The 2020 Census Is Crucial To CRE, And It's Less Certain Than Ever
The decennial U.S. census is best known for determining how many seats each state gets in the House of Representatives (and thus, electors in the Electoral College), but it is a linchpin of data gathering for both the public sector and private companies, and it is critical for numerous commercial real estate functions.
From the U.S. Department of Housing and Urban Development down to municipal governments, census data is crucial in determining how much money is spent building and maintaining affordable housing, as well as where that money gets sent. All told, the census determines the location of $1.5 trillion of annual federal spending. Though statewide population counts are what determine seats in the House, the granular details of each small census tract affect the real estate industry far more.
A confluence of circumstances has made the 2020 census a more complex animal than ever before — pandemic-induced difficulty in door-to-door follow-up campaigns, political pressures from President Donald Trump’s failed bid to include a citizenship question and the introduction of an online form all are unprecedented situations. All of them could possibly increase the risk of an undercount or other inaccuracies.
“There’s more reason for concern this time around than ever before,” former Claritas Chief Demographer Ken Hodges told Bisnow. “One problem we’re seeing is that with the pandemic, at about the same time that the census was going into the field, people were moving to places where they were not supposed to be counted.”
The 10-year census is especially important due to its use as a demographic baseline for the American Community Survey. The ACS is gathered every year and includes much more detailed information, but only 3.5 million to 4 million people are asked to participate every year.
The U.S. Census Bureau claims that is enough to make statistical models that can be representative of the entire population, but it depends on the decennial census to figure out how many people in each cross-section of the population — ethnic group, income bracket or metropolitan area, for instance — participates in the ACS.
A relative undercount of a demographic or metropolitan area means a decade of compromised accuracy for the ACS, which would unleash a cascade of consequences.
Claritas, for which Hodges still performs consulting work, is one of multiple companies that take massive, publicly available datasets, mostly from the ACS and the USCB and pull out data for use in corporate, location-based strategies, like a national retailer determining where to put new stores (or which stores to close).
“Every business of any size is using that privately made software,” George Washington University professor of public policy Andrew Reamer said. “They do not make a decision without using analysis based on that software and the ACS.”
Other companies, like affordable housing developer Pennrose, then take that refined data and turn it into insights that can be more easily digested by those retailers or real estate companies.
“We obviously want to build housing where people are going and not become over-invested where people are leaving,” Pennrose Senior Vice President Tim Henkel said. “We’re also looking at heavy trends, so if you own a portfolio in a state that has an [incoming] migratory pattern that looks to continue for 10 years, we know there would be more competition and we might divest from that portfolio. We haven’t done that specifically yet, but it’s a possible hypothetical.”
States had discretion over which census tracts that contained a certain percentage of low-income households (or neighbored low-income tracts) were designated Qualified Opportunity Zones in the much-discussed program created by the Tax Cuts and Jobs Act of 2017. Such designations, made in 2018, relied on decennial census data rather than from the ACS, meaning the 2020 census could give fuel to critics that complained of tax breaks given to areas that had already improved dramatically over the intervening years.
One of the key outcomes of the census, especially for affordable housing development, is the awarding of Low Income Housing Tax Credits from HUD. The dispersal of such credits can often be the difference between a project getting funded or not, and the credits are affected by state, metropolitan area and individual census tract measurements of population density, average income level and the designation of Fair Market Rent. Rental costs are one of the questions on the ACS, and its answers are the lone determinant of an area’s FMR, Reamer said.
How census data-determined funding is parceled out varies based on the program. States have discretion to declare certain tracts Qualified Census Tracts and/or Difficult Development Areas, which increases the amount of LIHTC credits for an affordable housing project by 30%.
“One state might say, ‘We’re going to give all of our credits to projects next to train stations this year,’ and another might say, ‘We’re only giving credits to projects in very poor areas,’” Henkel said. “Every 10 years, we have to look very carefully at which tracts our projects are relying on that may no longer get the 30% basis boost because its economics have improved, or which now qualify that didn’t before.”
The largest sum of money dependent on census data that affects the commercial real estate industry goes to HUD’s Section 8 vouchers, which allow low-income families to pay rent in certain market-rate housing, Reamer said. A census tract’s FMR determines how much a voucher is worth before HUD sends it directly to households.
Local governments of large and dense cities get to determine where to direct funding from the Community Development Block Grants/Entitlement Grants. Still other programs rely on a combination of state and local input, while being propped up by decennial census and ACS data.
Beyond financial allocations, census data is at the heart of any geographically focused effort at every level of government. A state or city government may conduct its own research or contract with private firms to add to its understanding, but as Philadelphia Deputy Director of Planning and Zoning and Philadelphia Planning Commission Executive Director Eleanor Sharpe told Bisnow, “We swim in a sea of census.”
“We’re always using census data to determine strategy: income, ethnicity, commute times and destinations, [etc.],” Sharpe said. “It informs everything we do.”
