Spatial Interaction of Crime Incidents in Japan1

large empirical literature has developed around the estimation and testing of economic models of crime. Recently, spatial interactions of crime incidents have become one of the research areas in economics in connection with the progress of spatial econometrics. But these models of crime based on aggregated data relied heavily on cross-sectional econometric techniques, and therefore previous attempts do not control for unobserved heterogeneity. Cornwell and Trumbull (1994) show that using panel data gives us a large model class for criminal data. Although there also exist several works on spatial panel models, homogeneity of variance across time is assumed in these models. But with the progress of Bayesian techniques such a model wit