Understanding Housing Market Dynamics through a Joint Model of Time-on-Market Duration and Listing Outcome of Residential Property Listings

The maximum time-on-market (TOM) of a residential property listing is an important assumption in the housing market–clearing process of a land-use and transport interaction (LUTI) model, but it is rarely studied empirically. This paper introduces an innovative approach to estimating the listing outcome and TOM through a discrete-continuous joint model. A multinomial logit model and a competing risk hazard model are estimated simultaneously to model the probabilities of different listing outcomes and TOM, respectively. Two joint models are developed, one for the initial listings and another for relistings following the termination of the initial listings. The results indicate statistically significant correlations between the termination outcomes and their corresponding TOM durations. Both models reveal that a higher listing price can increase the likelihood of termination outcomes. Factors such as parking spots, attached garages, and nearby subway stations can negatively influence the probability of terminating the initial listing. Additionally, more washrooms and transit stops nearby can shorten the TOM, whereas higher listing prices and a lack of garages extend it. This study challenges the conventional assumption of a uniform TOM for all sellers in LUTI models. Integrating this joint model into a LUTI microsimulation can enhance its ability to reflect real-world housing market dynamics and provide insights into seller behavior.