A brand choice model that incorporates both reference and observed prices is proposed for frequently purchased products. The model is composed of a probability-of-purchase component and a reference-price-formation component. Empirical testing of the model using coffee UPC scanner panel data demonstrates that for two of the three brands, the model predicts probability of purchase better than do standard demand models that utilize only current observed brand prices.

Author notes

Russell S. Winer is Associate Professor, Owen Graduate School of Management, Vanderbilt University, Nashville, TN 37203. This work was supported by the Faculty Research Fund of the Graduate School of Business, Columbia University.