Simon-Pierre Boucher
All research

UQO Working Paper No. 10· 29 pages

The Assessment Gap in Quebec: Vertical and Horizontal Inequity in Municipal Property Valuation

Quebec's Cheapest Homes Pay 65% Too Much Tax

Are municipal assessments equitable? Matching 522,769 sales to the roll shows systematic regressivity — 99% of municipalities fail the IAAO uniformity standard.

Simon-Pierre Boucher — contact@spboucher.ai

522,769

Sales matched to the assessment roll, across 625 municipalities

−0.34

Elasticity of the assessment ratio w.r.t. price (Cheng fixed effects)

−0.08

Elasticity under the measurement-error-robust Clapp rank IV

99%

Municipalities failing the IAAO uniformity standard (PRB < 0)

+65%

Excess property tax paid by the median bottom-decile dwelling

26

Median municipal COD — versus the IAAO ceiling of 15

Abstract

Quebec taxes every dwelling in proportion to its assessed value, redrawn on a triennial roll that must by statute reflect market conditions at a single reference date. If assessments are regressive — cheap homes overvalued relative to expensive ones — the effective tax rate silently falls with wealth. Matching 522,769 residential sales (2021–2026) at the parcel level to the assessment roll in force at sale, this paper delivers the first province-wide audit of property-assessment equity in Canada.

Within the same municipality × roll × sale-year cell, the elasticity of the assessment ratio with respect to price is −0.34 under Cheng fixed-effects estimation, and remains −0.08 under Clapp's measurement-error-robust rank instrument — regressivity is real, not a statistical artifact. The quantile profile shows failure concentrated at the top: β falls from 0.87 at the 10th percentile to 0.49 at the 90th.

The consequences are stark. Ninety-nine percent of municipalities have a negative price-related bias, 95% fall below the IAAO vertical-equity band, and the median municipal COD of 26 far exceeds the IAAO ceiling of 15. The median dwelling in the bottom local price decile pays roughly 65% more property tax than uniform assessment would imply, while the top decile pays about 5% less. Montréal is the lone progressive large market, with PRB +0.07 in all six years.

Key Results

  1. First province-wide equity audit in Canada

    Parcel-level matching of 522,769 sales to the triennial assessment rolls (median match distance 0.6 m) produces the first Canadian province-scale test of assessment equity.

  2. Regressivity survives measurement-error correction

    The Clapp rank IV, designed to purge the attenuation bias that inflates naive ratio studies, still finds γ = −0.083 — regressivity is genuine, not spurious.

  3. Failure is concentrated at the top

    Quantile regressions show β falling from 0.87 at the 10th percentile to 0.49 at the 90th: expensive homes are the ones assessed furthest below market.

  4. Near-universal IAAO standard failure

    99% of municipalities show negative price-related bias and 95% fall outside the IAAO vertical-equity band; the median COD of 26 breaches the uniformity ceiling of 15.

  5. Exact intra-municipal tax-shift calculation

    Under Quebec's exemption-free ad valorem rule, the bottom local price decile overpays about 65% while the top decile underpays about 5% — a silent redistribution.

  6. The Montréal exception

    Montréal is the only progressive large market, with a positive PRB of +0.07 in all six sample years, possibly reflecting borough-level composition.

  7. Subgroup anatomy of the gap

    Regressivity is strongest for plexes (γ = −0.49), single-family homes (−0.45), high land share (−0.50) and old dwellings (−0.43); condos are mildest at −0.12.

Data

Matched sales–roll snapshot

745,119 residential transactions (Jan 2021 – Jul 2026) matched at the parcel level to the assessment roll in force at sale; every retained sale reproduces the roll value exactly.

Assessment-roll attributes

Total, land, and building assessed values, lot and floor areas, year built, unit count, CUBF use code, roll vintage, and the statutory market-condition date.

Final estimation sample

522,769 sales in 625 municipalities and 2,884 municipality × roll × sale-year cells, after residential filtering, match-quality screens, 1/99% ratio trimming, and a 20-sale cell minimum.

Methodology

IAAO ratio-study battery

Median assessment ratio, COD, PRD, and PRB with percentile-bootstrap confidence intervals, computed for every municipality with at least 100 usable sales.

Cheng (1974) log-log fixed effects

ln AV regressed on ln SP with 2,884 absorbed market-timing cells (AbsorbingLS), standard errors clustered on 625 municipalities; γ = β − 1 measures vertical inequity.

Clapp (1990) rank instrument

A three-valued instrument built from within-cell rank agreement of ln AV and ln SP, estimated by 2SLS on demeaned data, provides the conservative measurement-error-robust bound.

Quantile and heterogeneity regressions

Quantile regressions on within-cell demeaned data (τ = 0.10–0.90) and separate fixed-effects estimates by dwelling class, age, land share, roll lag, market size, and sale year.

Horizontal-inequity regressions

The absolute deviation of the log ratio from its cell median is regressed on property characteristics to measure dispersion in assessment quality among comparable homes.

Exact tax-shift computation

Ratio relative to the cell median by within-cell price decile yields exact percentage over- and under-payment under Quebec's exemption-free ad valorem property tax.

Reproducibility

  • Five-script end-to-end pipeline (sample, IAAO stats, regressions, figures, tables) runs in about 5 minutes on an Apple Silicon laptop with roughly 6 GB peak RAM.
  • A reusable src/wp10 package separates IAAO diagnostics (iaao.py), econometric models (models.py), sample construction, and a validated journal-calibre plot style.
  • Every CSV the scripts emit is committed under results/reproduced/, and 10 LaTeX tables plus 10 publication figures (300 dpi) feed the 29-page paper directly.
  • The paper builds with latexmk from modular LaTeX sections via a Makefile.