Abstract: We assess the extent and cost of misallocation in agriculture in less-developed countries comparing the analysis at the plot and farm levels. Using detailed data from Uganda, we show that the plot-level analysis leads to extremely large estimates of reallocation gains, even after adjusting for measurement error and unobserved heterogeneity. These results reflect two empirical limitations of the plot as unit of analysis: excess measurement error and near constant returns to scale production estimates. We find limited evidence of substantial measurement error at the farm level.
Keywords: plot, farm, misallocation, measurement error, agriculture, distortions.
JEL Classification: O4