The British Phonographic Industry (BPI) recently released a study suggesting that illegal downloading cost the British music industry 1.8 billion pounds over the past three years. Echoing the sentiment of the US content industry, BPI Chairman Peter Jamieson said that “when people share music files illegally, they are stealing the future of the British musicians and the people who invest in them.”
Jamieson is right if filesharing reduces the net penetration of contemporary music in Britain in a way that significantly reduces the net profits of music producers. But the rhetoric of “stealing” and “theft” seems unhelpful in this context, and not because people have a right to free music. Rather, this rhetoric seems unhelpful because copyright—like other forms of intellectual property protection—aims to strike a balance between the public’s interest in information and the private sector’s claim to compensation. It cannot be doubted that these interests chafe against each other. But one can, and should, question the notion that all leakage of copyrighted content is inconsistent with the interests of the copyright holder. Sampling, for instance, is widely thought to be a reliable method for stimulating purchasing patterns.
But what do we really know about the causal impact of illegal filesharing? Stan Liebowitz, an economist who has written a number of papers about the economics of filesharing, believes that filesharing does indeed have this effect. In Testing Filesharing’s Impact by Measuring Record Sales in Cities, Liebowitz analyzes city-specific record sales data suggesting that illegal downloading is causally related to depressed CD sales.
This study suggests that filesharing reduces by just over 1—1.27 to be precise—the number of albums purchased by the average record purchaser (in 2003 specifically). The Liebowitz study is a serious attempt to identify the true causal impact of filesharing, but it is vulnerable to a number of objections. For present purposes, consider just two. First, the study does not identify a causal relationship between actual downloading activity and decreases in record sales because reliable data on downloading patterns is hard to come by. This study, in other words, did not have treat reduced record sales as a dependent variable because reliable data on the relevant independent variable—illegal downloading patterns—is not available. Instead, Liebowitz treats the level of internet penetration as a rough estimate of filesharing in large cities. Internet penetration may be the only readily available proxy, but it hardly seems ideal. The number of individuals with access to the internet is always likely to be much higher than the number of people trading music illegally. Moreover, accurate numbers for rates of internet penetration in large cities do not indicate what the average propensity to engage in illegal filesharing might be. (Those interested in this issue should read Liebowitz’s paper. He makes an interesting argument for regarding the lack of data on average filesharing propensity as irrelevant.)
Second, even if all of Liebowitz’s assumptions are methodologically sound, the policy implications of his work are far from clear. For instance, if one examines purchasing patterns from 1973 through 2004, the decline in record sales that Liebowitz attributes to filesharing hardly seems startling. From 1973 through 1991, per capita record sales, with the exception of 1988 and 1990, were well below 4 for the average record purchaser. Between 1993 and 1994, per capita sales rose from 4.5 to 5.5 per person, and then remained relatively steady until 1999. In the interim period between 1999 and 2003, average sales-per-person dropped from 5.5 to 4. Per capita record sales have never remained near 5.5 for more than three sequential years, and given earlier drops and increases in demand for music, it is an open question whether the harm of filesharing really accounts for reduced record sales. For those with somewhat rusty statistical skills, Liebowitz’s paper gives the impression that reduced record sales could in part be the result of filesharing. But the alternative explanations—(1) that sampling induces sales, (2) that the state of the economy impacts willingness and ability to purchase records, and (3) that estimates of the entertainment industry’s growth provide a dubious baseline against which to judge post-Napster CD sales, just to name a few—seem too intuitively plausible to dispense with on the basis of the analysis he provides. To his credit, Liebowitz explicitly points out that his analysis is to be taken with a grain of salt, since “the future can always prove you wrong.”
Economists could certainly provide a more nuanced assessment of Liebowitz’s work than I have here been able to provide. But what can’t be gainsaid is that accurately measuring the economic impact of filesharing on music sales is a difficult task. In addition to adopting a defensible metric for measuring lost revenues, it is also important to have a firm sense of what normal market fluctuations in music purchasing patterns look like. This is important because downward departures in music sales should not be attributed to illegal filesharing when they are really due to decreased demand and associated market phenomena.