An Agent-based Simulation Research About The Relationship Between Information Lag And Investment Return

In order to study how the information-lags affect investors' return,this paper constructs an Agent-based artificial financial market through statistical simulation method.The market only deals single stock whose valuation is determined by the sum of many independent information sources,where the value of each information source is a random walk process.And there are many virtual investor called “Agents” on the market.After the the information source is updated,Agents need to experience a duration before they receive the innovation.The duration obedience exponential distribution,but the expected values of the distribution are different among all Agents,so that some Agents tend to receive innovation earlier than others on average.In the basic framework,Agents estimate asset's value based on their own information,and determine weights of cash and stock based on the difference between their estimation and market price.Then they place orders in the market according to their demand and supply.New equilibrium price were established through their simulated trading.Through our analysis about the price change process,I find that the simulated price can better reflect the stock's real value,but the return sequence has time series relevance which not appears in the difference sequence of real value.Besides,the volatility of price sequence is much smaller than real value sequence.Agents' invest return is negatively correlated with their information-lag.After,we have modified the framework of first experiment,let Agents have learning ability.Each Agent will estimate a restricted regression model.In the model,the dependent variable is sum of all information source,the independent variable is price of last period and it's estimate value of the stock.Then Agents trading based on these estimated values,and new price has been established.According the modify,Agent with longer information-lag tend to estimate real value based on price of last period,and Agent with shorter information-lag tend to estimate real value based on it's own information.Loss of Agent with longer information-lag has reduced in this experiment,but the basic conclusion is not changed compared with the previous experiment.In another experiment,we add Agents represent inverse investors and technical traders into the framework of experiment 1,and we found that they can acquire outstanding positive return.But with the growth of those investors,the excess returns will eventually disappear.Finally,we add Agents represent noise traders on the basis of first experiment,they make decisions based on the difference between the market price and the price of the previous period.In this case,the autocorrelation property of returns sequence disappears,and volatility increased significantly.Meanwhile,wealth differences greatly reduced among each group of Agents,however,it still has a significant correlation with the information receiving lag.