I think it all comes around to how the funds are distributed.

Researcher gets grants from the federal government to enable their work. They do the work, write the paper, get it accepted to a conference. They're required to attend and present to get it published. So they have an excuse to buy flights, hotels, expense food all on government dime. And the conference is put on, in part, by other researchers, who aren't going to use their own funds to put on the event. So they charge people to attend, and those who want to get published have the largest incentive to attend, so they can be charged the most.

I only did 2 years of graduate research and attended a handful of conferences (unpublished unfortunately)... I could have this wrong, but I'm pretty sure this is the way, at least in the computer science field.