The disentanglement of adverse selection from ex ante moral hazard remains an empirical challenge. Our comment dissects a natural experiment proposed by Chiappori and Salanié (2000) to test for ex ante moral hazard. Firstly, we argue that their test, as proposed, is too simple and too general to enable reliable inferences about the existence of ex ante moral hazard to be drawn and the reported negative coefficient does not rule out moral hazard. Secondly, their analysis strongly suggests that their proposed instrument (inherited bonus malus) is endogenously determined and therefore does not satisfy the technical requirements of a natural experiment.