A Heretical View of File Sharing

[ Two economists conclude that illegal “Downloads [of music] have an effect on sales which is statistically indistinguishable from zero” — contrary to what the music industry has claimed, based on a simplistic “single-bullet theory” of why Top 10 CD sales have decreased over the last four years. –doclalor ]

April 5, 2004, New York Times


The music industry says it repeatedly, with passion and conviction: downloading hurts sales.

That statement is at the heart of the war on file sharing, both of music and movies, and underpins lawsuits against thousands of music fans, as well as legislation approved last week by a House Judiciary subcommittee that would create federal penalties for using what is known as peer-to-peer technology to download copyrighted works. It is also part of the reason that the Justice Department introduced an intellectual-property task force last week that plans to step up criminal prosecutions of copyright infringers.

But what if the industry is wrong, and file sharing is not hurting record sales?

It might seem counterintuitive, but that is the conclusion reached by two economists who released a draft last week of the first study that makes a rigorous economic comparison of directly observed activity on file-sharing networks and music buying.

“Downloads have an effect on sales which is statistically indistinguishable from zero, despite rather precise estimates,” write its authors, Felix Oberholzer-Gee of the Harvard Business School and Koleman S. Strumpf of the University of North Carolina at Chapel Hill.

The industry has reacted with the kind of flustered consternation that the White House might display if Richard A. Clarke showed up at a Rose Garden tea party. Last week, the Recording Industry Association of America sent out three versions of a six-page response to the study.

The problem with the industry view, Professors Oberholzer-Gee and Strumpf say, is that it is not supported by solid evidence. Previous studies have failed because they tend to depend on surveys, and the authors contend that surveys of illegal activity are not trustworthy. “Those who agree to have their Internet behavior discussed or monitored are unlikely to be representative of all Internet users,” the authors wrote.

Instead, they analyzed the direct data of music downloaders over a 17-week period in the fall of 2002, and compared that activity with actual music purchases during that time. Using complex mathematical formulas, they determined that spikes in downloading had almost no discernible effect on sales. Even under their worst-case example, “it would take 5,000 downloads to reduce the sales of an album by one copy,” they wrote. “After annualizing, this would imply a yearly sales loss of two million albums, which is virtually rounding error” given that 803 million records were sold in 2002. Sales dropped by 139 million albums from 2000 to 2002.

“While downloads occur on a vast scale, most users are likely individuals who would not have bought the album even in the absence of file sharing,” the professors wrote.

In an interview, Professor Oberholzer-Gee said that previous research assumed that every download could be thought of as a lost sale. In fact, he said, most downloaders were drawn to free music and were unlikely to spend $18 on a CD.

“Say I offer you a free flight to Florida,” he asks. “How likely is it that you will go to Florida? It is very likely, because the price is free.” If there were no free ticket, that trip to Florida would be much less likely, he said. Similarly, free music might draw all kinds of people, but “it doesn’t mean that these people would buy CD’s at $18,” he said.

The most popular albums bought are also the most popular downloads, so the researchers looked for anomalous rises in downloading activity that they might compare to sales activity. They found one such spike, Professor Oberholzer-Gee said, during a German school holiday that occurred during the time they studied. Germany is second to the United States in making files available for downloading, supplying about 15 percent of online music files, he said. During the vacation, students who were home with time on their hands flooded the Internet with new files, which in turn spurred new downloading activity. The researchers then looked for any possible impact in the subsequent weeks on sales of CD’s.

Professor Oberholzer said that he had expected to find that downloading resulted in some harm to the industry, and was startled when he first ran the numbers in the spring of 2003. “I called Koleman and said, ‘Something is not quite right – there seems to be no effect between file sharing and sales.’ “

Amy Weiss, an industry spokeswoman, expressed incredulity at what she deemed an “incomprehensible” study, and she ridiculed the notion that a relatively small sample of downloads could shed light on the universe of activity.

The industry response, titled “Downloading Hurts Sales,” concludes: “If file sharing has no negative impact on the purchasing patterns of the top selling records, how do you account for the fact that, according to SoundScan, the decrease of Top 10 selling albums in each of the last four years is: 2000, 60 million units; 2001, 40 million units; 2002, 34 million units; 2003, 33 million units?”

Critics of the industry’s stance have long suggested that other factors might be contributing to the drop in sales, including a slow economy, fewer new releases and a consolidation of radio networks that has resulted in less variety on the airwaves. Some market experts have also suggested that record sales in the 1990’s might have been abnormally high as people bought CD’s to replace their vinyl record collections.

“The single-bullet theory employed by the R.I.A.A. has always been considered by anyone with even a modicum of economic knowledge to be pretty ambitious as spin,” said Joe Fleischer, the head of sales and marketing for BigChampagne, a company that tracks music downloads and is used by some record companies to measure the popularity of songs for marketing purposes.

The industry response stresses that the new study has not gone through the process of peer review. But the response cites refuting statistics and analysis, much of it prepared by market research consultants, that also have not gone through peer review.

One consultant, Russ Crupnick, vice president of the NPD Group, called the report “absolutely astounding.” Asked to explain how the professors’ analysis might be mistaken, he said he was still trying to understand the complex document: “I am not the level of mathematician that the professors purport to be.”

Stan Liebowitz of the University of Texas at Dallas, author of an essay cited by the industry, said the use of a German holiday to judge American behavior was strained. Professor Liebowitz argued in a paper in 2002 that file sharing did not affect music sales, but said he had since changed his mind.

The Liebowitz essay appeared in an economics journal edited by Gary D. Libecap, a professor of economics at the University of Arizona, who said that his publication was not peer reviewed, though the articles in it were often based on peer-reviewed work. Professor Libecap said he attended a presentation by Professor Strumpf last week, and said the file-sharing study “looks really good to me.”

“This was really careful, empirical work,” Professor Libecap said.

The author of another report recommended by the industry said that the two sets of data used by the researchers should not be compared. “They can’t get to that using the two sets of data they are using – they aren’t tracking individual behavior,” said Jayne Charneski, formerly of Edison Media Research, who prepared a report last June that she said showed that 7 percent of the marketplace consists of people who download music and do not buy it. That number is far lower than the authors of the new study estimated. “There’s a lot of research out there that’s conducted with an agenda in mind,” said Ms. Charneski, now the head of research for the record label EMI.

Leave a comment