The source of those stats is not clear enough in my opinion.

Is the "per-stream" measure a result to try and have a comparable basis between the various actors, or is it really the only underlying factor in the calculation (beside the country which is mentionned as another factor)?

My guess would be that services would pay per minute-stream, not per stream. Otherwise, artists that tend to have longer form would not be willing to participate (I'm a big fan of Keith Jarrett, for instance, whose solo concerts mostly consist of tracks of around 40 minutes). You could also imagine hybrid models with bounds that de-linearize certain factors.

Looking at a very aggregate average is a recipe for erroneous conclusions...