Modelling a property's time to sale by inflating its previous selling price

Abstract

The modelling of the time to sale for residential property is complex because of complicating factors such as properties changing their price or being withdrawn from sale, changing market conditions, the presence of submarkets, whether the property represents ‘good value’, and seller motivation. This paper analyses a data set of modern properties that contains information about their previous selling price. This information is used to value the properties instead of the standard method of regressing on the physical characteristics of the properties. A survival analysis approach is then used to model the time to sale with this time being adjusted to take into account changing market circumstances and alterations to the list price during this period. It is found that the hazard rate for selling a property is approximately constant for the first 9 months a property is marketed.

Keywords:

• statistics
• survival analysis
• housing
• bidding

Acknowledgements

The paper has been considerably improved by the referees’ comments, especially with regard to the proportional hazards modelling.