Information value of property description: A Machine learning approach

This paper employs machine learning to quantify the value of “soft” information contained in real estate property descriptions. Textual descriptions contain that traditional hedonic attributes cannot capture. A one standard deviation increase uniqueness a based on this leads 15% sale price model and 10% repeat sales model. The effects appear arise through two channels: unobserved quality housing unit, market power unit relative competing properties. be driven entirely by unit. Further, an annual index ignoring our measure overstates prices between 23% mistimes stabilization following Great Recession. Similar, but smaller effects, are observed for index.