Abstract: During the last decade a large amount of empirical works have shown that spatial effects matter for regional growth and convergence in Europe. Spatial econometric models have substantially improved the standard growth regression approach. A drawback of those models, however, is the assumption that, apart from some remarkable exceptions, physical distance is the only determinant of the spatial weight matrix, which incorporates the topology of the system and which is necessary to implement the spatial econometrics techniques. In the present paper we try to go beyond such limitation, by adopting a matrix with a weighting structure that also incorporates legal and institutional distances. Within this framework, the expected impact on the growth rate of a region, stemming from the neighbouring ones, is likely higher (lower) if the specific regions share a similar (different) socio-institutional framework. Preliminary results suggest that the variables considered have the expected signs and the inclusion of both institutional controls and heterogeneous country-specific weights provides a higher convergence rate than the standard Mankiw, Romer and Weil specification.