A REEXAMINATION OF THE RELATIONSHIP BETWEEN PREFERENCES AND MOMENT ORDERINGS BY RATIONAL RISK AVERSE INVESTORS

This paper shows that a flaw exists in the logic behind the previously stated theoretical connections between utility theory and moment preferences. In fact, no such relationship exists. There is also a flaw in the logic that postulates that approximate normality can justify moment (e.g., mean-variance) methods for investors' preferences (it cannot), or even that exact normality can justify the use of mean variance analysis consistent with the von Neumann-Morgenstern rationality axioms except for the quadratic utility. A flaw also exists in attempts to analyze the effect of moment changes "ceteris paribus" since "all else the same" is "all the same." Finally, arguments that claim that empirical evidence supports the use of moment approximations to investors' preferences are logically questionable. In our discussion, we have questioned the logical underpinning of the empirics and provided references to numerous empirical illustrations in which the exact opposite relationship has been found.