This study uses cross-sectional data of 342 small-scale lowland rice farmers in Northern Region of Ghana to analyze the impact of the adoption decision of bund construction and seed dibbling on net returns, input demand and output supply. Matching was conducted based on Mahalanobis distance combined with propensity score. Balancing tests by checking the mean standardized absolute bias in the matched sample were conducted as well as sensitivity analysis to check for hidden bias due to unobservable selection. The empirical results of impact assessment using propensity score matching controlling for self-selection bias suggest that input demand is significantly higher for adopters of bunds, but not statistically different for adopters and non-adopters of dibbling seed. However, output supply and net returns were not found to be statistically different for adopters and non-adopters of bunds. Adopters of dibbling were found to have higher output supply while no statistically significant difference was found for net returns of adopters and non-adopters of dibbling. The results were found to be relative insensitive to hidden bias.