![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
This study examines the change in crime observed in the weeks following the release (exogenous shocks to video game play) of top-selling video games between 2006 and 2011. We find that the release of violent (Mature-rated) video games is associated with an increase in overall crime in the weeks following release. Crime increases for both youth and adults following blockbuster Mature-rated releases, but the increase in crime among youth is approximately four times greater (8%) than the increase among adults (2%). Conversely, we find that the release of best-selling nonviolent (Everyone-rated) video games is not associated with a change in crime in the weeks directly following the release. Our results suggest that the release of violent video games increases crime in the United States, at least in the short-term, especially among the under-17 population for whom Mature-rated games are explicitly labeled as not “suitable.” Interestingly, our results are completely moderated in U.S. counties that forbid alcohol sales, which suggests that alcohol is a necessary channel through which exposure to violent video games contributes to crime.
Acknowldgement
We thank Larry Kenny, Kay Stice, Burcin Unel, and participants at the 2015 Conference on Empirical Legal Studies and 2015 Eastern Economic Association Annual Conference as well as workshop participants at the University of Florida for their valuable comments. The authors are grateful for the financial support of Emory University and the University of Florida.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
2 Essential Facts About the Computer and Video Game Industry – Published by The Entertainment Software Association.
3 Looking at top-selling game release dates allows us to capture the biggest “shocks.” We consider this setting to be a type of natural experiment which allows us to hold all else constant while only changing video game exposure. This methodology is powerful and is seen across disciplines. See for example Drago et al. (Citation2009) which uses an exogenous shock to prison sentences to examine the effect of incarceration on the likelihood of recidivism. Beerthuizen et al. (Citation2017) is the only paper we are familiar with that use the event study method in a setting similar to ours. Those authors observe a decrease in crime among Dutch youth in the weeks after the release of Grand Theft Auto 5.
4 We conduct sensitivity analysis in which we examine all release dates for games that had lifetime sales exceeding 1,000,000 units. The results are directionally and statistically identical (see section 4.3.4).
5 Wii Sports is a notable exception with relatively few units sold in the first few weeks but 80+ million units sold over the life of the game.
6 We validate this supposition – video game releases result in abnormally high interest in video games – by examining Google searching activity around the release of top-selling games. In untabulated tests, we find that the month a game is released corresponds to the highest search volume for that game and searching drops off precipitously immediately (beginning the following month). Thus, interest in a game appears to be heavily concentrated immediately upon its release. This gives us confidence that examining release events is appropriate.
7 24 events in our main sample, and 59 events in the expanded sample.
8 States are not required to report information for use in the NIBRS database.
9 We choose these age categories because the Entertainment Software Rating Board (ESRB) has established 17 as the cutoff age for Mature-rated games.
11 Not surprisingly, prior literature supports the notion that less parental guidance is perhaps the largest contributing factor to the higher levels of crime in urban areas (Glaeser & Sacerdote, Citation1999).
13 Both Cunningham et al. (Citation2016) and Dahl and DellaVigna (Citation2009) concede that there is a short-term arousal effect associated with violent media. However, they argue that in non-laboratory settings this effect is dominated by both incapacitation and catharsis.
14 Oddly, Jacob and Lefgren (Citation2003) also find that violent crimes increase when school is in session. They argue that the high concentration of children leads to this increase (more fighting).
15 The effects of incapacitation are also observable in a non-crime setting. Stinebrickner and Stinebrickner (Citation2008) find that when students cohabitate with other students who own video games, they engage in less studying activity – consistent with the idea that video games incapacitate them and keep them from engaging in other activities, such as studying.
16 See for a more detailed description of video game ratings.
17 Prior research finds that globally, different video game content rating regimes are generally uniform in classifying games at the particularly nonviolent and particularly violent ends of the spectrum. There is considerable discrepancy in ratings for games in the intermediate range of violence, however. Accordingly, we focus on games that most rating regimes and independent raters would classify as clearly violent or clearly nonviolent (Dogruel & Joeckel, Citation2013; Funk et al., Citation1999).
20 Many anecdotes and articles exists to support parents’ willingness to purchase M-rated games for their children. For example see http://www.bostonglobe.com/lifestyle/style/2013/01/08/parents-cringe-even-they-hand-over-credit-cards-for-call-duty-and-other-violent-video-games/BjldlLVWH2kJuIKY1fR7QP/story.html
23 States are not required to submit data to NIBRS. Thus, not all states are included in the database.
24 See for details and Appendix A for code definitions.
25 Vgchartz.com.
26 Sales data are incomplete. For example, Madden NFL 2007 has weekly sales reported for the Xbox 360 but not for the Playstation 2 even though twice as many games were sold for the Playstation (overall).
27 This is not to suggest that we expect crime spurred by video game violence only occurs in the first weeks after the release. However, the strength of our event study approach is in identifying changes in crime rates around date-related shocks. Given that high levels of video game play occur in the weeks after release (Bavelier & Davidson, Citation2013), our ability to identify crime associated with video games is strongest in the very short run after release.
28 Results are very similar when we do not use a count but use indicator variables instead (1 if one or more games has been released, 0 otherwise).
29 We obtain essentially the same results when using Newey-West standard errors using a lag of one day, as well as seven days.
30 Several types of crime do not change after the release of any type of game: Confidence (tabulated), Incest, Bribery, and Gambling. We note that these types of crimes are less prevalent in popular violent video games.
31 Percentages are calculated by raising to the coefficient value obtained from estimating our regression equation.
32 Calculated as .
33 This is particularly important for the Grant Theft Auto 4 release, given that it overlaps with the Mario Kart Wii release. See .
34 Out of 3,221 counties, 148 are ‘dry’.
35 This is the same regression specification used in the Jackknife procedure in section 4.2.
36 There are 3 Mature game releases for years 2006 and 2007.
37 These coefficients remain significant at 5% level or better.