This depends on how it is designed. I have two different power supplies, one has a mains connector with earth connection, the other one has only live and neutral.
I tried to measure resistance from output side to the input side to see if it was completely isolated (no path for the current) or if they have implemented some sort of reference to anything on the input side.
The one with the ground connector had the outer ring ("0V") connected with a 1 MegOhm resistor to mains earth. So in this case the output was referenced to ground and the output voltage was measurable with respect to ground. (I tried that and I measured the 19VDC from mains earth to output (inner ring))
The other one is different, it had no measurable connection from output to either neutral or live. I tried measuring the voltage with respect to ground, but on every pin (+12V and +5V and 0V in this case) I got the same voltage (314mVDC). This means that there is no reference to ground in any way, otherwise I should have been able to measure the output voltage at least with some offset, but all pins had the same voltage to ground.
My measurement influenced the circuit in the second case (the multimeter created a connection to ground which was not there before).
If there is a grounding pin (in the sense of earth), I would think, that 0V of the output and ground are referenced to each other. If you mean the third pin you see sometimes in a laptop charger, those are often used for communication to identify the charger and their capabilities (you can get 45W, 60W and 90W chargers, so the laptop has to limit the battery charging currents in some cases).
So you see that there are different approaches on how to implement this. I am not a designer of power supplies, so I cannot comment on which way is better, safer or needed by regulations.