Determinants of Agricultural Productivity Among Smallholder Farmers in Tanzania: Evidence from the National Panel Survey, 2021

Abstract

Agricultural productivity in Tanzania remains low, limiting food security and rural livelihoods. To inform evidence-based interventions, this study analyzed determinants of smallholder staple crop harvests using Wave 5 of the National Panel Survey with 237 households having complete socio-economic, farm input, and environmental data. A generalized linear model with a Gamma distribution and log link was used to account for the right-skewed, strictly positive nature of harvest data. Stepwise models were estimated separately for socio-economic, farm input, and environmental/institutional factors. The socio-economic model showed positive effects of asset ownership (β = 0.0125, p < 0.1) and negative effects of older household heads (β = –0.0159, p < 0.05; log pseudolikelihood = –1804.43, AIC = 15.28, BIC = –967.21, deviance = 355.92). The farm input model identified plot size as strongly positive (β = 0.216, p < 0.01) and irrigation as negative (β = –1.701, p < 0.05; log pseudolikelihood = –1802.23, AIC = 15.28, BIC = –895.19, deviance = 351.53), indicating slightly better fit than the socio-economic model. The environmental/institutional model performed poorly (log pseudolikelihood = –1840.04, AIC = 15.59, BIC = –830.58, deviance = 427.15). The full combined model, integrating all variables, provided the best overall fit (log pseudolikelihood = –1775.08, AIC = 15.15, BIC = –889.35, deviance = 297.22), with positive effects of plot size, asset ownership, household size, and organic fertilizer, and negative effects of irrigation, hired labor, and older household heads. Findings indicate that socio-economic and farm input factors jointly drive productivity, emphasizing interventions to strengthen household resources and improve input efficiency.