We tested a theoretical model with the Marshallian inefficiency (H1) and threat of eviction (H2) hypotheses having opposite effects on land productivity on sharecropped plots. The model also assumes that kinship contracts may eliminate or reduce the Marshallian inefficiency (H3) and threat of eviction (H4) effects on land productivity. Our empirical findings were consistent with H2 and H4 being true. We found higher land productivity on sharecropped plots than on share tenants’ own plots and higher land productivity on sharecropped plots of nonkin than of kin tenants. The nature of the data allowed controlling for unobservable household characteristics through household fixed effects and for observable plot characteristics. Analyses with and without plot characteristics revealed that these findings were stronger with plot characteristics than without them. Based on the plausible assumption that observable plot characteristics are positively correlated with unobservable plot characteristics this strengthens our conclusion. The results are also supported by first-order stochastic dominance analysis. Sharecropped plots’ output value distribution unambiguously dominated the output value distribution from share tenants’ own plots. Nonkin sharecropped plots’ output value distribution also first-order stochastically dominated the output value distribution from kin sharecrop plots.