A five-season field experiment with smallholder farmers in sub-Saharan Africa shows that new agricultural technologies are hard to adopt not because farmers lack information, but because such technologies require multiple interdependent decisions that take years of costly trial and error to get right.
Editor's note: You can listen to this podcast on Spotify, Apple Podcasts, or wherever else you get your podcasts.
Development economists have long puzzled over why smallholder farmers in sub-Saharan Africa are slow to adopt new agricultural technologies. A new paper argues that the standard explanations – lack of credit, lack of access to inputs – miss a bigger, underappreciated problem: adopting a new technology is rarely a single decision, but a web of interdependent choices that farmers must get right all at once. Karen Macours and Rachid Laajaj join us on VoxDevTalks to discuss their findings, drawn from a unique five-season field experiment with smallholder farmers.
Why agricultural productivity matters in Africa
The stakes for improving farm productivity in the region are enormous.
"Fifty percent of employment in sub-Saharan Africa is in the agricultural sector, that's up to two thirds of employment when you take the agro-food sector combined, so increasing profits in that sector is one of the most obvious ways to increase income at a large scale for large number of people." Macours
Because so many farmers are smallholders, raising profits in this sector is not just an income issue but a food security one. Agricultural productivity in the region has historically been low, and is now under mounting pressure from environmental degradation, climate change, and geopolitical shocks that drive up input prices.
The hidden complexity of farming decisions
Macours and Laajaj argue that existing models underestimate how complicated technology adoption really is.
"We try to put ourselves in the shoes of the farmers, and think of how many interdependent decisions they have to take. They have to decide tons of things, from which seeds they use, which fertilisers, which spacing, maybe they do some crop rotation." Laajaj
Seed choice, fertiliser type, spacing, crop rotation, and weather all interact with one another, so the best decision in one area depends on choices made in others. Adopting a single new practice can mean readjusting everything else a farmer already knows works. This is especially true for sustainable agriculture practices such as conservation agriculture, which combine no-tillage, crop rotation, and residue management – each requiring its own learning curve.
Designing a five-season farm trial
To unpack this complexity, the researchers moved beyond standard adoption studies. Farmers were randomly assigned small experimental plots, divided into six subplots of five by five metres, each testing different combinations of inputs under integrated soil fertility management. Crucially, the team followed farmers for five seasons, tracking not just what happened on the trial plots but how farmers adjusted their own plots in response. Treatment-group farmers, Macours notes, had the chance to "think a little bit like scientists", directly comparing outcomes under controlled conditions before applying what they learned elsewhere.
Profits fall before they rise
One of the study's most striking findings is what happens to farmers' profits once they start experimenting.
"In the first season, we clearly showed that, in fact, profits go down significantly." Macours
Despite this, farmers kept experimenting season after season, gradually reducing their losses until profits turned positive – for some. By the end of the five seasons, a subset of farmers had achieved positive profits from their new practices, while others had not, suggesting the practices being tested were not equally suited to everyone. Both farmers who found success and those who did not kept adjusting their approach, reflecting a genuine, ongoing process of trial and error rather than a single leap to a 'correct' technique.
How skilled farmers learn faster
The researchers tracked two dimensions of farmer learning: know-how about agricultural practices, and experimentation – how frequently farmers changed what they were doing.
"The high-skilled farmers clearly learn faster, both learn, but the high-skilled farmers have a steeper learning curve, and they're very quick at learning." Laajaj
High-skilled farmers experimented more and absorbed knowledge faster, but also made more mistakes along the way – such as reusing hybrid seeds across seasons, a practice that works with traditional seeds but not improved ones. As a result, they suffered greater losses early on before catching up. Low-skilled farmers, by contrast, adopted later but benefited from watching their higher-skilled peers first, allowing them to skip much of the costly trial-and-error phase.
Sharing the costs of experimentation
This dynamic points to a powerful spillover effect. Network data collected in the study shows that both low- and high-skilled farmers are more likely to say they are influenced by high-skilled farmers, evidence of a positive externality: the 'hero' experimenters bear the early costs of learning, and everyone around them benefits. Laajaj suggests this has implications well beyond the trial itself, including for debates about rural brain drain: if skilled farmers who migrate to cities for higher wages take these spillover benefits with them, rural communities may lose more than is captured by individual wage differentials.
Policies to support farmer learning
Given how costly experimentation is – and how essential it has become as farmers face a fast-changing climate – the researchers see a role for policy in lowering that cost.
"Is there a way of providing some kind of financial buffer, so that people have the space to experiment while they're not sacrificing the food they need to put on the table." Macours
Because the trial shows it can take several seasons for profits to recover, any such support would need to last more than a single season. Other options include demonstration plots that let more farmers observe experimentation directly, and citizen-science style initiatives that pool farmers' collective learning. Longer term, the researchers point to education systems that build farmers' capacity to learn from observation, expanding the pool of 'high-skilled' experimenters beyond a small few.
Reference
Laajaj, R, and K Macours (2026), "The complexity of multidimensional learning in agriculture," Econometrica, 94(2): 465–503.