This study develops a hedonic pricing model to improve the estimation of locational effects by leveraging clustering algorithms. Specifically, we examine and compare the predictive accuracy of four models: a baseline specification without locational effects, a standard model with traditional administrative boundaries, one using constrained K means clustering, and another employing a constrained spectral clustering. Focusing on Berlin’s housing market (2021–2022), our results show that spectral clustering improves predictive accuracy across standard metrics (root mean square error (RMSE),R squared (R2), mean absolute error (MAE)). To the best of our knowledge, this is the first study to apply spectral clustering to estimate locational effects within a hedonic pricing framework. Beyond the methodological contribution, we also address Berlin’s regulatory and institutional housing context and highlight how improved modelling of locational effects can enhance fairness in property taxation by providing more accurate assessments of locational value.
Gianluca et al. (Thu,) studied this question.