Essays on online reviews

Abstract

The first chapter examines how hotel reviews affect the consumer's expected utility based on the reduced-form logit model and the random coefficient model, respectively. By using the data from Alitrip.com, we have four findings. First, the volume of reviews plays a crucial role in determining the impact of the room price on the expected utility. More reviews can alleviate the negative marginal effect of price on the utility and mitigate the substitution effect caused by the increase in the room price. Second, sensitivities to the average rating of reviews vary across heterogeneous consumers based on results of the random coefficient model. Third, cross-price elasticity exhibits larger for hotels with similar hotel characteristics, including the average rating and the volume of reviews. Finally, the results of counterfactual experiments show that the average rating of reviews is the main driver for the dispersity of the predicted market shares of hotels, comparing to the volume of hotel reviews.

The second chapter presents a rational expectation model of review manipulation to investigate the effect of review fraud in the motion picture industry. Using the data of Douban.com, we measure review fraud of movies by examining the difference in the ratings of reviews between the opening week and off-theater period. Our model provides testable implications for empirical study. We find that movies with high quality are less likely to commit review frauds. Moreover, promotional reviews do not help to improve box office performance. Finally, it is costly for the producer to post a fake review and the cost of posting a fake review depends on movie characteristics.

The third chapter uses textual analysis approach to study online movie reviews. In particular, we identify the aspects of reviewer profile that can lead to the difference in review texts, movie ratings, and therefore the box office revenues. Using data collected from Douban.com, we find that the higher quality of a movie, the less discrepancy of the review text over time. This, in turn, leads to better revenue performance. We also find reviewers with popularity and passion on art are more critical on ratings. Moreover, movies with film stars, award nomination, and less market competition attract more active reviewers to movies.