[AnimalMuppet]

The paper's data set is from 1870 to 2015, which is 145 years. If that's not close enough to "the long run" for the theory to match reality, then the theory is either wrong or useless.