|Budget Amount *help
¥3,900,000 (Direct Cost : ¥3,900,000)
Fiscal Year 1998 : ¥700,000 (Direct Cost : ¥700,000)
Fiscal Year 1997 : ¥3,200,000 (Direct Cost : ¥3,200,000)
In this study, we constructed an artificial market model of a foreign exchange market, which is a new agent-based approach. Using this model, we analyzed emergent phenomena of markets, which traditional economic theories can not explain well. Our study consists of three steps : fieldwork in markets, construction of a multiagent model, and computer simulation.
First, we asked 30 dealers about their forecast factors, prediction methods, and learning mechanisms by inter-views and questionnaires. The acquired data were analyzed from viewpoints of learning theory in cognitive science. The results reveals that the dealers' prediction methods are always changing and that the dealers may interact with others by imitating others' prediction methods.
Second, based on the field data acquired, we constructed a multiagent model of a market using genetic algorithms. In this model, dynamics of prediction methods is described on the basis of analogy from biological evolution process.
Finally, four emergent phenomena of markets were analyzed using the simulation results of the model. First, we analyzed rate bubbles in 1990 and 1995, and found that the bubbles were caused by the trend-following behavior of market participants and convergence of forecasts. Next, the results of analysis also revealed that the idea of "the phase transition of forecast variety" could explain three emergent phenomena in markets : negative correlation between rate fluctuation and trading volume, peaked and fat-tailed distribution of rate changes, and phenomena of contrary opinions.