Another mystery :-)

I would not doubt the predicted STEEM for the time being.
I would agree that the number of burned STEEM must be considered. Likewise the burned SBD, but there are certainly not many of those.

However, I doubt that over the entire period of your graph the number of average observed new STEEM is constant at 88191. I can't imagine that. Especially in view of the different numbers of burned STEEM.

This is perhaps where I would take another look.