We propose a geoadditive negative binomial model (Geo-NB-GAM) for regional count data which allows us to simultaneously address some important methodological issues, such as spatial clustering, nonlinearities and overdispersion. We apply this model to study location determinants of inward...