关键词: 模型不确定性, 极大极小期望效用, 投资组合有效前沿, CAPM

Abstract: The paper takes model uncertainty into account in studying portfolio selection problems,and proposes a definition of robust portfolio frontier along the line of mean variance analysis We find that model uncertainty has unequal effects on portfolio weights of risky assets when risk-free asset is inaccessible,which leads to under diver-sified portfolios We also find that the two fund separation theorem no longer stands under robust portfolio frontier and the same is true for the zero-beta CAPM model However,when there is a risk-free asset in the economy,we find that model uncertainty has equal effects on portfolio weights of risky assets,and that the two fund separation theorem still holds,as all robust frontier portfolios can be generated by the risk-free asset and the market portfolio.The CAPM model also can hold under equilibrium except for a change that there is an additional factor representing loading of model uncertainty,and the excess rate of return of any portfolio can be divided into riskpremium and uncertainty premium.

Key words: model uncertainty, max-min expected utility, portfolio frontier, CAPM