AN EMPIRICAL-INVESTIGATION OF THE ADVERTISING SPENDING DECISIONS OF AMULTIPRODUCT RETAILER

In a multiproduct retailing context an approach for testing marketing mix variables is presented that should be of interest to retailers and researchers alike. Traditional model building for decision evaluation purposes suffers from a serious drawback in that the underlying model is assumed to be known. The present approach incorporates the search for the ''true'' model as an intrinsic part of the method. Based on up to date modeling perspectives, the approach can be divided into three steps: (1) the use of multiple time series analysis to assist in mode l specification; (2) the choice of functional form, model estimation a nd testing; and (3) the use of statistical and simulation methods to t est decision making hypotheses. The approach is applied to three years of a western wear retailer's weekly sales and advertising data; multi ple time series analysis is used to specify a simultaneous equation mo del that captures demand interdependencies. After subjecting the model to several stability tests, it is then used as a basis for testing se veral interesting hypotheses concerning the retailer's advertising dec ision making behavior: The key insight for multiproduct retailers is t hat advertising budget allocation decisions across products and media are probably more important than decisions affecting the level of the total budget since the former have a greater impact on profit contribu tion.