with Pierluigi Balduzzi, Tommaso Bonafini, and Roberto Savona, 2026, working paper.
[Paper] We exploit the fact that factor betas regulate both the time-series (TS) systematic volatilities of asset returns and the cross-sectional (CS) dispersion of asset risk premia. This leads to a link between the TS and the CS fit of a linear factor model. Specifically, we factorize the CS R2 into two components: the first component captures the factor model’s average TS fit across test assets, while the second component reflects the risk-return trade-off of the model’s optimal factor portfolio (OFP) relative to the tangency portfolio of the test assets. Our factorization explains why adding factors to a model may reduce the TS fit and why the introduction of weak, but profitable factors, can improve pricing. Our results provide an interpretation for the longevity of the CAPM and for the skepticism regarding the empirically-motivated pricing factors proposed in the literature; i.e., the “factor zoo.”