Parser combinators define a recursive parsing implementation, as well as declaration syntax. While it's very flexible in terms of what it can parse, it has no fixed guarantees on making progress on the token stream or having a bounded lookahead size before they can parse anything.

A hand written LL(1) parser, (for a suitable grammar) is much more efficient than a parser combinator, afaik most production languages use a hand-written parser, with parser combinators not really being used in high-effort implementations.

I have no extensive experience in surveying parser generators (only using them, and they seemed to be fast enough), I'd guess it's up to the implementation on how efficient parsing code is, with my ballpark assumption being that it's worse than handwritten code.