US metropolitan markets differ in demographic composition, in economic indicators and in local and regional factors, all of which imply potentially different marketing phenomena such as consumer segment sizes, consumption patterns and responses to marketing actions. We investigate whether and how this heterogeneity explains variation in an interdependent system of consumer response sensitivities to the price, promotion and distribution elements of the marketing mix. Primarily, four questions of research interest are examined - (i) the nature of the relationship between different aggregate response sensitivities and the implications therein, (ii) the usefulness of aggregate area-wide macroeconomic indicators such as inflation, unemployment and poverty in the context of a marketing problem, (iii) the usefulness of incorporating geospatial information into the analysis, and (iv) the use of a moments based approach against making distributional assumptions in a maximum likelihood context. To address these questions, we estimate a distribution-free generalized spatial three stage least squares model that allows for asymmetric influences between neighboring market pairs. Beer category sales data across 49 major US metropolitan markets are analyzed. We find a pattern of strong inter-dependence among aggregate response sensitivities that substantive insight into potential consumer segments. Ignoring inter-dependence mischaracterizes response sensitivity in two-thirds of the markets sampled. Further, on a standalone basis, neither area-wide economic indicators nor geospatial information help the analysis, but in conjunction, they vastly improve model fit (by almost 40%), explained variance (by over twice), parameter significance and consequently, substantive implications.

KEYWORDS: Response sensitivity, simultaneous equation models, spatial analysis, economic indicators, instrumental variables