Accounting for non-exposure bias, self-selection, and heterogeneity in production technology: evidence from rice cultivation in Ghana

This study applied stochastic metafrontier whilst correcting for non-exposure and selection bias to assess the adoption of improved rice varieties on output and technical efficiency of Ghanaian households. Varietal awareness was estimated to account for non-exposure bias and adoption using treatment effect. The exposure and adoption rates of improved rice varieties were 82.5% and 67.2%. Adoption was influenced by rice projects, agricultural extension, higher yield motive, and irrigated production. Application of herbicides, fertilizer, seed, labour and farm size raised rice output amongst adopters. The difference in metafrontier technical efficiency of adopters (42.7%) and non-adopters (44.5%) was statistically insignificant, albeit adopters had higher metatechnology ratio (0.909) compared with nonadopters (0.785). Therefore, adopters applied the best production technology than nonadopters. Weeding twice with herbicides, managing plot water levels and agricultural extension raised the technical efficiency amongst adopters. This study recommends cultivation of improved rice varieties whilst improving technical efficiency.