Visible environment and invisible label: Does stigmatization exist in old communities in Hangzhou, China?

Old residential communities in Chinese cities are byproducts of urbanization and often suffer from poor living conditions and other negative issues. Due to prolonged neglect in maintenance and management, these communities are susceptible to external stigmatization. However, existing research rarely examines the effect of stigma on housing prices in such neighborhoods. To address this gap, this study explores the influence of stigmatization on housing prices in old urban communities, with a particular emphasis on the mediating role of the neighborhood environment. Utilizing street view image data and deep learning methods, this study constructs six indicators of the neighborhood environment: greenness, openness, walkability, imaginability, enclosure, and complexity. Based on these indicators, a hedonic pricing model and a mediation effect model are developed to empirically examine the complex relationships among old communities, neighborhood environment, and housing prices. To address the potential limitations of linear models, a robustness check was conducted by using a Gradient Boosted Decision Tree (GBDT) model combined with SHAP analysis. The results indicate that a significant stigma effect exists in old communities, leading to an overall decline in housing prices by 11 %. Further analysis reveals that this stigma effect can be categorized into a direct effect of psychological labeling (−9.6 %) and an indirect effect mediated by environmental factors (−1.4 %). The non-linear model produced highly consistent results, with the indirect effect mediated by the neighborhood environment accounting for 16.7 % of the total effect. These findings highlight the dominant role of social perception over physical conditions in shaping housing prices. The study offers empirical support for stigma theory in urban housing markets and provides practical implications for neighborhood regeneration strategies. © 2025 Elsevier Ltd.