Using a new transaction level dataset on new vehicle purchases from J.D. Power, this paper documents violations of assumptions underlying cost-of-living theory and constructs a new price index that address these violations. First, we show that there are significant differences in consumer characteristics at different points in a typical product cycle, which suggests profit maximizing behavior through intertemporal price discrimination. When the composition of consumers changes each month, observed prices do not represent the purchases of a typical consumer at each point in time. Consumer segmentation violates assumptions necessary for measuring cost-of-living and can introduce bias into matched model price indexes. Second, we construct a price index that uses product age to control for consumer heterogeneity at different points of a product's life cycle. Specifically, we compare the price of a vehicle at a particular point in its life cycle to its previous-year's model at the same point in its life cycle, which we implement by assuming stability in the pattern of model year introduction and comparing each vehicle to its model year equivalent as of 12 months prior. Finally, this year-over-year price index does not convey information about short-run fluctuations in new vehicle prices, so we reincorporate highfrequency price change into the index by using a novel filtering technique. This new index averages 0.24 percentage points lower growth than the current CPI for new vehicles on an annual basis.