When excess returns are used to estimate linear stochastic discount factor (SDF) models, researchers often adopt a normalization of the SDF that sets its mean to 1, or one that sets its intercept to 1. These normalizations are often treated as equivalent, but they are subtly different both in...