spb/wp4_uqo

1% ==========================================================================2\section{Introduction}3\label{sec:introduction}4% ==========================================================================56The theory of hedonic prices, formalized by \citet{Rosen1974} and rooted in7the characteristics approach to consumer demand developed by8\citet{Lancaster1966}, posits that differentiated goods can be described by a9vector of objectively measured attributes and that the market price of such a10good equals the sum of the implicit valuations consumers place on each11attribute. In the four decades since Rosen's seminal contribution, hedonic12models have become a workhorse of applied microeconomics, with extensive13applications in real estate \citep{ChattopadhyayMachemehl2006}, automobiles,14environmental valuation, and, more recently, the sharing economy.1516The rapid global expansion of Airbnb and competing short-term rental (STR)17platforms has created a uniquely rich laboratory for hedonic analysis. Unlike18traditional hotel room inventories, which are relatively standardized within19a star-rating tier, Airbnb listings exhibit enormous heterogeneity along20structural, locational, amenity, and reputational dimensions. Each listing21is, in effect, a bundle of characteristics chosen and priced by an individual22host operating in a decentralized marketplace. The platform's publicly23accessible data---including detailed descriptions, amenity lists, review24scores, host profiles, calendar availability, and geographic25coordinates---provides researchers with an unusually granular view of the26hedonic price surface.2728A growing empirical literature has exploited this richness.29\citet{WangNicolau2017} estimate hedonic price equations for a cross-section30of Airbnb listings in 33~major cities, finding that property type, location,31host characteristics, and online reputation are significant price32determinants. \citet{Gibbs2018} extend the analysis to include amenity33bundles and find that premium amenities such as pools and hot tubs34command substantial price premiums. \citet{Zhang2017} and \citet{Chen2020}35document similar patterns across Chinese and U.S.\ markets, respectively,36while \citet{BenitezAurioles2018} and \citet{Dogru2017} examine European and37Turkish markets. \citet{PerezSanchez2018} apply geographically weighted38regression to Spanish Airbnb data, demonstrating that implicit prices vary39spatially. \citet{GunterOnder2018} study Vienna's market structure,40while \citet{Magno2018} investigate the role of host experience in shaping41pricing outcomes.4243Despite these contributions, several gaps remain. First, most studies rely on44a single econometric specification---typically ordinary least squares (OLS)45applied to a log-linear price equation---and therefore cannot capture46heterogeneity across the conditional price distribution. Second, the47potential for complex non-linear interactions among listing characteristics48is seldom explored, and when machine learning methods are employed, the49resulting models are treated as black boxes rather than as complements to50structural hedonic inference. Third, spatial dependence in listing51prices---a first-order concern in any market where location matters---is52frequently acknowledged but rarely modeled explicitly using spatial53econometric techniques.5455This paper addresses all three gaps by developing a comprehensive56multi-method hedonic pricing model for Airbnb listings in Quebec, Canada.57Our central research questions are:58%59\begin{enumerate}[leftmargin=2em]60    \item What are the implicit prices of structural, amenity, host,61          reputation, locational, and text-derived characteristics of62          Airbnb listings in Quebec?63    \item How do these implicit prices vary across the conditional price64          distribution, from budget to luxury segments?65    \item To what extent do non-linear interactions and spatial dependence66          alter the hedonic price surface estimated by conventional linear67          models?68    \item Which listing attributes contribute most to predictive accuracy,69          and does the importance ranking differ between econometric and70          machine learning frameworks?71\end{enumerate}7273Our contribution is fourfold. First, we assemble a comprehensive hedonic74dataset for Quebec---a market that, despite its significance as a major75North American tourism destination, has received relatively little attention76in the Airbnb pricing literature. Our sample encompasses 5,000 listings77across Montr\'{e}al, Quebec City, the Laurentians, the Eastern Townships,78Charlevoix, and dozens of smaller municipalities. Second, we construct a79feature space spanning seven attribute categories---structural80characteristics, amenity indices, host attributes, reputation signals,81locational variables, NLP-derived text features, and spatial82controls---producing 49 modeling features. Third, we implement a layered83estimation strategy combining OLS with heteroskedasticity-robust standard84errors (HC3), quantile regression at five percentiles, an ensemble of seven85machine learning models interpreted through SHAP values86\citep{Lundberg2017}, and spatial econometric models including spatial lag,87spatial error, and spatial Durbin specifications \citep{LeSagePace2009,88Anselin1988}. Fourth, we demonstrate how estimated implicit prices aggregate89into a price decomposition that benchmarks any listing against the market,90providing actionable intelligence for hosts and platform managers.9192The remainder of the paper is organized as follows.93Section~\ref{sec:data} describes the data sources, variable construction,94and summary statistics. Section~\ref{sec:methodology} presents the95econometric and machine learning methodology.96Section~\ref{sec:results} reports the main empirical results.97Section~\ref{sec:robustness} discusses robustness checks.98Section~\ref{sec:conclusion} concludes.99