I'm trying to use the methodology proposed in the article "Taming the Factor Zoo: A Test of New Factors" to evaluate whether some new factors can have significant explanatory power on asset prices beyond the set of previously proposed factors. In the paper, they assess a new potential factor by looking at the significance of its estimated SDF loading. However, little is said about the sign of the SDF loadings and the connection to marginal utility.
From what I've been able to find in other papers, only the absolute value and significance of SDF loadings tend to be considered to say something about the pricing power of the factor. Is it not possible to make more nuanced inferences addressing things like the sign of the SDF loading and the connection to marginal utility?
For instance, if I were to test a new factor capturing company controversy level, how would I interpret a significant and positive SDF loading on this factor? What if it's negative? How does the sign relate to marginal utility?