# Municipal valuation and property tax — v1.2 (2026-10-14; v1.1 2026-10-05; v1 2026-09-30)

Property-tax and valuation figures for every organized municipality in Maine, from Maine Revenue Services (MRS),
keyed to `unit_id` so they join to registration, election, census and geography tables. Three sources:

| File | What it is |
|---|---|
| `state_valuation_by_town.csv` | **State valuation** (the equalized full value MRS certifies each year, used for school funding, county taxes and revenue sharing) for every municipality, **2007–2026**, one row per town × year, in dollars. Source: MRS *State Valuation History 2007–2026* spreadsheet. 9,819 rows, 490 towns. |
| `state_valuation_county_state.csv` | County totals (`level=county`), unorganized-territory totals by county (`level=unorganized_territory`), and the state totals (`level=total`: STATE TOTALS, STATE TOTALS (NET OF UT), UT TOTALS) from the same spreadsheet, 2007–2026. 620 rows (v1.1 had 939: see v1.2). |
| `full_value_tax_rate_by_town.csv` | **Estimated full value tax rate** (mills — tax per $1,000 of full value; the property-tax burden on a comparable basis across towns) for every municipality, **1991–2024**. `adjustment` records how MRS computed it in that era (1991–1997 pre-Homestead, not TIF-adjusted; 1998–2007 Homestead and TIF adjusted; 2008–2024 Homestead, BETE and TIF adjusted). Source: MRS *Full Value Tax Rates* spreadsheet. 16,575 rows. |
| `full_value_tax_rate_county_state.csv` | County averages and the state weighted average mill rate, same years. |
| `mvr_by_town.csv` | **Municipal Valuation Return statistical summary**, 2009–2024, one row per town × year: what each town itself reported to MRS — certified ratio, local tax rate, tax commitment, taxable land / building / personal-property valuation, homestead and BETE exemption counts and values, TIF captured value and revenue, motor-vehicle and watercraft excise, utility / industrial / dam / generation valuation (where printed), parcel count, taxable acres, and the date and cost of the last revaluation. Source: MRS *Municipal Valuation Return Statistical Summary* PDFs, sections 1, 2, 3 and 11. 7,785 rows. `correction_note` (new in v1.1) records the few cells corrected from the printed figure and why. |
| `mvr_county_totals_as_printed.csv` | The county totals as printed in those PDFs (used to check the parse; see FLAGS). |
| `parse_mvr.py`, `build_valuation.py` | Code that built v1 (`parse_mvr.py` reads the MVR PDFs; `build_valuation.py` builds the tables). |
| `fix_v1.1_valuation.py`, `fix_v1.2_valuation.py` | The v1.1 and v1.2 corrections (each re-runnable on the previous version's folder). |
| `README.md`, `FLAGS.md` | This file; known issues and judgement calls. |
| `raw/` | MRS source files as downloaded: `state_valuation_history_2007-2026.xlsx`, `state_valuation_2026.pdf` (scanned 2026 certification), `full_value_tax_rates_2013-2024.xls` (despite the name it covers 1991–2024), and `mvr_2009.pdf` … `mvr_2024.pdf` (16 files). |

**Definitions.** *State valuation* is MRS's estimate of 100% market value, certified two years after the assessment
date (the "2026 state valuation" reflects April 1, 2024 market conditions). *Local (municipal) valuation* in
`mvr_by_town.csv` is what the town's own assessor carries on the books; `certified_ratio` is the town's declared ratio
of local to market value (1.00 = assessed at full value). `tax_rate` is the local mill rate expressed as a decimal
(0.02567 = $25.67 per $1,000 of local valuation); `commitment` is the total property tax levied that year.
The *full value tax rate* removes differences in assessing practice: commitment ÷ state valuation, adjusted for
exemptions and TIFs, so towns can be compared.

**Column notes for `mvr_by_town.csv`.** The MVR layout changed in 2015: 2009–2014 print land / buildings / personal
property / *total real & personal*, `industrial_*` and `dam_powerhouse`; 2015–2024 print `total_taxable_valuation`
directly, `electrical_generation`, and drop the industrial columns. `total_taxable_valuation` is filled from
`total_real_personal` for 2009–2014 so the series is continuous. Section 11 (parcels, acres, revaluation) is
self-reported and uneven — see FLAGS. `report_note` marks towns printed with "not reported" / "***".

