Asymmetric persistence of accounting income is often tested in a regression of changes in earnings on lagged changes in earnings, including an interaction term for negative changes (see Basu [1997] or Ball et al. [2009] for a recent overview). In this note we propose an alternative, but closely...