It's difficult to tell what's going on in that video because he doesn't "reset" the voices back to voice 1 while demonstrating the pan effect. A riff or chord in the video played on the OB-6 might start with voice 3 but might start with voice 6 on the OBX8. Also, with 8 voices compared to 6 there will have to be a different pan scheme applied. I would say the above video doesn't reveal much about the pan spread differences between the synths.
I did some OB-6 tests one day-
I calibrated at 0% pan spread, both L and R inputs measuring -12.0 dBFS. These are my results after the 1.6.4 update:
pan at 50%=
voice 1 L -12.0, R -13.7
voice 2 L -14.1, R -12.1
voice 3 L -12.4, R -18.2
voice 4 L -18.0, R -12.3
voice 5 L -12.2, R -25.5
voice 6 L -25.6, R -11.8
pan at 100%=
voice 1 L -12.0, R -17.3
voice 2 L -17.8, R -12.1
voice 3 L -12.4, R -37.9
voice 4 L -38.0, R -12.3
voice 5 L -12.2, R - silence
voice 6 L - silence, R -12.3
At 50% the pan progression (one side, alternate voices) starts at -2dB, then -4dB, then -7.5dB
At 100% it starts at -5.5dB, then -20dB, then silence. In all cases one side of the stereo field is kept near -12dB.
I'd like to see this information for the OBX8. I suspect it's a similar scheme but with two more voices accounted for. But I could be wrong, it happens often. If there is a different thing going on with the OBX8 stereo field I'm interested in understanding. (I won't be getting the OBX8 so I can't check it out myself.)