**Examples.** Yarmouth: state valuation $1.50 bn (2007) → $1.53 bn (2016) → $3.46 bn (2026); full value tax rate
16.69 mills (1991), 17.00 (2010), 12.76 (2024); 2024 MVR: certified ratio 63%, local rate $25.67, commitment
$44.7 m, taxable valuation $1.74 bn, 1,870 homestead exemptions, 3,793 parcels.

Sources (MRS): state valuation history — https://www.maine.gov/revenue/taxes/property-tax/state-valuation
(`2026_historic_state.valuation.xlsx`); full value tax rates — https://www.maine.gov/revenue/taxes/property-tax/municipal-services
(`2024 FULL VALUE TAX RATES HISTORY with percentage increases.xls`); MVR statistical summaries —
https://www.maine.gov/revenue/taxes/property-tax/municipal-services/valuation-return-statistical-summary (one PDF per year, 2009–2024).

## What changed in v1.2 (2026-10-14)

- **`state_valuation_county_state.csv`: the 319 unnamed "COUNTY TOTALS" rows are gone (939 → 620 rows).** Each repeated a named county row of the same year without the county's name, and 2025 had only 15 of the 16 (Somerset's was printed one row low in the MRS workbook), so summing `level == "total"` double-counted. What is left adds up: in every year the 16 county rows plus UT TOTALS equal STATE TOTALS, the 16 county rows equal STATE TOTALS (NET OF UT), and the unorganized-territory rows equal UT TOTALS. (Audit II, GC-11.)
- **`full_value_tax_rate_by_town.csv`: 24 rates of 0.00 are now blank**, with the reason in `note`: Frye Island 1991–1997 (it became a town in 1998), Long Island 1991–1992 (part of Portland until July 1993), Codyville Plt 2003 and 2006–2018, and Glenwood Plt 2018 (MRS computed no rate). A 0-mill rate would plot as a real rate. (GC-13.)
- **`mvr_by_town.csv`: Frankfort 2023 business equipment $23,254,989 → $803,111**, with a `correction_note`. The printed row proves it: production machinery $0 + business equipment + other $0 must equal the printed total personal property of $803,111, and land & buildings $117,255,350 + $803,111 = total taxable $118,058,461, which × 0.01225 is the printed commitment $1,446,216. MRS's printed Waldo County business-equipment total includes the slip and stays as printed. (GC-12.)
- The other cells the audit questioned (Jonesboro 2011 excise, Portland 2015 land/building, eleven motor-vehicle excise figures about 1/100 of normal, Carroll Plt and Orrington 2021 $4 / $50 gaps) were checked against the MVR PDFs in `raw/`: each is printed exactly so by MRS and nothing in the printed rows proves the right figure, so they are **flagged, not changed** (FLAGS #14).
- Script: `fix_v1.2_valuation.py <v1.1 folder> <out folder>`.

## What changed in v1.1 (2026-10-05)

- **Somerset County 2025 state valuation corrected (33 towns).** MRS's history workbook prints the Somerset 2025 column
  one row out of place (each town carried the next town's figure: Anson $143.95m, Skowhegan $239.8m, Cambridge $213.75m).
  Re-read and checked line by line against the 2025 state valuation printed in MVR 2024 Section 12: Anson **$235,500,000**,
  Skowhegan $1,530,750,000, Cambridge $44,100,000, West Forks Plt $39,650,000 (blank in v1). Towns now sum exactly to the
  county total ($7,480,950,000). Every other town-year 2008–2025 (8,763 rows) matches MVR Section 12 to the dollar.
- **Penobscot Nation 2008** state valuation: printed 0 in the workbook → blank.
- **MVR corrections** (proved by commitment = tax rate × taxable valuation): Whiting 2010 tax rate 0.100 → 0.0100;
  Pownal 2014 personal property 99,965,090 → 0 and total taxable 199,930,180 → 99,965,090 (land+building total re-keyed
  into the personal-property columns; tax rate 0.0345 was right); Pleasant Ridge Plt 2009 commitment 7,816,740 → 781,674;
  Andover 2009 tax rate 0.0102 → 0.0120. Masardis 2012 flagged, not changed. `mvr_county_totals_as_printed.csv` is
  unchanged (MRS's printed totals include the slips — see FLAGS).
- FLAGS #2 rewritten (the West Forks diagnosis was wrong), Washington 2008 explained, Franklin 2005 tax rates flagged.