Volumetric Demand and Set Size Variation

Abstract

To improve product planning, supply chain, and pricing decisions for packaged goods, brand managers must understand drivers of both overall category ("primary") brand-specific ("secondary") demand. A key reason for this is that their decision space can involve adding (or removing) not just one but several products to their current assortment. That is, they must consider not only composition, but assortment size. Although state-of-the-art Multiple Discrete Continuous Models (MDCM) can explain simultaneous demand for multiple varieties of products, they can unwittingly encode assumptions that hinder accurate demand forecasting across assortments of different sizes. Whereas classic MDCMs impose (in the absence of binding budget constraints) a monotonically increasing relationship between category demand and assortment size, empirical and behavioral research suggests that smaller assortments can often yield equal or higher sales; this in turn suggests a potentially positive effect of set size on the baseline utility of the outside good.
To that end, we develop a new model that retains the brand-level fidelity of MDCMs but enables a flexible relationship between assortment size and primary demand. Two large-scale choice experiments in disparate categories (chocolate bars and air fresheners) demonstrate the proposed model's ability to predict demand for market-like scenarios, while analogous MDCMs over-predict primary demand by 40%-80%.
Moreover, the proposed model is computationally tractable using standard Bayesian machinery, allowing scalable inference for real-world category management.

Keywords: Choice Models, Demand Analysis, Volumetric Demand, Multiple Discrete Continuous Models, Bayesian Estimation

JEL Classification: M3, C8, C9