Two Harvard Business School economists tackled the effects of digital piracy in a new working paper titled “File-Sharing and Copyright” (download the PDF here). In the paper, Felix Oberholzer-Gee and Koleman Strumpf address the impact file-sharing has had on the incentive to create new works. Oberholzer-Gee and Strumpf are most well-known in music industry circles for their 2004 paper “The Effect of File Sharing on Record Sales – An Empirical Analysis.” In that paper, the two concluded file sharing can explain at most a “tiny fraction” in the decline of in recorded music sales.

This section lays out the paper’s goal:


Copyright exists to encourage innovation and the creation of new works; in other words to promote social welfare. The question to ask is thus whether the new technology has undermined the incentives to create, market, and distribute entertainment. Sales displacement is a necessary but not a sufficient condition for harm to occur. We also need to know whether income from complementary products offset the decline in income from copyrighted works. And even if income fell, welfare may not suffer if artists do not respond to weaker monetary incentives.



Some or all of the paper’s concepts will be familiar to those who follow academic research and the public policy aspects of file sharing. For them it is a worthy read and a refresher on recent research into a topic dear to the music industry’s heart. For the less initiated, the paper is a good introduction to the widespread and heated debate over the effects of file sharing on businesses, artists and copyright’s role in public welfare.

The paper has some flaws that detract from one of its main conclusions. Central to the authors’ assessment of artists’ incentive to create is the increase in the number of sound recordings released annually in the U.S. “Since 2000,” they wrote, “the number of recordings produced has more than doubled. In our view, this makes it difficult to argue that weaker copyright protection has had a negative impact on artists’ incentives to be creative.” Oberholzer-Gee and Strumpf pointed to the increases in the number of new books published and new albums released (new books up 66% from 2000 to 2007, new albums up over 100% since 2000 and worldwide feature films up 30% since 2003). “Weaker copyright is unambiguously desirable,” they later wrote, “if it does not lessen the incentives of artists and entertainment companies to produce new works.”

If an increase in the number of new albums is a barometer for artists’ incentive to create, one should understand the many causes for that increase. If album releases went up only 30%, for example, would that change their opinion? Some of the increase can be attributed to the arrival of new digital distribution services. One cannot ignore the licensing deals signed by U.S.-based digital distributors that have opened the U.S. market to labels and artists from other countries. And some new releases are merely old recordings repackaged. Neither factor can be viewed as an indication of an increase in artists’ desire to create.

According to Nielsen SoundScan, the number of digital-only albums released in the U.S. was 49,370 in 2008, 25,159 in 2007, 24,720 in 2006 and 16,580 in 2005. The rise in digital-only releases can be attributed in part to the influx of foreign catalogs as well as previously out-of-print domestic recordings and other titles thrown into the market in hopes of capitalizing on the level playing field offered by digital distribution. An indication of the expanse of catalogs now available in the U.S. is the number of distributed labels listed at IODA’s website: 11,871, including 115 from Latin America and 24 from Africa. It’s safe to say these are the types of labels that did not have distribution in North America at the beginning of the decade.

The rise in digital-only albums is also the result of the ease of releasing titles to digital stores. The low cost of digital releases has prompted labels to repackage their catalogs for easy consumption – the Rhino Hi-Five releases are a good example – and reissues of previously out-of-print titles. These digital-only releases are mainly the result of low costs combined with widespread distribution (though a fear of losses from file sharing has clearly not impeded their growth). The key point here is that many new albums are not new recordings and thus are not indicative of a greater incentive to create.

Some reasons for the increase in albums are a better fit for the authors’ conclusions. The rise in the number of digital-only EPs has not been dampened by file sharing. Since EPs tend to be classified as albums by SoundScan, the greater number of EPs adds to the number of new albums released each year. Even though they are entering an increasingly crowded market, artists continue to add to retailers’ immense catalogs. Some amateur recordings, which would not have been commercially released just a few years ago, now have access to the marketplace via new services and technologies. The existence of piracy has not stopped their release. Even so, it could be argued that in an absence of piracy, the cost efficiencies and powerful services given to independent musicians would lead to explosive growth in the amount of music released commercially.

