Working Paper: NBER ID: w4702
Authors: Shmuel I. Kandel; Robert F. Stambaugh
Abstract: A plot of expected returns versus betas obeys virtually no relation to an inefficient index portfolio's mean-variance location. If the index portfolio is inefficient, then the coefficients and R- squared from an ordinary-least-squares regression of expected returns on betas can equal essentially any desired values. The mean-variance location of the index does determine the properties of a cross- sectional mean-beta relation fitted by generalized least squares (GLS). As the index portfolio moves closer to exact efficiency, the GLS mean-beta relation moves closer to the exact linear relation corresponding to an efficient portfolio with the same variance. The goodness-of-fit for the GLS regression is the index portfolio's squared relative efficiency, which measures closeness to efficiency in mean-variance space.
Keywords: Portfolio Management; Asset Pricing; Mean-Variance Analysis
JEL Codes: G11; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Inefficient index portfolio (G11) | misleading regression outcomes (C34) |
| Improved efficiency in index portfolio (G11) | more accurate mean-beta relation (C46) |
| Index portfolio approaches efficiency (G11) | fitted GLS mean-beta relation converges to exact linear relation (C51) |
| Higher relative efficiency of index portfolio (G11) | more accurate representation of expected returns in relation to betas (C46) |