Suppose we are modeling a process of P as:
dP = h P (M - P) dt + s P dz
where h is the mean reversion parameter and M is the long term mean level.
In certain cases, you may revert to the mean more quickly when P is above the mean vs. when P is below the mean.
For example, if you need to increase staff to accommodate some increase in demand, this would take more time than laying off staff when demand has fallen off.
Does there exist a model which explains / models this phenomenon?