Crist,
If I understand your last comment about PWM being "more aggressive", I would assume that the rpm is somewhat higher. I am not sure I would have predicted that. However flying faster of course would cost you more energy. Note that PWM doesn't have much to do with the commutation, except that it should be faster than the commutation. For 8000-9000 rpm the ESC is commuting in the 2000Hz range for our 14 pole motors. Actually I am not sure if it is 1/2 or 2x 2000Hz, but it is somewhere in this range. I read on the CC Forum that the new "Out Runner mode" automatically adjusts the PWM up as the prop rpm goes up.
Naively, a higher PWM has two effects, one good and one not-so-good
The good concerns the motor. I would say "ideally" the PWM rate would be fast enough so that the motor "feels" like it is being driven by a nice continuous DC current (this is ignoring the commutations). A low PWM means that the motor is being driven by discrete pulses of full voltage--sort of like hitting it with a hammer (an exaggeration). For the motor, the former is more efficient, since the current is smoother (almost constant and equal to the average current) while in the latter case you get higher peak currents. But of course you get the same average current since we would be turning the prop at the same rpm. Since motor heating goes as the square of the peak current *the duty cycle, you make less heat (in the motor) with a higher PWM. Any heat you make is power not going to the prop.
The not-so-good concerns the ESC. In the analysis I use above, I normally assume the battery current is either full on (during the on part of the PWM drive) and full off in the off period. So if we are running 70% throttle, we are pulling current 70% of the time from the battery, and are off 30% of the time (so you have 70% duty cycle). Now the reality isn't like that. As the ESC turns on the FET's, they go from basically full off (infinite resistance) to full on (almost zero resistance), but for a short period during the transition,they have a finite resistance, and this is what heats up the ESC. A higher PWM rate means more transitions per second, so the FET's will be in this transitional phase more than at a lower rate. Again, more heat means power not going to the motor (plus ESC heating).
I seem to think that the "good" of high PWM tends to beat the "bad" of high PWM. It isn't a lot, and from some data Igor Burger gave a year or two ago, it looks like at most a 5% effect.
I don't think this is inconsistent with your results, if you were indeed flying faster. It goes to show you that there are a lot of variables in this business, and it is hard to really change only one thing at a time. At least we don't have to worry about the combustion process of an IC engine too!