Censorship of U.S. Movies in China (extended version), with Tim Groseclose, and Tyler Cowen

In this paper we analyze the extent to which the Chinese government bans U.S. movies.  A challenge for our analysis is that, just because a movie was not shown in China, that does not mean it was banned. Instead, the reason might have been that the movie was not economically viable—that is, no Chinese distributors thought they could make money from it.  By introducing a structural econometric model and by observing the revenue of U.S. movies that were actually shown in China, we can estimate, for any given U.S. movie, the probability that it would be economically viable.  In turn, this allows us to estimate, for any movie that was not shown, the probability that it was banned.  Our sample contains all U.S. movies that were released between 1994 and 2019 and generated U.S. sales of at least $10 million.  According to our estimates, if a U.S. movie has characteristics similar to the median movie in our sample, then the probability is approximately 0.91 that the Chinese government will ban it.  If a U.S. movie has characteristics similar to the median blockbuster movie in our sample—where “blockbuster” means generating at least $170 million in U.S. sales—then the probability is approximately 0.32 that the Chinese government will ban it.  During our sample period, approximately 28 percent of the movies shown in China were produced by U.S. studios.  We estimate, however, that if the Chinese government had not banned any U.S. movies then the latter number would have risen to about 68 percent, and Chinese sales of U.S. movies would have increased from approximately $22.6 billion to $45.1 billion.  In addition, we estimate that: (i) approximately 9.1 movies in our sample were banned because they contained lgbtq themes, (ii) 44.9 were banned because they contained occult themes, and (iii) 98.9 were banned because they contained controversial adult content (i.e. they were rated R in the U.S.).

The newest version of this paper can be found here. Data and replication code will be available later once the paper gets published.