Does video game violence decrease violent crime?

Abstract

Public concerns about potential harms of violent video games are often based on a positive association between exposure to video game violence and aggression found in psychology experiments. Whether this aggressive tendency manifests itself in terms of actual criminal behavior seems unclear. In this paper, I utilize temporal variation in aggregated violence exposure using data containing weekly unit sales of the top 30 game titles in the United States from 2005-2014, and analyze the impact of this exposure on assaults committed by young males. I find that weeks with high intensely violent video game sales have a lower number of assaults. The effect becomes more noticeable during the 9 P.M. – 12 A.M. time window over the weekend. One million additional sales of intensely violent games reduces assaults by 2.15% during that time period. I interpret this finding as a product of time substitution, where people play games that they have recently bought for longer hours, which draws them away from risky outdoor activities. As a result, crime drops in response to the absence of proactive criminals on the streets. A highly pronounced impact among young males can be explained through self-selection: they disproportionately commit impulsive crimes, and also identify as frequent purchasers and avid players of intensely violent games. In further evidence of the time substitution hypothesis, crime rates are higher in the week preceding a new release, when gamers are likely to reduce their playing hours of old games in anticipation of beginning the new game. The presence of time substitution in this natural experiment reconciles with the opposite findings from the lab. Gamers cannot commit crimes spurred on by their heightened aggressive tendencies while simultaneously gaming.