1\section{Institutional intuition and the frictionless benchmark}2\label{sec:benchmark}34A house is sold through a sequence of costly events: a listing, visits and offers from buyers who arrive at random, a5negotiation in which the seller's willingness to wait and the buyer's alternatives both matter, and, for most buyers, a financing6step whose outcome is uncertain and takes time. During this sequence the seller pays a carrying cost that differs across7sellers: mortgage interest and maintenance for all, the cost of a second residence or a bridge loan for those who have already8moved, and the psychological or financial cost of an unresolved sale for those under distress. The evidence reviewed in9Section~\ref{sec:literature} shows that these costs translate into prices and durations in a systematic way: motivated,10relocated and foreclosed sellers obtain lower prices and sell faster, and the discount they accept is larger when the market is11thin. The purpose of the model is to make this logic explicit in a setting where the arrival of buyers, the acceptance decisions12of sellers of different urgency, the price that results from bargaining, and the entry of buyers that determines how many of13them there are per listing, are all determined jointly.1415Two benchmarks are used throughout, and they answer different questions. The first is the most patient seller in the same16market. Sellers who differ only in their carrying cost face the same buyers, the same matching technology and the same17bargaining protocol, so the gap between their expected transaction prices isolates the price of urgency. The \emph{urgency18discount} of a seller with carrying cost $c$ is defined relative to the most patient type,19\begin{equation}20D(c)\equiv\frac{\bar p(c_L)-\bar p(c)}{\bar p(c_L)},21\label{eq:Ddef}22\end{equation}23where $\bar p(c)$ is the expected transaction price conditional on a sale. This is the model counterpart of the forced-sale24discount measured by \citet{campbell2011forced} or \citet{andersen2017firesales}: a comparison among houses that are identical25except for the circumstances of their owners. Proposition~\ref{prop:urgency} shows that $D(c)$ is nonnegative and increasing in26$c$ under weak conditions, so that it can be called a discount without further qualification.2728The second benchmark is a frictionless market. Consider the same buyers, with the same entry cost $k$, the same distribution $G$29of match-specific valuations $w$ and the same discount rate $r$, but suppose that buyers meet listings instantly and that each30house trades at a uniform price $p^*$ at which the buyer either buys or walks away. With free entry, buyers enter until the31expected gain from an immediate purchase equals the entry cost:32\begin{equation}33\int (w-p^*)^+\,dG(w)=rk\qquad\Longleftrightarrow\qquad H(p^*)=rk,34\label{eq:pstar}35\end{equation}36where $H(v)\equiv\int(w-v)^+dG(w)$ is the expected surplus function that recurs throughout the paper.\footnote{The entry37cost is a stock and the surplus flow is a present value; equating the flow equivalent $rk$ to the expected surplus of a single38immediate purchase is the Walrasian counterpart of the free-entry condition of the search economy, where the value of an active39buyer equals $k$. Details are in Appendix~\ref{app:proofs}.} Sellers in this benchmark trade immediately at $p^*$ and pay no40carrying cost, so $p^*$ does not depend on $c$, on tightness, on bargaining power or on matching efficiency. The \emph{market41liquidity gap} is defined as42\begin{equation}43\Gamma\equiv\frac{p^*-\bar p_T}{p^*},44\label{eq:Gammadef}45\end{equation}46where $\bar p_T$ is the transaction-weighted mean price of the search economy. It would be natural to call $\Gamma$ the liquidity47premium of the housing market, and the sign convention $D=(p^*-p)/p^*$ suggests that it should be positive. It is not: search48delays and carrying costs push transaction prices below $p^*$, but in a search market with bargaining a seller also has the option49to wait for a buyer with a high match-specific valuation and to capture part of that valuation, an option that the uniform-price50benchmark rules out. Proposition~\ref{prop:arrival}(iv) shows that $\Gamma$ is positive in thin markets and negative in51sufficiently thick ones. The paper therefore reports $\Gamma$ as a decomposition and reserves the term liquidity discount for52$D(c)$ and its transaction-weighted average $\bar D=\int D(c)\,\omega(c)\,dc$, where $\omega$ is the composition of transactions53across seller types. Two further objects are used: the \emph{seller's liquidity cost} $L(c)=c/(r+\delta+\eta(c))$, the expected54discounted carrying cost paid over a listing spell with sale hazard $\eta(c)$ and withdrawal hazard $\delta$, and the55\emph{reservation-value gap} $V(c_L)-V(c)$ between the most patient and a type-$c$ seller. A buyer's willingness to pay to avoid56search is a distinct object, measured by the buyer's value $U$, and is not reported as a price discount.5758The distinction between the two benchmarks matters for interpretation. Because the model has no quality heterogeneity within a59market segment (quality is a scale factor, Lemma~\ref{lem:scale}), every price gap it produces is a gap among identical houses. In60data, the comparison of forced and unforced sales also requires conditioning on quality and condition, and61\citet{clauretie2009foreclosure} show that about a third of naive foreclosure discounts disappears once this is done. The urgency62discount $D(c)$ is the part that survives such conditioning; the model is silent on the part that does not.63