CIPC Study on P2P File-sharing Doesn’t support its Policy Recommendation

Last week, the Canadian Intellectual Property Council released a commissioned report on the effects of music file-sharing in Canada.  The report, titled The True Price of Peer to Peer File-Sharing, is based on a study conducted by Dr. George Barker, director for the Centre for Law and Economics at the Australian National University.  Dr. Baker revisited survey data gathered by Industry Canada between 2005 and 2008, which served the basis for a study by Birgitte Andersen and Marion Frenz, titled Don’t Blame the P2P File-sharers: The Impact of Free Music Downloads on the Purchase of Music. In their study, Andersen and Frenz invested the motivations behind P2P downloads in detail, and found,

a positive sampling or market creation effect of P2P activity on CD album purchases. Downloads linked to a ‘hear before buying effect’ stimulate sales of CD albums. As expected, we also found a negative CD market substitution effect. Where P2P activity was motivated because CD albums were perceived as being too expensive, such downloads displaced CD sales. The positive sampling or market creation effect and the negative sampling effect cancel one another out; i.e. neither one of them is significantly large than the other.

Based on these findings, they concluded that P2P file-sharing is not to blame for the decline in CD markets.

The study commissioned by the CIPC claims to refute Andersen & Frenz’s conclusions, and CIPC quickly issued a press release proclaiming that the study “Reinforces Need for Copyright Reform—Peer to Peer Downloading is Harming Music Sector”.  I have now read the CIPC study and I’m afraid that it is not entirely clear to me that this study refutes Anderson & Frenz’s conclusions.  Moreover, even if it did, the study fails to support the policy prescriptions the CIPC advocates. Let’s start with the first issue.

Does the CIPC study refute A&F’s

Here are some of the key findings in the CIPC study:

• three out of four people who download some or all of their music using P2P,  say they would purchase music through legal means if P2P were not available;
• “hardcore” downloaders who only access music through free P2P downloading would legally purchase one third of their music if P2P were not available;
• given the portion of music  that all downloaders surveyed say they would purchase from legal sources, and the average price of music at the time the survey was conducted, downloaders would spend in excess of $175 more per year on music, adding up to 1.1 billion dollars of increased music sale (annually).

As a preliminary comment, this estimate of the “true price” of P2P file sharing should be taken with a grain salt. Even the study acknowledges that this estimate is “seems to be on the high side. [And] therefore invites further work on the underlying data.”  I agree, but this acknowledgment is an understatement, because $1.1 billion is about twice the annual revenue of the Canadian recording industry. The suggestion that without P2P revenues would be three times higher than they currently are doesn’t seem very credible.  In fact, it is highly implausible if only because it is substantially higher than the industry revenue in 1998 or 1999, the year Napster was born.  In other words, we know that recorded music sales dropped over the last decade or so, but the unanswered question is how much of this drop is attributable to P2P.  But even if we attribute the entire decline to P2P (and no serious study suggests that) the claim that without P2P sales would be significantly higher that they were before the onset of P2P doesn’t make much sense.  This, of course, doesn’t mean that P2P causes no harm, only that the present estimated harm is exaggerated.

Now, does the CIPC study actually refute the A&F study? I don’t think so.   The A&F study identified several possible effects of file sharing, some negative for the industry while others positive.  It tried to quantify each of them and finally found that the positive and the negative cancel out each other.  Their conclusion may be right or wrong, but that is not my issue right now.  In contrast, the CIPC study assumes that file sharing has only one effect: to substitute authorized sales, and then proceeds to quantify this effect.  The CIPC study is premised on a basic, but incomplete, model of downloaders’ behaviour:

Economics predicts the decisions whether to acquire music illegally (participation decision) and how much (activity decision) will depend on access to relevant resources including computers and broadband. It will also depend on the legal sanctions for piracy and preferences or attitudes to criminal behaviour. In this regard, the economic analysis of criminal behaviour assumes piracy is a function of the probability of being caught (p) and the sanction applying if one is caught (s). (p. 13)

It also explicitly states that:

Relevant economic theory would suggest that CD purchases and pirated music are substitutes. However, given there are important differences in their nature, they are best thought of as imperfect substitutes. This means individuals can compensate for a reduction in their use of CDs by an increase in the use of pirated music copies. (p. 14)

As a starting point, there’s nothing wrong with this assumption, the model and its prediction.  But only as a starting point.  The assumption that “legitimate copies and pirated copies are … substitutes” is definitely a sensible one, and one that should inform any serious debate on the effects of unauthorized copying. However, the modern economic literature is rich enough, both theoretically and empirically, to suggest that substitution is not the only effect, and that under some conditions unauthorized copying may actually have positive effects on sales (e.g., some people sample new artists by downloading some of their works and if they like it go and buy more of it) or no effect at all (e.g., some people download music that is otherwise not available on commercially available CDs). The debate is whether P2P file sharing leads only to substitution, whether other effects take place, and what the overall combined effect is.  The CIPC study assumes that substitution is the only effect and totally ignores the possibility of other effects.  But assuming that something doesn’t exist does not refute its existence.  Unfortunately this is a serious shortcoming, because it provides a very incomplete, and potentially distorted, picture of the true effects of P2P.  Consider the finding that those who obtain music both from P2P downloads and by purchasing CDs would replace 49% of their P2P downloads through paid purchases if P2P were not available (p. 24).  The CIPC study focuses on the 49% and identifies a significant substitution effect.  But it neglects to consider the effect of the other 51%.  That is, in the absence of P2P these individuals will not obtain 51% of the songs of currently have.  What’s the effect of that? Tough question.  If many of those downloaded songs were downloaded for sampling, and some of these samples led to actual paid purchases, eliminating P2P will have some negative effect on sales.  If those 51% consisted primarily on songs that aren’t commercially available in Canada, eliminating P2P will have no effect on their sales and will only decrease the users’ welfare.  What will be the overall effect? I don’t know.  A&F made some findings in this regard, which may be right or wrong.  However, the CIPC study doesn’t even acknowledge the possibility of any effect other than substitution.  This is regrettable.

