I do not believe that there is any theoretical difference. By "theoretical", I mean using principles of work, energy, etc. without friction. The original post asks about efficiency. So, I'm assuming jamdaaman agrees that there's no theoretical difference.
If somebody presented experimental data that showed a measured difference in practice, I could be convinced.
How? Chain tension, sag, and related things would be different, depending on the direction of gravity. Imagine a very lightly loaded drivetrain, with an extremely heavy chain. I could be convinced that all the minute frictional losses would change with gravity direction in a measurable way.
Edited to add: Similarly, I could be convinced that frictional losses differ in the case of a horizontal arrangement, depending on which span of chain is driving the load, top or bottom.