Sources and methods

Consumer spending, from Form 10

Source. Florida Department of Revenue Form 10, sales and use tax returns by county and kind of business (kind code), every year from 1994, plus the current preliminary month. Each figure has three measures: gross sales, taxable sales and tax collected. Gross sales lead on this site.

Month of sales. Form 10 prints the month a return was filed. Sales are reported the month after they happen, so each figure is shown in the month before its return month.

Kind codes. DOR renumbered and renamed its kind codes at the start of 2002. Returns filed through December 2001 use the older list and later returns the newer one; each group on a county page names the codes it takes from each list, and the full crosswalk is in the workbook's Kind codes sheet. The groups are visitor spending (hotels, restaurants, bars, admissions), resident spending (grocery, general merchandise, clothing, furniture) and big ticket spending (motor vehicles, boats, building materials). Every other business type is in a full table on each county page.

2002 and 2003. DOR's Form 10 file for those years is incomplete: most business types have no figures from December 2001 to November 2003. Those months are missing rather than zero.

Form 9. DOR also publishes county totals as Form 9. Every county month is compared with it. Where the two differ the difference is published in the workbook with its reason, and any month more than 5% apart is marked on the chart. Form 10's business types must add up to Form 10's own printed total to the penny, or the build fails.

Real sales. Deflated by the US Bureau of Labor Statistics consumer price index for all urban consumers: the Miami-Fort Lauderdale-West Palm Beach series for Miami-Dade, Broward and Palm Beach; Tampa-St. Petersburg-Clearwater for Hillsborough, Pinellas, Pasco and Hernando; the South Region for every other county. The metro series are published every other month; the months between are interpolated, and months before a metro series begins use the South Region. Each county's series is named on its page and in the workbook.

Tourist tax, converted to sales

Collections are tourist development tax dollars collected on short term rentals of six months or less, and on food and beverage where a county or city taxes it, as each source reports them. Miami Beach levies a city resort tax instead of the Miami-Dade County tourist taxes and has its own page.

Sales are collections divided by the tax rate in effect for the month of sales, from the rate table, which cites the ordinance or code section behind every rate. A rate change therefore does not show up as a change in sales; each change is marked on the charts. A series that blends levies with different rates carries no single rate and is shown as collections only.

Month of sales is the month the stay or meal happened. Where a source reports only the collection or distribution month, the sales month is derived with the lag given for that source and marked as derived.

Year over year change as reported compares collections with the same month a year earlier. Change in sales compares sales, which is the same as the rate adjusted change in collections.

HB 5601. Where a report groups small jurisdictions together for disclosure reasons, the group is shown as a combined area and its members are listed in the series name.

Miami-Dade. The county's monthly collections by municipality run from May 1988 to May 2023. Annual county totals for every levy come from the tourist tax appendix of each adopted budget through the latest closed fiscal year. June 2023 onward is a pending public records request. Form 10 hotels, restaurants and bars are the current monthly series.

Economic data

Labor force, employment and unemployment. US Bureau of Labor Statistics, Local Area Unemployment Statistics (LAUS): labor force, employment, unemployment and the unemployment rate for Florida, every county and every city, town or village of 25,000 or more, monthly, full history, read from the BLS Florida series file. LAUS is a federal and state program: the county and city estimates are produced by FloridaCommerce's Labor Market Statistics under BLS methodology and published by BLS, so both are cited on every county and city page. Counties and cities are published not seasonally adjusted only; the state is shown both ways. Counties and the statewide map, each county with its own page listing its cities; find a city or town.

Validation. Every figure BLS prints in its release tables for these areas, the county table for the latest 14 months and the state tables for every month since 1976, is compared with the parsed data at refresh and again on every build, and any difference fails the build. BLS publishes no monthly release table for cities, so the current year of every city series is read back from the BLS Public Data API and compared the same way. The latest month of every area is in one of those comparisons or the build fails. Each city is placed in its county by its Census place code (data/laus_city_counties.csv), and the build fails if a city has no county.

Blanks and annual figures. A month BLS prints as not available is left blank with BLS's footnote; a month with no row in the file is left blank and says so. Nothing is interpolated. Annual figures are the annual averages BLS publishes, carried with BLS's note where a year is averaged over fewer than twelve months (2025, with October not collected). A year BLS has not averaged is not listed. The last 12 months view charts the twelve months ending at the latest month against the same months a year earlier.