Does the CIPC Study Support its Policy Recommendation

Based on the identified substitution effect, the study concludes that

removing P2P file-sharing, for example through stronger copyright laws, would increase music purchasing, music industry sales, artist revenues and, by implication, increase industry employment, economic growth and government tax revenues. Conversely, these results show that allowing P2P file-sharing through weak copyright law reduces music purchases, music industry sales, artist revenues, industry employment, GDP and government tax revenues. (at p. 5)

There are several problems with this conclusion.  First, as noted above, failing to consider the possibility of positive effects of file sharing undermines the credibility of this conclusion.  Second, the results of the study are based on answers to survey questions, but the actual question from the Industry Canada survey didn’t quite ask how respondents would behave if P2P file sharing were removed through stronger copyright laws or otherwise.  It asked something different.  Here’s the question:

Considering the songs that you downloaded for free through P2P networks during 2005,

a) what % would you have purchased as paid music sites if they were not available through P2P;

b) what % would you have purchased as part of a music CD if they were not available through P2P?

Note, that respondents were not asked what they would do if P2P did not exist, or if it were illegal, or if penalties were higher.  They were only asked what would they do if the specific songs that they had downloaded were not available through P2P.  I do not know if the answers would be different, but it is important to note that the question that was actually asked differs from the question that the CIPC study ultimately seeks to answer.

More importantly, even if the question asked is equivalent to asking what would you do if P2P didn’t exist, the answers are not immediately relevant to the policy question at hand, namely the legal response to P2P, unless we assume that we can devise a legal rule that will completely annihilate P2P.  But as the experience of the last decade suggests, P2P strives worldwide despite being outlawed and despite heightened enforcement. Therefore, as a basis for rational law-making, the results of this study aren’t very informative.  Given that P2P file sharing exists despite the fact that in many cases it is already illegal, and because it isn’t likely to immediately and completely disappear even if Parliament accepts all of the CIPC proposals, the important question is whether, if any, changes in current law would limit its use, to what extent, and at what social cost.  It is entirely plausible that any marginal tightening of the law will have a very small marginal effect on downloaders’ behavior, if any.

But let’s assume the the CIPC study is flawless, and that it’s clear that P2P file sharing only harms the sales of sound recordings, that the harm is significant and losses are substantial.  Would that support the policy changes advocated by the CIPC?  Not necessarily.

Consider the following example.  Most Canadians have reliable supply of safe, clean and cheap tap water in their homes, although many of them also consume bottled water.  Suppose that the Bottled Water Industry commissions a study that flawlessly finds that 100% of respondents would purchase bottled water instead of using tap water if tap water supply were banned.  Bottled water sales would sky-rocket; employment in the bottled-water sector would increase (after all, distributing water in bottles is much more labour intensive than delivering them in pipes).  Government revenues from taxes on the sale of bottled water, on the profits of the companies, the dividends of their shareholders and the income of the employees would increase too.  But would that justify banning tap water?  Probably not.  Industry interests notwithstanding, Canadians as a whole are probably better off without such a ban.

Now, let’s move away from water and closer to the industries that are directly affected by copyright.  Suppose that the publishing industry commissions a flawless study that shows that lending books negatively affect the sales of books and that closing down all libraries and banning all other forms of book lending and sharing among individual would, paraphrasing the CIPC study, increase book purchasing, book industry sales, author revenues and, by implication, increase industry employment, economic growth and government tax revenues.  Would such a study justify a copyright reform along those lines?  Not necessarily.  Canada, and Canadians, probably gain from copyright laws that guarantee the availability of libraries and the right to lend and freely read books more than they loose from whatever negative effect that such rules might have on industry revenues.  I say “probably” because our laws are based on such an assumption, not on any hard empirical evidence to that effect.  Therefore, serious studies attempting to measure such various effects are certainly welcome.  But if we had a study that focuses only on how libraries affect sales, without considering the enormous social benefits they create, if would provide a very weak basis to change our policies.  Likewise, measuring the effects of P2P by focusing only on the costs and benefits for the record companies, are as helpful for public policy as the hypothetical studies about book lending or tap water.

We wouldn’t decide to get rid of tap water on the basis of studies that show a positive effect of such move on the bottled-water industry, and (hopefully) we will not get rid of public libraries even if it were shown that publishers will sell more books as a result.  Likewise, we ought not base our policy choices with respect to file-sharing solely on the base of studies that measure the effect on the sales of recorded music.  We deserve more than that.