I'm a professor (in mathematics), and there are some reasons why I personally might say "yes" to review a paper. (1) If the paper is so interesting that when I see it I think "I really, really want to read this paper!", then I am likely to peer review it because as a reviewer when I read it, I get to ask any question and the author will take my questions and concerns very seriously. (2) I owe the editor who is asking me to peer review the paper (or maybe even the author of the paper), e.g., if the editor put a lot of work into ensuring a paper I once wrote got properly reviewed. (3) Even if I say "no" to fully reviewing a paper, as an expert in the area I can often provide quick valuable feedback about the relative value of the claimed contribution, and who is most likely to be suited to review it; I'll do a quick informal review like that in some cases.

From the outside, it is easy to forgot the extreme extent to which academics are motivated by (or view themselves as motivated by) principles of "good citizenship". This is selection -- the people who aren't this way, often do not get hired or promoted. Also, many academics (at least in pure math) view motivation for money as a "lower order term". It's a good thing for them, because for much of your long career in academia there is little you can do to impact your salary, besides applying for a job somewhere else. For me it's always been: each year you get some maximum possible merit raise of between 0% and 4%, depending on external economics that the department has no control over. Academic book royalties might also raise your yearly salary by 2%. Being highly money-motivated in some parts of academia would end up being very frustrating indeed. E.g., even when I've got a big NFS grant, that doesn't change my salary one bit; instead, it changes how many students and/or postdocs I could support.