The Shape of Advertising Response Functions Revisited: A Model of Dynamic Probabilistic Thresholds

Abstract

Prior work in marketing has suggested that advertising —levels beneath which there is essentially no sales response—are rarely encountered in practice. Because advertising policies settle into effective ranges through early trial and error, thresholds cannot be observed directly, and arguments for their existence must be based primarily on a "statistical footprint," that is, on relative fits of a range of model types. To detect possible threshold effects, we formulate a switching regression model with two "regimes," in only one of which advertising is effective. Mediating the switch between the two regimes is a logistic function of category-specific dynamic variables (e.g., order of entry, time in market, number of competitors) and advertising levels, nesting a variety of alternative formulations, among them both standard concave and S-shaped responses. A sequence of comparisons among parametrically related models strongly suggests: that threshold effects exist; that market share response to advertising is not necessarily globally concave; that superior fit cannot be attributed to model flexibility alone; and that dynamic, environmental, competitive, and brand-specific factors can influence advertising effectiveness. These effects are evident in two evolving durables categories (SUVs and minivans), although not in the one mature, nondurable category (liquid detergent) studied.advertising, econometric models, product management, switching regression, dynamic models

Optimal Dynamic Advertising Policy for New Products

“…Perhaps in the future researchers can use more sophisticated approaches to build advertising into the diffusion and explore what optimal policies those models yield. A good place to start on this is the Vakratsas et al (2004) paper. Finally, note that we did not focus on the introductory advertising in this research.…”

Goodwill and dynamic advertising strategies

“…This literature advocates the important role played by the shape of the advertising response function on dynamic advertising strategies (e.g. Mahajan and Muller, 1986;Mesak and Darrat, 1992;Sasieni, 1989;Vakratsas et al, 2004). In a seminal paper, Sasieni (1971) analytically shows for a monopoly case that if marginal returns of sales to advertising are not increasing, then there exists an "optimal" expenditure level that allows a firm to maintain its optimal sales level.…”

Editorial: Errors in the Variables, Unobserved Heterogeneity, and Other Ways of Hiding Statistical Error

“…For example, consider the theory that marketing activities have threshold effects. We should observe s-shaped advertising response functions (Little 1979, Vakratsas et al 2004. We should observe turning points in new product life cycles (Golder and Tellis 2004).…”