Economics & Finance / Economic Theory / Development Economics

90% of Africa's Farm Productivity Gap is Just Measurement Error?

Douglas Gollin, Christopher Udry

For years, economists believed that massive inefficiencies in African agriculture—like land being held by the wrong farmers—were the main reason for low productivity. A famous study using data from Malawi even claimed that simply reallocating land could triple total output. But what if that conclusion was built on a statistical illusion? In this video, we break down a groundbreaking 2019 paper by Douglas Gollin and Christopher Udry, which uses detailed plot-level data from Tanzania and Uganda to challenge the entire narrative. They found that up to 90% of the observed productivity dispersion across farms is actually due to measurement error and unobserved heterogeneity (like soil quality or local weather shocks), not true misallocation. The key insight? The same farmer manages multiple plots at the same time. If markets were the problem, he'd allocate resources inefficiently between his own plots. But he doesn't—so any variation between those plots must be noise. By cleaning out that noise, the real efficiency loss from misallocation turns out to be modest. This isn't just a technical debate. It has huge implications for policy: instead of pushing politically difficult land reforms, maybe we should focus on improving data quality, reducing agricultural risk, and spreading better farming techniques. Watch to see how the authors pulled off this clever identification strategy, and why it flips conventional wisdom on its head.