Market response models for frequently purchased branded goods : a comparative study

Econometric research in marketing, now entering its third decade, has generated a large number of diverse models and estimation methods. While this research has shown considerable potential in reducing the uncertainty about the nature of market response, it provides few guidelines about which models and estimation methods should be used by marketing managers. This dissertation compares a representative number of econometric models in order to provide guidance about the choice between models. The study restricts itself to comparing aggregate market response models for frequently purchased branded goods and concentrates on the application areas of market response analysis and short term forecasting. Data from three New Zealand markets are used in the comparison. The literature review critically examines the development of market response models over the last two decades and arrives at a representative set of models to be included in the methodology. The methodology focuses on two aspects of modelling market response: the comparison of basic market response specifications and the comparison of models of carryover effects. The central issue is whether the more sophisticated models and estimators are better than the simpler ones. The results lead to the conclusion that there is little justification for using more sophisticated market response models either for market response analysis or short term forecasting. However, there is strong support for using the more sophisticated simultaneous equation estimation methods when analyzing market response. The general conclusion of the choice between carryover effect models is that, among the models tested in this study there does not appear to be one correct model for all brands. Secondly, the more sophisticated nonlinear estimation procedures are preferred to estimate the more complex models. While some of these conclusions are consistent with previous studies some are contradictory highlighting the need for further research to confirm, refine or reject these conclusions. The study's contribution to marketing science is that it is one of the first to systematically compare market response models and their estimators using a broad set of criteria and data from a number of markets.