Consumer price index. Used only to restate sales in constant dollars; see Real sales above.

Business patterns

Establishments, employment and payroll. US Census Bureau, County Business Patterns and ZIP Code Business Patterns, from the Bureau's complete data files: establishments, employment in the week of March 12 and annual payroll by industry for Florida, every county from 1986 and every ZIP code from 1994, through 2023. NAICS sectors and subsectors from 1998 and two digit SIC before; ZIP codes carry establishments by sector and all industries employment and payroll. Counties and the statewide map; ZIP codes.

Validation. For every year and industry the 67 counties are added up and compared with the Bureau's published Florida row, at refresh and again on every build, and any difference fails the build. Establishments agree exactly through 2016; from 2017 the Bureau drops any county cell with fewer than three establishments, so the counties may fall short of the state by at most two establishments for each county without the code, and never exceed it. Employment and payroll agree exactly wherever every county publishes the code and none is withheld. ZIP detail is compared with the ZIP totals file the same way.

Withheld and noisy figures. Where the Bureau withholds employment and payroll it prints an employment size range, which is shown in place of the figure. From 2017 the Bureau adds noise to employment and payroll and flags each figure as low, medium or high noise; the flag is shown on each card. Establishment counts are never withheld or noised.

Economic Census

Sales by place. US Census Bureau, 2017 and 2022 Economic Censuses, geographic area series: sales (sales, shipments or revenue), establishments, firms, employees and annual payroll for retail trade (NAICS 44-45) and accommodation and food services (NAICS 72), sector to four digit industries, for Florida, every county and every place the Bureau publishes, 703 places in all. Read from the flat file package behind each data.census.gov table. Counties and places.

Validation. Every figure and flag is compared with the Bureau's API dataset file for the same census, the file behind the Census Data API, at refresh and again on every build; any difference fails the build.

Comparability. The 2022 NAICS regrouped retail trade below the sector, so retail subsectors and industries are shown for each census but not compared; the retail sector and everything in accommodation and food services compare directly. Withheld figures keep the Bureau's flag.

Population

Population estimates. University of Florida, Bureau of Economic and Business Research (BEBR), Florida Estimates of Population: the state's official April 1 population of Florida, every county, every municipality and each county's unincorporated area, as published in the edition for that year. Counties run from 1972 and municipalities from 1979. The census count stands in a census year. BEBR's revised intercensal estimates for the state and the counties (Special Population Reports 7 and 11, 2001 to 2009 and 2011 to 2019) are a second series. Counties and the statewide map; municipalities.

Editions. BEBR's own workbooks for 2011 to the latest edition are read directly, and a new edition is picked up from BEBR's Estimates and Projections page when it appears. Editions before 2011 are read from the Florida Legislature's Office of Economic and Demographic Research, which compiles every BEBR edition into one sheet per year and names the booklet each was taken from. Where both hold an edition, every county and municipal figure is compared and must agree; the 2011 and 2012 sheets of the compilation reprint a few municipalities with the Census Bureau's later corrections, marked as such, and the booklet figure stands.

Validation. Every statewide total BEBR printed (the counties must add to it) and every county total in each booklet (the municipalities and the unincorporated area must add to it) is checked to the person at refresh and again on every build from the committed copy of each printed total; any difference fails the build. Years with no printed statewide total online (1972 to 1978) carry county figures only and no state row. The 2011 edition prints Duval's first three municipalities against each other's figures; the pairing is corrected from the compilation of the same edition and noted on each card.

Names. One series per municipality under its current name: Lake Worth Beach (Lake Worth until 2019), Greenacres, Golf, Ocean Breeze and Islamorada, Village of Islands. A municipality in two counties (Fanning Springs, Flagler Beach, Longboat Key, Marineland) is shown whole on the municipalities page and as a part on each county's card. A municipality that was dissolved keeps its series to the last edition that printed it, as printed, including a zero where BEBR printed one.

