Land degradation, in the form of soil erosion and nutrient depletion, threatens food security and the sustainability of agricultural production in many developing countries. Governments and development agencies have invested substantial resources in promoting soil conservation practices, in an effort to improve environmental conditions and reduce poverty. However, very limited rigorous empirical work has examined the economics of adopting soil conservation technology. This paper investigates the impact of stone bunds1 on crop production value per hectare in low and high rainfall areas of the Ethiopian highlands using cross-sectional data from more than 900 households having multiple plots per household. We use modified random effects models, stochastic dominance analysis (SDA) and matching methods to ensure robustness. The parametric regression and SDA estimates are based on matched observations obtained from nearest neighbor matching using propensity score estimates. This is important because conventional regression and SDA estimates are obtained without ensuring the existence of comparable conserved and non-conserved plots within the distribution of covariates. Here, we use matching methods, random effects and Mundlak’s approach to control for selection and endogeneity biases that may arise due to correlation of unobserved heterogeneity and observed explanatory variables. The three methods used herein consistently show that plots with stone bunds are more productive than those without such technologies in semi-arid areas but not in higher rainfall areas, apparently because the moistureconserving benefits of this technology are more beneficial in drier areas. This implies that the performance of stone bunds varies by agro-ecological type, suggesting a need for the design and implementation of appropriate site-specific technologies.