The "force spectrum" method is an "envelope / statistical" solution. It is the same numeric solution that is used for "response spectrum seismic" evaluations. This method may not be accurate (compared to time history), but it should be conservative.

"Time history" on the other hand is very precise - assuming you know the loading "precisely", and all of your other input data are correct/precise. It is very easy with time history to get the "wrong" answer (because of some uncertainty in the input).