could be right...
I was going off my historical observations from my work on flash memory, but after some thought I guess on a uP you would not expect it to go down much. What I was thinking is that the FET's get much weaker, which should result in lower power... but in a full CMOS clocked circuit that will not happen. I suspect in the SRAM sense amplifiers, and any dynamic logic you would see a drop in power though... but that should be very small compared to the CVF power... and would appear as part of the "leakage" power.
Thanks for the clarification.
--Alan