The Census Bureau released its Q2 2020 Housing Vacancy Survey (HVS) estimates on July 28. Housing analysts look to this survey each quarter not just for its estimate of the vacancy rate, but also as the only timely measure of homeownership rates and household growth in the US – two key metrics for measuring the current state of housing demand. This Q2 data was to be a first look at the impact that COVID and the related lockdowns had on homeowners and renters.

Unfortunately, the new information doesn’t shed much light on how the pandemic has affected vacancies, households, or homeownership rates. The problem is that COVID forced the Census Bureau to substantially change the way it conducted the survey in the second quarter, which means that the data isn’t comparable to earlier measures. For example, the HVS estimate for the number of households in Q2 of 2020 was 2.4 million higher than the estimate for Q1 which, if it were accurate, would mean there had been more household growth in that quarter than in any year on records dating back to 1965. The change in the estimate for the number of homeowner households was even more unlikely, with the second quarter estimate 4.8 million higher than the first quarter’s, which would be several times more growth than in the largest full year of growth in homeowners ever recorded by the survey. All of this as we entered the pandemic and lockdowns. For this reason, we suggest readers disregard any recent stories that use HVS data to say the US homeownership rate soared in Q2 of 2020 to heights not seen in years, because, unfortunately, this is not a valid conclusion from these data.

The HVS prepared a detailed FAQ explaining the ways in which these and other changes forced upon the survey impacted the Q2 results and gives examples of the most likely implications. According to the FAQ, when COVID lockdowns began in March, the Census Bureau stopped all in-person surveying and interviews for the HVS, which made it difficult for survey takers to accurately determine whether a housing unit was occupied or vacant. As a result of the change in method and accuracy, the HVS estimate for the number of households (occupied housing units) in Q2 is not comparable to earlier estimates, so trends in household growth measured in that quarter are not valid.

The pandemic also significantly lowered the rate at which households responded to the HVS survey questionnaire, which has implications for the homeownership rate reported for Q2. According to the FAQ, response rates for renters appear to have been affected more than owners, which likely had the effect of increasing the homeownership rate an unknown amount. Without knowing how much of the change in homeownership rates in Q2 was real and how much was due to the uneven changes in survey response rates, and without the normal, in-person visits to get more information from those who fail to respond to the survey, the HVS homeownership rates in Q2 are not comparable to those from earlier quarters, and we can’t determine how this quarter’s data impact trends in homeownership rates.

Analysts at the Census Bureau have advised users to consider a ‘break’ in the series in Q2 of 2020, meaning do not compare Q2 data to any earlier data. It is likely that either this data will be revised or there will be another break in the series when HVS can resume in-person interviews and when survey response rates are less volatile. So it appears that we are in for a period of time where trends in housing metrics obtained from the HVS – such as homeownership rates, vacancy rates, and household growth – will be difficult to determine and largely unknown.