We investigate the unique contractual arrangement between a large Ethiopian sugar factory and its adjacent outgrower associations. The only significant difference between the sugarcane production on the factory-operated sugarcane plantation and on the outgrower-operated plots is the remuneration system and thus, the incentives to the workers. We compare the productivity of the factory-operated plantation with the outgrower-operated plots based on a new cross-sectional plot-level data set that includes all plots that are operated by the sugar factory and its adjacent outgrower associations. As sugar-cane production depends on various exogenous factors that are measured as categorical variables (e.g. soil type, cane variety, etc.), we estimate the production function by a nonparametric kernel regression method that takes into account both continuous and categorical explanatory variables without assuming a functional form and without imposing restrictions on interactions between the explanatory variables. In order to obtain meaningful productivity measures, we impose monotonicity in input quantities using the constrained weighted bootstrapping (CWB) method. Our results show that outgrower-operated plots have−ceteris paribus−a statistically and economically significantly higher productivity than factory-operated plots, which can be explained by outgrowers having stronger incentives to put more effort into their work than the employees of the sugar factory.