This article extends the conventional spatial autoregressive efficiency model by including firm characteristics that may impact efficiency. This extension allows performing the typical inference in spatial autoregressive models that involves the derivation of direct and indirect marginal effects, with the latter revealing the nature and magnitude of spatial spillovers. Furthermore, this study accounts for the endogeneity of the spatial autoregressive efficiency model using a lag spatial lag efficiency component, which makes inference to be performed in a long-run framework. The case study concerns specialized Dutch dairy farms observed over the period 2009–2016 and for which exact geographical coordinates of latitude and longitude are available. The results reveal that the efficiency scores are spatially dependent. The derived marginal effects further suggest that farmers’ long-run efficiency is driven by changes in both their own and their neighbors’ characteristics, highlighting the existence of motivation and learning domino effects between neighboring producers.