The Quebec cluster — Laval, Gatineau, and Quebec City

2026-07-18 · wave 24 · ESTIMATE lane, deepening mode · fifth and sixth cities measured; completes the initial Quebec picture.

What was measured

Same provincial open-roll host, same MAMH schema, same parser as waves 20–21:

City Units Land share (all) Land share (residential) Vacant-land class (self-check)
Laval (Montreal's suburb) 154,876 38.4% 35.5% 97.1%
Gatineau (Ottawa's Quebec side) 103,287 37.1% 34.0% 98.3%

The finding: Quebec's urban land shares cluster at ≈ one-third

Montreal 33.2%, Laval 35.5%, Gatineau 34.0% (residential) — metro core, suburb, and a different CMA entirely, all within 2.3 points of each other, with Sherbrooke (27.3%) modestly below. Compare BC's measured 65–85%. Two readings, both useful:

  1. Substantive: Quebec's housing markets are systematically less supply-constrained (plex traditions, permissive multiplex zoning, flat geography, weaker speculative demand), so structures dominate value nearly everywhere — the w21 supply-constraint thesis, now with four in-province confirmations.
  2. Methodological: such tight clustering across four independent municipal rolls also suggests Quebec's assessment convention allocates land conservatively as a practice. Distinguishing 1 from 2 is exactly academic question Q5, and these six parsed rolls are its ready-made dataset.

The Gatineau bonus: Gatineau is the measurable half of the Ottawa–Gatineau CMA — the Ontario side computes land values and publishes none (readiness map, w14/w22). One metro, one river, two transparency regimes: the sharpest possible illustration of the "publish the split" ask.

The measured spectrum (six cities)

Vancouver 75% → Victoria 69% → (national join 53%) → Laval 38% → Gatineau 37% → Montreal 38% (33% res.) → Sherbrooke 30%. The incidence machinery's national assumption remains bracketed, now from six directions.

Wave 26 addendum — Quebec City, and the measurements become a dataset

2026-07-19 · wave 26 · seventh city measured; the roll work is now a committed, reproducible dataset.

Quebec City's full 2026 roll (RL23027_2026.xml, 273MB, 175,478 assessment units) parsed the same way:

Segment Units Land value Building value Land share
All units 175,478 $39.68B $73.32B 35.1%
Residential (usage 1xxx) 159,649 $28.56B $51.76B 35.6%
Non-residential 8,762 $9.58B $21.52B 30.8%
Vacant land (9xxx) 7,067 $1.55B $0.04B 97.6% (self-check)

The cluster holds — remarkably tightly. Every major Quebec metro now sits within 2.4 points on residential land share: Montreal 33.2%, Gatineau 34.0%, Laval 35.5%, Quebec City 35.6% — with only smaller Sherbrooke (27.3%) below. The provincial capital, a metro core, a suburb, and a border city under one assessment regime all allocate about a third of residential value to ground. That tightness sharpens both w24 readings at once: Quebec's housing supply conditions really are systematically looser than BC's, and such uniformity across independent municipal rolls keeps the allocation-methodology question (Q5) alive.

One local wrinkle: Quebec City is the first measured city where non-residential land share (30.8%) falls below residential (35.6%) — everywhere else commerce sits on relatively pricier ground. A capital city's stock of large institutional and government structures on ordinary land would do exactly this; flagged for the Q5 dataset rather than interpreted further.

The reproducibility deliverable. The parser used across waves 20–26 is now committed as tools/parse_quebec_roll.py, and all seven cities' aggregates live in data/land-shares.csv with per-row provenance (units, values, source URL, retrieval date). The parser was cross-validated by re-running Laval and Gatineau from the raw provincial XML on 2026-07-19: unit counts and shares reproduced wave 24's published figures exactly. This is the open dataset the academic brief promises referees — anyone with Python can regenerate every Quebec number in it from the province's own servers.

The measured spectrum (seven cities)

Vancouver 75% → Victoria 69% → (national join 53%) → Laval 38% → Gatineau 37% → Montreal 38% (33% res.) → Quebec City 35% → Sherbrooke 30%. The incidence machinery's national assumption remains bracketed, now from seven directions.

Sources

  1. MAMH / Données Québec open rolls: RL65005_2026.xml (Laval, 241MB), RL81017_2026.xml (Gatineau, 132MB), downloaded and parsed in full 2026-07-18 (re-validated from scratch 2026-07-19); RL23027_2026.xml (Quebec City, 273MB, 175,478 units), downloaded and parsed in full 2026-07-19. https://www.donneesquebec.ca/recherche/dataset/roles-d-evaluation-fonciere-du-quebec
  2. Prior measurements: analysis/montreal-land-share.md, analysis/sherbrooke-land-share.md, analysis/vancouver-land-share.md, analysis/victoria-land-share.md.
  3. Machine-readable aggregates: data/land-shares.csv; parser: tools/parse_quebec_roll.py.

Groundshare — a proposal in open development. Every number traces to a cited public source with its retrieval date; corrections are published, not erased. Rebuilt 2026-07-19 from the repo's research files.