I'd like to start this by saying how incredibly unqualified I am to discuss this on a serious analytical level; I simply find the subject fascinating. That subject is, as indicated in the title, the presence of emergent economic structures in MMOGs, and it's covered in (part, but also in) great detail at this article at Gamasutra. I'll be referencing this (or at least using it as a starting point) as I go into digressions on a topic that I feel is interesting enough that I think I may have already written about it. I'm not sure, though.
Anyway, some perspective. I don't have a lot; I'll be the first to admit that, at least in the realm of MMOs. My only real experience with them was a browser-based Java MMO (Runescape) I enjoyed in my teens, that for all its simplicity, has seemed to do pretty well for itself. I haven't found the time or inclination to sink my free time into that kind of endeavor since then, but one of the things I know I found incredibly interesting at the time (and perhaps moreso, in retrospect) was the nature of the player-driven economy. I suppose what was interesting about it was the fact that it was player-driven.
I understand that there are more robust economies out there - the article referenced above uses both EVE Online and WoW as touchstones, and given their player bases it is not hard to see why. Their sizes alone should account for substantial evolution of a social economy. However, my benchmark can only be my experience.
The article linked above was a publication on "faucets" - those inputs and outputs of the economy. One of the first points made was that in virtual worlds, inputs have no real limit to them. While certain items can be limited in availability, most base items (fish, ores, wood, etc.) are effectively unlimited, unlike the Real World. This means that (with time as effectively an unlimited resource) gold can simply come from nowhere, gold that ostensibly retains as much value as the gold that already exists - inflation is not an issue since prices remain largely fixed. This seems to work, economically speaking, because large-scale increases in resources generated usually represent increases in player base, which represents a counterbalancing drain on the resources available.
The author also makes a point about specialization being an archaism in terms of virtual worlds. I think that I might diverge at this point, but I will do so with the caveat that this may simply represent my inexperience or limited awareness of the genre. One of the tools I've seen game developers use to combat this (although I'm not sure this is necessarily viewed as a weakness versus strength issue) is by making skills interdependent and by minimizing the amount of cross-pollination seen when training one skill. For example, in my experience, mining was most effective to train and use when one simply "gave up" his ores acquired - both reducing the impact on the economy and minimizing the opportunity to cross-pollinate (with skills like smithing). It was also in the player's best interest to specialize, as the ability level required to fully use all aspects of one resource (for example mining-smithing-crafting) was prohibitive enough to make it most financially viable to do one. Or maybe I'm simply talking out of my posterior. I can't really say.
The most interesting thing about the economy I'm familiar with, though, was the presence of true limited-input, available-to-all items, and it's a thing that I don't know that the author of the article addressed. One of the things I saw in the Runescape economy was the presence of limited-release "holiday" items (like New Years' hats or Santa Claus caps), and they were truly interesting because they served as benchmarks for the value of other items. Since they were unable to be increased in number in the game (and in fact could be removed, by turning them into gold or simply dropping them and letting them fade away), they maintained a relatively even value (all things considered). When new items were introduced, you could see how prices fluctuated and changed based on their percentage of the value of a party hat, for example.
I think that a large part of this is simply me looking up with wonder saying "Wow, gee, I can't believe these things exist." It's really quite exciting and quite interesting (anthropologically? sociologically?) to see that things of this complexity can really exist when we give enough people a sandbox and some time.