Any idea if this technique could be used for dividend distributions on chain?

Specifically I’m wondering if Merkle roots can be efficiently calculated on chain? Will run tests myself soon too.