There is near universal agreement that estimates and inferences from spatial regression models are sensitive to particular specifications used for the spatial weight structure in these models. We find little theoretical basis for this commonly held belief, if estimates and inferences are based on the true partial derivatives for a well-specified spatial regression model. We conclude that this myth may have arisen from past applied work that incorrectly interpreted the model coefficients as if they were partial derivatives, or from use of misspecified models.

We attempt to clarify a number of points regarding use of spatial regression models for regional growth analysis. We show that as in the case of non-spatial growth regressions, the effect of initial regional income levels wears off over time. Unlike the non-spatial case, long-run regional income levels depend on: own region as well as neighbouring region characteristics, the spatial connectivity structure of the regions, and the strength of spatial dependence. Given this, the search for regional characteristics that exert important influences on income levels or growth rates should take place using spatial econometric methods that account for spatial dependence as well as own and neighbouring region characteristics, the type of spatial regression model specification, and weight matrix. The framework adopted here illustrates a unified approach for dealing with these issues.

We apply a Bayesian hierarchical Poisson spatial interaction model to the paper trail left by patent citations between high-technology patents in Europe to identify and measure spatial separation effects of interregional knowledge flows. The model introduced here is novel in that it allows for spatially structured origin and destination effects for the regions. Estimation of the model is carried out within a Bayesian framework using data augmentation and Markov Chain Monte Carlo (MCMC) methods, related to recent work in Frühwirth-Schnatter and Wagner (2004 ). This allows MCMC sampling from well-known distribution families, and thus provides a substantial improvement over MCMC estimation based on Metropolis-Hastings sampling from non-standard conditional distributions. Hemos aplicado un modelo de interacción espacial de Poisson Bayesiano jerárquico a la siguiendo la pista a documentación que menciona otras patentes entre patentes de alta tecnología en Europa para identificar y medir efectos de separación espacial de flujos de conocimiento interregionales. El modelo aquí introducido es novedoso en cuanto a que admite efectos de origen y destino espacialmente estructurados para las regiones. La estimación del modelo se lleva a cabo bajo un marco Bayesiano usando métodos de aumento de datos y Monte Carlo basado en cadenas de Markov (MCMC), relacionado con investigación reciente de Frühwirth-Schnatter y Wagner (2004). Esto permite el muestreo MCMC de familias de distribución bien conocidas, proporcionando así una mejora substancial sobre estimaciones MCMC basadas en muestreo Metropolis-Hastings de distribuciones condicionales no estándar.