Do Asset Demand Functions Optimize Over the Mean and Variance of Real Returns? A Six-Currency Test

Working Paper: NBER ID: w1051

Authors: Jeffrey A. Frankel; Charles Engel

Abstract: International asset demands are functions of expected returns.Optimal portfolio theory tells us that the coefficients in this relationship depend on the variance-covariance matrix of real returns.But previous estimates of the optimal portfolio (1) assume expected returns constant and (2) are not set up to test the hypothesis of mean-variance optimization. We use maximum likelihood estimation to impose a constraint between the coefficients and the error variance-covariance matrix. For a portfolio of six currencies, we are able statistically to reject the constraint. Evidently investors are either not sophisticated enough to maximize a function of the mean and variance of end-of-period wealth, or else are too sophisticated to do so.

Keywords: No keywords provided

JEL Codes: No JEL codes provided


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
Investors do not optimize their asset demands according to the mean and variance of real returns (G11)Likelihood ratio test rejects the constraint of mean-variance optimization (C52)
Rejection of the hypothesis of mean-variance optimization (G40)Investors lack the sophistication to maximize their end-of-period wealth (G11)
Rejection of the hypothesis of mean-variance optimization (G40)Investors may be employing a more complex intertemporal utility function (D15)
Expected returns and variance-covariance matrix of returns (G17)Coefficients of the asset demand functions (G19)
Parameters of the asset demand functions depend on expected returns and risk aversion (D11)Actual behavior of investors deviates from theoretical expectation (G41)

Back to index