Because of the random timing of the stimulus payments, Parker, Souleles, Johnson, and McClelland have a compelling natural experiment, and their estimates suggest that the stimulus payments had large effects on households' spending on automobiles (see especially their Table 7, which indicates that roughly one-third of the stimulus payments were spent on new vehicle purchases). It is natural to interpret this evidence as implying that the stimulus payments raised aggregate spending on new cars relative to what it would have been otherwise by roughly one-third times the total amount of the stimulus payments, or about $66 billion. Suggest two reasons that the true effect could have been much smaller.