Bayesian spatial econometrics: a software architecture

Abstract Bayesian approaches play an important role in the development of new spatial econometric methods, but are uncommon applied work. This is partly due to a lack accessible, flexible software for estimation models. Established probabilistic struggles with specifics econometrics, while classical implementations do not harness flexibility modelling. In this paper, I present layered, objected-oriented architecture that bridges gap. An implementation bsreg package allows quick and easy models, remaining maintainable extensible. demonstrate benefits approach using well-known dataset on cigarette demand. First, show posterior densities yield better insights into uncertainty non-linear Second, find earlier studies overestimate spillover effects distance-based connectivities scaling error, highlighting need tried tested software.