High efficiency is nice, but not essential.
Consistent efficiency is, IMO, the best goal. While you can adjust to compensate for variations from batch to batch, this is going to introduce other variations in the beer you produce. Many homebrewers are happy with some variation batch-to-batch, and this is part of the charm of any homemade product. However, to consistently make the best product possible, maintaining as much control as possible over all steps of the process is essential. This way you can adjust one or a few parameters at a time and have confidence that you know what effect your adjustment had.
One of those parameters, of course, is the efficiency itself. Others have pointed out some side effects of tweaking efficiency, such as changing the rate of tannin extraction. It seems plausible to me that if you want more of the grainy character of the malt, extracting less from a larger quantity may have that effect. I know from brewing coffee that changing the amount of coffee can have unexpected effects. For example, it was a major revelation when I found that the key to avoiding harsh, sour bitterness was to INCREASE rather than decrease the coffee-to-water ratio. This is probably like the tannin effect---you lower your extraction efficiency and leave behind some of the unpleasantness. Of course for beer, the process is more complex (unless you're roasting your own coffee), so things like mash temperature will enter this equation prominently.
In general, I suspect that a high efficiency---one nearing the limit of your process---will probably be more repeatable. My reasoning is that a high efficiency means you "got everything right," while a lower efficiency may come from a number of different errors. This is just a guess, though. My own efficiency is mediocre and variable, so I don't speak with authority here.