In 2013, David Lowery, a singer and guitarist for the band Cracker, was aghast with the music service Pandora. In a now infamous blog post, Lowery lambasted Pandora for paying him only $16.89 in royalties despite playing Cracker’s 1993 hit song “Low” over 1 million times.
But Lowery’s attitude towards Spotify, the music streaming giant, is very different. He calls himself a “Spotify agnostic.” The 7-year-old Swedish company — which has an $8.5 billion valuation and has seen the number of its active users grow from 15 million to 60 million since 2010 — has a business model different from Pandora’s, one Lowery sees as revolutionary for the entire music industry. “Music streaming, as opposed to ownership, is the future,” he said in a phone interview last week. According to the Recording Industry Association of America, total streaming revenue is up from $0.5 billion in 2010 to $1.9 billion in 2014 — and now amounts to more than a quarter of all earnings for the music industry.
The debate over the economic impact of streaming is still being hashed out. The biggest flashpoint occurred last year, when pop megastar Taylor Swift decided to pull her catalogue from Spotify. Swift, writing in The Wall Street Journal, claimed streaming had “shrunk the numbers of paid album sales drastically.” In response, Spotify CEO Daniel Ek pointed out that the company had paid $2 billion in royalties since 2008, and claimed streaming was promoting music sales overall.
But perhaps the Swift-Spotify spat is much ado about nothing. While streaming is profoundly changing how music is consumed, services like Spotify could be neither cannibalizing nor boosting music sales. That is the conclusion of published, but not-yet peer-reviewed, research by two economists, Joel Waldfogel and Luis Aguiar.
They set out to determine, in their words, “not only whether streaming displaces sales but, if so, at what rate.” To do so they measured the growth of Spotify streaming and the sales of digital music in 21 countries for an eight month period in 2013, and on all music sales (including physical ones) in the U.S. from April 2013 through March 2015. What they found corroborated Swift’s claim, at least in part: more streaming led to fewer digital track sales. Specifically, they estimate that for every 137 streams there was one fewer sale.
Coming to this number took some major data jiu-jitsu, however. At first, the economists compared song- or artist-level streaming with sales of the equivalent singles or albums, and they saw a positive correlation — as if Spotify were stimulating sales. But this is deceiving, they concluded. “People stream stuff that’s popular at the same time people are buying stuff that’s popular,” Waldfogel said. Streaming, sales and even piracy all move in tandem, so it’s difficult to analyze the effect of one activity on another at such a granular level.
Instead, the researchers thought a better approach would be an aggregate analysis — to look at how streaming activity as a whole affected sales as a whole. They looked at total Spotify streams for the U.S., Canada and 19 European countries. To get those numbers they examined the total streams of the top 50 or top 200 songs in a country by week, covering the period from April 2013 to March 2015, and then used that data to estimate total streams by country.
In a given week and in a given country, the economists estimate the total streams of the top 50 songs on Spotify represented between 7.7 percent and 10.8 percent of all Spotify streams in that country. Those Spotify streaming totals were then analyzed alongside digital download sales through April 2013, using data from Nielsen, to measure the rate of sales displacement (which produced the takeaway figure I mentioned earlier: on average, 137 more streams equals one fewer sale). For a more current analysis, the economists examined the effect of Spotify streaming on U.S. sales — of both digital downloads and physical albums — from April 2013 through March 2015, using data from Billboard, and they found a slightly smaller rate of displacement.
Will Page, director of economics at Spotify, is respectfully skeptical of the study. “There are several methodological shortcomings,” he said in an interview on Tuesday. The researchers had to restrict their international analysis to eight months in 2013 because of the limited data from Nielsen; that’s before Spotify’s growth really took off in 2014, he pointed out. (Waldfogel was also quick to acknowledge that they didn’t have all the relevant data.)
Furthermore, Page sees the decline in music sales as a structural shift in the industry that would have happened regardless of the rise of streaming. “The business of [music] ownership has been in decline long before Spotify came along,” he said. “Is it fair to assume people were going to continue buying in the first place?” He used an analogy from a different industry, comparing Netflix’s growth alongside the collapse in DVD sales.
While the study backs up Swift’s claim that Spotify displaces record sales, the service’s overall impact on artists and the industry is more complicated. As the company is keen to point out, it’s paying billions in royalties. While Spotify’s exact songwriting royalty rate is not publically known, and can vary depending on the deal struck with music labels, the company claims its per-stream payout is “between $0.006 and $0.0084.” That too is disputed: David Lowery, the musician, thinks this is too high; from surveying a couple of “medium-size indie catalogues,” he thinks the per-stream rate is around $0.005.
For their analysis, Waldfogel and Aguiar took Spotify at its word, and used the median per-stream point ($0.007) to estimate any new revenue generated through streaming. They found that revenue roughly offsets the revenue lost from any drop in sales. In other words, they think Spotify has had a revenue-neutral impact on the music industry, though whether Spotify has cut into artists’ earnings is a different question not directly addressed in this paper.
There is one thing Page and Waldfogel do agree about: legal streaming’s effect on piracy. In the working paper, the economists compared the growth of Spotify streaming to the weekly volumes of torrents for 8,000 top artists, using data from Musicmetric, a music analytics company. Every additional 47 streams displaced one pirated download, they concluded. The economists even cited Page’s own work showing Spotify lowered unpaid downloads. But as Waldfogel said, “reducing piracy is not the same as increasing revenue.”
And even if Spotify isn’t making the music industry any poorer, and its effect on revenue is a wash, it still might be creating winners and losers. Lowery has experienced both. For his band Cracker, which had a handful of hits, a majority of revenue (he estimates 97 percent) comes from sales — and Spotify streams — of those songs. Lowery sees Spotify as a good thing for Cracker: “Probably a fair number of those people [streaming] wouldn’t buy the album, or even the track. I look at that and see it as additional revenue.”
But Lowery said his other band, the alternative rock group Camper Van Beethoven, is a loser in Spotify’s emergence. The band, Lowery said, is more “album-oriented and culty.” In the pre-streaming world, diehard fans would have purchased Camper Van Beethoven’s music, he believes; but with Spotify, they can get it cheaper or even for free.
Spotify may not be changing the overall size of the music industry’s revenue, but it’s still shaping how the money is divvied up.