Assessing Purchase Timing Models: Whether or Not is Preferable to When

Two types of purchase timing models-those which model purchase incidence (whether or not) and those which model interpurchase time (when)-are examined. We show that modelling purchase incidence is almost always preferable to modelling interpurchase time. We also derive upper bound R2's for purchase incidence models. These R2's serve as useful benchmarks for assessing the goodness-of-fit of empirical choice models.