Structural Estimation of Expert Strategic Bias: The Case of Movie Reviewers

We develop the first structural estimation of reputational cheap-talk games using data on movie reviews released in the US between 2004 and 2013. We identify and estimate movies' priors, as well as movie reviewers' abilities and strategic biases. We find that reviewers adopt reporting strategies that are consistent with the predictions of the literature on reputational cheap-talk. The average conservatism bias for low prior movies lies between 8 and 11%, depending on the specifications of the model. The average conservatism bias for high prior movies ranges from 13 to 15%. More- over, we find a significant, albeit small, effect of the reputation of the reviewers on their strategies, indicating that incentives to manipulate demand in order to prevent reputation updating are present in this industry. Our estimation takes into account and quantifies potential con icts of interest that might arise when the movie reviewer belongs to the same media outlet as the film under review. Out-of-sample predictions confirm that movie reviewers do have reputational concerns.