A somewhat downbeat assessment of the current state of gold farming emerges from a discussion with Dr Jack Qiu from the Chinese University of Hong Kong; who has been a regular observer of gold farming and other informal sector activities in China.  I report this as, and appreciate that it is, conversational rather than "hard" evidence.

Jack observes four phases of Chinese gold farming.

1. Globalisation (c.2003-c.2006)
The growth that we are all aware of, serving the global and regional games market

2. Localisation (c.2006-c.2007)
A very rapid expansion of gold farming serving the local Chinese market, spreading West from the Southern/Eastern coastal origins of gold farming.  In particular, related to Shanda’s Legend games.  At least some of the livelihoods created in this way were "playbourers" who received payouts from Shanda itself.

3. Partial Decline (c.2007-c.2008)
A severe decline of gold farms ("workshops") serving the local market for the Legend games; particularly due to a furore over the business model that Shanda had been using, and a notorious case of "gold slaves" in which students at a South China university were found to be forced by their professors to undertake unpaid "internships" if they wanted to pass their classes.  These involved gold farming either all night (male students) or all day (female students).  As evidence of the decline, Jack cites the example of  his own home city of Wuhan (popn. c.6m), which had some 4,000 "workshops" that grew up during the second phase to serve the local market but which now has few, if any.

4. Current State (c.2008-date)
Gold farms serving the global market remain as, too, do farms serving the local market.  The latter particularly represent a very "footloose" form of capitalism.  Footloose in the traditional sense of readily closing down operations in one place and opening up in another that has low wage and low rent costs and has recently obtaining decent broadband access.  And footloose in the sense of games.  The workshops will target a new or growing game, and try to find ways to make money out of it for a month or so.  If it succeeds, they stick with it for the moment. If not, they move on to the next game.

One further conclusion I noted from our conversation: there is a whole lot of data about gold farming that never makes it across the dual barriers of geography (it stays in China) and/or language (it stays in Chinese).