As a recently "retired" student, this is my perspective:
You seem to be tacitly implying that the efficiency is somewhere around "very poor". If so, I am certainly not surprised.
I found very early on in my mathematics education that if I went to a mathematics lecture in which the material was entirely or almost entirely new, I would comprehend very little of what was said. I would perhaps be able to follow the first few (largely preliminary) definitions and arguments given by the professor, but after that I would get progressively more lost. By the end of such a lecture, I was usually looking at something vaguely reminiscent of Greek, and not at all appealing :)
So what did I do? I took up the habit of learning things entirely before the lecture started. My goal was always to come into the lecture hall having a solid grasp of all the topics covered in that day's section. I found that this helped enormously.
I don't think it's fair to say that "lectures are worthless". But, at least for me, the primary benefit from lectures was seeing a proficient expert work examples that I already had a rough idea of how to solve. I don't think anyone but an intellectual daredevil could simply walk into a graduate-level mathematics class and learn the material cold, with no background whatsoever. Very few people absorb concepts with such rapidity, and even fewer retain them after such brief exposure.
I definitely believe that lectures have the potential to bestow massive intellectual benefits on their students, but those students have to engage at an appropriate level. Passive listening, at least in mathematics, seems rarely to work (if at all).