Abstract Low-input production systems aim at mitigating agriculture's environmental impact with a lower level of chemical inputs. However, comparing the performance of low-input systems to conventional ones, particularly in terms of productivity and yield, is challenging due to selection bias. First, we often lack observational data on low-input systems. Then, when available, the comparison between the two production systems is challenging due to potential endogeneity in input use and selection bias. To effectively develop policies promoting the adoption of low-input systems and assess their impact, for example, on pesticide use and yields, it is crucial to employ an econometric framework that addresses these issues. This article proposes an endogenous switching approach combined with control functions to tackle selection bias and input endogeneity simultaneously. Using unbalanced panel data on Swiss wheat production, which includes both low-input and conventional systems, our framework allows us to analyze the differentiated role of inputs as well as their price elasticity for both conventional and low-input farming systems.