3-20：厦门大学马成虎教授讲座预告.pdf

Mean-Preserving-Spread Risk Aversion and Shadow-CAPM: A Dynamic Equilibrium Asset Pricing Model By Chenghu Ma WISE, Xiamen University, China Abstract: This paper concerns the viability of the classical Sharpe-Lintner’s CAPM as an equilibrium model in dynamic settings. Specifically, we will explore the implications of mean-preserving-spread risk averse behavior assumption on investor's sequential portfolio choices, and to examine the extent to which the classical CAPM or conditional CAPM sustained as an equilibrium model for asset returns in dynamic settings. The followings are a brief summary to what we have discovered through this research: (a) In resembling Markowitz's (1952) mean-variance theory for portfolio choices in static setting, MPS-RA investors will optimally hold portfolio in a so-called shadow efficient frontier. Similar to Markowitz (1952, 59)'s static MV efficient frontier, analytic expression of the shadow efficient frontier is obtained. (b) Black-Tobin's mutual fund separation extends to dynamic setting, yet with several critical distinctions concerning the composition of the separating portfolios. (c) The shadow-efficient frontier evolves over time, so are the separating mutual funds. Investor at different time spot or at different state of nature may face different shadow frontiers, and may thus end up holding different mutual funds. (d) As a final product, the so-called equilibrium shadow-CAPM as a dynamic extension to the classical Sharpe-Lintner’s (static) CAPM is obtained in a Lucas (1978) exchange economy with Markov uncertainty. The shadow-CAPM reduces to the conditional CAPM if and only if the P/E-ratios are constant as is the case in IID economies. The empirical violation of a constant P/E-ratio suggests the shadow CAPM rather than the classical CAPM as a dynamic equilibrium asset pricing model. The newly derived shadow CAPM is potentially useful to resolve several well documented puzzling empirical findings in finance. These include the “equity premium puzzle” and the “existence of market anomalies”. The former casts doubt on the expected utility as a normalized assumption on representative agent; and the latter suggests a different performance measure in risk management to the conventional Sharpe-ratio.