Yesterday, I wrote a post for my Globe and Mail blog — and cross-posted it here — about some comments that University of Texas economist Stan Liebowitz made on his website with respect to file-sharing and its effect on CD sales. Prof. Liebowitz took issue with a recent study that found file-sharing actually increases CD sales among downloaders, but overall has no effect.
In my post, I raised a number of questions about Prof. Liebowitz’s response to the study — as well as his response to the Oberholzer-Strumpf study from 2004, which found something similar — and he sent me an email late last night responded to some of those questions. I asked him if I could post some of his email here and he agreed. Here are some of the relevant parts of his message:
“You neglected to tell your readers that I don’t just criticize the Industry Canada study because the emphasized results (the positive impacts) are implausible. I also provide the reasons why their statistical setup will be biased in a positive direction. If you want to know about the problems with the Oberholzer-Gee/Strumpf study, check out this paper which is was intended to be understandable by people with your type of background.”
I wondered in my post whether Prof. Liebowitz takes into account only sales of full CDs or whether he includes CD singles, ringtunes, downloads from iTunes and so on. Here is his response:
“I always examine “albums.†The use of albums and exclusion of singles makes little difference since in the time periods of the analysis (the last year of data the paper of mine that you mention) singles were a very small component of the market. Digital downloads were too small to measure and ringtones would have been excluded had the data existed, but it did not. Ringtones are a separate market since they do not serve the same consumption purpose as album purchases.”
I also raised the issue of whether the data Prof. Liebowitz was using referred to unit sales or revenues, since the latter would have fallen as average prices have dropped over the past few years.
“I make clear in my papers (which I presume you have not looked at) that I am talking about units and not revenues. One reason for that is that there are no reliable revenue statistics. Nielsen SoundScan, which is the data source on sales for the study of mine that you mentioned, doesn’t provide revenues. The revenue statistics from the RIAA just take the unit sales and multiply them by list price, so they are not actual revenue statistics.”
I’d like to thank the professor for responding. We still disagree about the extent to which file-sharing is solely responsible for the downturn in CD sales, but I appreciate his attention to detail and the amount of time he has devoted to his research.
Update:
Prof. Liebowitz sent me an email to say that he has updated his comments at his website after having taken a closer look at the details of the Industry Canada study. He says that he still believes that it is flawed, but not as badly as he first assumed.
Update 2 (Nov. 19):
Birgitte Andersen, one of the researchers who did the Industry Canada study, sent me an email and said that she has posted a response to some of Prof. Liebowitz’s comments here.