B1065
Title: Estimation of spatially clustered panel data models
Authors: Raffaele Mattera - Sapienza University of Rome (Italy) [presenting]
Roy Cerqueti - Sapienza University of Rome (Italy)
Pierpaolo Durso - University of Rome Sapienza (Italy)
Vincenzina Vitale - Sapienza, University of Rome (Italy)
Abstract: The heterogeneity in panel data models - given by spatial variation, unknown clustering structures, or a combination of both - is a well-documented empirical phenomenon in social and economic sciences. In particular, neglecting the underlying clustering structures can lead to misleading results, such as overlooking the existence of cluster-specific relationships that are crucial for more informed decision-making. The challenge of estimating panel data models with unknown clustering structures is tackled. Recognizing the relevance of spatial heterogeneity in the framework, an algorithm is proposed that employs a spatial penalty to enhance the identification of spatial clustering in the panel data. By integrating the spatial dimension with the information of the regression results, the procedure offers a helpful approach for estimating spatially clustered panel data models. The proposed iterative algorithm is comprehensively discussed and its properties and empirical performance are highlighted through various illustrative examples.
