Advertising, learning, and consumer choice in experience good markets

This dissertation is an empirical examination of the effects of advertising on consumer behavior in experience good markets. The introduction starts by reviewing both the theoretical and empirical economics literature on advertising, focusing on work that concerns qualitative effects of advertising. Effects are broadly categorized as to whether or not they involve informing consumers about a product's inherent characteristics.;The first chapter introduces empirical arguments regarding these different effects of advertising. We argue that in general, informative effects of advertising should relatively affect inexperienced users of a brand while non-informative or "prestige"-type effects should affect inexperienced and experienced consumers more equally. This identification argument is applied using household level panel data tracking purchases and advertising exposures for a new brand of Yogurt. Standard discrete choice models allowing for consumer heterogeneity indicate that these advertisements primarily affected the purchase behavior of inexperienced users, suggesting that the effects of these particular advertisements were primarily informative.;The second chapter begins a more structural analysis of the question by introducing a dynamic learning model of consumer behavior. This model distinguishes between observable characteristics of a product, known to consumers before purchase, and unobservable characteristics, not generally known to consumers prior to purchase. Consumers learn about these unobservable characteristics through both experimentation with the product and through informative advertising. We also allow for "prestige" effects of advertising by permitting a measure of a brand's advertising level to enter directly into a consumers utility function. Because current decisions affect future states of information in this model, optimal behavior involves dynamic decision-making by the consumer.;The third chapter uses the data of Chapter 1 to estimate this structural model. We find strong evidence that these consumers did learn through their experiences with the product. Additionally, we find a strong, significant informative effect of advertising and an insignificant prestige effect of advertising, supporting the conclusions of Chapter 1. The structural estimates are then used to evaluate welfare implications of an alternative advertising regulatory regime. Although we find significant value of the information contained in advertising, this benefit is outweighed by its costs.