Understanding the spatial variation in housing prices plays a crucial role in topics ranging from the cost of living to quality-of-life indices to studies of public goods and household mobility. Yet analysts have not reached a consensus on the best source of such data, variously using self-reported values from the census, transaction values, tax assessments, and rental values. Additionally, while most studies use micro-level data, some have used summary statistics such as the median housing value. Assessing community price indices in Los Angeles, we find that indices based on transaction prices are highly correlated with indices based on self-reported values, but the former is better correlated with public goods. Moreover, rental values have a higher correlation with public goods and income levels than either asset-value measure. Finally, indices based on median values are poorly correlated with the other indices, public goods, and income. Housing is the most important asset and largest expenditure category in most households’ budgets. Accordingly, accurate data on the value of homes is a lynchpin in many economic studies. For example, because housing accounts for about 30 percent of households’ expenditures, housing costs play a key role in computing geographic comparisons in the cost of living, such as the US ACCRA index, as well as intertemporal indices of inflation. An accurate representation of home values is also a critical step in many empirical studies of local public goods.

Through the process of capitalization, home values are deeply intertwined with spatial public goods such as school quality, crime, air quality, hazardous waste sites, and green spaces, and the taxes that pay for them. The differences in housing values associated with differences in these public goods and taxes have long been used by economists to infer people’s demand for such goods. 1Prominent examples of such “hedonic” methods include applications to intercity quality-of-life measures (Albouy 2010, Blomquist, Berger and Hoehn 1988), education (Black 1999, Figlio and Lucas 2004), crime (Bishop and Murphy 2011), racial segregation (Bajari and Kahn 2005), air quality (Chay and Greenstone 2005, Grainger 2011, Smith and Huang 1995), superfund sites (Gamper-Rabindran, Mastromonaco, and Timmins 2011, Greenstone and Gallagher 2008), cancer risks (Davis 2004), and property taxes (Palmon and Smith 1998).