Purposeful Structural Model Design

Abstract

Structural models enable marketing researchers to evaluate counterfactual decisions such as pricing, market entry, and platform design, yet their credibility depends critically on choices made prior to estimation. In strategic settings, the mapping from primitives to outcomes can be locally flat, so distinct mechanisms generate similar behavior, while equilibrium responses attenuate policy effects and limit their detectability. This paper develops a framework linking identification geometry to counterfactual feasibility, showing that model complexity redistributes identifying information across competing mechanisms and that statistical insignificance can reflect equilibrium constraints rather than lack of data. The results imply that credible structural analysis requires aligning model design with the decision margin and the informational content of the data, rather than relying solely on estimation advances.

Keywords: Structural Models, Counterfactual Analysis, Model Design, Identification, Strategic Interaction, Detectability, Entry Games

JEL Classification: C18, C52, C57, M31