A Bayesian Hierarchical Model for Consumer Demand

Mary Beth Broadbent. In4mation Insights, Needham, MA

We estimate a fully Bayesian model of demand elasticities for a large number of consumer packaged goods produced by a specific manufacturer. The hierarchical nature of the model allows us to estimate elasticities even when the item is new or has limited price history -- an important consideration when roughly 70% of the product list turns over each calendar year. We present a summary of the results and recommend a number of pricing changes across the manufacturer's product portfolio.