spb/wp9_uqo Public

1% Author: Simon-Pierre Boucher — contact@spboucher.ai2% ============================================================================3\section{Data}4\label{sec:data}5% ============================================================================67\subsection{Source and coverage}8\label{subsec:data_sources}910The analysis draws on a de-duplicated snapshot of the Canadian Multiple Listing Service11(MLS), \texttt{realtor\_mls\_unique.duckdb}, containing 172{,}019 unique active12``for-sale'' listings. Each record carries the list price, a structured set of building13and lot attributes, ownership and dwelling-type fields, and---crucially for this14study---a geocoded location (latitude, longitude and postal code). Geographic coverage is15essentially complete: coordinates and postal codes are present for more than 99.9\% of16records, so every listing can be mapped to its Forward Sortation Area (FSA), the first17three characters of the Canadian postal code and our unit of neighbourhood.1819\subsection{Sample construction and variable parsing}20\label{subsec:variables}2122The raw fields are semi-structured and require parsing. Bedroom counts reported as23``3~+~1'' (main plus lower level) are summed. Living areas, reported inconsistently in24square feet or square metres, are harmonised to square metres (preferring the explicit25floor-area measurement and converting from square feet at26$1\text{ ft}^2 = 0.0929\text{ m}^2$). Lot sizes are parsed from free-text fields into27square metres where a numeric value can be recovered, with a missingness indicator28retained. We restrict the sample to residential dwellings (houses, condominiums, plexes,29townhouses and apartments) with a strictly positive list price and non-missing core30structural fields, and we trim the extreme 1\% tails of price and living area to limit the31influence of data-entry errors and ultra-luxury outliers, a standard precaution in hedonic32work with listing data \citep{malpezzi2003hedonic}. The resulting estimation sample33contains \textbf{140{,}931 dwellings}, of which 82{,}334 are houses and 57{,}857 are34condominiums.3536The dependent variable is the natural logarithm of the list price. Structural regressors37are: $\ln$ living area; counts of bedrooms, full bathrooms and half bathrooms; parking38spaces; storeys; and an indicator for the availability of lot information together with39$\ln(1+\text{lot area})$. Categorical controls comprise the dwelling type (house,40apartment/condo, row/townhouse, duplex, triplex, fourplex, manufactured home, other), the41ownership form (freehold, condominium/strata, leasehold, etc.), and the broad listing42category. The neighbourhood fixed effect is the FSA; FSAs with fewer than 25 listings are43pooled into a province-level residual category so that each absorbed effect is estimated44from a reasonable number of observations.4546\subsection{Descriptive statistics}47\label{subsec:summary_stats}4849Table~\ref{tab:summary_stats} reports the descriptive statistics. The median dwelling50lists for roughly \$565{,}000 and offers about 125~m$^2$ of living space, two full51bathrooms and three bedrooms. Both price and price per square metre are strongly52right-skewed---means exceed medians throughout---motivating the log transformation53illustrated in Figure~\ref{fig:price}, whose right panel is approximately symmetric and54underpins the semi-logarithmic hedonic specification.5556\input{../results/tables/summary_stats}5758\begin{figure}[t]\centering59\includegraphics[width=\textwidth]{fig_price_dist.png}60\caption{Distribution of list prices. Panel (a) shows the raw price (truncated at \$3M for61readability) with the sample median marked; panel (b) shows log price, the dependent62variable, which is close to symmetric.}63\label{fig:price}64\end{figure}6566The provincial composition is dominated by Ontario, Quebec, British Columbia and Alberta,67which together account for the large majority of listings; the Atlantic provinces and the68Prairies are present but thinner, and Prince Edward Island and the territories contain no69listings in this snapshot. Median price per square metre ranges from roughly \$2{,}000 in70Newfoundland and Labrador to over \$6{,}100 in British Columbia71(Figure~\ref{fig:province}), foreshadowing the dominant role of location documented below.7273\begin{figure}[t]\centering74\includegraphics[width=0.82\textwidth]{fig_province_ppm2.png}75\caption{Median price per square metre of living area by province. Grey extensions show76the mean--median gap, reflecting right-skew from high-value listings.}77\label{fig:province}78\end{figure}79