‘A Brush with Catastrophe’: Inside the 2020 Census Meltdown

Twenty days from now, the U.S. Census Bureau will stop its once-in-a-decade count of people living in America, and there are good reasons to think that the results will be at best incomplete—and at worst a miscount that will warp American politics and government for the next decade.

An estimated 60 million American households had yet to be counted when the Trump administration decided that, against the wishes of demographers in the Census Bureau, it would slash a full month from the typical two-plus months allotted to following up with households that didn’t respond. And it’s also shortening the clean-up period that follows, during which the Bureau checks its information for accuracy and removes duplicate counts. That process normally takes five months; this year, the Bureau will have only three. Throw in a continued effort by the administration to come up with a separate count of undocumented immigrants that could be used to exclude them from the numbers used for redistricting, as well as the first time the Census is integrating online-collected data into the in-person count, and you can begin to see the predicament.

“It has a pretty good chance of being the least accurate” census in modern memory, said William H. Frey, a University of Michigan demographer and senior fellow at the Brookings Institution. “This may be enough of a brush with catastrophe that people will realize we don’t want to get this close to it again.”

Why does it matter? Aside from apportioning House seats, the Census provides the statistical basis for many governmental funding decisions for the next decade. Each year, roughly $1.5 trillion in federal funds get redistributed according to census data.

With undercounted populations disproportionately made up of Black Americans, Latinos, lower-income households and immigrants, an undercount would mean not only less political representation for those groups, but also fewer resources going to those communities. “It is a political issue, but it’s also just an issue of access to the resources of our country,” said Frey.

What could go wrong? How is this year’s census shaping up differently than past counts? And what are the short- and long-term effects of everything we’re about to encounter? To sort through it all, POLITICO spoke with Frey on Thursday. A transcript of that conversation is below, condensed and edited for length and clarity.

Stanton: You've studied the census. You’re a demographer. Walk us through how a census normally works, and how the pandemic has changed the way the census is working this year.

Frey: Normally, the census is the culmination of 10 years of hard work by people at the Census Bureau and their partners around the country. It’s a huge civic-engagement exercise that people take a lot of pride in. Census Day — April 1st every 10 years, when the census begins to be administered — is a time of celebration and public events and civic leaders getting everybody to fill out their census forms. The results come in, and at the end of December, they are released to the states and used to reapportion House seats and draw congressional and state legislative districts, which is the official reason for the census.

Then, over the following year, more detailed data come out for smaller geographic areas, different populations and subgroups — like where people are living or where populations are growing — and get used by researchers for the next decade. It’s used for all kinds of planning. I’m a demographer, so I’m biased, but I think it’s probably one of the most important government functions there is. There’s usually not a lot of politics around it; it’s usually a feel-good kind of operation. That’s the kind of census that we were hoping to have.

Stanton: I sense there’s a “but” coming.

Frey: Yes. This year, the Census Bureau was hit with a huge impediment: the pandemic. On April 1, Census Day, most people were worried about the pandemic. You saw very little about it in media, because coronavirus cases were popping up, and places were closing down and people were worried about how they were going to get to work or if they were going to have work. The census was not on people’s minds. More importantly, from the operational standpoint, the Census couldn’t conduct the kinds of operations they normally would.

The pandemic really threw a monkey wrench into the whole process. Around the middle of April, the Census Bureau decided that it had to postpone a lot of its follow-up activities. Normally, they do something called NRFU — “non-response follow-ups.” That’s when they follow-up with people who did not fill out their census forms. Normally, that would take place between the middle of May and the end of July, which allows the Bureau to take the time between the end July and the end of December — when they have to report the results of the census to the president — to deal with all sorts of logistical issues and make sure there weren’t double counts. This is the biggest survey in the world, and they only get one chance every 10 years to get it right.

Under normal circumstances, they would have begun that clean-up at the end of July. But in April, they decided that they had to push back everything because of the pandemic: Rather than have the NRFU go from mid-May to the end of July, they shifted it to go from mid-August until the end of October. As a result, they thought they would not be able to report the results to the president by the end of December and requested that Congress give them a four-month lag so that they would have until the end of April.

But a bigger issue came at the end of July, when President Trump came out with a memorandum that said he wanted to figure out how many undocumented immigrants were included in the census counts of each state in order to exclude them for the purposes of congressional reapportionment. Now, if you've been following this issue, you know that he’s been trying to do something like this for a long time. Initially, he wanted to put a question on the 2020 Census, something like: “Are you a citizen or are you not a citizen of the United States?” It got struck down by the Supreme Court and he wasn’t able to do that. This is a way around that to try to accomplish the same goal. It’s probably not legal, and there are many lawsuits about it, but it takes a while for things to go through the courts. So that’s one part of the bombshell.

Then, the other shoe dropped on August 3, when the president said that all of this planning that the Census Bureau did to postpone the delivery of results so that they could ensure the numbers were good and they could follow-up with non-responses — he wants to rush that, and they still need to get those results in to him by the end of December. And now, the Census Bureau has to stop following up with non-responsive people at the end of September, a month early, which will give them three months to fix everything, instead of the normal five or six. Of course, they cannot do the same kind of diligent job in just three months.

People who previously worked at the Census Bureau — including four prior Census directors under both Democratic and Republican administrations — have come out and said it’s not a good idea. You’re going to wind up with results that severely undercount large parts of the population — the “hard to count” populations, including people of color, immigrants, poor people, very young people, renters and a whole list of people who are going to be undercounted and underrepresented in terms of political clout as well as services and government funds, which use the Census as a benchmark for the next 10 years. Something like $1.5 trillion of federal funds annually get redistributed on the basis of these Census counts. It’s a real problem.

