Testing for Regime Switching: A Comment