“Oh, great,” you’re probably saying upon seeing the headline, “one more thing in this crazy administration that I can usually take for granted, that I now have to worry about.” And while it’s true that the risk that Republicans somehow screw up the 2020 Census isn’t as tangible a threat to people’s day-to-day-lives as the threat of, say, deportation, or losing your health coverage, it’s more of an over-the-horizon risk. But the risk of not adequately counting everyone—whether it’s just through bureaucratic sloth, or through actual malign intent—has some real perils as possible outcomes.
For one thing, the Census is used to allocate the government’s financial resources. Funding for a particular function, or grants to state and local governments for a certain purpose, might depend on how many people are in a particular place. Many of the most basic functions of government—whether it’s paying for upkeep of roads or plans for new road construction, or allocating money to school districts for hiring teachers and maintaining buildings—depend on accurately knowing in advance how many people will be using those services. Figuring out how many people there are, and where they are, is what the Census does every ten years (and, just as importantly, that decennial count serves as a baseline for annual estimates on how the population in every place has kept growing or shrinking).
And for another thing, the Census is used to allocate actual political power. It’s the basis for the House reapportionment process every 10 years. A precise count of how many people are in each state might determine whether or not your state gets another House seat (or gets one taken away) if your state is right on the cusp. States will actually get in protracted legal battles over Census measurement techniques if there’s a House seat at stake (as seen, for instance, in the fight between Utah and North Carolina in 2001 over where Mormon missionaries serving overseas should be considered domiciled).
Often, the most vulnerable populations, the people who are most likely to need and benefit from government services, are also the hardest to count. The homeless are the most extreme example (since the Census works by matching people with addresses) but it’s true in immigrant communities as well, where residents might be suspicious of talking to random government officials knocking on their door, even if those officials speak the language and are from the community. Even non-immigrant communities where there’s a lot of poverty are hard to count, because there’s simply a lot of moving from place to place because of evictions, temporarily bunking with extended family, and the like. Even a reasonably well-done Census runs the risk of undercounting those communities (like, for instance, the 2010 Census, which New York City claimed vastly undercounted the population in Queens).
Unlike most of the other things that the federal government does, the obligation to perform the Census is spelled out right there in the Constitution: Article I states “Representatives and direct Taxes shall be apportioned among the several States ... according to their respective Numbers ... . The actual Enumeration shall be made within three Years after the first meeting of the Congress of the United States, and within every subsequent Term of ten Years.” Of course, that’s all it says; it doesn’t give any specifics about how the enumeration needs to be done, but in a nation of more than 300 million people, it has become a huge undertaking.
The Census isn’t just a matter of “Okay, let’s send out some forms and see who sends them back.” It’s often compared to a military mobilization, in that the government has to rapidly absorb hundreds of thousands of new employees, train them in new technology, deploy them to various parts of the country, and then demobilize them just as quickly a few months later. (In fact, one unfortunate side effect of a generally healthy economy is that there might not be a large enough pool of people eager to take on short-term employment; compare now to 2010, when the Census was actually a big short-term boost to the economy by hiring a lot of unemployed people.)
With the Census two years away, the ramp-up is already happening, and the Washington Post’s Catherine Rampell explains how arduous those preparations are:
While the enumeration itself doesn’t happen until 2020, planning begins years in advance. The Census Bureau must try out questions. It must test-drive technologies — especially important this time around, because for the first time, the questionnaire will be administered primarily online.
The bureau must also figure out which community groups can help with public outreach and what kind of messaging will be most effective in getting people to stand up and be counted. This last task is especially challenging in an era marked by record distrust of government.
So far, what we’ve seen in terms of preparations isn’t very encouraging. The Census traditionally goes through actual “dress rehearsals” in the years leading up to the real event, and the dress rehearsals this time have either been outright cut—two out three field tests scheduled for 2018 have been cancelled, and the first planned ad campaign to encourage participation has been delayed—or else pared back in scope. The one remaining field test, for instance, is more focused on the new technology and less on question-testing and community outreach.
Part of the problem with the preparations is simply the issue of funding and bureaucratic entropy. The 2018 Trump administration budget requested $1.5 billion for the Census (for 2018 alone), but that’s $350 million short of what the Obama administration said the Census Bureau would need, and there’s no guarantee that Congress will even pay for that level of funding.
The total cost of the 2020 Census is now estimated around $15.6 billion, an increase of 27 percent over previous estimates. Part of that is simply keeping up with the nation’s fast-growing population, but the escalating costs are also associated with playing catch-up with deferred projects that got put off because of the same chronic underfunding earlier in the decade. One example is the need to build a database of every address in the nation using satellite imagery (which is a big savings from the usual way of doing that, which is having workers fan out across the country to physically examine every address); tests on that system have fallen behind, meaning the need for more field workers instead.
Related to that (and compounding the bureaucratic confusion) is that the Census Bureau is in a leadership vacuum for now, and the proposed new leadership may not be much help. The previous director, John Thompson, resigned in June (he was near the end of his expected term anyway, though frustration over funding issues may have hastened his departure), and there is no proposed appointee to replace him. There’s also no permanent person in the deputy director position (which has been filled on a temporary basis by a career civil servant), and the proposed appointee here may wind up doing more harm than good.
