A Flexible Method for Demand Forecasting with Structural Decomposition

• Activity Number: 604 - Bayesian Inference in Discrete Choice Analysis of Consumer Behavior
• Type: Topic Contributed
• Date/Time: Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Mingyu Joo* and Chul Kim and Dongsoo Kim
• Companies: UC Riverside and Baruch College (CUNY) and Ohio State University
• Keywords: Demand Forecasting; Counterfactual; Machine Learning

Abstract

Demand forecasting is a key to optimize future marketing variables. Although existing time-series forecasting methods present reasonably high face validity, the model parameter estimates in such models may embed confounds from environmental factors and trends. Therefore, such models may not clearly distinguish between consumers’ responsiveness to firms’ marketing actions from extraneous variables. This paper proposes a simple modeling framework to decompose consumer responses and environmental trends using a theoretical demand function derived from an Engel curve. We aim to show that the model parameters of the demand function allows counterfactual predictions of treatment effects, and separate descriptive predictions of environmental factors can be merged to improve forecasting accuracy.