The impact of early DVD release on box office revenues

Abstract

In recent years, film studios have accelerated the release of films on DVDs, prompting theatre owners to express concerns about the possible negative impact of the early release of DVDs on a film's box office revenues. This study examines the impact of releasing a film on DVD before the film ends its theatrical run. Based on a sample of the top 50 grossing films in the USA over the period 2001 to 2005, our results indicate that roughly a quarter of the films appear on DVD before exiting the theatres, and that studios release these films on DVD after they have been in the theatres for an average of 20 weeks. By delaying the DVD release until the twentieth week, studios minimize the impact on box office revenues, as the average films take in 95% of its total box office growth by the seventh week in distribution. Regression results indicate that an early DVD release reduces box office revenues by 46%, or $303 989, which represents an average of 0.21% of a film's total box office revenue.

Notes

¹For example, see Waterman and Lee (2002).

²The study is available on the MPAA's website at http://www.mpaa.org/researchStatistics.asp.

³See Belson (2006).

⁴The National Association of Theater Owners (2006) issued a position article containing quotes from over 30 motion picture executives and directors arguing that a long theatrical window is vital to maintaining the financial health of motion picture theaters.

⁵See Ressner (2005).

⁶The data are available online at http://www.variety.com.

⁷By week 21, only 82 of the original 250 films were still in the theaters. After week 21, the average weekly box office revenue begins to increase because only the most popular films remain in the theaters beyond this point. For example, in week 30 (40), there were 11 (3) observations with an average box office revenue of $893 215 ($1 042 490).

⁸The omitted genre is Documentary.

⁹See Nelson et al. (2001).

¹⁰We tested for the inclusion of a set of yearly dummy variables. Based on the results of an F-test, we were unable to reject the null hypothesis that the coefficients as a group were not equal to zero, so they were excluded from the final model.