New testing approaches for mean–variance predictability

We propose parametric tests for serial correlation in levels and squares that exploit thenon-normalityoffinancialreturns.Ourtestsarerobusttodistributionalmisspecification.Furthermore, our mean predictability tests can be robustified against time-varyingvolatility. Local power analyses confirm their gains over existing methods, while MonteCarlo exercises assess their finite sample reliability. We apply our tests to quarterlyreturns on the five Fama–French factors for international stocks, whose distributionsare mostly symmetric but fat-tailed. Our results highlight noticeable differences acrossregions and factors and confirm the numerical sensitivity of the usual tests to influentialobservations