An Extended Mathematical Theory of Social Learning and Information Cascades in Nigerian Agricultural Technology Adoption

Agricultural technology adoption in rural Nigeria, including improved seed varieties, precision fertiliser application, and conservation tillage practices, proceeds through social learning processes in which farmers update technology beliefs by observing neighbours' adoption decisions and outcomes, creating information cascades that can either accelerate beneficial technology diffusion or trap communities in suboptimal practices. This dissertation develops an extended mathematical theory of social learning and information cascades specifically adapted to the network structure and information environment of Nigerian rural farming communities, making three original theoretical contributions. The first contribution develops a Bayesian social learning model on a random network with the degree distribution estimated from social network surveys of 2,400 farmers in Kogi, Nasarawa, and Taraba States, and proves conditions for the emergence of rational information cascades that generate adoption lock-in independent of technology quality. The second contribution introduces an original concept of network-conditional learning efficiency that quantifies the information aggregation loss induced by network topology, proving that it is bounded below by a function of the network's spectral gap. The third contribution analyses the social planner's optimal seeding strategy for technology demonstration plots to maximise adoption diffusion speed subject to a budget constraint, deriving a tractable approximation algorithm with a proved performance guarantee. The theoretical framework is validated against a three-year agricultural technology panel dataset from the study communities. Keywords: social learning, information cascades, technology adoption, Bayesian learning, Nigerian agriculture.