Tuesday, February 13, 2007

Study: P2P effect on legal music sales "not statistically distinguishable from zer

A new study in the Journal of Political Economy by Felix Oberholzer-Gee and Koleman Strumpf has found that illegal music downloads have had no noticeable effects on the sale of music, contrary to the claims of the recording industry.

Entitled "The Effect of File Sharing on Record Sales: An Empirical Analysis," the study matched an extensive sample of music downloads to American music sales data in order to search for causality between illicit downloading and album sales. Analyzing data from the final four months of 2002, the researchers estimated that P2P affected no more than 0.7% of sales in that timeframe.

The study compared the logs of two OpenNAP P2P servers with sales data from Nielsen SoundScan, tracking the effects of 1.75 million songs downloads on 680 different albums sold during that same period. The study then took a surprising twist. Popular music will often have both high downloads and high sales figures, so what the researchers wanted was a way to test for effects on albums sales when file-sharing activity was increased on account of something other than US song popularity. Does the occasionally increased availability of music from Germany affect US sales?

The study looked at time periods when German students were on holiday after demonstrating that P2P use increases at these times. German users collectively are the #2 P2P suppliers, providing "about one out of every six U.S. downloads," according to the study. Yet the effects on American sales were not large enough to be statistically significant. Using this and several other methods, the study's authors could find no meaningful causality. The availability and even increased downloads of music on P2P networks did not correlate to a negative effect on music sales.

"Using detailed records of transfers of digital music files, we find that file sharing has had no statistically significant effect on purchases of the average album in our sample," the study reports. "Even our most negative point estimate implies that a one-standard-deviation increase in file sharing reduces an album's weekly sales by a mere 368 copies, an effect that is too small to be statistically distinguishable from zero."

The study reports that 803 million CDs were sold in 2002, which was a decrease of about 80 million from the previous year. The RIAA has blamed the majority of the decrease on piracy, and has maintained that argument in recent years as music sales have faltered. Yet according to the study, the impact from file sharing could not have been more than 6 million albums total in 2002, leaving 74 million unsold CDs without an excuse for sitting on shelves.

So what's the problem with music? The study echoes many of the observations you've read here at Ars. First, because the recording industry focuses on units shipped rather than sold, the decline can be attributed in part to reduced inventory. Gone are the days when Best Buy and others wanted a ton of unsold stock sitting around, so they order less CDs. The study also highlighted the growth in DVD sales during that same period as a possible explanation for why customers weren't opening their wallets: they were busy buying DVDs.