Estimating Differential Lag Effects for Multiple Media across Multiple Stores

Abstract

More than 30 years of study have provided ample evidence that advertising in one period can continue to influence sales performance in subsequent periods. However, these carryover, or lagged, effects of advertising have generally been studied (1) only for the total advertising budget or (2) using estimation techniques that implicitly assume that different advertising media have the same lag effect. We develop a model that estimates the influence on sales when different advertising media are allowed to have different lag structures. A methodology is presented and illustrated by an application to weekly data from three stores of a large national retailer. Of the two media studied, radio had longer lagged effects than did billboards. The results were consistent for all three stores. This research yields important insights for theory, as well as for the practice of advertising campaign planning and budget allocation.