Rising Rents: Forecasting Housing Inflation at the Metropolitan Level

Abstract

I utilize housing data available at the metropolitan level to build a panel regression model forecasting shelter inflation as measured by the Bureau of Labor Statistics's Consumer Price Index. I achieve an in-sample R2 of 0.54 on data from 18 metropolitan regions from 2002-2022 and, using a panel model regressed on data from 2016-2022, forecast national housing inflation peaking in mid-2023 at 8.5% and falling to 5.9% by the year's end. I identify one year lagged home price and rent growth as the most consistent predictors of shelter inflation, locally and nationally, with the second lag in rent growth, acceleration in the growth of housing stock, and rental vacancy rates also showing statistically significant impacts. As part of two separate investigations, I find that local land use regulation has a statistically significant negative impact on housing supply elasticity in metropolitan areas, but that geographic land constraints do not. Second, I find that while there is relatively little variation in housing inflation between wealthier and poorer counties, lower income counties have experienced slightly higher levels of home price and rent growth since 2018. Uniting these three analyses is the use of newly available home price and rent data at the metropolitan and county levels from Zillow.