I think it HAS to be that the Kill-a-Watt is not measuring a non-linear load properly. If the PS was burning 150 W internally, a small fan would NOT be able to cool the hot components effectively. Saying that 2 75 W bulbs in the housing would not burn up is not germane. There are several components in the PS that will be generating all the heat. Switching transistors, diodes and maybe transformers are what will get hot. Real Kill-a-Watt meters are SUPPOSED to properly measure non-linear loads up to some reasonable crest factor, like 10. Probably most non-PFC corrected power supplies won't be too much worse than a crest factor of 10 (I hope).
So, you might check this with a real electromagnetic wattmeter or a Kill-a- Watt that is known to handle non-linear loads accurately.
Or, if you want to go nuts, get a digital scope or other scope, digitize the current waveform and numerically integrate the true power draw.
Jon