Suppose you find, as research indicates, that in the cross-section regression of the CCAPM, the coefficients of factor loadings on the Fama-French model are significant predictors of average return factors (in addition to consumption beta). How would you explain this phenomenon?