This essay first appeared at Vox EU on Nov 14, 2011.
Since the appearance of Napster a dozen years ago, intellectual property rights have effectively been substantially weakened. Unfettered stealing has caused revenues to the recorded music industry to tumble by roughly a third. While this reversal has prompted academic chin-scratching over its cause, most observers now agree that consumers’ ability to steal music undermines firms’ ability to sell it. The woes of the recorded music industry, along with fears of the movie, television, and book industries, have led many to advocate stronger legal protections for intellectual property, including private lawsuits, and threats of disconnection from Internet service.
While the question of whether stealing undermines selling is of vital interest to the recorded music business, it is arguably not the only important question for evaluating the success of intellectual property rules. The purpose of copyright laws is to provide incentives for the creation of new works. Weakened intellectual property protections present a threat consumers as well as producers: if producers cannot appropriate sufficient revenue to cover their costs, they may stop bringing new products to market, causing harm to consumers as well as producers. Keeping this in mind focuses attention on a different question: what has happened to the volume – and quality – of new works since Napster?
Documenting the volume of high-quality work is challenging, but in recent work I have developed three approaches to characterizing the evolution of music quality over time. The first approach makes use of critics’ retrospective best lists, such as Rolling Stone’s list of the 500 best albums of all time. Inclusion of a work on such as list means that a work has quality above some time-constant threshold. Each list can be transformed into an index showing the number of listed works originally released in each year. I have 88 such lists from Anglophone sources, together covering the period between 1960 and 2010. These lists’ associated indices can be combined statistically to produce an overall index of music quality. As Figure 1 shows, the index rises from 1960 to 1970, then falls, remaining low during the 1980s. The index rises during the first half of the 1990s, then falls in the second half. Following 2000, the decline stops, and the index is flat. The critic-based index provides no evidence of a decline in recorded music quality since Napster.
The second and third approaches make use of information on the service flow of recorded music, based on sales and radio airplay, by time and vintage. For example, I observe the share of music on US radio in 2008 that was originally released in 2008 and each preceding year. I observe the analogous vintage distribution for music aired in each year back to 2004. I also observe coarse data on music sales by year and vintage, so I observe the share of music sold in, say, 2010 that was originally released in 2010 and in each previous year. I observe analogous sales data back to 1970.
In any given year, shares vary across vintage for two reasons. First, music depreciates: music sells less, and is played less on the radio, as it gets older. Second, vintages may differ in their quality. If the quality of music had declined after 2000, then we would expect elevated use of older vintages after 2000, controlling for deprecation. Thus, if I can account for deprecation, the remaining variation in use by original vintage should reflect vintage quality. To this end, I regress the log share of music used in a year t from vintage v on a flexible set of age dummies (to account for depreciation), along with a flexible set of vintage dummies. The coefficients on vintage dummies show the evolution of vintage quality over time.
Results are interesting. The airplay data allow me to infer vintage quality back to 1960 and exhibit the following pattern (in Figure 2): quality rises from 1960 to 1970, then falls and remains flat from 1980 to about 1999, with a small bump up in the mid-1990s. After 2000, vintage quality rises sharply, reaching levels not seen since the 1970s. The sales data, which produce an index going only back to 1970, suggest a similar story: vintage quality declines from 1970 to 1980, is flat to 2000, then rises after 2000. See Figure 3.
Two points are in order about these results. First, the three approaches, while independent of one another, produce broadly similar patterns from 1960-2000. Second, none of the three approaches shows a decline in quality following 2000; and two of them – based on actual music usage – show substantial increases.
These results are at first blush puzzling: how could the flow of high-quality music continue despite sharp reductions in firms’ ability to appropriate revenue. The puzzle’s resolution may lie in the observation that technological change has not only reduced effective demand; it has also reduced costs of bringing music to market. The costs of creation, promotion, and distribution have all been revolutionized by new information and communications technologies.
While the traditional purveyors of recorded music – the major record labels – have suffered since Napster, good music has continued to make its way to market. Independent labels account for a growing share of successful music, measured both by critical acclaim and sales. While many producers of recorded music have been made worse off by changes in technology, there is no evidence that the volume of high-quality music, or consumers, have suffered. Many caveats are in order: chiefly, it is not clear whether the results extend to other content industries where creation remains quite expensive. But the results are a reminder that an evaluation of copyright policy should be cognizant of its effects on both producers and consumers.
References
Liebowitz, Stan J. “File Sharing: Creative Destruction or Just Plain Destruction?”
Journal of Law and Economics, vol. 49, no. 1, April 2006, pp. 1-28.
Oberholzer‐Gee, Felix and Koleman Strumpf. “The Effect of File Sharing on Record Sales: An Empirical Analysis.” Journal of Political Economy. Volume 115, Issue 1, Page 1–42, Feb 2007
Rob, Rafael and Joel Waldfogel. “Piracy on the High C’s: Music Downloading, Sales Displacement, and Social Welfare in a Sample of College Students.” Journal of Law & Economics. Volume 49, Issue 1, Page 29-62, Apr 2006.
Waldfogel, Joel. “Bye, Bye, Miss American Pie: The Supply of New Recorded Music since Napster.” NBER Working Paper 16882. 2011a.
Waldfogel, Joel. “Copyright Protection, Technological Change, and the Quality of New Products: Evidence from Recorded Music Since Napster.” NBER Working Paper 17503. 2011b.
Zentner, Alejandro, “Measuring the Effect of File Sharing on Music Purchases.” The Journal of Law and Economics. Volume 49, Issue 1, Page 63–90, Apr 2006.
Digitopoly 
These are very interesting results. My intuition tells me that the increase in the quality of music is not just related to the total supply of music released but to the dissemination of the music via the internet rather than through record label gatekeepers. For instance, before napster and other file sharing websites for a band to bring a song to the market, they had to go through a record company. This didn’t just introduce entry costs, it introduced a kind of sorting effect that based the eventual release of new music on the expectations of a record company executive instead of on the preferences of the eventual listeners. While these two things may have been highly correlated, they are by no means identical. Today, on the other hand, music is often circulated around the internet fairly rapidly as single files or as an EP, allowing the listeners themselves to decide what bands will succeed and what bands will fail, essentially eliminating the middle-man record company executive and more fully aligning the success of bands with the preferences of listeners, making the music that eventually gets made into an album of higher quality according to your indexes which rely on the revelation of the preferences of the same listeners that caused the albums to end up being made.
This is interesting indeed. I see your data sources are radio airplay and sales. I assume radio doesn’t include internet radio services, streaming and so forth, and that sales don’t include digital downloads. Also, as you mention, sales have dwindled and more music is accessed and consumed through p2p. Is there a way of augmenting your analysis using some of these data to give a wider picture of consumption broken down by vintage?
In order to validate your conclusion about the impact of technology on music creation and distribution (i.e. technology has lowered barriers to entry for new talent), you’d need to check the origin of the contemporary acts/records that are selling, or being rated highly by critics (or at least check the concentration ratios for sales and airplay for different vintages).
My suspicion is that the majority of these are still funded and distributed by the music industry using conventional channels, which, to some extent, would undermine claims about the transformative impact of technology on the mainstream. If anything, music moguls could claim that they are releasing great music nowadays (as per your analysis), and that they’d release even more if it wasn’t for all those pirates syphoning away revenues from A&R.