This paper sifts through explanations for the weakness of the out-of-sample evidence on the Phillips curve relative to the in-sample evidence, focusing on the output gap-based models. One explanation could be that, even when the model are stable, out-of-sample metrics are less powerful than...