spb/wp9_uqo Public

1% Author: Simon-Pierre Boucher — contact@spboucher.ai2% ============================================================================3\section{Introduction}4\label{sec:introduction}5% ============================================================================67Housing is the largest asset class on household balance sheets and the collateral behind8most household debt, and its price is the single most consequential relative price most9families ever face. Yet a dwelling is the archetypal heterogeneous good: no two houses10are identical, and the most important attribute---\emph{where the dwelling stands}---is11not a scalar that can be read off a listing sheet. The hedonic approach pioneered by12\citet{lancaster1966new} and \citet{rosen1974hedonic} resolves this heterogeneity by13treating the dwelling as a bundle of attributes, each commanding an implicit price14determined in equilibrium by the interaction of buyers' marginal willingness to pay and15sellers' marginal cost of supply. Regressing the (log) price of a dwelling on its16measurable characteristics recovers these implicit prices, and the resulting estimates17provide the empirical backbone for constant-quality house-price indices18\citep{bailey1963regression,hill2013hedonic}, property assessment and mass appraisal19\citep{mccluskey2013prediction}, mortgage valuation, and the welfare analysis of local20amenities \citep{black1999better,chay2005does}.2122How much of a home's value is the structure, and how much is the location? The question23is old---it is the empirical content of the realtor's adage ``location, location,24location'' \citep{kiel2008location}---but its answer matters far beyond folklore.25Assessment authorities must apportion value between land and improvements; property-tax26reform hinges on the incidence of the land component; the macro literature has shown27that land, not structure, drives both the level and the growth of aggregate housing28wealth \citep{davis2007price,knoll2017no}; and the dispersion of location values across29cities is the fingerprint of supply constraints and agglomeration30\citep{saiz2010geographic,gyourko2013superstar,combes2015empirics}. Credible answers31require holding the structural bundle fixed while letting location vary---precisely what32a hedonic model with fine geographic fixed effects delivers.3334This paper estimates such a model for the Canadian residential market at national scale.35Using a cross-section of \textbf{140{,}931} active MLS listings spanning the nine36provinces present in the data---from British Columbia to Newfoundland and Labrador---we37decompose dwelling prices into structural attributes, dwelling type and ownership form,38and a rich set of \textbf{1{,}153 neighbourhood fixed effects} defined at the Forward39Sortation Area (FSA) level. To our knowledge this is among the most geographically40comprehensive single-equation hedonic exercises assembled for Canada, where the hedonic41tradition has remained largely metropolitan in scope42\citep{desrosiers1996shopping,haider2000effects}. It is made feasible by an absorbing43least-squares estimator in the lineage of \citet{abowd1999high} that sweeps out the44high-dimensional location effects without materialising thousands of dummy variables45\citep{guimaraes2010simple,correia2017reghdfe}.4647Our main findings can be summarised as follows. The purely structural model explains 46\%48of the variation in log prices; adding province fixed effects raises this to 57\%, and49replacing them with FSA fixed effects lifts it to \textbf{77\%}. Neighbourhood location50\emph{alone} therefore accounts for roughly thirty percentage points of explanatory51power---more than all structural attributes combined, and a listing-level counterpart to52the large land shares found in aggregate data \citep{davis2007price,knoll2017no}.53Conditional on location, the living-area elasticity is estimated at 0.51--0.62: a 10\%54larger dwelling sells for about 5--6\% more. Each additional full bathroom commands a55premium of 11--15\%, whereas the number of bedrooms is economically negligible once floor56space is held fixed---a recurrent finding in the hedonic literature57\citep{sirmans2005composition}. The implicit prices are remarkably stable across58alternative samples, trimming rules and dwelling types, and the model attains an59out-of-sample $R^2$ of 0.76 with a median absolute valuation error of 16\%, a level of60accuracy competitive with the mass-appraisal benchmarks reported in the valuation61literature \citep{mccluskey2013prediction,mullainathan2017machine}. We map the estimated62neighbourhood premia and show that the highest-valued FSAs---concentrated in the City of63Vancouver and the Greater Toronto Area---trade at more than triple the national-median64level net of structure. A battery of additional analyses sharpens the picture: the65neighbourhood effects absorb 82\% of the spatial autocorrelation in raw residuals66(Moran's~$I$ falls from 0.46 to 0.08); floor space exhibits clear diminishing returns;67the location component of price decays with distance to the nearest major metropolis, as68the monocentric tradition predicts \citep{alonso1964location,muth1969cities,mills1967aggregative};69and the implicit prices vary sensibly across the price distribution and across provinces70while transferring well out-of-region.7172The contribution is threefold. \emph{First}, we provide the first (to our knowledge)73listing-level structure-versus-location decomposition spanning the entire Canadian74market, quantifying the neighbourhood as its single largest priced component.75\emph{Second}, we bring the high-dimensional fixed-effects technology of modern applied76microeconomics \citep{abowd1999high,correia2017reghdfe} to a national housing77cross-section and validate the design directly, showing that more than a thousand78absorbed neighbourhood intercepts eliminate the bulk of spatial dependence that79parametric spatial models are built to address80\citep{dubin1988estimation,basu1998analysis,gibbons2012mostly}. \emph{Third}, we supply a81transparent accuracy benchmark for the automated-valuation debate: every coefficient of82the model is inspectable, yet its held-out performance is competitive with the83mass-appraisal standards reported in the literature84\citep{clapp2003semiparametric,mccluskey2013prediction}.8586The remainder of the paper is organised as follows. Section~\ref{sec:literature}87positions the paper in six related literatures. Section~\ref{sec:data} describes the88data and variable construction. Section~\ref{sec:methodology} presents the empirical89strategy. Section~\ref{sec:results} reports the main results.90Section~\ref{sec:robustness} provides robustness checks, heterogeneity analyses, and91out-of-sample validation. Section~\ref{sec:discussion} interprets the findings against92the literature and discusses implications, limitations and future work.93Section~\ref{sec:conclusion} concludes.94