Licensed businesses, from DBPR

Source. The Florida Department of Business and Professional Regulation posts its active licensure base as flat files: public food service and public lodging establishments (Division of Hotels and Restaurants, one file per district, refreshed weekly) and retail alcoholic beverage licenses (Division of Alcoholic Beverages and Tobacco, statewide). The lodging files list one row per rental unit, so a group or collective vacation rental license appears once per unit; rows are collapsed to one per license number before anything is counted. Counties and cities.

Active. A license is active when DBPR lists it in the file. Temporary event food vendors are in the file but not in DBPR's summary and are not counted. An alcohol license is counted at primary status 20, Current, the basis of the Division's own county table; other statuses are in the download.

Validation. On every refresh and every build the count of each license type is compared with DBPR's statewide summary (HR400A) and county summary (HR400C), both dated the first of the month. The extract is from a different day, so the build allows 1.5% or 15 licenses, whichever is more, and fails beyond that. Group and collective vacation rental licenses list units in more than one county and are checked statewide only. The retail alcohol file is compared with the Division's all licensees file exactly, series by series and county by county; DBPR's last alcohol count by county and series is dated July 1, 2021 and is carried in the published counts download for reference.

Cities. Every address was geocoded with the US Census Bureau batch geocoder (current benchmark) and placed in a Census incorporated place by point in polygon; a license outside every incorporated place is unincorporated. Where the geocoder could not place an address, the city name as filed is matched to a place name and the license is marked as placed by name. Each page states how many licenses sit on each basis. A multi-county license is placed in the county holding most of its units. DBPR publishes no GIS layer of its licenses.

Openings and closures. Food service and lodging openings are new licenses by the month DBPR approved them, from the new establishment files, initial licenses only; the active files carry no issue date. Alcohol openings are first issue dates of licenses still on the books, so the further back, the more closures have thinned them. A month inside a published file with no new license is zero; a month no file covers is blank, with the reason below. Closures: DBPR publishes no closure date (an expiration date in the active file is a renewal date), so a license is recorded as closed in the month it leaves the active file, from the first extract on.

Months with no published new license file

New business formations

Source. Florida Division of Corporations (Sunbiz) bulk corporate data files: the quarterly file of every entity on record, plus the daily files of records added or changed since it, read from the Division's public SFTP server. Each record is one corporation, limited liability company, limited partnership or trust with its filing type, principal address and file date. An entity is counted once, in the month of its file date, which is the day it was formed in Florida or, for an out of state entity, registered to do business here. A daily record replaces the quarterly record for the same document number. Florida, counties, cities and ZIP codes.

Validation. The Division publishes new filings by entity type and calendar year, 2005 to the latest full year. The count of records on file by file year and type must agree with every one of those figures within 2 percent or 10 filings, at refresh and again on every build, or the build fails; the comparison, pair by pair, is the Validation sheet of the workbook. Most pairs agree within 0.2 percent. The Division's table for its latest year is a snapshot taken early in the following year, before filings received in December and processed in January were numbered, so for that year only the comparison may instead use the records whose document number was issued within the year; the pages count every record. Five pairs are recorded exceptions with their reasons (declarations of trust in 2005, 2006, 2007 and 2010, and foreign limited partnerships in 2021: the Division's figures count documents the corporate data file does not hold, while every neighbouring year agrees within a few filings). General partnerships and limited liability partnerships are in a different bulk file and are not compared or counted.

Geography. The ZIP code of the principal address decides the county and the city: every Florida ZIP code is mapped to its county and to the US Postal Service city name for that ZIP (GeoNames postal code file, CC BY 4.0). A city is a postal city, not a municipal boundary. An entity whose ZIP code is outside Florida, or whose state or country says so, is counted statewide under "outside Florida"; one with no usable ZIP code is "not matched". County and city pages count Florida addresses only.

Months. Every month from January 1990 to the month of the last daily file is present, a month with no filings as zero, so trailing 12 month totals and annual sums are defined everywhere. The latest month is preliminary until the next quarterly file: a filing recorded a few days after its file date reaches the daily file late. Entities formed before 1990 and still on file, and filings dated after the last daily file (a delayed effective date), are counted on each page but not charted.

Economic data: bank deposits

Source. Federal Deposit Insurance Corporation, Summary of Deposits: every office of every FDIC insured institution in Florida with the deposits booked there as of June 30, for every survey from 1994 to 2026, read from the BankFind Suite API with the office's address, the coordinates the FDIC geocodes for it, its county, city and ZIP code, and its office type. The FDIC's bulk files are no longer posted, so the API is the source of record. Deposits are in thousands in the source and in dollars here.

