Skip to content

Pre-release Piracy Hurts the International Box Office

January 18, 2012

Although the academic literature seems to agree that music piracy displaces music sales (Rob and Waldfogel 2004, Liebowitz 2006, Zentner 2010, among other studies), there is a tension in the academic literature over whether movies also experience sales displacement caused by piracy.  Video files are much larger than music files and a pirated movie is not necessarily a good substitute for the theater-going experience.  Published studies in the economics and IS literature have found a range of estimates, from movie piracy having no effect on sales to nearly 1-for-1 sales displacement (where each pirated download is a lost sale).

My coauthor Joel Waldfogel and I have recently finished a paper that we think will help to inform the issue.   We find that international pre-release piracy (piracy that occurs after the world premiere of a film but before a country’s premiere) caused a decrease in sales in the 2005 international box office of at least 7%, although the true decrease may be larger in magnitude.  To be clear, if the results from our sample of movies/countries are generalizeable, this means pre-release piracy caused a decrease of at least $1 billion in the 2005 international box office.  These results should be informative to governments considering policies aimed at mitigating copyright infringement.

The paper can be found here.


From → Uncategorized

  1. Brady Kelly permalink

    First of all, I think that the methodology you use for your “marquee result” is very interesting, and a welcome addition to the literature. I have a few questions about your US results, though – while I understand your frustration at all the blogs ignoring your primary results in favor of your secondary, admittedly weaker results, I think the US results are still worth talking about.

    Based on your description and on the variables in Table 4, your sole indicator of lost revenues due to piracy is a steepening of the week-to-week revenue decrease for each film (as compared to before/after the rise of BitTorrents). Is your reasoning here that because pirated copies of a film are increasingly available as the film spends time on the market, people will watch more pirated copies of a film as ‘weeks-since-release’ increases? If so, and if I’m understanding your methodology correctly, there may be a problem with your results beyond merely being a weak indicator.

    Your methodology implicitly assumes that potential consumers are not forward-looking – that because piracy increases as ‘weeks since release’ increases, it must be the case that box office revenues are increasingly affected as ‘weeks since release’ increases. But if potential consumers forgo visiting the theater in the early weeks of a film’s release because they expect to pirate the film later, I think that this would invalidate your result as an unbiased estimator. Insofar as most films eventually become available in pirated form, it seems reasonable that most consumers who pirate movies are aware from week one that a pirated copy will become available, even though there is not a pirated copy available in the film’s first week(s). Luan and Sudhir (2007) find that consumers are forward-looking with regard to DVD sales (i.e. consumers sometimes forgo seeing a film because they expect to see the DVD instead), and so I expect consumers to be even more forward-looking when the required wait time is even shorter.

    Not only would this kind of revenue-snatching behavior be missed by your methodology, this phenomenon might actually attenuate any effect that your methodology would otherwise pick up. Since the result that your method misses is decreased revenue in the early stages of a film’s release, the natural outcome of this is a flattening of the week-to-week decline in revenues. Because your test for piracy is a steepening in the week-to-week decline in revenues, this phenomenon could be highly problematic for your result. If I’m thinking about this correctly, this attenuation would exist if and only if those who pirate are relatively more concentrated toward early weeks of a film’s release (with regard to the week in which they counter-factually see the film) than are other consumers (with regard to the week in which they actually see the film). Considering what we know about demographics of those who pirate and the demographics of those who see a film early, this assumption seems highly likely. Most disconcertingly, the phenomenon described in the third paragraph above should hold true regardless of differences in film-going behavior, also leading to underestimation of lost box office revenues.

    Again, I really enjoyed the main analysis and results of your paper, but I’m having trouble seeing a way around this issue for US results. Is there some confounding factor I’m not thinking of? Am I misunderstanding the premise? If not, this secondary section could certainly detract from the perception of your primary results. Either way, I’d love to hear your thoughts – film economics is a big area of interest for me.

    • Hi Brady –

      First let me apologize for never getting back to you on this. I use this blog largely as a feed for my main website and missed that you had commented on my post (I’ll approve it stat!).

      Second, you make an excellent point. The US section is not well identified at all for several reasons, and yours is probably the most important. We’re actually reworking the paper with a new identification strategy and those US results will be gone in the new version since they were never very clean.

      Feel free to contact me by email if you want to discuss media economics further.

      P.S. I forgot to mention, you are right that piracy seems to be more concentrated in the early weeks. I can’t remember if I included this in the paper or not, but here’s an interesting fact from the international results (which are better identified). The first week of lag is actually the worst in terms of negative correlation with box office returns. That is, if you believe that our story is causal, the first week of lag causes the biggest decrease in returns. Each additional week causes a greater decrease, but none so much as that first week. So you’re right that it’s non-linear.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: