

We find occurrences of threshold-related content in 83 % of food and beverage advertisements for children and in 71 % of advertisements for general audiences. To examine the use of thresholdrelated content in practice, we study a sample of 370 television advertisements. We show that the resulting demand functions give firms incentive to provide threshold-related information. ” To understand the economic rationale for this phenomenon, we develop a theory of endogenous preferences in which information about threshold payoffs (which we interpret as being present over the course of human evolutionary history) induces sudden shifts in demand.

In particular, much of the imagery and repetitive thematic content seen in advertisements seem “psychological ” in nature, as opposed to “informative. I exemplify my approach to assessing the quality of mechanistic evidence in economics with an analysis of two models of rational addiction.Ĭertain aspects of advertising–especially on television–are not easily explained with conventional economic models. Still, claiming the actuality of the represented mechanisms requires establishing that crucial assumptions of these models are descriptively adequate. I show that economists mainly rely on mathematical models to represent possible mechanisms (i.e., mechanisms that could produce a phenomenon of interest). Furthermore, I discuss the differences between biological and social mechanisms and mechanistic evidence across the disciplines. I show that mechanistic evidence allows for discriminating econometric models representing genuine causal relations from accidental dependencies in data. Both studies apply plausible estimation techniques but report inconsistent results. I analyze a recent econometric controversy regarding the tax elasticity of cigarette consumption and smoking intensity. I argue that the program of evidential pluralism, which originated in the context of medicine and encapsulates to the normative reading of the Russo-Williamson Thesis that causal claims need the support of both difference-making and mechanistic evidence, offers a ground for resolving empirical disagreements. The fragility of econometric results undermines making conclusive inferences from the empirical literature. The results of econometric modeling are fragile in the sense that minor changes in estimation techniques or sample can lead to statistical models that support inconsistent causal hypotheses.
