Endogeneity and moments in time series momentum’s predictability test

The predictability test is critical in developing momentum trading strategies. In time-series momentum (TSM) tests via linear predictive regressions, the classical and Newey–West t-tests have size distortions because of endogeneity and a lack of enough finite moments. To tackle these issues, this paper proposes a new test that features a model of the error correlations, weighted least squares estimation, and the random weighted bootstrap method. Simulations confirm its accurate size and good power. Empirically, we revisit the evidence in (J. Financ. Econ. 135 (2020) 774–794) and find that the TSM predictability detected by the new test is much more widespread than that by the Newey–West t-tests.