We show that the main nonparametric identification finding of Abbring and Van den Berg (2003b, Econometrica) for the effect of a timing-chosen treatment on an event duration of interest does not hold. The main problem is that the identification is based on the competing-risks identification...