It puts the Census Bureau in a horrible situation. For those of us who analyze data and want to do research—and even more importantly, for people who are planning policy or allocating federal funds—this puts everybody in in a huge predicament. It puts the whole country in a bad situation.

Stanton: So, the deadline moved from the end of October to September —we’re talking about two or three weeks from now. When they moved up that deadline, how far along was the count?

Frey: When the Census Bureau announced it was moving up the deadline, 60 million households still needed to be accommodated. At the same point in the 2010 census, only 47 million households needed to be followed up. It’s a pretty stark difference. But the question isn’t just “How many households?” It’s “What are the demographic attributes of those households?” These are hard-to-count populations, disproportionately people of color, especially young people. Around half of the U.S. population under age 16 identify as people of color.

Stanton: When we look at the undercounted population, how many people are we talking about just not being included in the Census data?

Frey: It’s hard to say. In the 2010 Census, which was thought to be a very good census, 2 or 2.5 percent of Black Americans were undercounted. More than 1 percent of Hispanics were undercounted. But that’s nationally. There are specific neighborhoods across the country that can be severely undercounted. American Indians on reservations are also severely undercounted, and they will be especially impacted by this non-response follow-up issue, because they need to have more attention paid to them, and are more likely to lack internet access to respond online. They were undercounted by something like 4 or 5 percent in the 2010 Census, but that could be much bigger this time around.

Stanton: You noted that this has political implications beyond the reapportionment of House seats. When you think about this particular moment in time, which could be an inflection point on matters of race in America, I wonder if an undercount could actually end up countering a lot of the progress we’ve seen — making some of these problems worse in terms of less representation, less investment and less of a sense of being included in the body politic.

Frey: Yeah, I think that’s right. The census numbers are used to plan where housing is built, where hospitals are built, where schools are built. These are public — and even private — investments. If you’re running a big corporation, you'd like to know where your labor force lives before you open up a plant or facility somewhere. And if you get that wrong and you’re leaving out these people who are going to be your potential workers or customers, that’s a problem. It is a political issue, but it’s also just an issue of access to the resources of our country.

Stanton: Given your studies of past censuses, is there any historical precedent or a similarity to a past census that is useful in thinking through the issues we’re facing right now?

Frey: The only thing that comes to mind is the 1920 Census, which occurred at the end of a decade of immigration and a large degree of urbanization. When Congress got these results — and especially the rural Congressmen — many realized “I’m not going to have a job next year if they reapportion the Congress and all these urban areas get seats.” And Congress did not use the results of the 1920 census; the 1910 Census reapportionment was used up until the 1930 census.

Stanton: Wait, how did they get away with that?

Frey: I don’t know the legalities there in detail, but it wasn’t because it was a bad census; it was because they didn’t like the results.

Stanton: It sounds a little bit like we’re going through an inside-out version of that right now: pre-engineering an inaccurate census for political reasons because the administration thinks it won’t like the results of an accurate count.

Frey: Yeah, I think that is probably some of the motivation behind this. Going back to when the citizenship question was being proposed for the census, a Republican operative named Thomas Hoffeler, who is now deceased, made the case that if you could use “voting-age citizens” as the basis for reapportionment or at least redistricting, this would strongly benefit Republicans, because the younger part of the population is the most racially diverse. So if you could count only citizens ages 18 and older, you leave out all the non-citizen population as well as a large number of people of color and the under-18 population.

Many people think that was the main reason why the Trump administration wanted to have this citizenship question, because then he could have numbers of all voting-age citizens in the United States, down to the neighborhood block level. And that data could be sent out along with the redistricting info, and states could decide if they wanted to use it for redistricting—though it would take a court decision to allow that.

Stanton: There was not a citizenship question on census forms. So how would you actually go about identifying non-citizens and excluding them from the count?

Frey: This is a key question. There are two problems with the Trump proposal: one, it’s probably illegal; and two, how do you get that number if it were legal?

A few weeks after the Supreme Court ruling last year, when they told Trump he couldn’t he couldn’t put this citizenship question on the census or at least had to come up with a better reason for including it, he basically said, “We don’t have time to do that, but I would like to commission the Census Bureau to make arrangements with other government agencies to get statistics on various populations — especially the Department of Homeland Security, which has information on refugees and people applying for citizenship, the Social Security Administration, and the State Department, where people apply for visas.”

They figure that if you can get all of these different agencies to put their pieces of a puzzle together, they could figure out who is a citizen and maybe even get some clue as to who is an undocumented non-citizen. In the most recent version of it, he has the Census Bureau reaching out to states to see if they want to cooperate with, say, driver’s license information and stuff like that. It’s really a hodgepodge, and there are lots of holes in it. It’s making it more difficult for the Census Bureau to actually do its job of counting people.

Stanton: Given all of that, is this likely to be the least accurate Census ever?

Frey: [Pause] I don’t know if I can answer that. I don’t go back to the days when they counted enslaved people as three-fifths of a person. But in modern memory, I would say it has a pretty good chance of being the least accurate.

Stanton: If we look at the 2020 Census and everything that has gone differently than it was supposed to, are you concerned about whether that will be a precedent for future census efforts?

Frey: I try to be hopeful. Irrespective of today’s politics, I’m a demographer. This century, we have a younger, more racially diverse, more inclusive group of young Americans moving into voting age or, for some, into their prime labor-force ages. And they’re going to steer the country in, hopefully, different direction.

You know, this may be enough of a brush with catastrophe that people will realize we don’t want to get this close to it again. As someone who’s worked now for many censuses and had close relationships with people at the Census Bureau, it’s a very good agency. The people there are our top statisticians. They care about doing their work well. This politicization is putting them in an awkward situation, but they’re going to make the best out of it.