The end of the citizenship question is just the beginning.
The imbroglio over the Trump administration’s unsuccessful effort to add a citizenship question to the 2020 census has passed. But that debate was only a foreshadow of what may become the most divisive political battle so far this century: After next year’s census, states will need to draw new legislative districts, and the courts have not resolved some major questions about how this may be done.
The census has only one constitutionally mandated purpose: apportioning seats in the House of Representatives. All of the states also use federal census data to apportion their legislatures. A series of landmark Supreme Court decisions in the 1960s ruled that both federal and state legislative districts had to be of (at least roughly) equal populations, a principle referred to as “one person, one vote.” Although controversial at the time, this principle is now broadly acknowledged.
But when counting the number of “persons” in a district for this purpose, does it make sense to include those with no votes to cast? Or is the ideal more accurately stated as “one voter, one vote”?
Apportionment has traditionally been carried out on the basis of total population, including non-citizens, many of whom are not even in the country legally. This gives extra political power to heavily immigrant areas on the basis of residents who are not eligible to vote. And the courts have never resolved the question of whether a state could, if it wanted to, draw its districts on the basis of citizen (or citizen-voting-age) population rather than total population.
The Scale of the Problem
To understand how the uneven distribution of non-citizens affects representation, we must look at apportionment on three distinct levels:
1) Allocating seats in the U.S. House of Representatives among the states.
2) Drawing election districts for those seats within each state with more than one representative.
3) Redistricting the state legislatures.
Although the census does not ask about citizenship, the Census Bureau’s American Community Survey does. A review of such data from 2014 and 2015 shows significant deviations from “one voter, one vote” at each of the three levels of apportionment.
1) There are significant differences in non-citizen populations among the states, ranging from 14 percent in California to less than 1 percent in West Virginia. California, the first sanctuary state, has five or six more members of the House (and consequently Electoral College votes) than it would if apportionment were based only on citizen population.
2) There are similar differences within states. For example, non-citizens make up 17.6 percent of New York City’s population but only 5.2 percent of the rest of New York State. As a result, New York City has one more seat in the U.S. House than it would if a “one voter, one vote” apportionment were followed. Similarly, about 10 percent of Florida residents are not citizens, but those non-citizens are heavily concentrated in Broward and Miami-Dade Counties, where 20 percent of the residents are not citizens. This shifts almost a full House seat to those two counties away from the rest of Florida. Such distortions are not as common in states with low non-citizen populations, but one study using 2015 data found that the Democratic share of the House would fall from 43.2 percent to 42.3 percent if only citizens were counted, and slightly further if only voting-age citizens were. Such differences may not seem large, but even a small deviation can be material in a closely divided legislature.
3) These uneven distributions within states also affect state legislatures. Due to the discrepancies described above, New York City has two or three more seats in the New York state senate and at least five more seats in the state assembly than it would have if apportionment were based on the citizen population. In Florida, apportioning on the basis of gross population shifts three or four seats in the state house of representatives and one or more seats in the state senate to Miami-Dade and Broward counties and away from the rest of the state. Under citizen-only redistricting, the study based on 2015 estimated, the Democratic share of state houses nationwide would fall from 46.4 percent to 45.2 percent, and the Democratic share of state senates from 44.4 percent to 43.4 percent. The effect would be particularly dramatic in states with high non-citizen populations, with, for example, swings to the Republicans of more than two points in the state houses of Texas, Nevada, and New York.
Further, all of the foregoing figures are based on Census Bureau data from 2014 and 2015. As the foreign-born share of the population continues to grow, so will the deviations from “one voter, one vote.”
The Battle to Come
There will be no citizenship question on the 2020 census, but there are other data sources available to legislators if they decide to make a push to address these problems. President Trump has issued an executive order requiring federal agencies to send any relevant data to the Census Bureau. Citizen numbers can also be extracted from the annual American Community Survey, which could be expanded or improved to provide more accurate estimates. In the recent conflict, opponents of adding the citizenship question to the full census argued that these alternative data sources could yield very accurate citizenship numbers, though it is likely that these same forces would flip that position if there were an effort to apply “one voter, one vote.”
If it is accepted that the Census Bureau can produce accurate citizenship data without the benefit of a question on the 2020 census, what are the legal prospects of applying “one voter, one vote” to the three levels of apportionment?
1) The Constitution requires that seats in the U. S. House of Representatives be allocated among the states based on “the whole number of persons” in each state. It has always been assumed that that means the gross population. However, in a recent Wall Street Journal opinion article, constitutional attorneys David Rivkin and Richard Raile argue that the original understanding of that phrase was to include only permanent residents, a category that excludes many non-citizens, especially illegal immigrants. Whether justices committed to originalism and textualism would accept Rivkin and Raile’s argument is unclear, and the Supreme Court’s liberals will presumably reject it because it would overall benefit Republicans, so reforms to this part of the process may require a constitutional amendment. Such an amendment is discussed in my book Fifty States, Not Six. Of course, any chance of enacting that or any other amendment will require a reform of Article V’s outdated and now impossible hurdles (see here for a proposal to realize that).
2) However, once seats in the U. S. House have been allocated, a different constitutional provision comes into play. The Constitution explicitly provides that Congress and state legislatures shall determine the “Times, Places and Manner of holding Elections for Senators and Representatives.” This authority includes redistricting, an understanding that was strongly re-affirmed by the Supreme Court in the partisan-gerrymandering cases, so Congress may be able to require that redistricting be conducted on a “one voter, one vote” basis, even if the total number of House seats given to each state is based on total population. Absent action on the federal level, the same clause arguably would authorize a state’s legislature to draw its own House districts this way.
3) The U. S. Constitution says nothing directly about the apportionment of state legislatures. However, as noted above, the Supreme Court has interpreted the Equal Protection Clause of the 14th Amendment to require that all houses of state legislatures by apportioned on a “one person, one vote” basis. In its 2016 decision in Evenwel v. Abbott, the Supreme Court ruled that the Equal Protection clause did not mandate that apportionment be based on the citizen voting-age population. However, that case did not determine whether it was permissible for a state to choose to apportion this way. Given that the principle of “one person, one vote” is incoherent when applied to people who can’t vote, and that the Supreme Court tends to give more leeway when it comes to state (as opposed to federal) districts, a state would have a strong case that excluding non-citizens is permissible.
Democrats can be expected to resist any drive toward true equal representation by arguing that it will deprive “communities of color” of political power. But deviations from “one voter, one vote” do not directly benefit non-citizens, who cannot vote anyway. Instead, they give disproportionate political power to citizens who live in primarily urban areas with large non-citizen populations.
Here we reach the heart of the conflict. Are we first individual citizens of our republic, each with equal legal standing, or are we primarily members of political ethnicities, competing to acquire political power by reducing that of our fellows through depriving them of equal representation? The leftist urban practitioners of identity politics who benefit from the current deviations from “one voter, one vote” see themselves as virtuous representatives of the non-citizen neighbors who bulk up their political clout. They will undoubtedly decry any push toward “one voter, one vote” equal representation as racist. But this only demonstrates yet again the old leftist principle that some are more equal than others.