Two Essays on Marketing Communication Strategy

The dissertation consists of two essays exploring how user-generated and firm-controlled marketing communication strategies influence marketing performance. The first essay, titled "Sales Impact of Reviews with Verified and Non-Verified Purchases," examines 1) whether the impacts of VPR and non-VPR change over a book's lifespan and 2) whether such time-varying impacts are contingent upon book authors' reputation as bestselling authors. I estimate a Bayesian dynamic linear model using 74,743 online reviews of 304 books. The estimation accounts for the endogeneity in the effect of online review measures using instrumental variables. The results show that VPR valence has a more pronounced overall impact on book sales rank than non-VPR valence for books published by bestselling authors. More importantly, the difference between the impacts of VPR and non-VPR valence is more significant when the book is new, diminishing later in the book's lifespan. In contrast, for books published by non-bestselling authors, the impacts of both VPR and non-VPR valence are not significantly different. These findings suggest that publishers and booksellers should emphasize VPR early on for books by bestselling authors and that platforms should consider the impact of VPR across time and brand strength in review management. The second essay, titled "Bias in Voluntary Disclosure of Advertising Expenses: Consequences and Remedies," investigates the extent to which voluntary disclosure of advertising expenses leads to biases in estimating advertising effectiveness and, more importantly, a solution to decrease such biases. Given that many firms stopped disclosing advertising information in the annual financial statement since the implementation of SEC's Financial Reporting Release No. 44, most research on advertising use roughly one-third of public firms that voluntarily disclose advertising expense to examine advertising effectiveness. Hence, there is a possibility that nondisclosed advertising expenses may systematically bias the results reported in previous studies. In this essay, I compare the firms with and without advertising disclosure on Compustat and find that these firms are significantly different in terms of firm size, financial leverage, profitability, and industry. Demonstrating the systematic bias from these differences, I propose using Automatic Machine Learning techniques to predict nondisclosed Compustat advertising with Kantar Ad$pender and other Compustat variables. After generating the predicted Compustat advertising, the authors use firms with voluntarily disclosed Compustat advertising and firms with predicted Compustat advertising and replicate previous research on advertising elasticity, advertising effectiveness on firm value, the effect of financial leverage on advertising, and the excess returns of portfolios classified by advertising intensity. The replications show that compared to the firms with voluntarily disclosed advertising, the firms with predicted Compustat advertising enjoy significantly higher returns of advertising on sales and firm value while facing higher systematic risk related to advertising expenses and a more negative effect of financial leverage on advertising. The results indicate that firms strategically choose to disclose or keep their advertising information to maintain their competitive advantages in product market and financial market.