On the relevance of spatial distribution of events for seismic hazard evaluation