The city of Philadelphia uses census data to make crucial decisions about its school district, public health and other policies central to how a city operates. The city’s planning, zoning and housing bodies likely have the most direct impact on real estate, such as matching a percentage of LIHTC funding for projects in targeted areas.
Of the census data it uses, the Philadelphia Department of Housing and Community Development pays particular attention to census tracts federally designated as Racially or Ethnically Concentrated Areas of Poverty for a raft of policies and projects meant to address economic and racial inequalities. DHCD saw a 6% decline in homeownership in ACS data from 2006 to 2016, which DHCD Senior Program Manager Mark Dodds said was the largest drop over that time period for any major U.S. city. His department started tailoring some programs to slow or reverse that trend.
“We want to make sure we know which demographics are most impacted, and we use census data to see who’s the most cost-burdened in terms of renters,” Dodds said. “What we were doing there is assessing the relationship between ethnicity and access to jobs and housing, so we see where assistance is needed the most.”
The Undercount Risk
Everyone interviewed for this article agreed the risk of an undercount is higher this year than an average census. Many (but not all) states have carried out campaigns to raise awareness and participation in the census, as have plenty of cities like Philadelphia with its Philly Counts program. California has spent over $1M since 2017 for its own program, the Christian Science Monitor reports.
“There is a huge fear of an undercount, specifically when it comes to COVID-19,” PHCD Executive Director Melissa Long said. “Undercounting means we don’t get a clear picture of who and where the needs are. There’s not enough funding at all to meet the need, so we need to be as strategic as possible.”
The USCB was forced to suspend its in-person follow-up operation when the coronavirus outbreak prompted a virtually nationwide shutdown, and only resumed in waves over the summer. The door-knocking campaign targets areas that have low rates of voluntary response, primarily low-income areas both rural and urban, making it critical for the ACS to be able to sample accurately from the decennial census data.
If there is a mismatch between the number of tax records in a given area and the number of census respondents, that is one of the indicators an in-person follow-up is needed, Reamer said. Without one, the USCB guesses, or “imputes,” about the non-respondents based on the demographics of those nearby who did respond. In low-income areas, people are less likely to be identified as missing from the count if, say, they don’t have a consistent home address or haven’t filled out tax returns.
“The census can impute people when they know you’re missing, but they can’t impute the people they don’t know are missing, and that’s where you get the undercount,” Reamer said.
Before the pandemic, census employee hiring was lagging behind USCB targets in most of the country. In spring, census workers raised alarms about insufficient protection against the spread of disease. Whether on the worker side or the respondent side, fear of the coronavirus could lead to substantially fewer face-to-face surveys.
In August, about 240,000 part-time workers were hired for the final stretch, but reports in at least one state indicate the Trump administration’s attempts to shut down operations days earlier than a federal judge’s temporary injunction required have translated to layoffs in September. While the coronavirus caused the collection timeline to be pushed back and raised concerns, deadlines for the USCB to report its findings are closer to the count’s end date than ever before. Typically, the bureau wraps up decennial census collection by July 30 and submits its findings to the president by Dec. 31. Though this year's deadline for collection was delayed to Oct. 5, the deadline to submit to the president is unchanged, despite the Census Bureau's request for an extension. Whether that will affect accuracy is an open question.
The Trump administration’s policies and attitude toward immigrants and people of color, and the proposal to add a citizenship question, may have an indirect impact on the census count, as those communities may have more fear and distrust of anyone knocking on their door to collect data for the federal government. Multifamily landlords have had some opportunity to assuage those concerns.
“How do you make sure that those that are in your properties are counted to make sure they get represented fairly? That’s something we deal with on the ground,” Henkel said. “We have supportive services personnel who interact with our managers at the property, and mostly we make sure that census takers have access, that signage is put up, and to educate residents about what is happening and what they could be asked.”
The addition of online responses has been popular — around 80% of responses have come online, Reamer said — but could introduce more scrutiny into the security and reliability of the data.
“The census makes a very powerful pledge to confidentiality, and they take it very seriously, even if some people out there may not believe it,” Hodges said. “But with some new technology, even if the census [is] anonymizing the data for individuals, companies with data on demographics within very small areas could have powerful software that could re-identify individuals.”
The USCB is attempting to combat such invasions with a system called Differential Privacy, in which certain individuals are artificially moved to be identified in a different census tract so as to make a process of elimination impossible, Hodges said. While that may be reassuring for individuals concerned about their private information, the exact method the USCB will use to “introduce noise” is confidential as well, and thus impossible to evaluate.
The implications of an undercount carry the most urgency in the public sector, where Medicare and Medicaid distributions could be a life-and-death issue, on top of all the housing programs. But for the commercial real estate industry, even those outside of affordable housing, the effects would still be felt.
“I would imagine that there are companies interested in what kinds of people are living in a very specific neighborhood, and then you’d be more at risk for any type of census error, whether due to the pandemic, early wrapping up of data collection or differential privacy,” Hodges said.