717 DATA - FIRST DATA DROP (Lancaster County, PA and neighboring markets) Built 2026-09-26/27 (America/New_York). Public sources only. UTF-8 CSVs with header rows. No rows invented. FILES AND ROW COUNTS (data rows, excluding header) openings.csv 242 openings / incoming concepts (news) closings.csv 153 closings, wind-downs, WARN sites (news + PA WARN merged) filings_flow.csv 72 FLOW: fictitious names, dissolutions, foreign registrations from public notices developments.csv 38 developments / investments warn_events.csv 26 PA WARN notices: Lancaster, York, Lebanon, Berks, Dauphin, Chester permits.csv 32 county planning commission (LCPC) plan reviews listed publicly listings_confirm.csv 8 listing confirmations for places already in other files snapshot_metro.csv 10 BLS metro unemployment rate + total nonfarm jobs (5 MSAs) registered_entities_lancaster.csv 73444 STOCK: active registered entities, Lancaster (PA Open Data) registered_entities_york.csv 54188 STOCK: active registered entities, York (PA Open Data) registered_entities_lebanon.csv 13884 STOCK: active registered entities, Lebanon (PA Open Data) registered_entities_berks.csv 51944 STOCK: active registered entities, Berks (PA Open Data) _sources_log.csv 55 every source touched, with result ok|paywalled|no_public_data|error|blocked README_DATA.txt DATE WINDOW Change logs (openings, closings, filings, WARN, developments, permits): 2025-01-01 through 2026-09-26. Actual coverage: openings.csv date_noted 2025-01-03 .. 2026-09-24 closings.csv date_noted 2025-01-03 .. 2026-09-17 filings_flow.csv notice/filing date 2025-01-07 .. 2026-09-25 warn_events.csv notice month 2025-02 .. 2026-08 (all listings on the PA WARN page from Jan 2025 through Sep 2026 were checked); effective dates 2025-01-31 .. 2026-10-03 developments.csv first_announced 2025-01-01 .. 2026-09-16 (some large items announced earlier are included because their status changed); last_status_date up to 2026-09-17 permits.csv filed 2026-07-14 .. 2026-08-17 (all for the LCPC meeting of 2026-09-14) Snapshots: registered_entities_*.csv retrieved 2026-09-26 (~23:47 ET); PA Open Data dataset last updated 2026-09-02. snapshot_metro.csv retrieved 2026-09-26; unemployment rate latest = Jul 2026 (preliminary), prior = Jun 2026; total nonfarm jobs latest = Aug 2026 (preliminary), prior = Jul 2026. Not seasonally adjusted (BLS LAUS + CES series). Metros are kept separate: Lancaster, York-Hanover, Lebanon, Reading, Harrisburg-Carlisle. Dates: YYYY-MM-DD with date_precision=day, or YYYY-MM-01 with date_precision=month when only the month is known. FILINGS REALITY No county-indexed DOS change feed exists. Stock = Open Data. Flow = notices + news + WARN. file.dos.pa.gov (PA DOS business search) was blocked by a Cloudflare challenge and was not bypassed. No DOS lookups were completed. The dos_file_number_if_found values in filings_flow.csv (31 of 72 rows) come from exact normalized-name matches against the data.pa.gov registered-entities extract, not from DOS lookups. For fictitious-name rows the number is the registering entity's file number, not a number for the fictitious name itself. Sole proprietors often never appear in DOS. This is not a complete list. The Open Data snapshot is STOCK and must not be read as "new filings". OPEN DATA NOTES (registered_entities_*.csv; data.pa.gov dataset xvd7-5r2c "Registered Businesses in PA Current by County") - There is no status column. The dataset is published as "Current", so all rows are treated as active. - The county field (shortcountyname) appears to be derived from the geocoded registered-office address, not from where the business operates. - The source has one row per officer. Rows were collapsed to one per filing_number/business_name, which match 1:1 (Lancaster 138,007 raw -> 73,444; York 94,371 -> 54,188; Lebanon 25,096 -> 13,884; Berks 91,857 -> 51,944). - Officer names and roles were deliberately left out (no owner lists). - address_looks_local: Y if the 5-digit ZIP is in that county's ZIP set, built from the US Census 2020 ZCTA-to-county relationship file (any overlap) plus GeoNames postal codes for the county, which add PO-box ZIPs. If the ZIP is valid but outside the set -> N, unless the city is a USPS/GeoNames place name in that county (then unknown). No usable ZIP -> Y if the city is a county place name, else unknown. Because county is address-derived, N mostly flags geocoding or ZIP errors. Results: Lancaster Y 72,889 / N 246 / unknown 309; York 53,688/184/316; Lebanon 13,701/86/97; Berks 50,983/346/615. WARN NOTES - The PA DLI WARN page lists notices by month heading and does not publish the day. notice_date = YYYY-MM-01 (month precision). - effective_date = first date listed; the raw effective-date text is kept in notes (for example, "beginning X; ending Y"). - 18 Lancaster/York/Lebanon/Berks WARN sites are in closings.csv (warn=Y). 2 were merged into existing news rows (Giant Direct, 235 N. Reservoir St; WILX Logistics, Rapho Township); 16 were added with W- ids. Closing status comes from the WARN effective/ending date and is not independently confirmed. Layoff notices have status=unknown because the facility was not stated to be closing. NEWS NOTES - LNP/LancasterOnline is metered. News rows rest on free headline/lede text (RSS) and public pages; metered article bodies were not read. - LNP Local Business coverage before April 2026 is thin because the site rate-limited requests (HTTP 429) on the older result pages. - Statuses are as of the article date unless a later source is cited in notes. - Municipality: spec spellings (Lancaster City, Manheim Township, ...). Where the source only named a postal town and no address could be geocoded, municipality = "not stated" and the location hint is kept in notes. Where an address could be geocoded, the municipality came from an OSM Nominatim geocode plus PennDOT municipal-boundary point-in-polygon (method noted per row). Park City Center was verified to be in Lancaster City (the whole mall footprint falls inside the city boundary). GAPS AND CAVEATS - Permits are fragmented. The City of Lancaster CityView portal requires an account and has no public listable permit or business-registration search; the city's public GIS permit layers stop at 2020 and its land-development layer at 2023. The Lancaster County Planning hub layer only holds the 2026-09-14 LCPC meeting (32 reviews). Those are county advisory reviews, not municipal permits. Other municipalities were not swept. Applicant, owner and person names are withheld in permits.csv. - First-ring news (York, Lebanon, Berks) is thinner than Lancaster; Dauphin and Chester are context only (WARN and large items). - A registered office is not the shopfront: many entities use an agent or home address, and the county comes from that address. - Sole proprietors and many small operators are missing from DOS and Open Data. - Statuses are as of the article/notice date; follow-up was limited. - listings_confirm.csv is a small sample (8 rows), not a Google/Yelp sweep. - There was no full sweep of DCED project announcements, EDC Lancaster CRA, or ROCK Real Estate market reports; only items reached via news were captured. - PA WARN only covers employers and events above the federal/state WARN thresholds. See _sources_log.csv for every source touched, with its result and notes.