Asset Pricing Research On Some Classical Stock Return Anomalies With Agent-based Computational Finance

Since 1980s many stock return anomalies not be explained by classical financial economics have been found. Specially, momentum, long-term reversal, excess volatility and excess comovement are much more well-known than others. Because of their strong impacts on asset pricing and risk management, academics and practitioners pay much attention to them. As far as theoretical research is concerned, they have been focuses until now.The dissertation depends on agent-based computational finance (ACF) mainly to check up these anomalies and test whether the hypotheses on how they form are scientific. ACF is different from traditional approach, which adopts bottom-up modeling strategy and emphasizes heterogeneous agent's interaction and design. Simulation data are obtained through building and running artificial stock markets.In detail the hardcore of the dissertation is composed of three related parts below:(1) SRA-ASM simulation model is developed and used to examine applicability of investor type and the values and effects of important parameters.(2) The expectation modes of chartist and fundamentalist are redesigned with the help of current literature. Then a multi-factor return equilibrium model (DMF-REM) is operated, which explains above anomalies in a united framework.(3) A multi-risky asset simulation model MRA-ASM is created on almost the same investor assumption in DMF-REM. It supplies quantitative test about anomalies and discusses their formation conditions from the general view of market supply and demand equilibrium and investor interaction.In a word the dissertation illustrates that, as for stock return anomalies, ACF is good at dealing with proposition that traditional approach disables and also could match traditional approach as well as possible for improving analysis ability to complicated problems.