Competitive reaction- and feedback effects based on VARX models of pooled store data

We apply a model that accommodates dynamic phenomena in demand- and reaction functions. The latter functions capture reactions to actions as well as to consequences of actions. We estimate a fixed effects VARX model with dynamic and interactive effects for multiple brands based on pooled time series and cross-sectional data for two product categories. The Impulse Response Analysis (IRA) results for one category (tuna) under different scenarios show that the inclusion/exclusion of competitive reaction- and feedback effects matters a lot, consistent with a high degree of competitive interaction in this market. We find that the role of cross-brand feedback effects is greater than the role of traditional competitive reaction effects. Intrafirm effects (internal decisions) also play an important role. In a decomposition study we show that the exclusion of these effects may either increase (up to 12%) or decrease (by as much as 50%) the net unit sales effect of a 20% price reduction. By contrast, in the second category (shampoo), where brands have distinct positions, the exclusion of these effects matters very little.