GENERIC ADVERTISING AND THE STRUCTURAL HETEROGENEITY HYPOTHESIS

This paper adduces and tests the hypothesis that generic advertising r esponses are dynamic, i.e., are subject to change over time due to cha nges in target audiences, managerial expertise, copy quality or other time-related factors. Specifically, the authors consider the structura l heterogeneity hypothesis from the perspective of three alternative e conometric models that permit random and systematic time-varying respo nse: the Prescott-Cooley model, the return-to-normality models, and th e stochastic-trend model. A distinguishing characteristic of the model s is the presence or absence of heteroscedasticity. Based on pretests, which failed to detect heteroscedasticity, a modified version of the stochastic-trend model is selected for hypothesis testing. Results bas ed on data of the first 15 years of the Ontario fluid milk campaign su ggest advertising responses are dynamic. Estimated advertising elastic ities decline more or less monotonically over this sample period, from a high of 0.020-0.031 in the initial 1973-74 theme period, to a low o f 0.0004-0.009 in the final 1986-87 theme period. The apparent declini ng effectiveness of the Ontario fluid milk campaign is consistent with wearout theory, and suggests that program managers may want to reasse s marketing strategies to identify possible ways to improve performanc e. Given the importance of advertising elasticities in normative decis ion models and the growing evidence of structural heterogeneity, model s that permit parameters to change over time should provide an improve d basis for program assessment and resource allocation.