, a county and two school districts). Every count now reads the Comptroller's scores alone.
November 13, 2026 — New York City: debt service through the General Debt Service Fund was read from the wrong row
From February 20, 2026, when our New York City import was released, until November 13, 2026,
the figure we stored for New York City's debt service through its General Debt Service Fund
came from the wrong row of the Comptroller's table, the Hudson Yards debt service row
printed below it: $58,655,000, $70,546,000 and $84,332,000 for fiscal years 2016 to 2018 and
$0 for every later year we held, where the Comptroller reports between $3,377,653,000 and
$5,137,881,000 a year ($3,851,631,000 for 2026). Until October 30, 2026, our sums of New
York City's spending and of its debt service included this row (the same fault, fixed that
day, counted many New York City figures twice), so those dollar sums were too high by the
three amounts above for fiscal years 2016 to 2018 and unchanged for later years, and the
figures built on them, per-resident spending and the debt service share of spending, moved
with them. Since then no page has shown it and no total or chart has counted it; our MCP
server returned it, ranked and per resident too, when the figure was asked for by name. We
found the error on November 13, 2026, when we began comparing our New York City figures with
the Comptroller's second printing of them (see Cross-Checking Against Source Totals); each
year now reads the General Debt Service Fund's own row, and 2016's, which the report we
import no longer prints, comes from the Comptroller's 2025 report.
November 12, 2026 — Three New York villages that changed their fiscal year: one year mixed two reports
From March 8 or 9, 2026, when town and village account-level figures first reached our
site, until November 12, 2026, the pages for three New York villages showed revenue and
spending for one year that mixed two of the village's reports. Each village moved its year
end to December 31 and filed, under the same year, its 12-month report and a separate
report for the seven months that followed. Our import kept one figure per account for the
year, whichever came later in the Comptroller's file, so the year's totals matched neither
report and changed when the Comptroller's file listed its rows in another order. The
Village of Kiryas Joel's page showed fiscal year 2021 revenue of $17,861,308 ($19,888,003
from the Comptroller's March 2026 file) and spending of $33,174,826, where its 12-month
report has $20,708,112 and $28,588,738. The Village of Schoharie's showed $1,491,392 and
$1,310,446 for fiscal year 2020, where its 12-month report has $1,415,365 and $1,810,532;
the Village of Middleburgh's showed $1,288,203 and $1,658,252 for fiscal year 2021, where
its report has $1,248,733 and $1,436,584. The per-resident figures for those years were off
by the same shares. Until November 6, 2026 these figures were also counted twice (see that
correction). Our MCP server, which reads the same rows, gave the same figures. No other
year or government is affected: in the Comptroller's files we hold, fiscal years 1995 to
2025, no other town, village, city, county, school district or fire district files two
reports under one year.
We found the error on November 5, 2026, while comparing our totals with the Comptroller's
own. Each year now shows the village's 12-month report, and each page's Data Notes give
the seven-month report's dates and totals, which are in no year shown. The Comptroller's
summary file shows the seven-month report for that year instead, so for these three years
our figures and the summary's differ. An import now stops if a government files two
reports under one year and it cannot tell which one is the 12-month report.
November 11, 2026 — Florida cities and counties: other general taxes were left out of fiscal years 2013 to 2020
From April 3, 2026, when Florida's local finances first reached our site, until November 11,
2026, the general government taxes charted on 100 of our 476 Florida pages (57 cities, 18
towns, 1 village and 24 counties) left out one of the Department's accounts, other general
taxes, in fiscal years 2013 through 2020. 409 yearly figures on those charts were too low or
missing: 402 rise with the fix, and 7, on 3 pages, appear where the chart had no figure for
the year. Miami Beach's page showed $219,540,282 of general government taxes for fiscal year
2019, where the figure is $393,019,127; Okeechobee County's showed $24,565,274 for that
year, where it is $32,510,830. On 14 pages a year's figure rises by more than a tenth, and
on 5 of those by more than a quarter. No figure for fiscal years 2021 through 2025 changes,
so the latest-year figure printed above each chart and the spending per resident at the top
of each page stay as they were, and no page gains or loses a charted category.
Across the state the figures left out came to between $167.6 million (fiscal year 2013) and
$286.1 million (fiscal year 2019) a year, and to $1.83 billion over the eight years: $1.65
billion for cities, towns and villages and $182 million for counties. Our MCP server, which
reads the same rows, gave the same figures.
