spb/wp10_uqo Public

1% Author: Simon-Pierre Boucher — contact@spboucher.ai2% ============================================================================3\section{Discussion}4\label{sec:disc}56\subsection{Mechanisms}78The evidence assembles into a coherent mechanism story with three layers.910\paragraph{1. Hedonic difficulty, not assessor idiosyncrasy.} The11heterogeneity profile tracks the difficulty of the valuation problem with12almost embarrassing fidelity: condominiums (homogeneous, comparable-rich)13are nearly equitable; old houses, plexes, cottages and high-land-share14properties (heterogeneous, comparable-poor, unanchored by construction15costs) are severely regressive. This is the pattern predicted if assessors16run reasonable hedonic models whose unexplained component is largest ---17and most value-correlated --- where the housing bundle is hardest to price18\citep{amornsiripanitch2022residential, gloudemans2011fundamentals,19bostic2007land}. Regression toward the mean in any imperfect valuation20model over-values the cheap tail and under-values the expensive tail;21our quantile profile, with proportionality holding at the bottom and22collapsing at the top ($\beta(0.90) = 0.49$), shows Quebec's failure is23concentrated where comparables are thinnest.2425\paragraph{2. Staleness amplifies the gap.} The doubling of the26province-wide PRB in 2021--2022 shows that triennial rolls interact27perversely with fast markets: when prices move 30--40\% between reference28dates, the cross-sectional dispersion of appreciation --- which is itself29correlated with price segment --- loads directly into measured inequity.30Jurisdictions cannot control housing cycles, but they control revaluation31frequency; annual-cycle systems mechanically cap this channel32\citep{berry2021reassessing, ross2012assessor}.3334\paragraph{3. Scale helps: the Montr\'eal exception.} Montr\'eal ---35the largest sales sample, a dedicated in-house assessment service, dense36comparable markets in every segment --- is the only large jurisdiction37with progressive assessments, in all six years. Mid-sized and small38municipalities, many relying on contracted regional evaluators with39limited modelling capacity, are uniformly regressive. The thin-market40mechanism of \citet{mcmillen2008thin} operates within a single provincial41rulebook: identical statutes, divergent outcomes, sorted by market depth42and assessment resources. We note one caveat: Montr\'eal's measured43progressivity may partly reflect within-city composition (its cell is the44whole city, pooling boroughs), and borough-level analysis is a natural45extension.4647\subsection{Policy implications}4849Three levers follow directly from the diagnosis, in increasing order of50ambition. First, \textbf{shorten the cycle}: moving from triennial to51annual (or indexed) rolls would eliminate the staleness component that52dominated 2021--2022; several U.S. states and British Columbia already53operate annual cycles. Second, \textbf{pool assessment capacity}: the54regressivity gradient by municipality size argues for regional or55provincial mass-appraisal services --- scale is the cheapest known56technology for assessment quality. Third, \textbf{audit vertically, not57just horizontally}: Quebec's oversight regime monitors median ratios and58CODs but sets no explicit PRB requirement; adding the IAAO vertical59standard to the ministry's audit criteria, with published60municipality-level statistics like those in this paper, would make the61equity dimension visible and contestable. Because low-priced-segment62owners appeal least \citep{avenancio2022assessment, plummer2014evidence},63supply-side correction --- better models --- will do more than64demand-side remedies.6566\subsection{Limitations}6768Four limitations bound the interpretation. First, sale prices below69\$50{,}000 are absent from the source registry and some residual70non-arm's-length transfers may survive our screens; the robustness71battery (dropping sub-\$100k sales, tightening trims) shows the72conclusions survive, but the bottom-decile tax-shift magnitude should be73read as an upper bound on that decile. Second, our data cover 2021--2026,74an unusually turbulent market; the within-market elasticity is stable75across the window, but levels statistics from calmer periods would differ.76Third, we observe no appeals, renovations between assessment and sale, or77conditions of sale; some of what we call horizontal noise is genuine78unobserved quality change. Fourth, without owner demographics we cannot79speak to the distributional incidence across income or racial groups as80\citet{avenancio2022assessment} do --- linking these assessments to census81tract characteristics is the obvious next step.82