Okay, here is my "extensive" analysis of the two methods in terms of F2L. LL is pretty much all in the same direction. Petrus has the upper hand, on that one, though.
Step 1: Cross vs. 2x2x2
2x2x2 has more freedom, seeing as you have 8 options over 6 (but Fridrich users generally aren't color-neutral).
Cross generally has easier-to-see patterns, making it more accessible to beginners.
In terms of move count, we're about the same: 5-7 moves (correct me if I'm wrong).
Step 2-3: 2x2x3 + Bad Edges vs. CE Pair 1-2
Technically, CE pairs have more freedom in terms of choices, but Petrus also has three faces free for turning.
CE Pairs all have generally the same finger tricks, making them easy to pick up fast.
Once again, doing the 2x2x3 extension is harder to see [psuedo]-optimal solutions. Also, watching Johannes's cube simulator fastest solve seems to indicate a slight pause between 2x2x2 and 2x2x3. This, however, is compensated for the fact that the 2x2x3 phase with bad edges is on average shorter (in move count) than doing two CE pairs.
Step 4[-5]: Rest of the F2L
Petrus seems to have the upper-hand here. It can constantly use RU, while Fridrich may take a couple extra faces. The cases are pretty much the same between the two methods.
From this, I argue that the two methods are pretty much balanced in terms of solving time. However, over the past couple of years, Fridrich has enjoyed greater popularity, hence more work has been done on it. In the long run (theoretically speaking), I think that which method is "faster" just depends on the specific scrambles and the cubers themselves. In the words of Jessica Fridrich, "we should not be comparing systems but cubists."
