Except that, by and large, those reviewers aren't being paid. So the 'outsourcing' argument isn't quite it. If it were there would be no problem with open access.

It is more about brand. They're outsourcing the need to determine quality by buying into a brand. That brand is kept afloat by unpaid labour on all sides, yet charges heftily for its brand. It is an understandable model, but not a very pleasant one. Unless you're an Elsevier shareholder, getting paid for your brand IP.