Check against the FDIC state table. The FDIC's Summary Tables report, National by States and Territories, prints Florida's offices, institutions and deposits for each year; its State by County report prints the same for each county. Both are fetched for every year and kept with the data. The parser refuses to write a year whose offices do not add up to them exactly, and every build repeats the comparison on the committed files and fails on any difference. The Florida page prints the two side by side.

Where deposits book. Deposits book at the office where the account is domiciled, a choice each bank makes, so one office or ZIP code can carry a region's deposits. Trends and office counts are the reliable read; a single year's figure for one place is not. The site therefore leads with change over time and with office counts, and publishes every office so a large figure can be traced to the office that carries it.

Openings and closures. An office opened in a year when its FDIC office number is in the area that June 30 and was not the June before, and closed when the reverse holds. An office that moves or is renumbered after a merger counts as a closure and an opening. Through 2010 the FDIC file leaves the office number blank for some offices; those are followed by bank and branch number instead, which a merger resets.

Cities and ZIP codes. City is the city the bank reports for each office, with spellings folded together by rule, not a municipal boundary. The spellings, the city each became and the rule applied are in the statewide deposits workbook. ZIP codes are the five digit code the bank reports for the office. Real change deflates each June 30 figure by the county's consumer price index for that June, the same regional series the county's spending uses.

Building permits

Residential building permits. US Census Bureau Building Permits Survey: new privately owned housing units, buildings and construction value authorized by building permits, by structure size (one unit, two units, three and four units, five or more), for every Florida county and every permit issuing place. Monthly from January 1988 and the Bureau's annual table from 1980 (county table from 1990). Counties and the statewide map; every place.

Which monthly figure. From 2013 the Bureau posts a revised monthly record for every month of a closed year, and those are the figures here. The current year's months are revised through the latest release and replaced by the final records when the year closes. For 1988 to 2012 the Bureau posts no revised monthly records, so the months are as first published, before late reports and corrections, and every row says so. A month the Bureau imputed because a place did not report carries the Bureau's source code and is marked imputed; nothing is interpolated here. From 2005 to 2021 the Bureau published monthly figures only for the places in its monthly sample, so a place surveyed annually, and a county whose offices were all annual reporters, has no monthly figure for those years; from 2022 every place is surveyed monthly.

Annual figures and validation. The annual figures are the Bureau's own annual table, never a sum made here, because the Bureau builds a place's annual figure from its annual report or an annual imputation where the place did not report every month. Where a place has a revised record for all twelve months, every month reported, and the Bureau built the annual figure from reports, the twelve months must add up to the annual table exactly, at refresh and again on every build, or the build fails. The county table must equal the sum of its places for every county and year it exists, or the build fails. Every other difference between a place's months and its annual table is published with the Bureau's reason.

Places. A place is a permit office: a city, town or village, or a county office, which issues permits for the unincorporated area or, where no city issues its own, for the whole county. The Bureau has renamed some offices over 46 years of files; the registry download lists every name used and the years it appeared, and the card says so where it matters.

Fiscal data: local government revenues and expenditures

Source. Florida Office of Economic and Demographic Research, Expenditures and Revenues Reported by Florida's County and Municipal Governments, published by the Florida Legislature's Office of Economic and Demographic Research from the Annual Financial Reports every county, municipality and special district files with the Department of Financial Services under s. 218.32, F.S. One workbook of revenues and one of expenditures per government, a sheet per local fiscal year (October to September, named for the year it ends in), by Uniform Accounting System account code and fund type. Last retrieved 2026-10-01.

Coverage. County files run from FY2005 (expenditures) and FY2006 (revenues); municipal files from FY2007 and FY2008; the consolidated City of Jacksonville and Duval County government files as a municipality from FY2005, so Duval has no county file and EDR's statewide county totals exclude it. EDR's statewide account totals run from FY2000. A year EDR publishes no sheet for is a gap, listed on the government's page; nothing is interpolated. A local fiscal year ends September 30; the Annual Financial Report is due to the Department of Financial Services nine months later, and EDR tabulates the reports over the following year, so a fiscal year is usually published one to two years after it ends.