The proposed deputy director is political science professor Thomas Brunnell, who has no governmental experience, or really any experience with federal statistics or managing a large organization. (He’d been considered for the director appointment, but it was clear he’d have trouble getting Senate confirmation for that job; deputy director doesn’t require confirmation. The director and deputy director positions are typically filled by civil servants who have decades of Census experience and have risen through the ranks in the Bureau.) On top of that, observers think his appointment would be a troubling politicization of the usually dry and nonpartisan job; Brunnell has appeared as an expert witness on behalf of Republican legislatures in a number of states, in cases involving lawsuits over gerrymandering.
Even Brunnell’s academic work seems to show his hand ideologically:
In his 2008 book, “Redistricting and Representation,” he argued that partisan districts packed with like-minded voters actually lead to better representation than ones more evenly split between Democrats and Republicans, because fewer voters in partisan districts cast a vote for a losing candidate. He has also argued that ideologically packed districts should be called “fair districts” and admits that his stance on competitive elections makes him something of an outlier among political scientists, who largely support competitive elections.
It’s still not clear whether Brunnell will ever actually get the job, based on the pushback the announcement received. If he does, though, there are a lot of subtle things he can do that would largely fly under the radar but might influence who participates in the Census. For instance, he would have control over the Census’s large advertising budget, which is spent on trying to encourage groups that are unlikely to participate in the Census to participate. He would have the discretion to, for instance, tip the advertising budget away from Spanish-language public service announcements.
He’d also have a lot of discretion with the wording of the actual Census questions, which is another looming problem for the 2020 Census. Part of the problem is simply that new questions need to be road-tested to see if they generate proper responses, or if the new questions generate a lot of confusion, and get skipped over or answered incorrectly. And if the Census Bureau is deferring its “dress rehearsals” over budgetary issues, that increases the risk of the problems not revealing themselves until the big event in 2010.
But the decision to include a particular question can have a big impact on whether or not people participate. One potentially huge one is the question of citizenship, and the Justice Department requested in December that the Census include a citizenship question on the 2020 Census form. The job of the Census, of course, is to count everyone in the country regardless of citizenship (the total amount of population is what, for example, puts the load on a city’s roads or sewer system, not the number of citizens). But the presence of a citizenship question is likely to deter participation by not just non-citizens, but anyone from a race or ethnicity who’s likely to feel targeted regardless of their actual citizenship status. (There’s no indication, yet, that the citizenship question will eventually appear on the 2020 form; for one thing, it’s a little late in the game to start adding questions.)
The Justice Department claims that knowledge of citizenship status is necessary for enforcement of Voting Rights Act Section 2 cases. However, the ACS or American Community Survey (an annual, ongoing survey that contacts many fewer people and provides estimates for the broader population, rather than contacting everyone in the country) has effectively replaced the Census “long form,” which, until 2000, was sent to a selected participants in the 10-year Census in order to get further information about education, income, housing costs, ancestry, and details like that) does ask about citizenship as part of its battery of questions, and ACS data has been used in VRA cases.
And dissuading participation by people who might be intimidated by the citizenship question would have the net result of undercounting, for instance, Latinos or Asians—generally, people more likely to vote Democratic. The undercount would not only have the effect of meaning less money for services allocated to the places they live, but it could even mean less representation for them in the House, in state legislatures, or anywhere where political power is divvied up by district (like city councils or school boards).
At the highest level, it could even mean a loss in representation for California. California is, according to some analyses, on the cusp of losing one of its seats in the House (which would take them down to 52, from the current 53). This may seem surprising, since California’s population (unlike other states at risk of losing seats, like West Virginia) is still considerably growing, but the issue may simply be that it’s not growing fast enough. Failing to count enough of the state's large Latino population could be what pushes California past the point where it loses a seat. (To avoid that end, California’s state government is planning to spend a lot of money on its own Census-related outreach.)
Of course, that risk is a two-way street. It could also, for instance, wind up hurting the state with the nation’s second-largest Latino population: Texas. The Lone Star State isn’t at risk of losing seats, but it is on the cusp between gaining two seats or gaining three seats. Not counting the state’s Latinos adequately could ensure that the state only gains two seats, so there might still be some pressure from certain red-state Republicans to keep that kind of question off the decennial Census form. (And the risk of lost seats certainly weighs on other red states, though in Alabama and West Virginia the Latino population isn’t much of a factor in that likely loss.)
(Which isn’t to say that Texas gaining only two seats would be chalked up as a long-term Democratic victory. It’s possible, say, that an undercount where the effects are focused on the Rio Grande Valley or in central Houston would affect the outcome at the redistricting, not reapportionment level. In other words, Texas would gain two seats, but they’d be two Republican seats, rather than two Republican seats and one Democratic seat, based on where in the state the population growth appears to have occurred. In addition, it could be a long-term harm if by, say, 2028, Texas has finally started to make its long-threatened shift into the Democratic column in the electoral college.)
Finally, there’s one other reason that the Republicans might not go all-out in an effort to sabotage the Census for short-term political gains, and that’s because the business world depends very much on its accuracy. (And also because the government gives them mountains of information for free that otherwise they’d have to pay for themselves through their own research). Groups like the National Retail Foundation and National Association of Realtors are pushing hard for better Census funding, for instance. Large businesses have decisions to make on where to market their products and how large the customer base for that product might be; they have to decide where to locate stores and how to build supply chains, and accurate Census data is what helps them do that. Outcry from them, on top of threatened red-state Republicans and, of course, Democratic communities at risk of undercounting, may cumulatively help reverse the Census’ slide toward dysfunction.