If a paper is EXTREMELY well written (which, out of hundreds of papers, has only happened once or twice), I might just suggest a different word or perhaps a comma or some formatting here or there. Still, reading very carefully, about 4 pages per hour is the max I can review.
 
If a paper isn't written QUITE so well, I'll also suggest moving phrases around within sentences, and probably MORE wording, punctuation, and formatting changes. Then I MIGHT be able to edit up to 3 pages per hour (but again, this is quite rare).
 
Most economists don't write very well (and don't know it, and don't care). There's usually poor organization, lots of unnecessary repetition, confusing changes of terms, etc. As soon as I realize that a paper isn't very well written, I start breaking up paragraphs into individual sentences and essentially rewriting each paragraph. Then I MIGHT be able to edit up to 2 pages per hour (though I just spent 2 hours on 1 page!).
 
I try of course to accurately present the ideas and to preserve what seem to be the important phrases used (though I insist on standardizing them where possible, which is one reason why feedback can be useful during the process). But within those parameters, I try to improve the writing as much as possible.
 
Finally, if organization is especially poor, I may have to move considerable text between paragraphs, reorganize subsections, etc. In that case, I'll probably be down to about 1 page per hour.

 

If I'm working in the file -- rather than on paper -- then it might go slower, especially if there are lots of formatting issues (for example, tables) -- or it might go faster, since it's easier to work with "clean" text to see what I've already suggested.
 
Some people don't want such thorough editing, and that's fine. But I can't do it. I do what I believe is required to make the writing better.

 

That not only takes considerable time, but it's a creative act which can't be stressed.