To measure globally, aggregate locally: Urban land, submarkets, and biased estimates of the housing stock

This article highlights the role of land and housing submarkets in the construction of aggregate house price indexes and in the estimated value of the housing stock. We document idiosyncratic house price appreciation and sales volumes across submarkets within metropolitan areas that result in a sample of sold homes that are representative neither of the housing stock nor its appreciation over time. Commonly used aggregate price indexes, like the Federal Housing Finance Agency (FHFA) and Case-Shiller indexes, fail to capture these local dynamics and produce indexes and estimates of the value of the stock that are biased. Our approach is to build "locally-pooled" indexes and then weight these local price indexes by their submarket's share of the housing stock to estimate our metropolitan-level index. This allows for more accurate movements in urban land prices, which is especially important in higher cost land markets. We show traditional globally pooled indexes exhibit significant bias. This may complicate research that makes use of the traditional house price indexes.