The price is determined solely as a function of excess demand with 价格的确定仅作为一个功能的需求过剩.docVIP

Professor Sheri Markose CF902 Artificial Stock Market Models: INTRODUCTION (Rationale for ASM : Pioneered by Brian Arthur et al at Santa Fe Institute (1977); Shu- Heng Chen(2001 , JEDC); Hommes et. al. (2007, JEDC) Traditional rational expectations models for stock market behaviour could not explain the boom bust stylized facts of stock market prices. Indeed, the traditional RE model results in the no trade result or that speculative trading does not exist as all agents must have HOMOGENEOUS RATIONAL EXPECTATIONS. That is, rational agents who form expectations of the stock price given the same information set as in the Fama type assertion on efficient market hypothesis (EMH) states: prices, P, that contain all (publically) available information, H, follow a martingale such that price changes are random and not serially correlated. If the theory of emergence or algorithmic undecidability of strong reflexive encodings is not explicitly considered, it is easy to become a victim of the fallacy of composition. It can be held that the global outcome of non-anticipative prices is the consequence of rational agents who believe this to be the case. That is, if i indexes all N agents, the following states that all agents have homogenous beliefs that the asset price is a random walk (i Ei (Pt+1 | Ht) = Pt , i= 1,2,……., N. In the case when assets are traded for purely speculative reasons, viz. for pecuniary gains based on price expectations, homogenous price expectations results in the non-existence of speculative trading. We have the paradox that with the cessation of trade, the price at t+1, Pt+1 , never gets determined. If one is to take the Samuelson (1965) view that ‘proper’ anticipation of prices implies taking conditional expectations, E(Pt+1| Ht), it is clear that this is without unique procedural content. In Arthur et. al. (1997) they make a case for heterogeneous multi-agent models where each agent uses genetic algorithms to arr