The Heterogeneous Pricing Effect of Textual Sentiment in Real Estate Listings: Evidence from Krakow, Poland

Abstract

Textual information contained in online property listings has become an increasingly important source of market signals, yet the economic value of linguistic features remains insufficiently understood. This study examines whether sentiment expressed in residential property descriptions is associated with asking prices and whether this relationship depends on the informational context of the listing. Drawing on signalling theory and information asymmetry, the study argues that sentiment should be viewed as a conditional signal rather than a universal price premium. The paper analyzes residential property listings from Krakow, Poland, and applies hedonic pricing models incorporating textual sentiment measures and interaction terms. The results show that the relationship between sentiment and price is heterogeneous rather than uniform. Positive sentiment is associated with higher asking prices particularly in contexts characterized by greater information uncertainty, including tenement buildings. The findings further indicate that seller type influences the level of sentiment expressed in listings but does not significantly moderate its price effect. Overall, the study contributes to the literature by showing that textual sentiment functions as a context-dependent market signal, with its value determined by the informational characteristics of the property and listing environment.