Essays on the Economics of the Motion Picture Industry

Abstract

My dissertation consists of three chapters that address important questions in the U.S. motion picture industry. In Chapter 1, I limit the attention to the theatrical market and use a linear model with fixed effects to examine the impact of online word of mouth (WOM) on box office revenue. The online word-of-mouth dataset is extracted from Internet Movie Database (IMDB), one of the most popular and informative movie websites in the U.S. Chapter 1 contributes to the literature by using basic natural language processing (NLP) techniques to construct text features including review polarity, subjectivity, readability, and similarity to the product description in a manner which has barely been discussed before. The evidence shows that 1) volume and valence of word-of-mouth communication in the last period are positively associated with the current movie sales; 2) reviews with extreme rating scores, no matter whether they are positive or negative, can attract attention and increase the movie demand; 3) the audience does read the reviews instead of relying only on summary statistics and subjective reviews with rich content, even at the cost of low readability, can potentially boost box office revenue; 4) the disadvantage of the insufficient marketing budgets for limited releases can be rectified through the impact of proper online word-of-mouth communication. Chapter 2 expands the scope to include both theatrical and home video markets in the study. The home video market generates more gross revenue than the theatrical market but has received surprisingly less attention from economic scholars. Using market-level data, I first conduct demand estimation separately for the two markets using logit and one-level nested logit models. This step allows me to quantify movie qualities, consumers' utility decay rates, seasonality in demand, and market expansion effect in order to understand how these two markets operate differently. Next, I pool all the sample movies from the two markets together and employ a two-level nested logit model to quantify consumers' substitution patterns between the two different viewership platforms. The results validate that the two-level nest structure is consistent with the maximization of a random utility function (McFadden 1978) and consumers do distinguish between watching movies in theaters and on home video. Chapter 3 builds upon the demand estimation results from Chapter 2 and studies the optimal time to release a movie under the sequential distribution setup. Movie distributors use what is known as the windowing strategy --- releasing a film in different venues over discrete periods of exclusivity to maximize consumption over the lifetime of a property. The trend toward shorter time lags in theatrical releases has caused controversy in the U.S. motion picture industry, necessitating the demystification of how distributors optimize release schedules for sequential distribution. I model distributors' windowing strategy as a one-shot sequential-move game with incomplete information. I follow the method of pseudo-backward induction proposed by Einav (2009) and further estimate the weekly cost of distribution and distributors' weighting coefficient on the two windows while taking the demand estimates as given. Eventually, by conducting counterfactual analyses in a two-player game, I find that higher cost of theatrical distribution, lower weights on the theatrical window, poorer movie quality with weak opening performance, or faster theatrical utility decay can all potentially rationalize the practice of shrinking theatrical window length.