A Scanner-Based Proxy for Manufacturer's Coupons in Brand Choice Models

A potentially serious threat to the validity of estimates of predictor variable impact in logistic models of brand choice is the omission of important explanatory variables from the empirical analysis. An obvious omission from most previous models has been manufacturer’s coupons, which are time consuming, expensive, and often difficult to measure. Further, the measures are lacking from most historical databases used to test consumer brand choice models. This paper proposes a proxy measure for manufacturer’s coupons which is readily calibrated from available scanner panel data. The empirical results show that the proposed proxy measure empirically exhibits the expected sign and is statistically significant in the case of the IRI instant coffee data. Further. the results show that the misspecification error in previous models from the omission of manufacturer’s coupons is not serious for the instant coffee data.