On-line Piracy and Recorded Music Sales David Blackburn

Ever since the introduction of Napster, the impact of file sharing on the music industry has been the focus of intense debate. The availability of songs on file sharing networks has two competing effects on sales that are likely to vary across artists. First, there is a direct substitution effect on sales as some consumers download rather than purchase music. Second, there is a penetration effect which increases sales, as the spread of an artist’s works helps to make the artist more well-known throughout the population. The first effect is strongest for ex ante well-known artists, while the second is strongest for ex ante unknown artists. Thus file sharing reduces sales for well-known artists relative to unknown artists. Taking account of this heterogeneity in estimating the effect of file sharing provides strong evidence of this distributional effect. Additionally, I find a large aggregate negative effect on sales not apparent in previous work that failed to account for the differential impacts on more and less wellknown artists. The overall negative impact of file sharing arises because aggregate sales are dominated by sales of well-known artists. Using my estimates of the effect of file sharing, counterfactual exercises suggest that the lawsuits brought by the RIAA have resulted in an increase in album sales of approximately 2.9% during the 23 week period after the lawsuit strategy was publicly announced. Furthermore, if files available on-line were reduced across the board by 30%, industry sales would have been approximately 10% higher in 2003. ∗Department of Economics. Email: blackb@fas.harvard.edu. An updated version of the paper may be found at http://www.economics.harvard.edu/~dblackbu/papers.html. I would like to thank Mariana Colacelli, Jan De Loecker, David Evans, Kate Ho, Joy Ishii, Larry Katz, Bryce Ward, and participants at the Harvard IO Workshop and the 2004 International Industrial Organization Conference for helpful suggestions. Special thanks to Gary Chamberlain, Julie Mortimer, and Ariel Pakes for their advice and encouragement. Additionally, I am indebted to Eric Garland and Adam Toll at BigChampagne and Rob Sisco at Neilsen SoundScan for providing access to themselves and their data, without which this project would have been impossible. I stake sole claim to any remaining errors.