A State-Space Approach to Capturing Market Dynamics for Frequently Purchased Products

• Activity Number: 355
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Eiji Motohashi*+ and Tomoyuki Higuchi
• Companies: The Graduate University for Advanced Studies and Institute of Statistical Mathematics
• Keywords: State-space approach ; Multinomial logit models ; Brand choice ; Marketing dynamics

Abstract

Marketers are often interested in how consumer preferences and/or promotion effects vary over time. Previous scholars have proposed many approaches to model such market dynamics in consumer brand choices. Parameters in a choice model may vary over time because of a variety of reasons such as change in market conditions. In this study, we develop a choice model that includes time-varying parameters (state variables) based on the state-space approach and show how we incorporate both invariant individual heterogeneity and variant marketing factors into the model. In empirical analysis, we use the particle filter method to estimate state variables and unknown parameters, and apply our methodology to scanner panel data in instant coffee category, which have been gathered at a super market in Japan. We find that the proposed model can obtain more appropriate inferences about brand choices than models that ignore time variation in parameters.