The cost of favoritism in network-based markets

This paper presents evidence showing that favoritism of close peers severely limits the ability of social networks to allocate a new agricultural technology. I use a three-year field experiment where a new and profitable seed variety was first delivered to a random subset of farmers in 82 Indian villages. The diffusion between farmers is then compared with demand that was revealed experimentally one year later. I find a large gap between adoption from neighbors and demand and this gap is largest for farmers that are the most socially distant from the original seed recipients. Applying an estimate of the causal impact of the technology to this adoption gap shows the costliness of caste-based favoritism. Specifically, the cost to the average farmer over the three-year period where diffusion was tracked is equivalent to about 20 days of casual work.