Spatial dependence of production choice: application of Bayesian Spatial Autoregressive Probit Model on smallholder rubber producers

In most smallholder agricultural production activities, farmers make choices about the types of output that they sell. This choice is influenced by various market and non-market factors, including the decisions of other farmers in the neighbourhood. Taking a sample of smallholder rubber farmers from Sri Lanka, this research is conducted to study their output choice with the special interest in measuring the impact of neighbours. We estimate a Spatial Autoregressive Probit Model using recently developed Bayesian econometric techniques. Results show that while social and physical capital and transaction costs have positive impacts on the choice of sheet rubber production, education and scale of production have negative impacts. A strong spatial relationship is found, and the farmers' choice is influenced by their neighbours. There are considerable amounts of spatial spill over effects, especially with respect to physical capital. Our findings reveal the possibility of central processing to overcome resource limitations, significant reductions in extension efforts in promoting good manufacturing practices by taking stock of the 'neighbourhood' effect present in farmer choices.