Worldcons Are the Wrong Size
Worldcons would be vastly less expensive per member if they were roughly half or roughly twice the convention's current size. (Approximately 4000-7000 members for North American Worldcons of the past twenty years; different issues apply to non-North American Worldcons.)
At their current size and desired space usage (large fixed exhibit hall areas for dealers/art show/fan tables/historical exhibits, large function rooms for big events like Masquerade/Hugos, lots of small and medium breakout rooms for a large, broad program), Worldcons have grown large enough to require large facilities with heavy fixed costs. (Even "one roof" facilities like the Chicago Hyatt.) However, the convention isn't large enough to spread those heavy fixed costs out, and thus the event costs too much per person.
Limiting membership to 2000 people would radically reduce the cost per member for the convention to organize it; however, it wouldn't change the actual practical membership cost, because the demand would exceed supply and in practice memberships would show up on the secondary market bid back up to the demand cost. (I'm aware of situations where events prohibit reselling memberships, like ComicCon, which has maxed its facilities and thus caps membership below actual demand. I don't like membership caps. They're difficult to enforce and make people mad at you.)
Increasing the attendance up to the 10-12K region would spread the fixed costs better and would lower the cost per membership, although of course the total cost of organizing the event (currently around $1 Million) would go up since there are variable costs per member as well. But increasing the total attendance for a Worldcon is deadly difficult for a different reason that also affects the cost of the convention.
Reinventing the Convention Annually
The second reason Worldcons cost so much more compared to (say) BayCon, Arisia, and Readercon is that they are all one-shots. You may think of Worldcon as an ongoing event, but it really is a series of 5000-person one-shot conventions. Every Worldcon is a brand new convention that is organized by a single organization from the ground up. Every Worldcon then is torn down after the event and disperses to the four winds, while the following Worldcon is in the process of starting up from scratch. There are no economies of scale to speak of. Everything has to be created anew every year.
Imagine if Readercon, after this year, dissolved and a completely different organization (possibly with some overlapping people) in, say, San Francisco organized next year's Readercon. Then that organization disbanded and a brand new Readercon in London happened the following year, followed by Vancouver, then Des Moines, then Sydney, so forth, so that year it was a new convention. It would be pretty difficult to keep the price down when you have to pay your start-up costs every year.
Westercon actually has a similar problem because it also is an ongoing series of one-shots, but the effect is slightly less noticeable because it's smaller and shorter.
If you held "Worldcon" in Anaheim (or Chicago, or Boston, or Atlanta, etc.) every single year, organized by the same group, the convention cost would come down because they could work out multi-year deals, achieve economies of scale, and work on growing the event (see above about how the in-bewteen size of the event makes it expensive). But how could you call it Worldcon with a straight face when it never goes anywhere but one city?
(I have read people who seem to think that rooting the convention and growing it in a single site would be just fine, and that people should just save up to attend, like the people who save for years to attend one of the mega-sized pop culture events. Alternatively, they think it should follow a fixed rotation of major US cities, as long as one of those cities is one near them, since "What's in it for me" is of course the most important thing in the world. Whoops, did I say that out loud?)
(Oh, and for all those Americans who moan about the horrible cost to travel to Europe or Australia, I suggest you just try and think about what it's like from the European and Australian point of view. For that matter, what about people like Peter Watts and Cheryl Morgan, who aren't even allowed to enter the USA thanks to idiotic US policies? Should they just lump it because you want Worldcon to be convenient to you every year?)
Worldcon as the Confederation Congress of the USA
I'd like to suggest that a lot of assumptions about why Worldcons in particular cost what they do and run the way they do are based on mistaken, albeit logical, assumptions about the nature of the organization. Worldcon isn't really a single convention. It's actually arranged more like the United States of America was between the Declaration of Independence and the Constitution of 1787: a set of semi-autonomous groups that cooperate when they must but who eye their neighbors warily and look out for their own interests before those of anyone else.
There is of course more to the issues than one article can address. Feel free to ask me anything about what I know about Worldcons. I've spent the better part of twenty years trying to include transparency on these issues. Worldcons are a very strange sort of convention compared to most SF/F conferences. More than once I've explained all of the constraints and conditions under which we run them and been told, "You're crazy! Nobody would ever do that!" Well, maybe they shouldn't, but we've been doing it this way now for over seventy years, and somehow it works.