Revisiting estimation methods for spatial econometric interaction models

This article develops improved calculation techniques for estimating the spatial econometric interaction model of LeSage and Pace (2008) by maximum likelihood (MLE), Bayesian Markov Chain Monte Carlo (MCMC) two-stage least-squares (S2SLS). The refined estimation methods derive parameter estimates their standard errors exclusively from moment matrices with low dimensions. For computation these moments, we exploit efficiency gains linked to a matrix formulation model, which generalize make more flexible use exogenous variables. To improve MLE restructure Hessian quadratic term in function. We also based MCMC estimator same restructuring. Finally, S2SLS presented this is first one solves problem collinearity among instruments. Several benchmarks show that estimators scale very well large samples can be used estimate models 100 million flows just few minutes. In addition methods, presents new way define feasible space allows verify consistency minimal computational burden. All developments indicate extension traditional gravity has become an increasingly mature alternative should eventually considered modeling approach origin-destination flows.