A Bayesian Belief Network to Infer Incentive Mechanisms to Reduce Antibiotic Use in Livestock Production

Efficient policy intervention to reduce antibiotic use in livestock production requires knowledge about the rationale underlying antibiotic usage. Animal health status and management quality are considered the two most important factors that influence farmers’ decision-making concerning antibiotic use. Information on these two factors is therefore crucial in designing incentive mechanisms. In this paper, a Bayesian belief network (BBN) is built to represent the knowledge on how these factors can directly and indirectly determine antibiotic use and the possible impact on economic incentives. Since both factors are not directly observable (i.e. latent), they are inferred from measurable variables (i.e. manifest variables) which are influenced by these factors. Using farm accounting data and registration data on antibiotic use and veterinary services in specialized finisher pig production farms, a confirmatory factor analysis was carried out to construct these factors. The BBN is then parameterized through regression analysis on the constructed factors and manifest variables. Using the BBN, possible incentive mechanisms through prices and management training are discussed.