Custodial Fund. DFS added the Custodial Fund column to the report in FY2021 for assets a government holds in a purely custodial capacity. Totals from FY2021 on include it, so they are not directly comparable with earlier years; each line's custodial part is carried separately in the tables and the workbooks.

Checks. In every sheet, each line's total must equal the sum of its fund columns, each category heading the sum of its lines, and the sheet total the sum of every line, to the cent. For every fiscal year EDR has published a statewide file for, the sum of every government's figure for each account code must equal EDR's statewide total for that code, and each government's own total and every one of its account lines must equal what EDR prints for it in that file. Where EDR's own two publications disagree (an amount coded under one account in a government's workbook and another in the statewide file, or a figure EDR revised in one and not the other), the disagreement is published with its reason in the workbook's Statewide differences sheet and on the government's page, and counted against a baseline: any new one, and any disagreement not recorded, fails the refresh and the build. Real figures are deflated by the fiscal year average of each county's consumer price index series, the same series as the county's spending pages; per resident figures use the population EDR prints with each year.

Housing data

Source. Shimberg Center for Housing Studies, University of Florida, Florida Housing Data Clearinghouse: every table the Clearinghouse publishes in its affordability, comprehensive plan data, population and household projections, parcels and sales, market rent trackers, income and rent limits and Assisted Housing Inventory topics, for Florida, all 67 counties, 502 cities and each county's unincorporated area. Each results page is read as the Clearinghouse serves it, table by table, with the Center's titles, column names, income bands, source lines and notes kept as published. Retrieved 2026-10-01; the Center updates most datasets once a year on a cycle ending June 30 and the inventory several times a year. Counties and the statewide map; the Assisted Housing Inventory.

Validation. Every county's figures are compared with the Clearinghouse's own results page for that county alone, row by row (52,179 rows); the Center's household estimates and projections for cities and the unincorporated area are summed against the county (22,482 sums); the cost burden summaries are compared with the 2025 rows of the projection detail (19,620 figures); every development on a city's inventory page is compared with its county page (3,158 rows) and the counties are summed against the statewide inventory. Any difference blocks the refresh, and the build check repeats the county comparison from the committed county pages. The inventory's funder summary and its property list are two vintages of the inventory; where they differ the difference is published, not failed on.

Blanks. A value the Center suppresses (fewer than 25 observations), prints as a dash, or codes as missing in the tenant characteristics is left blank and listed with its reason. Nothing is interpolated or inferred. The Center's population projections for cities are not constrained to the county total and are shown as published.

Assisted Housing Inventory layer. Every development with a location is written to a GeoJSON point layer in the format BusinessFlare GIS pages load, using the Center's published coordinates, or the US Census Bureau geocoder from the published address where the Center publishes none; each feature says which.

BusinessFlare city-level consumer spending estimate

What it is. BusinessFlare estimate from public sources: monthly consumer spending for 609 Census places (cities, towns, villages and census designated places) for which the Economic Census publishes sales, in two sectors, retail trade (NAICS 44-45) and accommodation and food services (NAICS 72), with the site's visitor, resident and big ticket groups cut from them with the county's kind code mix. The figures are Street Economics Pass data, served by the account service and never written into this site; what is free is that an estimate exists, how it is made, and its error.

Method. The estimate anchors on the sales the US Census Bureau published for this place in the 2017 and 2022 Economic Census, for retail trade and for accommodation and food services, and moves them month by month with the county's sales tax returns by kind of business (Florida Department of Revenue Form 10). After the 2022 anchor the place's share of the county moves with the growth of its commercial property on the Department's 2026 tax roll, as submitted by the county property appraiser, by use code: stores, shopping centers, supermarkets and auto dealers for retail; restaurants, bars and hotels for accommodation and food. The three spending groups follow the county's kind code mix within each sector. Every place with both anchors was carried from 2017 to 2022 the same way and the landing on the 2022 Census figure is the published error.

