If I understand correctly your problem / question, you are exciting with a PSD (Power Spectral Density) your structure, and the PSD has all its relevant energy in the range 20 to 2000 Hz. But your structure is so stiff, that the first, lowest natural frequency is higher than 2000 Hz. In this case your body will just behave as rigid for the sake of the vibration, that is there will be no amplification of any kind. The maximum displacement, velocity, acceleration of the exciting source will remain exactly unchanged throuhg your body.
In other words, the FEM analyis is of no interest, if the idealization is correct. The purpose of PSD input is to study the behavior of stress, displacement and so on in a body that gets excited by the input PSD.
The fact that you have a very small mass participation fraction could reflect the high natural frequency of your body. Does the mas participation augment if you, for example, just ask the modal solver to find the first, say 50 modes (with no limitation on frequency)?