Asymmetric risk and cropping-system economic performance in on-farm trials in Zambia’s Eastern Province

Abstract Smallholder farmers’ cropping-system performance depends on production and price variability. We evaluated crop diversification through legume rotations and intercropping, combined with no tillage and crop residue retention, under land preparation using either manual hand hoes or animal traction. A risk–value model was used to account for production and price risk with downside and upside deviations weighted differently. Our analysis estimated partial second moments using 1,169 plot-level observations from on-farm trials with 77 farmers across eleven cropping seasons (2011–2012 to 2021–2022) in six communities in Zambia’s Eastern Province. These estimates describe treatment performance among participating trial farmers under the study protocol and extension support. Rotation reduced downside deviations in cropping-system yield under both land preparation methods; under manual land preparation this reduction was accompanied by a lower conditional mean, but under animal traction the conditional mean did not differ significantly from the control. For gross margin, rotation under manual land preparation reduced downside deviations without a significant change in the conditional mean. Intercropping increased cropping-system yield but did not significantly affect the conditional mean or partial second moments of gross margin. Under manual land preparation and moderate risk aversion, intercropping gross margin ranked below no-tillage sole maize when only downside deviations were penalised but above it when upside deviations were also valued. Using multi-season on-farm trial data with observed prices, we showed that separating downside from upside variability can change the economic ranking of crop diversification practices.