Bayesian spatial probit estimation: a primer and an application to HYV rice adoption

Increasingly, spatial econometric methods are becoming part of the standard toolkit of applied researchers in agricultural, environmental and development economics. Nonetheless, applications in discrete‐choice settings remain few and despite its appeal, applications of the Bayesian paradigm in these settings are still fewer. We provide a primer to the Bayesian spatial probit with the objective of making accessible to non‐users a class of iterative estimation methods that have become fairly routine in Bayesian circles, offer an extremely powerful addition to applied researchers toolkits, and are essential in Bayesian implementation of spatial econometric models. We demonstrate the methods and apply them to estimate the ‘neighbourhood effect’ in high‐yielding variety (HYV) adoption among Bangladeshi rice producers. We estimate the strength of this relationship using a standard, spatial probit model and compare the policy conclusions with and without the neighbourhood effect included.