Learning from Past Market Outcomes: Evidence from the Music Industry

Abstract

We leverage unique features of music albums to investigate how musicians learn from current products when developing new products. We find that songs on a musician’s next album tend to be more similar to the songs that are more successful on that musician’s current album. This effect is stronger when the musician has less experience, and when the song on the current album is more novel (for that musician). Our findings suggest that musicians learn from the success of previous songs when developing new songs, and that learning is stronger if the musician has more need to learn, and when the song contains more new information.