Advertising, Movies and Crime




Mass media such as television, movies and video games are often blamed for social vices such as smoking, drinking, declining test scores, reduced cognitive development that later impairs one's ability to participate in the labor market, participation in gangs, road rage, etc. Increased public health concern in the spate of escalating youth crimes and high school shootings has policy makers threatening federal sanctions on advertising of violent movies, unless the movie industry changes its current marketing practices. This study first tests if there is a causal link between consumption of violent movies and crime using market-level field data. We do so by combining a structural movie-demand-model and a reduced-form model of crime using movie demand as inputs. Contrary to the extant experimental literature but akin to Dahl and DellaVigna (2009), our aggregate national-level field data analysis demonstrates that consumption of violent movies actually decreases violent crimes. This is because violent movies draw prospective criminals to the movies and incapacitate them from other violent crime-generating social activities. However, when we conduct our analysis at the disaggregate market level, our findings are quite mixed. Even after controlling for the self-selection of criminals to the movies but consistent with previous laboratory-study predictions, some markets see an increase in violent crimes. Also consistent with the Dahl and DellaVigna (2009) study, in some markets incapacitation results in a decrease in violent crimes. There are also markets where despite their appetite for violent movie content, these markets do not see a statistically significant change in violent crimes in response to consumption of violent movies.

We show that our mixed evidence is not simply the result of aggregation bias suffered by national-level analysis, but because key institutional features of the movie industry not reflected in aggregate data limit causal inference and therefore the policy recommendations that can be generated from them. Specifically, movies get sequentially released across markets, hence consumers across different markets face heterogeneous choice-sets that also vary over time. Causal associations in aggregate data are identified via endogenous temporal variation in movie sales and exogenous variation in movie content. While being able to account for seasonality in movie releases at the national level, aggregate data fail to account for across-market movie rollout. Furthermore, once we control for endogenous market-specific decisions by movie executives (ex. movie spell, advertising spend, etc.), our disaggregate results suggest that much of the variation in crime stems from spatial heterogeneity the combination of heterogeneous preferences, heterogeneous choice sets and heterogeneous responsiveness in crime amongst markets all of which are unobservable in the aggregate data.
Next, we evaluate the effectiveness of a federally issued sanction on advertising (ex. the extreme case of a nationwide ban) of violent movies in deterring crime. We show that a ban on advertising results in a net reduction in demand for violent movies, leaving more criminals to commit crimes. Furthermore, demand for non-violent movies also rises, which brings in different types of consumers to the movies, who then fall prey to other non-violent crimes. A federal ban on advertising of violent media, therefore, results in both a net reduction in demand for movie studios and a net increase in crime (both violent and non-violent crimes), the very thing that such regulation was designed to deter.
Contact information:
Dr. S. Puntoni