Weights. The county's commercial inventory on the Department of Revenue NAL tax roll (2026 preliminary roll (the 2026 final where the Department withdrew the preliminary)), by Census place through the Census Bureau's block assignment file, by use code family. The roll moves a place's share rather than setting it: after the 2022 anchor the share is the anchor share times the roll's growth ratio (the place's share of inventory built by today over its share of inventory built by 2022) raised to a power the back-test chose, 0.5 on jv for retail and 0.5 on jv for accommodation and food, from just value, building area and parcel count at powers 0 to 1.5 (0 being the anchor carried as is). From the year after the roll year the ratio drifts by the place's net new construction relative to the county's.

Validation. Every place with both anchors is carried from 2017 to 2022 with the method and the landing on the 2022 Census figure is measured: statewide median absolute percent error 15.7% over 902 place sectors, published with every estimate along with the place's own landing. A landing error over 25% puts a wide error flag on the place (275 places). The back-test is rerun on every build and the build fails if the statewide error worsens; every figure on the free pages is recomputed; and the build fails if any estimated figure appears in the static site. The one place with a direct series, Miami Beach, whose resort tax converts to room and food and beverage sales, lands within 7% of the accommodation and food estimate in every year from 2017 to 2025 except 2020.

Not used. Half-cent sales tax distributions and the communications services tax are fiscal flows, not sales. County Business Patterns by ZIP code and Sunbiz filings are kept by postal city, which does not map to a Census place; DBPR licenses carry a place only inside incorporated cities, which would give every census designated place a zero weight.

LocalFlare

What it is. One page for every incorporated city, town and village, every census designated place the BusinessFlare estimate covers, and every county, built from one data file per place at /data/localflare/places/<place FIPS>.json and /data/localflare/counties/<county FIPS>.json, indexed at /data/localflare/index.json. Each file carries every figure on the page with its value, period, source, source URL and retrieval date, the series behind each sparkline, and the full tables. The page, its PDF report and its Excel workbook print the file and nothing else, and the build fails if a printed figure differs from the file or a file figure differs from what the section it came from recomputes.

Sources. The cards read the sections of this site: BEBR population, BLS LAUS (county figures for a place under 25,000, labelled), the BusinessFlare estimate (method, anchors and error; the figures are pass data), the tourist tax where a county publishes it by city, DBPR licenses placed by geocoded address, Sunbiz filings by postal city, ZIP Code Business Patterns for the ZIP codes whose Census internal point falls inside the place, the Census Building Permits Survey, the Shimberg Center Clearinghouse, the Department of Revenue's distributions ledger, EDR's local government finances and the FDIC Summary of Deposits. Three sources are LocalFlare's own: the Florida Department of Revenue NAL tax roll (US Census Bureau TIGER/Line 2024 block assignment through the Census block assignment file; the 2026 preliminary roll for every county, the 2026 final for Citrus, Duval, Gilchrist, Hernando, Orange, Sarasota, St. Lucie, Taylor; retrieved 2026-10-03), summed by place and Department land use group with the median single family just value and the qualified sales of the latest twelve complete months on the roll; BusinessFlare GIS, Amendment 3 Impact (amendment3impact.com, retrieved 2026-10-03) for Miami-Dade places; and the BusinessFlare GIS workspace project list (gis.businessflare.net, retrieved 2026-10-03), which says which places have a live map to embed. Boundaries are the US Census Bureau TIGER/Line 2024 place and county files, simplified for drawing.

Not shown. A card a place has no data for is left out and the reason is listed under its sources. The preliminary roll carries no prior year value, and the 2025 final roll is no longer on the Department's portal, so the year over year change in just value is blank until a roll carries it. EDR's revenue and expenditure files carry no debt figure.

Monthly, annual and trailing 12 months

Every chart can show monthly figures, calendar years, fiscal years (October to September, named for the year they end in: FY2025 is October 2024 to September 2025), or trailing 12 month totals. A year with any month missing is partial: it is listed in the tables and the workbook with the months it lacks, and left off the annual chart. A trailing 12 month total exists only where all twelve months are published. Charts of series longer than five years open on the annual view.

Checks. Every build recomputes each figure printed on the site, and every figure in the Excel downloads, from the committed data, and fails on any difference. It also fails if a series has a month missing that is not explained, if a source was not retrieved within its expected cadence, or if the number of Form 9 differences rises. The monthly refresh validates new months against the totals each report prints, and Form 10 against Form 9, before committing.