Day 3: Research Session – Agriculture

Professor Kelsey Jack opened the session with, ‘Seasonal liquidity constraints and agricultural productivity: Evidence from Zambia’, which used a randomisation design to study how loaning maize to rural Zambian households affected consumption during Zambia’s hungry season, and found strong effects on consumption smoothing.

Next, Professor Jeremy Magruder, who has been working with Malawi’s Ministry of Agriculture and Food Security, presented his research using network theory to inform the design of agricultural extension programmes, with the intention of increasing the spread of knowledge through trainee farmers’ social networks.

The precise training used by Professor Magruder’s team encouraged Malawian farmers to use ridged fields, which improve rainfall collection and reduce fertilizer runoff. In control villages, village chiefs selected farmers to be trained in ridging- and often selecting farmers who were not well-connected to other farmers in the village social network. In the first set of treatment villages, the researchers instead selected for training the two farmers with the most social network connections to other farmers in the village; in the second set of treatment villages, training was provided to two farmers living in the densest part of the village.

This design reflected “threshold diffusion theory”, which states that people (in this case, farmers) adopt a new technology when enough (N) of their contacts have adopted that same technology. The theory can be interpreted as recommending prioritising the training of farmers with the most connections to other farmers, to optimise diffusion.

The study’s results found that, indeed, diffusion depended on the social connectivity of the trainee farmer (and, less so on the population density of the area around their house).

Finally, Lorenzo Casaburi presented experimental evidence from Kenya, regarding possible improvements to agricultural insurance models. Despite exposure to large, and harmful, risks, smallholder farmers around the world generally show low demand for insurance; to remedy this, he suggested, insurance products should be better aligned with farmers’ low liquidity, the seasonal variation in their need for expenditure, their present bias, and their low trust in insurers.

His research trialled three insurance models: In the first, farmers were offered insurance for an upfront payment at a fair (zero expected profit) price. In the second, farmers were offered insurance for an upfront payment at only 70% this fair price (a 30% discount). In treatments three and four, farmers were offered insurance ‘on credit’, their insurance payment being deducted from their final harvest revenues (group four was offered an additional cash payment equal to the discount paid to group 2). The first two (‘upfront payment’) treatments saw 13% and 33% take-up respectively; by dramatic contrast, the final (‘deferred payment’) groups saw 76% and 88% takeups.

To take these findings into policy, contract enforcement, and proper savings mechanisms for insurers, were recommended.

By Sally Murray, Country Economist, IGC Rwanda and Anne Laski, Country Economist, IGC Tanzania