BOSTON -- A new study by two researchers at Harvard Business School and the University of North Carolina, Chapel Hill, finds that sharing digital music files has no effect on CD sales. This is the first study that directly compares actual downloads of music files and store sales of CDs.

The authors, Associate Professor Felix Oberholzer-Gee of Harvard Business School in Boston and Professor Koleman Strumpf of the University of North Carolina, Chapel Hill, conclude that File sharing had no effect on the sale of popular CDs in the second half of 2002. While downloads occurred on a vast scale during this period 3 million simultaneous users shared 500 million files on the popular network FastTrack/KaZaA alone most people who shared files appear to be individuals who would not have bought the albums that they downloaded, say the authors.

File Sharing Cannot Explain the Decline in Sales of Music during This Period

Even in the professors most pessimistic statistical model, it takes 5,000 downloads to reduce the sales of an album by a single copy. If this worst-case scenario were true, file sharing would have reduced CD sales by 2 million copies in 2002. To provide a point of reference, CD sales actually declined by 139 million copies from 2000 to 2002.

More Popular CDs Benefit from File Sharing

The effect of file sharing on sales depends on the popularity of a release, according to the researchers. For the least popular albums (with sales of less than 36,000 copies) the authors found a small negative effect. In contrast, for the top 25 percent of albums (with sales of more than 600,000 copies) they found a positive effect: 150 downloads increase sales by one copy. This effect is particularly important because the profitability of the music industry depends almost entirely on the success of the most popular albums.

File Sharers Download a Small Selection of Songs

In this study, the researchers track downloads of songs on 680 popular albums. All albums in their study were included in one of the Billboard Charts in the fall of 2002. Although these albums do well commercially (the average release sells more than 150,000 copies), more than 50 percent of the songs on these albums are never downloaded. 75 percent of the songs are downloaded no more than two times, 90 percent are downloaded fewer than 11 times.

Songs from Top Current Albums Are Most Often Downloaded

Not surprisingly, not all types of music are equally popular among file sharers. Songs from albums that are on the Top Current Billboard Chart are most likely to be downloaded. Alternative Albums are the second most popular musical genre among file sharers, followed by Hard Rock and Catalogue albums. The least likely to be downloaded are songs in the categories Jazz, Latin, and New Artists.

Marketing Strongly Influences What People Download and What They Buy

The study finds strong evidence that the music industrys marketing campaigns continue to influence what individuals listen to. For example, showing a music video on MTV increases both the number of downloads and the legal sales of that release.

U.S. Has Largest Number of File Sharers

31 percent of all individuals who download music live in the United States. Other important countries are Germany with a 13 percent share of worldwide users, Italy with 11 percent, Japan with 8 percent and France with 7 percent. File sharers in the United States are particularly active. While they represent 31 percent of worldwide users, they download 36 percent of all files.

Germany Is the Most Important Foreign Supplier of Music Files

U.S. file sharers download files from all over the world. Only 45 percent of the files downloaded in the United States come from computers in the U.S. 16 percent of music files are downloaded from computers in Germany, 7 percent from Canada, 6 percent from Italy, 4 percent from the U.K. A legal strategy that focuses mostly on the United States is unlikely to change the supply of music files.

This Study Is Unique

Previous studies have relied on surveys to assess the effect of file sharing on music sales. This is problematic because it is impossible to know if survey participants truthfully respond to questions about an illegal activity. This study is unique in that it uses data from file-sharing servers, where the authors directly observed 1.75 million downloads during 17 weeks in the fall of 2002. Using statistical methods, they can then test if the sale of an album declines more strongly if that album is downloaded more often.

The full study is available at:
http://www.unc.edu/~cigar/papers/FileSharing_March2004.pdf