# Municipal valuation — FLAGS (v1.2, 2026-10-14)

1. **Units in the state valuation spreadsheet.** MRS prints 2007–2024 in thousands of dollars and 2025–2026 in
   dollars in the same sheet. Everything in `state_valuation_by_town.csv` has been converted to dollars; a check that no
   cell sat in the wrong unit passed for every municipality (the only trips were the state-total rows, which are
   large enough to look like dollars — they are thousands and were converted like the rest).
2. **Somerset County 2025 state valuation is printed one row out of place in the MRS workbook** (corrected in v1.1).
   In `2026_historic_state.valuation.xlsx` every Somerset town's 2025 cell holds the *next* town's value, West Forks'
   cell is blank and the county total sits one row below the "COUNTY TOTALS" line. v1 read the cells as printed and
   (wrongly) blamed a blank West Forks value for the $235.5m gap — that gap was Anson's value. v1.1 takes all 33 values
   from MVR 2024 Section 12 ("2025 State Valuation"), which agrees with the workbook shifted by one row and with the
   county total. A check of every county-year 2008–2025 against MVR Sections 12 (2013–2024 PDFs) found no other
   mismatch; 2007 and 2026 (not in any MVR) show no shift pattern on a neighbour-trend test.
   **Washington County 2008:** the towns sum to $4,150,000 more than the printed county total because MRS left the
   Passamaquoddy Nation line ($4,150,000 in 2008) out of that year's county total; in every other year it is included.
   **Penobscot Nation 2008** is printed as 0 (and the Penobscot county total excludes it); kept blank, not zero.
3. **Deorganized towns.** Bancroft (2015), Oxbow Plt (2017), Cary Plt, Atkinson and Codyville Plt (2019), Magalloway Plt (2021) and Drew Plt (2023) — years from the State Auditor's deorganization list — (and Madrid
   and Centerville, deorganized earlier) carry "deorganized" / "UT" in the source for the years after they joined the
   unorganized territory; those cells are blank with the source text in `note`. Their later valuation is inside the
   county's unorganized-territory line in `state_valuation_county_state.csv`.
4. **Passamaquoddy Nation** appears as one line (Washington County) in the state valuation and tax-rate sheets; the
   project's units are Indian Township and Pleasant Point separately, so this line has no `unit_id`.
5. **2018 MVR: Section 2 (personal property, homestead, BETE) is missing from the published PDF** — MRS printed
   Section 1 twice in its place. `homestead_*`, `bete_*` and personal-property columns are blank for 2018.
6. **Printed county totals that do not equal the sum of the municipal rows** (parsed values match the printed
   municipal lines; the discrepancy is in the source arithmetic): 2010 York TIF captured value; 2011 York TIF and
   Somerset BETE value (by $300); 2014 Penobscot commitment ($1.4 m); 2022 TIF tax revenue in nine counties
   (MRS's totals are lower than the sum of the rows); 2023 Hancock total land & building ($687 m). Also 2019
   Dennistown Plt land + buildings differ from its printed total by $20. Every other column in every year
   reconciles to the printed county totals to the dollar.
7. **Municipalities not reported.** A handful of towns each year filed no return or were printed as "not reported"
   (15 rows) or "***" (6 rows; Atkinson, Cary Plt and Codyville Plt after deorganization); see `report_note`. The
   MVR covers only organized municipalities (489 in 2009 falling to 482 in 2024 as plantations deorganized).
8. **Section 11 (parcels, acres, revaluation) is self-reported and messy** — dates like "5/29/1905" or "10/12/4670",
   acreage of 0 or 43 for whole towns, and revaluation dates missing for about a quarter of towns each year. The
   as-printed strings are kept alongside a normalised ISO date (two-digit years >30 read as 19xx). In 2012 the
   revaluation date column is largely blank (the source printed 1/0/1900 placeholders, dropped here). Treat parcel
   counts and acres as indicative.
9. **Tax rate as decimal vs mills.** `tax_rate` in `mvr_by_town.csv` is the local rate as printed (0.02567 =
   25.67 mills); `full_value_tax_rate` is already in mills. Certified ratio is stored as a fraction (0.93 = 93%).
10. **Not included:** MVR sections 4–10 (tree growth, farmland, open space, veterans and other exemptions) and
    section 12 (state valuation, duplicated by the spreadsheet). They are in the PDFs in `raw/` and can be added
    with the same parser. Years before 2009 are not published online by MRS.
