Location, location, structure type: Rent divergence within neighborhoods

Abstract

Housing rents are a large share of household budgets and make a large contribution to overall inflation. Hence rent measurement accuracy is necessary in these and other contexts. The location–location–location hypothesis suggests that rents within a neighborhood would grow at the same rate on average; prior research has both theoretically assumed, and empirically supported, this hypothesis. We show that, even within the same neighborhoods, rent inflation rates for different types of housing units often differ for years. Over the 2010s, apartment rents generally outpaced detached unit rents; this pattern reversed during the COVID-19 pandemic—providing nuance to prevailing narratives about the latter period, most of which are location-specific. These dynamics suggest that structural representativity is necessary for rent index accuracy and accurate inflation measurement. Even indexes based on careful geographical sampling, such as the United States Consumer Price Index’s (CPI’s) Owners’ Equivalent Rent component prior to 2023, may be biased by placing too much weight on apartment rents compared to detached unit rents. We demonstrate that this bias may be quite large, and offer recommendations — one of which was recently accepted by the CPI.