The increase in physical albums is evidence that Americans, in aggregate, are producing more albums. Albums with a physical release increased 28.5% from 2005 through 2008 – from 43,733 to 56,205. Albums released in the U.S. are more likely than releases from other countries to be available on CD. CD manufacturing in the U.S. is affordable and options are plentiful. Independent artists who sell albums at their concerts need to manufacture CDs. Because of these factors, one can look at the rise in CD releases as an indication that more albums are being released domestically than earlier in the decade. But because of the factors mentioned, there is uncertainty about what role lower costs, reduced barriers to entry and greater/lesser incentive to create play in the increase.

There is another problem with the authors’ take on complementary products and services. The economists see the rise in authors’ speaking fees as an example that file-sharing has enhanced ancillary revenue sources. There are two problems with this line of thinking. First, authors are not a good group to use when looking at the effects of file sharing. E-books are not traded online nearly as much as music and movies. The print media has a number of challenges that have nothing to do with file-sharing and everything to do with the loss of print advertising. Second, the world of book publishing is a “winner takes all” market, just as with music and movies. The high speaking fees commanded by a relatively tiny number of successful authors says nothing about the complementary revenue earned by the hundreds of thousands of authors who have yet to break into that exclusive club. To draw a parallel for music, the unknown and foreign artists who represent the increase in new album releases cannot command high ticket prices.

The “winner takes all” aspect of the music industry is exemplified by the touring industry. The authors cite findings on ticket prices from research by Princeton economist Alan B. Krueger. Although he had information on over 230,000 events over 20 years, Krueger admits the data is skewed toward more successful artists – he actually limited his sample to 1,275 artists that represented 75% of the dollar value of ticket sales reported in the Pollstar data. The use of hit-heavy data is fine if you assume only the top sellers are hurt by file-sharing (opinions on this are mixed). Only top-tier artists have the market power to raise ticket price to make up for lower album sales.

Less successful artists, the ones behind a portion of the influx of new releases, lack the ability to increase ticket prices at the same rate as the top performers. If they are indeed harmed by file sharing, touring is unlikely to make up the shortfall. There is much anecdotal evidence – even cited in the recent Page/Garland paper on P2P’s long tail – that says middle-tier touring artists are less well off today than they were ten years ago. It makes sense. Krueger’s work indicates the number of tickets purchased had fallen over the period he studied. With fewer tickets sold but higher prices on the most popular tickets, artists in the middle are no better off. If more of them are touring (because recorded music revenue is lower), they are competing harder for the same or less consumer spending. It’s the same thing with recorded music. There are more albums released but fewer albums sold. The hits get the same percentage of total sales – if not more – while a larger group fights for what’s left of consumers’ wallets.

It would be wrong to exclude all but the most popular touring artists from a discussion of non-recorded music revenue sources. Since artists of all levels of success should be part of the public policy debate about copyright, the authors should keep in mind the less-than-successful artists when determining the ability of complementary income to offset losses to file sharing.

One interesting point to the paper is that artists are not motivated purely by financial incentives. The authors’ use of Nielsen data accounts for only those new works that were created for direct commercial gain. But there is so much more out there. A brief examination of music blogs will reveal all sorts of works created without direct compensation in mind. DJs frequently post hour-long mixes and bedroom producers create and release mash-ups. In both cases, the use of the songs and samples are not cleared with the owners. If one takes into account the considerable amount of black market releases in addition to those registered with SoundScan, the result is a more accurate picture of artists’ total output and the incentives that drive them.

In spite of a few faults, the paper succeeds on many fronts. In particular, it does a good job addressing the actual monetary damages caused by file sharing and the incentive of trade groups to inflate those estimates. And the section on file-sharing studies is an excellent overview of the difficulties in gathering sufficient file-sharing data in order to estimate P2P’s displacement of sales. Those findings are summarized in Table 5 at the end of the paper.