Until fiscal year 2020 the Department's file reports other general taxes under one account,
numbered 319; from fiscal year 2021 it reports them under four (319.100, 319.200, 319.300
and 319.900). Our list of the Department's accounts was built from its 2025 list, which has
only the four, and our import passed over a figure filed under an account that was not on
our list, without reporting it. We found the error on November 10, 2026, by comparing each
government's revenue with the Department's own total for it: the comparison named 409
governments and years, all in fiscal years 2013 through 2020, whose revenue fell short by
this account's figure. We fixed it by adding the account as a line of its own, in the same
category as the four that replaced it, and reading the Department's files for fiscal years
2013 through 2020 again. Of the 3,783 governments and years in those files that we hold,
the revenue now differs from the Department's total by a dollar or more for one (Dade City,
fiscal year 2020, by $748). An import now stops if a figure of a government it reads is
filed under an account that is not on our list.
November 10, 2026 — Florida cities and counties: figures counted custodial and trust funds
From April 3, 2026, when Florida's local finances first reached our site, until November 10,
2026, our Florida figures added up every fund in the Department's file, including four kinds
that hold money for others: custodial funds, pension trust funds, investment trust funds and
private-purpose trust funds. A custodial fund holds money a government collects and passes
on: a county's tax collector collects property taxes for the school district and the cities,
and that money was counted as the county's own revenue and again as its spending when it was
paid over. A pension trust fund holds employees' retirement money: its contributions and
investment results were counted as the government's revenue, and its benefit payments as
spending. 340 of our 476 Florida pages change with the fix (215 cities, 48 towns, 11
villages and all 66 counties we hold): 7,345 yearly figures on their charts, in fiscal years
2013 through 2025, the latest-year figure printed above 714 of those charts, and the
spending per resident at the top of 265 pages, which falls on every one of them, by more
than a tenth on 111 and by more than half on 22. Alachua County's page showed $4,960 of
spending per resident for fiscal year 2025, where the figure is $2,049; Jacksonville's
showed $10,118 for fiscal year 2024, where it is $5,676. Miami-Dade County's page showed
$11,614,767,000 of general government taxes for fiscal year 2023, where the figure is
$4,144,557,000. On 10 pages a different set of categories is now charted, and the ranking of
Florida cities by spending per resident on our Florida page changed. 442 of the 7,345
changed yearly figures were too low, not too high: 239 are in fiscal year 2020, where impact
fees were missing from our copy (below), and 189 in fiscal year 2022, when a fund left out
could hold a loss, as many pension funds did: Pembroke Pines's page showed miscellaneous
revenues of −$8,989,736 for that year, where the figure is $49,221,466.
Across the state, counties' revenue for fiscal year 2024 falls from $111.8 billion to $59.7
billion and their spending from $103.1 billion to $53.6 billion; the revenue of cities,
towns and villages falls from $58.6 billion to $42.5 billion and their spending from $45.8
billion to $37.9 billion. For counties nearly all of it begins in fiscal year 2021, the
first year the Department's file reports custodial funds: for fiscal years 2013 through 2019
the fix moves counties' revenue and spending by less than half of one percent. Our MCP
server, which reads the same rows, gave the same figures, and its balance-sheet lines
counted the same funds: for cities, towns and villages in fiscal year 2024 they fall from
$147.0 billion to $91.3 billion.
The Department's file reports every account by fund group, twelve in all, and our import
added all twelve. The Department's own Total Revenue and Total Expenditures, in its report
for fiscal year 2024, leave these four out. We found the error on November 10, 2026, while
preparing to compare our Florida totals with the Department's, and fixed it by counting only
the eight fund groups a government runs for itself, in every year we hold, for revenue,
spending and balance-sheet figures, and by removing the figures that existed only in the
four left out. To do it we read the Department's files for fiscal years 2013 through 2020
again. That added the impact fees our copy of fiscal year 2020 lacked ($1,184,816,131 for
244 governments, where we had held $3,536,605 for 3), and it showed that the impact fees of
fiscal years 2013 through 2019 had been counted once, as we believed.
November 10, 2026 — Florida cities and counties: impact fees were counted twice
From November 4, 2026 until November 10, 2026, the "Permits, Fees, And Special Assessments"
figure on 263 Florida pages (180 cities, 38 towns, 8 villages and 37 counties) counted every
impact fee twice, in fiscal years 2021 through 2025. An impact fee is a one-time charge on new
construction that pays for the roads, parks, schools and utilities the growth needs. The chart
of that category showed 1,139 yearly figures that were wrong, and on 235 of the pages the
latest-year figure printed above the chart was one of them. On 168 pages more than a tenth of
at least one year's figure was the second copy, and on 93 more than a quarter. Palm Coast's
page showed $118,929,731 for fiscal year 2025, where the figure is $60,323,360; Osceola
County's showed $398,892,281 for fiscal year 2021, where it is $246,476,511. All but one of
the 1,139 were too high: Stuart reported a negative impact fee for fiscal year 2023, so its
page showed $2,653,514 where the figure is $3,733,042. No page gains or loses a charted
category. Across Florida's cities and counties the second copy came to between $1.65 billion
and $1.91 billion a year. Our MCP server, which reads the same rows, answered with the same
doubled fees, in its category figures and in its revenue totals for these governments.
