VoxEU Column

Video killed the radio star: Online music videos and recorded music sales

Artists and music labels have always had to negotiate with the platforms that provide their content to listeners. This column examines the ways that digital platforms, and YouTube in particular, have reshaped the music industry and finds that they fulfill an important promotional function. Because everyone can upload content to YouTube – unlike, say, to MTV – the promotional effect of free music availability outweighs the substitution effect. For artists, this makes any exposure valuable – even user-generated YouTube content promotes sales of the underlying song. 

Musicians and their labels have long had an ambiguous relationship with the platforms that provide their content for free. In the age of digitisation, this relationship has become even more fraught. The slogan “home taping is killing music” has served as a warning that private copies may substitute for purchasing music since the 1980s (Bottomley 2015), but digital platforms like Napster, Kazaa, and YouTube have made the scale of free content distribution much bigger, and the extent of rightsholders’ control much smaller. At the same time, content producers have been acutely aware of the promotional effect of having their music widely heard – the congressional payola investigation of 1959 uncovered music labels willing to pay substantial sums to get their music on rotation at radio stations, presumably to encourage sales (Coase 1979). 

In a recent paper, we examine the effect that YouTube has on music sales and find that YouTube works as an ‘anarchistic MTV’. Because everyone can upload content on YouTube – unlike MTV – the promotional effect of free music availability outweighs the substitution effect. For artists, this makes any exposure good exposure. Even cover songs or other user-generated YouTube content end up, on average, promoting sales of the underlying song. 

Music as an experience good

With some products, consumers can find out before buying if a product matches their tastes or delivers the promised quality. Other products have to be experienced for consumers to know whether they like it. This is the case for music – after all, tastes differ even among fans of the same genre. This is where radio, MTV, and YouTube come in, helping consumers to sample songs before buying.

But radio and MTV differ from YouTube in that they are closed platforms on which content providers control which songs are played, how often, and for how long. This sampling strategy could be designed to maximise content producers’ profits. On the one hand, interactive open platforms like YouTube leave content owners with little control over the amount of sampling, not only because each consumer can choose how often they want to listen to a song, but also because everyone can upload content. On the other hand, content owners are compensated for consumers listening to all versions of their copyrighted content, thanks to automated licensing that includes user-generated content.

Next to cat videos, compilations of ‘fails’, and countless ‘life hacks’, about one third of available YouTube content consists of music videos – official clips, user-generated, cover versions, live versions, etc. As such, YouTube functions as a sampling device for uploaded songs, but clearly some users watch music videos for free on YouTube instead of purchasing the music either online or offline. 

Aware of this possible substitution effect, the German rightsholders association GEMA (in English, the Society for musical performing and mechanical reproduction rights) entered negotiations with YouTube about royalties for their members, which consist of virtually all German artists (and due to international cooperation between national collection societies, virtually all artists worldwide). Negotiations were not successful, with YouTube arguing that GEMA asked for unreasonably high fees. As a result, YouTube started blocking virtually all videos containing music in April 2009. The result of this tussle between GEMA and YouTube was a fundamentally different YouTube experience for German users than for users in other countries. Measured in the number of restricted videos, no other country in the world had less access to popular content on YouTube than Germany – not even Turkey or Iran. 

The effect of the YouTube-GEMA controversy on sales of individual songs

We make use of the exogenously given differences among songs in Germany to estimate the impact of YouTube as a promotion (or substitution) channel on music sales. The exogenous variation comes from the fact that some songs had videos on YouTube prior to the GEMA controversy while others did not. The GEMA block reduced the availability of songs that were initially available on YouTube, while songs without ex-ante availability were not affected. When we compared sales in Germany of songs with videos to sales of songs without videos, we saw that blocking clips reduces sales of the corresponding song by about 5-10%. 

Figure 1 shows that this is mostly due to lower sales in the digital channel, while YouTube’s impact on physical record sales is more limited. Clearly, while official videos might serve as a close substitute for the song available for purchase, a poor live recording or a cover version might not provide listeners with the same experience. Interestingly, we find that blocking all videos – official and user-generated – has the same effect. Our results are not specific to YouTube. When the music industry entered the German market with their own music video platform, VEVO (or ‘video evolution’), in October 2013, sales of corresponding songs increased by about 10%, independent of YouTube.

Figure 1 Effect of the GEMA block on record sales

Note: Vertical axis: Demeaned logarithmic sales. Horizontal axis: Weeks prior/after 1 April 2009. Treatment: Average sales of songs with videos on YouTube prior to the block. Control: Average sales of songs without videos on YouTube prior to the block.

YouTube does not promote all artists in the same way. The effect of the GEMA block is especially strong for artists with ‘broad appeal’ (mainstream genres, language, and presence in the US market) as well as young, lesser known artists. The latter benefit the most: on average, we calculate that YouTube helps increase sales of newcomer artists about two thirds more than it helps to increase the sales of mainstream artists.

Policy implications

Our paper provides a way to understand the externalities that digital distribution platforms can exert on related markets, especially the distributional effects and resulting long-run incentives to produce content (Aguiar and Waldfogel 2018). We can also offer insight into specific policy efforts. In the US, rules governing compulsory licensing mandate that musicians receive royalties from interactive digital platforms, while radio broadcasts are traditionally exempt from licensing. The underlying assumption is that unidirectional media would only complement – not substitute for – record sales. For interactive media that lets consumers choose when and how often they want to consume music, it would be the opposite. Our findings suggest that interactive platforms can also fulfill an important promotional function. If replicated in other settings, such evidence might justify a reform of the existing rules (Lenard and White 2015), potentially extending the license fee exemption to digital platforms.


Aguiar, L and J Waldfogel (2018), “Quality Predictability and the Welfare Benefits from New Products: Evidence from the Digitization of Recorded Music”, Journal of Political Economy 126(2): 492-524.

Bottomley, A (2015), “‘Home taping is killing music’: the recording industries' 1980s anti-home taping campaigns and struggles over production, labor and creativity”, Creative Industries Journal 8(2): 123-145.

Coase, R H (1979), “Payola in radio and television broadcasting”, The Journal of Law and Economics 22(2): 269–328.

Lenard, T M and L J White (2015), “Moving music licensing into the digital era: More competition and less regulation”, Working Paper.



1,470 Reads