Using PAO and/or Ester based synthetics does result in significantly lower fluid temps - particularly under high loads in transmission/differential applications. However this is mainly due to the improved thermal conductivity of these fluids.
I have also seen consistent fuel savings with both synthetic engine oils - particularly the 0w-20/5w-20/0w-30 grades - and low vis, synthetic drivetrain fluids. I believe a lot of this is from a reduction in viscous losses during the warmup phase, which can be as much as 10-15 miles during cold weather. Note that coolant temps come to equilibrium much faster than oil temps in the engine/transmission/differential. The rapid rise in coolant temps is due to heating of the engine block from combustion gasses....
In addition, if you compare the VI's of synthetic lubricants to petroleum oils of the same SAE grade, the VI's of the synthetics can be much higher. So even though they may technically test out to the same SAE grade, the synthetic will easily meet the spec, while the petroleum oil barely will. For example, the CCS viscosity of a typical 10w-30 is about 5000-6200 Cp @ -13F and the VI's are in the 130-145 range. By comparison, the CCS viscosity of the PAO based, 10w-30/"ATM" formulation is only 3100 Cp @ -13F and it has a VI of 174. So pumpability of the 10w-30 synthetic is significantly better and it will provide improved fuel efficiency and lower fluid temps.
Finally, I also think that the PAO/Ester synthetics provide improvements in upper cylinder lubrication at the ring/cylinder interface, as well as in the very high pressure zones in the valve train. If you can avoid boundary lubrication conditions, friction in the valvetrain is significantly reduced.
I have not done enough controlled testing of Group III synthetics to know if they offer the same advantages of PAO's and Esters. But I intend to do this type of testing in the near future, using the Amsoil "XL" line of Group III oils.