Seminar

Structural Estimation of Expert Bias: the Case of Movie Reviews

Nicolas Dupuis (Toulouse School of Economics)

June 10, 2013, 12:30–13:30

Room MS003

Applied Micro Workshop

Abstract

We estimate the strategic incentives of experts to send biased messages using movie reviews released in the US between 2004 and 2013. We find results consistent with theoretical predictions: when driven by reputational concerns, less skilled movie critics bias their messages towards the prior of the market, thus behaving as trend followers; when consumers only observe an aggregate signal on the movie quality to assess the ability of the experts, critics have the additional incentive to garble this aggregate signal by discouraging demand and act as gurus. In order to identify experts' individual abilities and biases, we develop an original structural estimation which relies on equilibrium strategies played in the theoretical model. This allows us to compare reputational cheap-talk models with and without noisy realization of the state of the world, and to assess whether experts play in autarkic or global markets. We finally provide conterfactuals to identify the effect of simulatenous competition in expert reviews. Joint with Fanny Camara