I am facing the same problem with yours. I run PSS, SP transient analysis for my PA. At perfectly input/output matching condition, the gain of three cases should be approximately equal. But with transient simulation the gain is 5-6 dB lower than SP and PSS (SP and PSS results are correlated). When I change the integration method of transient simulation (gear, gear2, traponly, ...), the result also varies (output of PA varies from 150 mV ~ 700 mV). What simulation should be trusted?