Identifying and Estimating Brand Satiation Using Purchase Data: A Structural Hidden Markov Modeling Approach

Abstract

In product categories such as yogurt, cereal and candy, consumers are likely to be satiated after frequent consumption of the same brand, leading to variety-seeking and switching to other brands. Prior research has modeled satiation mostly using consumption and preference data, but most firms have access to only purchase data. Identifying satiation and estimating satiation effect using purchase data remain a significant challenge. We develop a Hidden Markov Model (HMM) based structural approach and identify and estimate satiation using scanner purchase data of yogurt brands. In this model, consumers temporarily stay in an unobserved satiation state. The results show that consumers may be occasionally satiated for a certain brand, and that the satiation probabilities differ significantly across brands. Our model explains and predicts consumer satiation and its effects better than benchmark models.