11. **MVR cells corrected from the printed figure in v1.1** (each noted in `correction_note`; the proof is that the
    corrected figures make commitment = tax rate × taxable valuation to the dollar): Whiting 2010 tax rate; Pownal 2014
    personal property / total taxable valuation; Pleasant Ridge Plt 2009 commitment; Andover 2009 tax rate (the town's
    printed rate was its 12.00-mill rate times its 85% certified ratio). MRS's printed county totals for Cumberland 2014
    and Somerset 2009 include the Pownal and Pleasant Ridge slips and are left as printed in
    `mvr_county_totals_as_printed.csv`, so those two counties no longer reconcile to the town rows.
12. **Masardis 2012 (not corrected).** Commitment $422,407 = 0.0177 × $23,864,829, exactly $2,000,000 less than the
    printed taxable valuation $25,864,829, whose components (production machinery $9,098,314 etc.) add up as printed.
    Either one valuation figure is $2m too high or $2m was not taxed; the PDF does not say which. Left as printed.
13. **Franklin County 2005 full value tax rates (kept as printed).** Seven towns have 2005 rates far from their 2004/2006
    values — Phillips 1.88, Rangeley 19.55, Rangeley Plt 10.04, Strong 25.37, Temple 5.37, Weld 4.23, New Sharon 18.13
    (e.g. Rangeley 8.35 → 19.55 → 10.34). These are exactly the figures in MRS's *Full Value Tax Rates History* workbook
    (the copy in `raw/` and the one on maine.gov on 2026-10-05); they are not a parse error and do not form a clean
    rotation among the seven towns, so they are not corrected. Treat Franklin 2005 rates for these towns as unreliable.
14. **MVR figures that look wrong but are printed so by MRS (v1.2, kept as printed).** Each was checked against the PDF in
    `raw/` on 2026-10-14; the parse matches the print, and nothing in the printed row proves a correction:
    - **Jonesboro 2011:** motor-vehicle excise $1,513 and watercraft excise $96,549. In other years it is about $100k
      motor vehicle and $1.3k watercraft, so the two are probably swapped. Jonesboro also shows $0 / $0 in 2012, 2016–19
      and 2021–22, which probably means "not reported" rather than zero.
    - **Portland 2015:** land $70,863,170 and buildings $7,221,736,550 (land is about $2.4 bn in every other year). The
      total ($7,292,599,720) and everything that depends on it are right; the split between land and buildings is not.
    - **Motor-vehicle excise about 1/100 of normal:** Corinth 2010 ($227), Lincoln 2010 ($9,887; its watercraft figure
      $754,128 looks like the motor-vehicle figure, so the pair is probably swapped), Orono 2015 ($4,205; watercraft
      $690,114, probably swapped), Presque Isle 2023 ($4,600), Columbia Falls 2012 ($321), Lovell 2015 ($290),
      Winn 2016 ($535; watercraft $84,471, probably swapped), Kenduskeag 2017 ($517), Machias 2018 ($1,583; watercraft
      $292,586, probably swapped), Parkman 2019 ($1,029; watercraft $153,712, probably swapped), Eastbrook 2024 ($1,441).
      Do not chart excise for these town-years without a caveat; MRS's printed county totals include them.
    - **Small printed arithmetic gaps:** Carroll Plt 2021 land + buildings ($14,476,983 + $5,344,853) is $4 less than its
      printed total ($19,821,840); Orrington 2021 personal-property parts add to $50 more than the printed total ($10,309,500).
15. **Frankfort 2023 business equipment corrected (v1.2)** from $23,254,989 to $803,111 — proof in `correction_note`
    (the printed row's own totals). MRS's printed Waldo County 2023 business-equipment total ($32,959,538) includes the
    slip and is left as printed in `mvr_county_totals_as_printed.csv`, so Waldo 2023 business equipment no longer
    reconciles to the town rows (by $22,451,878).
16. **Full value tax rate 0.00 (v1.2):** MRS's workbook prints 0.00 for Frye Island 1991–97 and Long Island 1991–92 (before
    they were towns) and for Codyville Plt 2003 and 2006–2018 and Glenwood Plt 2018 (no rate computed; Codyville's MVR shows
    a commitment in 2017–18). These 24 cells are blank in v1.2 with the reason in `note`. The county averages in
    `full_value_tax_rate_county_state.csv` are MRS's and were not recomputed.
17. **`state_valuation_county_state.csv` structure (v1.2):** the unnamed "COUNTY TOTALS" rows were removed (they duplicated
    the named county rows). Use `level == "county"` for county totals; `level == "total"` now holds only the three
    statewide lines.