Estimating the Security Market Line and Performance Measurement with Augmented Fama-Macbeth Regressions

This paper introduces an econometric theory which extends the Fama-MacBeth procedure for estimating risk adjusted returns by providing better estimators for security market line (SML) parameters. We show that the FM estimator violates necessary conditions for normality so that studies using it for inference tend to conclude that quot;beta is deadquot; when it may not be. We introduce methodology that (1) is robust to data generating processes; (2) satisfies the Dybvig-Ross critique; (3) shows how the FM estimator underestimates SML slope and; (4) show how puzzling empirical results proclaiming that the SML is flat, may be actually confirming an artifact of specification error in which the slope parameter appears to be zero, whenever its non-deterministic component is zero - even though its deterministic component is non-zero. To address this problem, we introduce simple augmented regressions - with beta as the sole explanatory variable - which provide estimators that stochastically dominate FM estimators, yield locally most powerful (LMP) tests and reject the null of zero slope more often. The methodology also generates a modified Jensen alpha to construct simple conservative econometric tests of portfolio performance measurement