en fr

Impact of advertising on brand's market-shares in the automobile market: a multi-channel attraction model with competition and carryover effects

Impact de la publicité sur les parts de marché des marques du marché automobile : un modèle d'attraction multicanal avec prise en compte de la concurrence et des effets retards de la publicité

Abstract : This article presents a new approach to measure the impact of multi-channel advertising investments on brands' market shares in the main segment of the French automobile market. We propose a multi-channel attraction model with adstock, in order to take into account the advertising car-ryover effect and the competition. This model allows to distinguish between short term and long term effect of the advertising. As, from a mathematical point of view, a vector of market shares is a composition belonging to the simplex space, i.e. subject to positivity and summing up to one contraints, we take benefit from the compositional data analysis (CODA) literature to estimate properly this model. We show how to determine the carryover parameters for each channel (outdoor , press, radio and television) in a multivariate way. We consider several model specifications with more or less complexity (cross effects between brands), including Dirichlet models, and we compare them using goodness-of-fit and prediction accuracy measures. We explain how to built confidence and prediction ellipsoids in the space of market shares. The impact of each channel on market shares is measured in terms of direct and cross elasticities. We conclude that in this market, radio only has a contemporaneous impact whereas outdoor, press and television have a large decay effect. Moreover, the advertising elasticities vary across brands and channels, and can be negative. It also turns out that positive interactions do exist between certain brands for certain media. Keywords: Market response model, fully extended multiplicative competitive interaction model, carryover effect, adstock, Koyck model, compositional data analysis, automobile market, multi channel advertising.