1\section{Related literature and mechanisms}\label{sec:lit}23\paragraph{Hedonic prices and depreciation.}4Quality-adjusted pricing of automobiles is the founding application of hedonic methods \citep{court1939hedonic,griliches1961hedonic,ohta1976automobile}, formalised by \citet{lancaster1966newapproach} and \citet{rosen1974hedonic} and extended to flexible functional forms and identification by \citet{pakes2003hedonic}, \citet{ekeland2004hedonic} and \citet{bajari2005hedonic}; \citet{triplett2006handbook} and \citet{hill2013hedonic} survey measurement practice. For used vehicles, the price of a given model falls with age and with accumulated mileage, and \citet{engers2009annualmiles} show that annual miles, not only odometer readings, drive used-car prices. The Quebec-specific evidence on depreciation is thin \citep{yanes2010depreciation}. Our hedonic models follow this tradition with asking rather than transaction prices and ask how far the flexible-form conclusions carry over.56\paragraph{Durable-goods markets, adverse selection and liquidity.}7Since \citet{akerlof1970lemons}, used-car markets have been the canonical setting for asymmetric information \citep{kim1985lemons,bond1982lemons,genesove1993adverse,hendel1999adverse,emons2009usedcars,peterson2014adverse}, with online evidence from eBay Motors \citep{lewis2011asymmetric,adams2011vettes}. \citet{stolyarov2002turnover}, \citet{gavazza2011frictions} and \citet{gavazza2014usedcar} model trading frictions and the resulting pattern of turnover and prices over the vehicle's life; \citet{larsen2021bargaining} measures bargaining efficiency in wholesale auctions and \citet{biglaiser2020intermediaries} the premium that dealers earn as intermediaries. Dealers in these models hold inventory precisely because buyers arrive stochastically; the markdown and exit hazards we estimate are the empirical shadows of that search process.89\paragraph{Dealer pricing and inventory.}10Retail price cuts are the standard tool for clearing inventory \citep{lazear1986clearance,pashigian1991sale,pesendorfer2002retailsales,aguirregabiria1999markups}, and the dynamic-pricing literature characterises optimal markdown paths for perishable or seasonal goods \citep{gallego1994dynamicpricing,elmaghraby2003dynamicpricing,sweeting2012perishable}. For cars, \citet{zettelmeyer2006internet} and \citet{scottmorton2001internet} document how internet referrals compressed dealer margins, \citet{busse2006cashback} how promotions pass through, and \citet{israeli2022marketpower} how new-car dealers vary prices with inventory. \citet{murry2020colocation} show that used-car dealers co-locate to attract searching buyers. Industry sources report that a substantial share of used listings receive price drops \citep{autotrader2024priceindex,iseecars2013pricedrops}, but to our knowledge no peer-reviewed study estimates markdown hazards or cut sizes on listing-level paths.1112\paragraph{Search, dispersion and asking prices.}13Price dispersion for homogeneous goods is the signature of costly search \citep{stigler1961information,diamond1971paradox,varian1980sales,burdett1983equilibrium,sorensen2000dispersion,lach2002dispersion,baye2006dispersion,kaplan2015dispersion}. The role of the asking price as a signal and a commitment device has been studied mainly for housing: time on market and list-price cuts \citep{knight2002listing,anglin2003tradeoff,genesove2001lossaversion,carrillo2012search,han2016askingprice}, momentum and strategic complementarity in list prices \citep{guren2018momentum}, and listing-based price indexes \citep{anenberg2017timely,lyons2019listprices}. Liquidity itself is measured with price impact in financial markets \citep{amihud1986bidask,amihud2002illiquidity}, with search-and-matching models in over-the-counter and housing markets \citep{duffie2005otc,lagos2009liquidity,krainer2001housingliquidity,head2014searchliquidity,ngai2014hotcold}, and with days'-supply or days-to-turn statistics in the car trade \citep{cox2022dayssupply,manheim2023methodology,cbb2026index}. Our index formalises the latter as a transparent multi-component measure.1415\paragraph{Survival analysis.}16We rely on discrete-time hazard models \citep{allison1982discrete,singer1993discrete}, the Cox model and its diagnostics \citep{cox1972regression,grambsch1994proportional,lin1989robust,therneau2000modeling}, flexible parametric and competing-risk alternatives \citep{royston2002flexible,fine1999proportional,putter2007tutorial}, and the econometric treatment of duration data with censoring and unobserved heterogeneity \citep{kiefer1988duration,lancaster1990econometric,heckman1984method}.1718\paragraph{Spatial competition and regional integration.}19Localised competition follows \citet{hotelling1929stability} and \citet{salop1979monopolistic}; empirical work measures spatial differentiation in retail gasoline \citep{houde2012spatial,pinkse2002spatial,chandra2011gasoline,byrne2019coordinate} and hotels \citep{kalnins2003hamburger}, and regional price integration through border effects \citep{engel1996border,gorodnichenko2009border,ceglowski2003lawofoneprice}; \citet{haan2010internetregional} document regional differences in Dutch used-car listing prices. Spatial inference follows \citet{conley1999gmm} and \citet{anselin1988spatial}.2021\paragraph{Identification of strategic interactions.}22Whether a dealer's price cut \emph{causes} a neighbour's cut is a reflection problem \citep{manski1993reflection,brock2001discrete,bramoulle2009peer,angrist2014perils,blume2011identification}. We use event-study and distributed-lag designs with the diagnostics developed in the recent difference-in-differences literature \citep{sun2021estimating,callaway2021difference,dechaisemartin2020twoway,goodmanbacon2021difference,borusyak2024revisiting,roth2022pretest,rambachan2023credible,roth2023trending}, cluster-robust inference \citep{bertrand2004howmuch,cameron2008bootstrap,cameron2011multiway,abadie2023clustering}, and multiple-testing control \citep{benjamini1995fdr,romano2005stepwise}; predictive comparisons use \citet{diebold1995comparing}, \citet{giacomini2006tests}, \citet{brier1950verification} and gradient boosting \citep{friedman2001greedy,ke2017lightgbm,athey2019machine,mullainathan2017machine}.2324\paragraph{What this paper adds.}25Relative to this literature the paper offers (i) the first listing-level daily panel of a whole regional used-car market with complete asking-price histories, (ii) hazard-based evidence on markdowns and exits with dealer-clustered inference, (iii) a competitor event study that distinguishes same-day synchronisation from lagged response, (iv) a precisely estimated null on regional premia, and (v) a transparent liquidity index. It does not add transaction prices, and we acknowledge that days'-supply measures already capture part of what the index measures.26