A fascinating paper from David Blackburn, a Harvard PhD student, on the economics of P2P file-sharing concludes that it does indeed depress music sales overall. But the effect is not felt evenly.

The hits at the top of the charts lose sales, but the niche artists further down the popularity curve actually benefit from file-trading.

Blackburn does a little mathematical magic to simulate what would happen if file-trading were reduced by 30%. [As you read, visualize the powerlaw curve that I use as my blog’s logo: “leftward” means towards the hits and greater sales.].

In this counterfactual world with 30% less file sharing, the lower 75% of the distribution of sales is shifted further to the left, while the top of the distribution increases its sales. This is what should be expected given the estimates from above. Artists who are unknown, and thus most helped by file sharing, are those artists who sell relatively few albums, whereas artists who are harmed by file sharing and thus gain from its removal, the popular ones, are the artists whose sales are relatively high.

This conclusion leads to further questions regarding the impacts that file sharing has had and will have on the recorded music industry. In particular, if file sharing essentially shifts sales away from established acts toward unknown acts, this has potentially very important implications for how talent is developed and distributed in the industry. As with the simple short-run effects of file sharing on sales, the direction of the impact is not clear. While one might guess that increasing the sales of new acts would lead to more investment in developing new talent, it is also possible that the investment in new acts is done as a fishing expedition to find artists who will sell millions of records. File sharing is reducing the probability that any act is able to sell millions of records, and if the success of the mega-star artists is what drives the investment in new acts, it might reduce the incentive to invest in new talent. This is, at its heart, an empirical question which is left to future work.

More here.