A robust approach to measure latent, time-varying equity in hierarchical branding structures

The literature suggests that brand equity can be split into two parts - an attribute-based equity and a non-attribute based one that captures consumer preferences beyond the utility offered by individual attributes. In addition to measuring attribute-based equity, firms deploying portfolios of products within complex branding structures often seek to measure the presence, distribution and evolution of these potentially heterogeneous non-attribute based unique branding associations - which we label ‘intangible equity’ – at each distinct layer of a product’s brand hierarchy. We develop and operationalize a robust and flexible Bayesian semiparametric model to first separate the attribute-based equity from intangible equity, to jointly estimate this multi-level intangible equity and to allow it to exhibit state-dependence using a random-walk prior. The model is empirically illustrated on syndicated US national beer sales data. We find significant, heterogeneous and temporally stable intangible equity presence across the brand hierarchy and highlight some substantive implications arising therein.