The Department's file prints each impact fee in two of its sections, under one identifier, and
our import added up every printing. Spending and the other revenue categories were not
affected. The Department's own Total Revenue for each government, which it publishes in a
separate report, is not added up the way our figures are: it leaves out the custodial and
trust funds, and it counts receipts we leave out of revenue, such as money a government moves
between its own funds and money it borrows. The lines in the Department's file, added up the
Department's way, equal that total to the dollar with each fee counted once for 118 of the 232
governments with an impact fee that are in both reports for fiscal year 2024, and with each
fee counted twice for none of them. The City of Lake Alfred's audited financial statements for
fiscal year 2024 report $5,588,976 of impact fees in its Impact Fee Fund, which equals one
copy to the dollar.
Fiscal years 2013 through 2019 were loaded from an earlier form of the file; their impact fees
($1.23 billion statewide in fiscal year 2019) run level with the corrected later years ($1.73
billion in fiscal year 2021), so we believe they were counted once. We found the error on
November 10, 2026, while preparing to compare our Florida totals with the Department's, and
fixed it by counting a figure once however many times the file prints it.
November 8, 2026 — North Carolina: two unincorporated communities showed a same-named town's or village's finances
From April 5, 2026, when we first showed North Carolina's local finances, until November 8, 2026,
the page for Fairview, an unincorporated community in Buncombe County, showed the Annual Financial
Information Report figures of the Town of Fairview in Union County, fiscal years 2012 through 2025,
and the page for Flat Rock, an unincorporated community in Surry County, showed those of the
Village of Flat Rock in Henderson County, fiscal years 2012 through 2025. Each page's per-resident
spending divided one place's spending by another place's residents: Fairview's page showed $325
for fiscal year 2025, where the town's own figure is $206; Flat Rock's page showed $2,714 for
fiscal year 2024, where the village's own figure is $816. The town's and the village's own pages
showed no finances at all, and our MCP server, which reads the same rows, answered the same way.
Our import matched each filer's name to our list of places by its bare name across every kind of
place, and a Census-designated place with the same name was found before the town or village,
which we store as "Town of Fairview (Union County)". The figures themselves were right; only the
page they sat on was wrong, for two of the 550 filers. The two communities' own figures, their
Census population, were right. We found the error on November 7, 2026, in a review of this page's
wording against the code, and fixed it by matching filers only to cities, towns and villages, and
by moving the two filers' figures to their own pages. A third community, Centerville in Franklin
County, holds the figures of the former Town of Centerville for fiscal years 2012 through 2016;
the town dissolved its charter on June 22, 2017 (Session Law 2017-43), and the Census Bureau now
lists the area as a Census-designated place, so those figures are the place's own and stay.
November 8, 2026 — New York cities and counties: category charts and debt service counted the trust and custodial fund
From January 27, 2026, when we first showed them, until November 8, 2026, the figures for a
category's revenue or spending in a year, which the charts on a place's page and our MCP
server's category answers are built from, counted the trust and custodial fund for 104 New
York cities and counties (51 cities, 53 counties): money a government holds or collects for
others, which we leave out of every total. The totals and per-resident figures on the same
pages were right; the category figures were not. A page charts its five largest categories,
chosen without the fund, so not every inflated figure reached a chart: today 99 of the 104
pages (48 cities, 51 counties) chart at least one. The pages' Debt Service line and
"Debt Service %" figure, and the county comparison charts' debt service, read the same way
(Saratoga County, 2002 through 2013, by at most $6,303 of $1,483,153). Our MCP server's
category and debt-service answers counted the fund from February 19, 2026. The "aid dependent"
share that colors the dots on the stress analysis chart and fills a column of each place's
stress trajectory page divided aid by revenue that counted the fund, so it read too low: in
fiscal year 2024, 59 of 115 city and county shares change, Glen Cove's from 10.0% to 22.8%
and New Rochelle's from 7.9% to 18.1%. Fiscal years 1995 through 2025 are affected.
In fiscal year 2024 the error touched 171 category figures, 90 of them on a chart, and 88 by
more than a tenth, 54 of those on a chart: New Rochelle's "Other Local Revenues" chart showed
$289,050,253 where the city's own revenue is $15,364,018; Glen Cove's showed $93,612,865 where
it is $3,916,427, 24 times too high, and Glen Cove's "General Government" spending $99,038,103
where it is $9,341,665.
Totals, rankings and per-capita figures were right, as were towns, villages, school districts
and every other state, whose data holds no such fund. Which five categories a page charts was
chosen without the fund, so no chart gains or loses a category. We found the error on
November 7, 2026, in a review of this page's wording against the code, and fixed it by giving
every sum one shared rule for which rows add up.
