The link between local institutional and market failures, rural poverty and environmental degradation suggests a win-win policy intervention: solve local ?constraints and achieve both poverty alleviation and environmental goals. However, designing such interventions is problematic since exposure to constraints is unobservable and responses can be heterogeneous. In this context we evaluate the ability of the world's largest land set-aside programme, the Sloping Lands Conversion Programme (SLCP) in China, to relax local constraints on off-farm labour markets and achieve these dual objectives. A farm household model in the presence of constraints is developed. This identi?es constrained and unconstrained households and predicts that the impact of the SLCP on off-farm labour supply will be larger for the former if constraints are relaxed. To test this, a novel empirical approach is employed which combines a switching regression with difference in differences. Applied to panel data, these features allow unobserved sample separation, into constrained and unconstrained households, and consistent estimation of the SLCP?s heterogeneous impact. Also identified is the impact on the probability of being constrained and the relative importance of constraints such as tenure security. We ?nd some mixed support for the win-win hypothesis in the case of the SLCP.