We revisit the long-standing empirical evidence of an inverse relationship between farm size and productivity using rich microdata from Uganda. We show that farm size is negatively related to yields (output per hectare), as commonly found in the literature, but positively related to farm productivity (a farm-specific component of total factor productivity). These conflicting results do not arise because of omitted variables such as land quality, measurement error in output or inputs, or specification issues. Instead, we reconcile the findings emphasizing the role of farm-specific distortions and returns to scale in traditional farm production. We exploit unique regional variation in land tenure regimes in Uganda in evaluating the role of farm-specific distortions. Our findings point to the limited value of yields (or land productivity) in establishing the farm size-productivity relationship. More generally, we demonstrate the limitation of using farm size in guiding policy applications.JEL classification: O4, O5, O11, O12, O13, E02, E13, Q15, Q18, C33, C55, D24.

This paper shows that using yields may not be informative of the relationship between farm size and productivity in the context of small-scale farming. This occurs because, in addition to productivity, yields pick up size-dependent market distortions and decreasing returns to scale. As a result, a positive relationship between farm productivity and land size may turn negative when using yields. We illustrate the empirical relevance of this issue with microdata from Uganda and show similar findings for Peru, Tanzania, and Bangladesh. In addition, we show that the dispersion in both measures of productivity across farms of similar size is so large that it renders farm size an ineffective indicator for policy targeting. Our findings stress the need to revisit the empirical evidence on the farm size-productivity relationship and its policy implications.