Asymmetric Advertising Response

Category: Marketing Seminar
When: 14 November 2017
, 12:15
 - 13:30
Where: Campus Westend, RuW 1.201
Speaker: Dr. ir. Maarten Gijsenberg (University of Groningen)

Julien Schmitt (ESCP Europe), Maarten J. Gijsenberg (University of Groningen), Jaap E. Wieringa (University of Groningen)

Managers frequently face the decision to intensify planned advertising actions and associated budgets in order to defend themselves against unexpected competition pressure. Conversely, and perhaps even simultaneously, they may be urged to cut back on advertising in order to reach short-term bottom-line objectives. The conundrum they face in doing so, is that they often do not have a precise idea of the respective impact on sales of these decisions. Indeed, while most advertising response models consider advertising increases and decreases as having symmetrical effects, we argue that this is not necessarily the case, and that effects can be asymmetric. Moreover, we acknowledge that effects for increases can even be different for first-time increases (first week of advertising) compared to subsequent increases (subsequent weeks of advertising). We shed light on this issue by adjusting (traditionally symmetrical) VAR time-series models to account for possibly asymmetric responses to advertising. Insights are based on the analysis of the weekly advertising decisions and sales outcomes over four years for nearly 250 consumer packaged goods brands from the UK, ranging from high-advertising, high-priced “premium mass” brands to low-advertising, low-priced “value niche” brands. Our results indicate that significant asymmetries exist in a) the size of immediate effects, b) the size of cumulative effects, and c) the speed with which cumulative effects occur. Patterns, in addition, are markedly diverse across the different types of brands. Finally, these asymmetries also show considerable long-term consequences for brands with regard to their market share evolution.


Keywords: Asymmetric, Advertising Models, Time Series