【摘要】 经典资产定价理论建立在投资者同质、完全理性、市场不存在交易成本、资本完全流动和无套利均衡会自发实现等假设的基础上，重点揭示了不同资产价格或收益率之间的关系，但忽视了财富约束、交易量和交易过程的作用。这些严重偏离现实的设定，导致经典资产定价理论的解释力严重不足。为发展解释力更强的模型，本文假设面临流动性约束和交易限制的投资者具有异质信念、有限理性，市场价格按撮合交易方式而非瓦尔拉斯均衡机制进行出清和调整，针对多风险资产市场构建了一个反映异质投资者群体交互作用的资产定价模型，得到了风险资产价格演化方程。在理论分析的基础上，本文构造出了收益率预测模型和投资组合选择模型，并在收益率预测精度与最优资产组合的盈利能力上与经典模型进行了对比，结果表明本文的模型具有更好的实战表现。在收益率预测精确度上，本文模型与经典的时间序列自回归模型相当，但比后者能够更加精准地识别收益率分布尾端的极值风险。而在组合盈利能力上，基于本文模型构造的HAM组合的日均收益率和考虑了风险折价后的夏普指数都显著优于其他经典组合。如果进一步拉长持有期，本文HAM组合在30～250个交易日的持有期下的累计收益率年化值随持有期限的延长而增加，波动率随持有期的延长而降低，并且综合表现显著优于相同持有期限下的其他备选组合，因此更加适用于持有期限较长的机构投资者。更多还原

【Abstract】 The classical capital asset pricing model(CAPM) and the associated portfolio selection models are built upon the assumptions of homogeneity and complete rationality of individual investors, as well as free liquidity and free arbitrage in market equilibrium. These assumptions differ significantly from the reality, making it hardly to generate accurate forecast for the yield of risky assets from the classical asset pricing models. As a complement to the classical theory, this study proposes a novel asset pricing and portfolio selection model that is derived naturally from a heterogeneous agent model(HAM) with the continuum of bounded-rational investors, liquidity constraint and the non-equilibrium price adjustment. Through deriving the price evolution equation, we find a four-fold decomposition of the temporal price variation for the risky assets, which consists of the variation arising from the bounded-rationality-incurred deviation between the market averaged yield rate and the yield rate implied by CAPM, the deviation between the market structure induced by the heterogeneity of investors and the homogeneous market structure, the mismatch between the demand and supply on individual risky assets, and the mismatch of the total demand and supply of risky assets incurred by the wealth effect. Combining the evolution equation for the yield rate of risky assets together with the real data of stock prices, we present an adaptive approach to forecast the stock yield rate and implement portfolio optimization. By applying both our adaptive approach and the classical CAPM-based approaches to the stocks in the China’s A-share market, we make a numerical analysis to compare the relative performance of different approaches in terms of both the forecast accuracy and profitability of the optimized portfolio. The results demonstrate that our adaptive approach outperforms the alternatives in a series of evaluation criteria, which proves the potential of the proposed HAM in the fields of asset pricing and portfolio construction.更多还原