Perfect foreseight simulations with large shocks

[jpfeifer]

A variable transformation is also the first thing I would try.