Constraints to Tanzanian Agricultural Development: Input Use in Households Under Non-Separability

This paper builds on the literature testing for labor market inefficiencies in developing countries. Empirical tests using a panel data survey from Tanzania first reject the homogeneity of family and hired labor, and then reject labor market separation or completeness. Further tests for the efficient allocation of manure among plots reject, revealing that agricultural households face considerable constraints in factor markets. All rejections, except hired harvest labor, are robust to the inclusion of household-specific effects, and control for heterogenous household preferences, and village-specific shocks. I incorporate high-resolution annual population estimates from the LandScan database, which uses satellite imagery to construct population estimates, and find that in areas with higher population density, less family labor is used and more hired labor is used.