The Efficiency of Trading in Social Networks: Experimental Measures from India

This paper studies whether trading in social networks allocates a new technology efficiently. I use a field experiment in Odisha India to compare decentralized trade of a new technology through networks with an approach where demand was revealed via door-to-door sales. While 84% of farmers are expected to gain from the technology, only 7% adopt in networks. Conversely, 40% of farmers adopt when demand is revealed in door-to-door sales. Using variation across the sample in estimated gains in revenue, I show that 63% of the gains from door-to-door sales are lost with decentralized trade through networks. Frictions preventing interactions between farmers from different social groups offer an explanation for the results. Sub-caste and surname association with suppliers are strong predictors of adoption in networks, but have no effect in door-to-door sales.