Saturday, November 10, 2012
Taking Agent Based Computational Economics to a new level
A greast little item in the SMH today about an economist applying theory to the question of the economies inside online games. A couple of interesting comments are included, For example, many would agree that; "Professor Varoufakis believes economics has ''hit the wall'' as a credible source of policy insight because its mathematical models are ''irrelevant'' or even ''dangerously wrong'', crippled by the lack of sufficiently accurate, complete and timely real life data." The kinds of events like asset bubbles and banking crises occur, despite theory, because of the interaction between agents tin the formation of prefernces that is ignored by traditioonal economics - and can't be reliably modeled. The interaction between agents is in pat transmitted by the fact that preferences are formed by expectation of future events - houses can be valued on the basis of their rental income (or the opportunity saving of not paying rent) or on the basis of their future value for esale. Bubbles grow as the latter becomes more important in valuation than the former. They pop when the gap between the two becomes so large that valuations start shifting back to the former. The article notes that; "Part of the aim is to avoid speculative bubbles and banking crises such as the one in Second Life that caused players to lose $750,000 in real money in 2007. He is also analysing the transactional data for its ''tremendous theoretical and empirical interest'' and writing a Valve blog." Question - do you think the makers of Stream could be encouraged to make a version of a game that could be played by economics students with the intention of testing the boundaries of collapse?