Skip Header
U.S. flag

An official website of the United States government

Is it Over Yet? Assessing the Effects of the COVID-19 Pandemic on the Housing Vacancy Survey Estimates for 2020 and 2021

Working Paper Number SEHSD WP2022-11
Jonathan Spader, Daniel Truver, Peter Mateyka, Patricia Holley, and Robert Callis


Rising COVID-19 case counts in early 2020 led to changes in the data collection procedures used for the Current Population Survey’s Housing Vacancy Survey supplement (CPS/HVS), an important source of information about vacancy rates and the homeownership rate in the United States. On March 20, 2020, the Census Bureau suspended in-person data collection attempts and shifted all data collection operations to telephone-based attempts, a change that remained in place through July 2020, when in-person data collection began to be gradually reintroduced. This paper examines the implications of these data collection changes for CPS/HVS estimates. First, the paper presents a series of nonresponse analyses that use auxiliary data sources to describe differences between responding versus nonresponding housing units, as well as to examine changes in these differences over time. The nonresponse analyses find significant changes in these nonresponse outcomes in the second and third quarters of 2020, effects that dissipate in the fourth quarter of 2020 and all four quarters of 2021. Second, the paper develops an alternative nonresponse weighting adjustment factor and examines the implications for CPS/HVS estimates of the homeownership rate, gross vacancy rate, rental vacancy rate, and homeowner vacancy rate. The results suggest that the observed changes in nonresponse outcomes likely contributed to elevated homeownership rate estimates for the second and third quarters of 2020. The vacancy rate estimates are not similarly sensitive to the alternative nonresponse weighting adjustment; however, the results illustrate the potential for the CPS/HVS vacancy rates to underestimate the actual levels of vacancy due to the weighting methodology’s assumption that all nonresponding housing units are occupied. Data users should consider these results when interpreting the CPS/HVS estimates of vacancy and homeownership for the quarters affected by the changes in data collection procedures.


Back to Header