Problems in the source data, and what GISBanker does about each.
Affects: American Community Survey (5-year) (2019–2021); County Business Patterns (2017–2021); Bureau of Labor Statistics (2018–2023); Regional Economic Accounts (2018–2023); Community Reinvestment Act Data (2018–2023); Home Mortgage Disclosure Act Data (all years)
In 2022 Connecticut's 8 counties were replaced by 9 planning regions with different borders. Sources kept reporting the old counties for a while: ACS before its 2022 release, County Business Patterns before 2022, BLS and BEA for 2018–2023, FFIEC CRA through 2023, and HMDA for every year (its API doesn't accept planning regions).
What GISBanker does: Figures published for the old counties are not converted, because the borders don't line up; those years show as unavailable for Connecticut counties. Connecticut tracts are not affected (see the next item), and state totals are.
Affects: American Community Survey (5-year) (2020–2021); Community Reinvestment Act Data (2022–2023)
Before the switch, Connecticut tracts carried their old county's code. Only that part of the code changed: every tract number appears before and after, so the tracts themselves are the same. ACS estimates for Connecticut tracts still shift more than other states' across the switch, likely because the survey's population controls moved from counties to planning regions.
What GISBanker does: Earlier releases are matched to today's tracts by tract number (ACS 2020–2021, CRA 2022–2023). Five water tracts with no population can't be matched and are left out. Treat Connecticut tract changes that span 2022 with extra caution.
Affects: American Community Survey (5-year) (2020–2022); Community Reinvestment Act Data (2022–2024); National Risk Index (all years); Rural-Urban Commuting Area Codes (all years); Opportunity Zones (all years); CDFI Fund Investment Areas (all years)
The Census Bureau revised Suffolk County's tracts in the 2023 ACS release, from 385 tracts to 371: 14 tracts (for example 1224.06 and 1460.01) were folded into neighbouring tracts. GISBanker's tract list follows the current ACS map, but sources built on the 2020 census keep the original 385 — FEMA's National Risk Index, USDA's rural-urban codes, Opportunity Zones, the CDFI Fund's Investment Areas — as do FFIEC's CRA files (checked through 2024).
What GISBanker does: Figures for the 14 retired tracts can't be matched to today's tracts and aren't shown (ACS 2020–2022, CRA 2022–2024, and their FEMA risk, rural-urban and Opportunity Zone rows). For tracts that absorbed them, earlier ACS figures cover a smaller area, so change across 2022–2023 isn't comparable.
Affects: American Community Survey (5-year) (2024)
The 2024 ACS block-group table includes block group 3 of tract 1460.01 (about 2,200 people), but the Bureau's tract table leaves that tract out.
What GISBanker does: The block group isn't shown, rather than placing it under a tract with no data behind it.
Affects: American Community Survey (5-year) (2023)
The 2023 ACS release has no row for tract 9544.01, while its neighbour 9544.02 is published. The two tracts' figures also swing sharply between releases (9544.01: 3,707 people in 2022, 6,088 in 2024).
What GISBanker does: Tract 9544.01 shows as unavailable for 2023; treat change over time for both tracts with caution.
Affects: American Community Survey (5-year) (2019); Community Reinvestment Act Data (2018–2021); Home Mortgage Disclosure Act Data (2018)
In 2019 Alaska split the Valdez-Cordova Census Area into the Chugach and Copper River Census Areas. Sources kept reporting Valdez-Cordova for a while: the ACS 2019 release, FFIEC CRA through 2021, and HMDA for 2018.
What GISBanker does: Valdez-Cordova figures can't be split between the two new areas, so those years show as unavailable for Chugach and Copper River.
Affects: Home Mortgage Disclosure Act Data (2018–2019)
In five places CFPB's HMDA data reports a negative dollar total for one kind of application — for example 342 withdrawn refinancing applications in Barnstable County, MA (2018) totalling −$845 million. Everything else for those areas is normal.
What GISBanker does: The number of applications is kept; the dollar total for that one cell shows as unavailable. Affected: Barnstable and Worcester counties, MA; El Paso County, CO; East Baton Rouge Parish, LA; and Louisiana's state total.
Affects: Community Reinvestment Act Data (2022)
FFIEC offers the 2022 all-banks file only in its original September 2023 release, while the per-bank file is the corrected January 2026 release. Added up, the banks' figures differ from the all-banks totals in 1,289 counties — 0.27% of small-business loans nationally.
What GISBanker does: All-banks totals are FFIEC's published figures; bank-by-bank figures are the corrected ones. Don't read the small difference between them as anyone's lending.
Affects: Community Reinvestment Act Data (2018–2020)
In FFIEC's own files for these years, the banks' figures added up differ slightly from the all-banks totals (0.01–0.08% of loans). Other years match county for county.
What GISBanker does: Both are shown as published.
Affects: American Community Survey (5-year) (2009–2019); Community Reinvestment Act Data (2018–2021)
Tract and block-group data before the 2020 ACS release (and before CRA's 2022 activity year) uses 2010 tract boundaries, and some 2010 tract codes now name different areas.
What GISBanker does: Those years aren't loaded for tracts and block groups; tract trends start with ACS 2020 and CRA 2022.