Testing for the presence of a broken linear trend when the nature of the persistence in the data is unknown is not a trivial problem, since the test needs to be both asymptotically correctly sized and consistent, regardless of the order of integration of the data. In a recent paper, Sayginsoy...