Back in 2014, I wrote a post about this 2012 paper that showed a very weak correlation between song complexity (as measured by changes in tempo) and the S&P500 Index. The weak correlation meant that there wasn't much to see, which is somewhat surprising. Our musical tastes reflect our mood, and we already know that the sharemarket is affected by our mood - it goes up more in the summer, and on holidays, etc.
So, I was interested to read this new article by Alex Edmans (London Business School), Adrian Fernandez-Perez (AUT), Alexandre Garel (Audencia Business School), and Ivan Indriawan (AUT), forthcoming in the Journal of Financial Economics (ungated earlier version here, and there is a non-technical summary on The Conversation). Edmans et al. use data on the 'valence' of each song from Spotify. As they explain, valence is a measure of the song's positivity, derived from a Spotify algorithm and measured on a scale of 0-1, where:
Songs with high valence sound more positive (e.g., happy, cheerful, euphoric), while songs with low valence sound more negative (e.g., sad, depressed, angry).
And apparently:
The songs with the highest valence in our sample are September by Earth, Wind & Fire (valence of 0.982), Here Comes Santa Claus (Right Down Santa Claus Lane) by Gene Autry (0.976) and Little Saint Nick - 1991 Remix by The Beach Boys (0.971). The songs with the lowest valence are: RMP by Trippie Redd (valence of 0.0333), I'll see you in 40 by Joji (0.0321) and Legion Inoculant by TOOL (0.0262).
Edmans et al. calculate the 'stream-weighted' average valence of Spotify's top 200 most streamed songs on each day from January 2017 to December 2019, based on the number of streams and the valence of each song. They then correlate that overall valence measure with the returns from the S&P500 Index, controlling for seasonal effects, the day of the week, U.S. holidays, and an index of sunshine (all of which might affect mood). First, they find that:
Consistent with music sentiment being responsive to mood swings of individuals, we document a strong positive association with U.S. holidays and weekends and a strong negative association with post-holidays, winter seasons, and bad weather conditions.
So, it appears that their Spotify-based measure of mood is plausibly correlated with factors that are known to affect mood. They then look at the relationship with sharemarket returns, and find that:
...a one-unit increase in weekly music sentiment leads to a 0.146 decrease in the S&P 500 returns the following week (alternatively, a one standard deviation increase in the weekly music sentiment leads to a 20 bps decrease in the following week returns).
I found this a little bit odd. They focused on the sharemarket returns the following week, rather than the contemporaneous sharemarket returns. The results for contemporaneous returns are available in the working paper version of the article, and show a positive correlation with sentiment. So, overall their results are consistent - a more positive mood (as picked up by musical tastes on Spotify) is associated with an increase in sharemarket returns this week, but then investors correct their overexuberance the following week, leading to a decrease in sharemarket returns the following week. And the reverse for a more negative mood.
Unfortunately, as with the 2012 paper, their results show correlations rather than causation. So, we can't say for sure that it is mood that is causing the sharemarket changes, even though we can tell a good story that explains it as a causal effect. The next step must be to try and find some exogenous change in mood or music tastes, to allow for a test of the causal relationship.
Read more:
No comments:
Post a Comment