9
We obtain land-share estimates from Davis et al. (2021), who publish land-share and structure
value estimates for various geographic designations for each year from 2012 through 2019.
When available, we use the 2019 county-level land-share estimates (which cover 85 percent of
the U.S. population). For the counties for which 2019 data are not available, we use their pooled
estimates which represent an average over the period 2012-2019 (covering an additional 13
percent of the U.S. population), which we update to 2019 based on state-level increases in land-
shares.
4
Land-share values are unavailable for 766 counties, but these counties contain less than
2 percent of the U.S population and are sparsely populated, with only 4.4 people on average per
square mile. Finally, we update the 2019 land-share estimates to 2021 based on metropolitan area
increases in home prices from 2019 to 2021, after netting out the 16.7 percent increase in U.S.
construction prices over this time period.
5
IV. Results
We estimate an aggregate U.S. housing shortage of 20.1 million homes in 2021, 14.1 percent of
the stock of existing homes. As reported in Table 1, our 20.1 million national housing shortage
estimate is several times larger than previous estimates, which relied on different definitions of a
housing shortage. For example, Kingsella and MacArthur (2022) and Khater et al. (2021) both
estimate a shortage of 3.8 million homes, and Rosen et al. (2021) estimate a shortage of 5.5
million homes. The White House reports a housing shortage of just 1.5 million homes. As
compounded annual growth rate of the housing stock based on the 2012-2016 ACS five-year pooled sample and the
2016-2020 ACS five-year pooled sample. U.S. Census Bureau, Housing Inventory Estimate: Total Housing Units in
the United States [ETOTALUSQ176N], retrieved from FRED, Federal Reserve Bank of St. Louis;
https://fred.stlouisfed.org/series/ETOTALUSQ176N, April 28, 2022.
4
We update the 2012-2019 pooled estimates to 2019 by assuming that the percent increase in the land-share in the
county from 2012-2019 until 2019 equals the percent increase in the land-share in the state from 2012-2019 until
2019.
5
We first calculate the 2021 home value (in dollars) for each county by increasing the 2019 home value by its
metropolitan area (using Census Bureau 2022a) percentage change in the Federal Housing Finance Agency’s
(FHFA) All Transactions House Price Index (HPI) (FHFA 2022). In the case that a county did not fall within a
metropolitan area, we applied the state level non-metropolitan area HPI change, following the methodology of NAR
(2022). We then calculate the 2021 structure value for each county by increasing the 2019 structure value obtained
from Davis et al. (2021) by the 16.7 percent increase in U.S. construction prices as measured by the Price Deflator
(Fisher) Index of New Single-Family Houses Under Construction (Census Bureau 2022b). The 2021 land value is
equal to the 2021 home value minus the 2021 structure value, which is then expressed as a share of the total 2021
home value. To validate our adjustment, we estimated the national value of housing stock, following the application
of the FHFA HPI values to counties, and compared our estimate to the 2021Q4 Z.1 Financial Accounts of the United
States from the Federal Reserve. The value of all real estate in Q4 of 2021 as estimated by the Federal Reserve
amounted to $75.4 trillion, while our estimate (limited to only residential real estate) amounted to $63.6 trillion. This
indicates that the remaining commercial real estate would be worth approximately $12 trillion, which is
approximately correct.