Beschreibung:
The growth of online platforms has raised questions about their power and the possibility that it could be exercised with bias, including by gender. Women account for about a fifth of the most successful artists at Spotify, prompting some concerns about bias. We explore the roles of female participation, along with promotion decisions at the platform - in particular playlist inclusion - in explaining the female share of successful songs and artists at Spotify in 2017. We employ two broad tests for gender bias. First, we ask whether songs by female artists are differently likely to appear on global playlists, conditional on the past success of the artists, song characteristics such as genre, and gender. Second, we test for bias in New Music Friday playlist ranking decision based on outcomes, asking whether songs by female artists stream more, conditional on their New Music Friday rankings. We find some evidence consistent with bias (in favor of women at Today's Top Hits as well as in the New Music rankings, and against women at some global playlists). These biases, however, do little to explain the low female share of streaming on Spotify, which we instead attribute to the relatively low share of female songs entering the platform.