November 7, 2026 — New York towns that share a village with another town: Part-Town Per Capita Spending
From March 9, 2026, when we first showed it, until November 7, 2026, the pages for 50 New York
towns that share a village with another town showed a "Part-Town Per Capita Spending" figure and
its chart, next to a note saying the figure was not yet available. The figure divides the
spending that serves only the area outside villages by the town's residents outside villages.
For these towns we do not know how many of the shared village's residents live in each town, so
the figure was meant to be withheld: our code held a list of the 53 towns by short name, but every
town's name also carries its county, so the list matched none of them. In the town the Comptroller
files the shared village under, the whole village was subtracted from the residents, so the figure
was too high; in the other towns none of it was, so the figure was too low. The error can be as
large as the shared village's population. The Town of Mamaroneck showed $4,872 for 2025, dividing
by 4,843 residents after the whole Village of Mamaroneck was subtracted, part of which lies in the
Town of Rye; the Town of Clay showed $232 and the Town of Cicero $423, each with the Village of
North Syracuse counted wholly in Clay.
The Town-Wide figure, which divides by all of a town's residents, was right, as were every other
town's figures, the villages' own pages and the comparability notes. We found the error on
November 7, 2026, while checking this page's new wording against the code.
November 6, 2026 — New York towns and villages: per-resident answers from our MCP server
From February 2026, when we first loaded town and village data, until November 6, 2026, our MCP
server, which AI assistants use to read this data, could divide a New York town's or village's
spending or revenue by either of two population figures for the same year: the Census Bureau's
estimate, or the Comptroller's figure, which repeats the 2010 or 2020 census count. Which one it
used depended on the order the database returned them. The two figures are a median of 1.6% apart
for towns and 4.7% for villages, and more than 20% apart in about 7% of years, so an answer could
be off by that much: for the Village of Scarsdale in 2024 it answered $4,947 per resident instead
of $5,014.
Our pages used the Census figure and were not affected, nor were cities, counties, school districts
or other states, which have only the Census figure. One chart year on our pages changes with the new
rule: Village of Edwards, 2013, from $684 to $834 per resident. We found the error on November 5,
2026, in a review of our code.
November 6, 2026 — Utah income tax rate
From April 19, 2026, when we published it, until November 6, 2026, the Tax Snapshot table on our
Utah state page and on the pages for places in Utah said "Utah imposes a flat individual income
tax rate of 4.5% under current statute." Utah had already lowered the rate to 4.45% for tax years
starting on or after January 1, 2026 (S.B. 60 of the 2026 General Session, signed March 23, 2026
and in effect from May 6, 2026), so the stated rate was 0.05 percentage points too high for 2026:
$5 on every $10,000 of taxable income.
The tax calculator does not yet compute Utah's income tax, so no calculated figure used the old
rate. We found the error on November 6, 2026, while reading Utah's current code for our
retirement-income research.
November 6, 2026 — New York towns and villages
From March 8 or 9, 2026 until November 6, 2026, town and village pages added together two
copies of the same Comptroller data for fiscal years 1997 to 2024, so their revenue and
spending figures were too high. Statewide in fiscal year 2023, spending read about 55% too
high for towns and 59% for villages, and revenue about 65% and 67%. Property tax figures
read about 1.65 times the correct amount for towns and nearly double for villages. Fiscal
year 2025 held only four funds and read too low.
The per-resident trend cards were also wrong for 2011 to 2015 and from 2020 on, because
they divided by two population figures added together. A town's cards showed about half
the correct amount (Town of Bethlehem, 2020: $368 shown, $736 correct). On a village's
card the dollars and the residents were both too high, which partly cancelled (Village of
Scarsdale, 2024: $4,233 shown, $5,014 correct).
About 1,490 pages were affected. Our MCP server, which AI assistants use to read this
data, gave the same too-high revenue and spending answers for towns and villages. Cities,
counties, school districts, other states and the tax calculator were not affected. We
found the error on November 5, 2026, by comparing our totals with the Comptroller's own.
Town and village totals are now added from one copy, across every fund except
Trust & Custodial. When we made the fix we compared them with the Comptroller's
summary totals for every town and village year where we hold both, 81,685 comparisons of
revenue and spending. One summary column was left out: a 2011 "Local Revenues" subtotal
that the summary lists beside the lines it adds up. 93% matched to the dollar. About 5%
were within $10, most of them in 2000 to 2008, when the summary is in whole dollars. About
2%, almost all in 2010 to 2019, matched to within $10 once trust funds were added, which
the summary then counted and our totals leave out. The remaining 97 (about 0.1%) are not
yet explained.