from the fun-with-data dept
Over the last couple of weeks, we’ve seen a bunch of folks on all sides of the debates about file sharing point us to two studies that have come out, one of which supporters of greater copyright enforcement insist prove their point, and another which seems to show (yet again) that file sharing has little impact on buying habits. I wanted to take the time to look through both studies before writing about them, and since they came out so close together, we might as well do one post looking at both. As I started to write up a single post about it all, it got really, really long. So I’m breaking it into three separate posts. One about the first study, which argues that file sharing hurts sales, one that looks at the second study, which argues that file sharing does not hurt sales (and may actually increase sales), and then a third post, which tries to reconcile the findings of the two studies, while also responding to some of the criticism.
The first study is called Gone in 60 Seconds: The Impact of the Megaupload Shutdown on Movie Sales and is by Brett Danaher and Michael Smith. Smith, especially, has a long history of producing reports that copyright maximalists love. For example, both he and Danaher were responsible for a study last year claiming that the Hadopi 3 strikes program had increased iTunes sales. Unfortunately, a quick review of that report raised serious questions about the basis for those claims, as an alternative hypothesis (related to the sale of new iPhones) showed much more compelling data. I see his name on a lot of research sent around by the maximalists. It’s worth noting, as well, that the study was effectively funded by the MPAA, since it was a project of a program run by Smith and funded by the MPAA. Oddly, the paper fails to disclose this tidbit.
This new study seeks to answer a question we’ve been asking over and over again: do any of these enforcement efforts actually increase sales? There has been evidence that greater enforcement has a small, but temporary, impact on decreasing infringement, but there was not that much data concerning actual sales. In fact, we’ve pointed to data (contradicting Smith’s other report) that suggested Hadopi had done little, if anything, to increase sales. However, the data here has been limited, in part because there are so many other variables at play, so it’s difficult to separate out the actual impact. Smith and Danaher try to use data from various studios to look at the impact on movie sales following the Megaupload shutdown.
To try to determine the impact of the shutdown of Megaupload, Danaher and Smith basically compare movie sales before and after the shutdown date in a few different countries that had very different Megaupload usage. For example, they (using Google Adwords data) suggest that Megaupload had 2% penetration in the US, but 17% in Spain. Then they look at what the impact was in terms of digital movie sales and rentals compared across the different countries, and whether or not there were more sales in countries that had more Megaupload usage. They use this to argue that the key difference is Megaupload usage. The end result is that countries that had more Megaupload penetration saw a greater increase in digital movie sales and rentals following the shutdown than the countries that had lower Megaupload penetration. As they note:
This difference is both statistically and economically significant. Our findings indicate that digital movie revenues for two studios were 6-10% higher over the 18 weeks following the shutdown (across 12 countries) than they would have been if not for the shutdown.
This chart highlights the basics:
While I’ve seen some criticism online of these findings, I actually think the basic research and methodology is fairly solid. Those who have jumped up and said “correlation is not causation” are ignoring the various methods that the researchers used to isolate the shutdown. However, I’m not sure that the conclusions are quite as meaningful as some have suggested. First off, we’ve seen very similar data when it came to decreases in file sharing after enforcement increases — but the impact has always been shown to be temporary, until people settle in on a new method for infringement. It would make sense that some users of such a service, who don’t want to go hunting for a new free option, will switch to an authorized service if it’s available. But if they become aware of other services, they might also switch back. The amount of time the impact lasts will be a key thing to watch.
Of course, the other key thing that is left out of the picture in this study is the role of authorized services in all of this. Part of the reason for the growth of infringement on Megaupload in the first place was the dearth of compelling, simple, non-annoying, authorized services. The industry has, finally, been trying to increase those, and so it could be that people who couldn’t find any legit services before looked around again after the shutdown and found newer, better services. While the authors of the report say the findings suggest that Megaupload usage decreases sales, a possible alternative explanation would be that the slow pace with which the industry rolled out authorized services was equally, if not more, responsible. Either way, this report is a useful contribution in understanding the impact.
In our next post, we’ll explore the second study that came out even more recently, which appears to come to a very different conclusion.